xiaxinhang/pr-agent
1
0
Fork 0
mirror of https://github.com/qodo-ai/pr-agent.git synced 2025-07-11 00:00:38 +08:00
Code Issues Packages Projects Releases Wiki Activity

Merge remote-tracking branch 'origin/main' into ok/gitlab_fix

Browse Source
This commit is contained in:
Ori Kotek
2024-06-03 14:15:46 +03:00
parent 76ff49d446 f8feaa0be7
commit 2b0dfc6298
91 changed files with 2626 additions and 1330 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
.github/workflows/build-and-test.yaml vendored
View File

@ -2,8 +2,11 @@ name: Build-and-test
on:
push:
branches:
- main
pull_request:
types: [ opened, reopened ]
branches:
- main
jobs:
build-and-test:

1
.github/workflows/docs-ci.yaml vendored
View File

@ -29,4 +29,5 @@ jobs:
mkdocs-material-
- run: pip install mkdocs-material
- run: pip install "mkdocs-material[imaging]"
- run: pip install mkdocs-glightbox
- run: mkdocs gh-deploy -f docs/mkdocs.yml --force

2
.gitignore vendored
View File

@ -1,4 +1,6 @@
.idea/
.lsp/
.vscode/
venv/
pr_agent/settings/.secrets.toml
__pycache__

4
.pr_agent.toml
View File

@ -1,7 +1,3 @@
[pr_reviewer]
enable_review_labels_effort = true
enable_auto_approval = true
[pr_code_suggestions]
summarize=true

2
MANIFEST.in Normal file
View File

@ -0,0 +1,2 @@
recursive-include pr_agent *.toml
recursive-exclude pr_agent *.secrets.toml

175
README.md
View File

@ -6,13 +6,16 @@
<picture>
<source media="(prefers-color-scheme: dark)" srcset="https://codium.ai/images/pr_agent/logo-dark.png" width="330">
<source media="(prefers-color-scheme: light)" srcset="https://codium.ai/images/pr_agent/logo-light.png" width="330">
<img alt="logo">
<img src="https://codium.ai/images/pr_agent/logo-light.png" alt="logo" width="330">
</picture>
<br/>
Making pull requests less painful with an AI agent
CodiumAI PR-Agent aims to help efficiently review and handle pull requests, by providing AI feedbacks and suggestions
</div>
[![GitHub license](https://img.shields.io/badge/License-Apache_2.0-blue.svg)](https://github.com/Codium-ai/pr-agent/blob/main/LICENSE)
[![Static Badge](https://img.shields.io/badge/Chrome-Extension-violet)](https://chromewebstore.google.com/detail/pr-agent-chrome-extension/ephlnjeghhogofkifjloamocljapahnl)
[![Static Badge](https://img.shields.io/badge/Code-Benchmark-blue)](https://pr-agent-docs.codium.ai/finetuning_benchmark/)
[![Discord](https://badgen.net/badge/icon/discord?icon=discord&label&color=purple)](https://discord.com/channels/1057273017547378788/1126104260430528613)
[![Twitter](https://img.shields.io/twitter/follow/codiumai)](https://twitter.com/codiumai)
<a href="https://github.com/Codium-ai/pr-agent/commits/main">
@ -20,83 +23,82 @@ Making pull requests less painful with an AI agent
</a>
</div>
### [Documentation](https://pr-agent-docs.codium.ai/)
- See the [Installation Guide](https://pr-agent-docs.codium.ai/installation/) for instructions on installing PR-Agent on different platforms.
- See the [Usage Guide](https://pr-agent-docs.codium.ai/usage-guide/) for instructions on running PR-Agent tools via different interfaces, such as CLI, PR Comments, or by automatically triggering them when a new PR is opened.
- See the [Tools Guide](https://pr-agent-docs.codium.ai/tools/) for a detailed description of the different tools, and the available configurations for each tool.
## Table of Contents
- [News and Updates](#news-and-updates)
- [Overview](#overview)
- [Example results](#example-results)
- [Try it now](#try-it-now)
- [Installation](#installation)
- [PR-Agent Pro 💎](#pr-agent-pro-)
- [How it works](#how-it-works)
- [Why use PR-Agent?](#why-use-pr-agent)
## News and Updates
### Jan 10, 2024
- A new [knowledge-base website](https://pr-agent-docs.codium.ai/) for PR-Agent is now available. It includes detailed information about the different tools, usage guides and more, in an accessible and organized format.
### May 31, 2024
### Jan 8, 2024
Check out the new [**PR-Agent Code Fine-tuning Benchmark**](https://pr-agent-docs.codium.ai/finetuning_benchmark/)
- A new tool, [Find Similar Code](https://pr-agent-docs.codium.ai/tools/similar_code/) 💎 is now available.
<br>This tool retrieves the most similar code components from inside the organization's codebase, or from open-source code:
### May 23, 2024
<kbd><a href="https://codium.ai/images/pr_agent/similar_code.mp4"><img src="https://codium.ai/images/pr_agent/similar_code_global2.png" width="512"></a></kbd>
We released a new version of [PR-Agent Chrome extension](https://chromewebstore.google.com/detail/pr-agent-chrome-extension/ephlnjeghhogofkifjloamocljapahnl), with two new features:
(click on the image to see an instructional video)
- PR-Agent filters 🎨
- Code suggestions interactions 🔗
### Feb 29, 2024
- You can now use the repo's [wiki page](https://pr-agent-docs.codium.ai/usage-guide/configuration_options/) to set configurations for PR-Agent 💎
See more [here](https://www.youtube.com/watch?v=v9bJ1frtPcg)
<kbd><img src="https://codium.ai/images/pr_agent/wiki_configuration.png" width="512"></kbd>
### May 21, 2024
Check out CodiumAI new project, [**Cover-Agent**](https://github.com/Codium-ai/cover-agent), that can automatically generate qualified tests to enhance existing test suites, aiming to increase code and behavior coverage efficiently.
## Overview
<div style="text-align:left;">
CodiumAI PR-Agent is an open-source tool to help efficiently review and handle pull requests.
- See the [Installation Guide](https://pr-agent-docs.codium.ai/installation/) for instructions on installing and running the tool on different git platforms.
- See the [Usage Guide](https://pr-agent-docs.codium.ai/usage-guide/) for instructions on running the PR-Agent commands via different interfaces, including _CLI_, _online usage_, or by _automatically triggering_ them when a new PR is opened.
- See the [Tools Guide](https://pr-agent-docs.codium.ai/tools/) for a detailed description of the different tools.
Supported commands per platform:
| | | GitHub | Gitlab | Bitbucket | Azure DevOps |
|-------|-------------------------------------------------------------------------------------------------------------------|:--------------------:|:--------------------:|:--------------------:|:--------------------:|
| TOOLS | Review | :white_check_mark: | :white_check_mark: | :white_check_mark: | :white_check_mark: |
| | ⮑ Incremental | :white_check_mark: | | | |
| | ⮑ [SOC2 Compliance](https://pr-agent-docs.codium.ai/tools/review/#soc2-ticket-compliance) 💎 | :white_check_mark: | :white_check_mark: | :white_check_mark: | :white_check_mark: |
| | Describe | :white_check_mark: | :white_check_mark: | :white_check_mark: | :white_check_mark: |
| | ⮑ [Inline File Summary](https://pr-agent-docs.codium.ai/tools/describe#inline-file-summary) 💎 | :white_check_mark: | | | |
| | Improve | :white_check_mark: | :white_check_mark: | :white_check_mark: | :white_check_mark: |
| | ⮑ Extended | :white_check_mark: | :white_check_mark: | :white_check_mark: | :white_check_mark: |
| | Ask | :white_check_mark: | :white_check_mark: | :white_check_mark: | :white_check_mark: |
| | ⮑ [Ask on code lines](https://pr-agent-docs.codium.ai/tools/ask#ask-lines) | :white_check_mark: | :white_check_mark: | | |
| | [Custom Suggestions](https://pr-agent-docs.codium.ai/tools/custom_suggestions/) 💎 | :white_check_mark: | :white_check_mark: | :white_check_mark: | :white_check_mark: |
| | [Test](https://pr-agent-docs.codium.ai/tools/test/) 💎 | :white_check_mark: | :white_check_mark: | | :white_check_mark: |
| | Reflect and Review | :white_check_mark: | :white_check_mark: | :white_check_mark: | :white_check_mark: |
| | Update CHANGELOG.md | :white_check_mark: | :white_check_mark: | :white_check_mark: | :white_check_mark: |
| | Find Similar Issue | :white_check_mark: | | | |
| | [Add PR Documentation](https://pr-agent-docs.codium.ai/tools/documentation/) 💎 | :white_check_mark: | :white_check_mark: | | :white_check_mark: |
| | [Custom Labels](https://pr-agent-docs.codium.ai/tools/custom_labels/) 💎 | :white_check_mark: | :white_check_mark: | | :white_check_mark: |
| | [Analyze](https://pr-agent-docs.codium.ai/tools/analyze/) 💎 | :white_check_mark: | :white_check_mark: | | :white_check_mark: |
| | [CI Feedback](https://pr-agent-docs.codium.ai/tools/ci_feedback/) 💎 | :white_check_mark: | | | |
| | [Similar Code](https://pr-agent-docs.codium.ai/tools/similar_code/) 💎 | :white_check_mark: | | | |
|-------|---------------------------------------------------------------------------------------------------------|:--------------------:|:--------------------:|:--------------------:|:--------------------:|
| TOOLS | Review | ✅ | ✅ | ✅ | ✅ |
| | ⮑ Incremental | ✅ | | | |
| | ⮑ [SOC2 Compliance](https://pr-agent-docs.codium.ai/tools/review/#soc2-ticket-compliance) 💎 | ✅ | ✅ | ✅ | ✅ |
| | Describe | ✅ | ✅ | ✅ | ✅ |
| | ⮑ [Inline File Summary](https://pr-agent-docs.codium.ai/tools/describe#inline-file-summary) 💎 | ✅ | | | |
| | Improve | ✅ | ✅ | ✅ | ✅ |
| | ⮑ Extended | ✅ | ✅ | ✅ | ✅ |
| | Ask | ✅ | ✅ | ✅ | ✅ |
| | ⮑ [Ask on code lines](https://pr-agent-docs.codium.ai/tools/ask#ask-lines) | ✅ | ✅ | | |
| | [Custom Prompt](https://pr-agent-docs.codium.ai/tools/custom_prompt/) 💎 | ✅ | ✅ | ✅ | ✅ |
| | [Test](https://pr-agent-docs.codium.ai/tools/test/) 💎 | ✅ | ✅ | | ✅ |
| | Reflect and Review | ✅ | ✅ | ✅ | ✅ |
| | Update CHANGELOG.md | ✅ | ✅ | ✅ | ✅ |
| | Find Similar Issue | ✅ | | | |
| | [Add PR Documentation](https://pr-agent-docs.codium.ai/tools/documentation/) 💎 | ✅ | ✅ | | ✅ |
| | [Custom Labels](https://pr-agent-docs.codium.ai/tools/custom_labels/) 💎 | ✅ | ✅ | | ✅ |
| | [Analyze](https://pr-agent-docs.codium.ai/tools/analyze/) 💎 | ✅ | ✅ | | ✅ |
| | [CI Feedback](https://pr-agent-docs.codium.ai/tools/ci_feedback/) 💎 | ✅ | | | |
| | [Similar Code](https://pr-agent-docs.codium.ai/tools/similar_code/) 💎 | ✅ | | | |
| | | | | | |
| USAGE | CLI | :white_check_mark: | :white_check_mark: | :white_check_mark: | :white_check_mark: |
| | App / webhook | :white_check_mark: | :white_check_mark: | :white_check_mark: | :white_check_mark: |
| | Tagging bot | :white_check_mark: | | | |
| | Actions | :white_check_mark: | | :white_check_mark: | |
| USAGE | CLI | ✅ | ✅ | ✅ | ✅ |
| | App / webhook | ✅ | ✅ | ✅ | ✅ |
| | Tagging bot | ✅ | | | |
| | Actions | ✅ | | ✅ | |
| | | | | | |
| CORE | PR compression | :white_check_mark: | :white_check_mark: | :white_check_mark: | :white_check_mark: |
| | Repo language prioritization | :white_check_mark: | :white_check_mark: | :white_check_mark: | :white_check_mark: |
| | Adaptive and token-aware file patch fitting | :white_check_mark: | :white_check_mark: | :white_check_mark: | :white_check_mark: |
| | Multiple models support | :white_check_mark: | :white_check_mark: | :white_check_mark: | :white_check_mark: |
| | [Static code analysis](https://pr-agent-docs.codium.ai/core-abilities/#static-code-analysis) 💎 | :white_check_mark: | :white_check_mark: | :white_check_mark: | :white_check_mark: |
| | [Global and wiki configurations](https://pr-agent-docs.codium.ai/usage-guide/configuration_options/) 💎 | :white_check_mark: | :white_check_mark: | :white_check_mark: | :white_check_mark: |
| | [PR interactive actions](https://www.codium.ai/images/pr_agent/pr-actions.mp4) 💎 | :white_check_mark: | | | |
| CORE | PR compression | ✅ | ✅ | ✅ | ✅ |
| | Repo language prioritization | ✅ | ✅ | ✅ | ✅ |
| | Adaptive and token-aware file patch fitting | ✅ | ✅ | ✅ | ✅ |
| | Multiple models support | ✅ | ✅ | ✅ | ✅ |
| | [Static code analysis](https://pr-agent-docs.codium.ai/core-abilities/#static-code-analysis) 💎 | ✅ | ✅ | ✅ | ✅ |
| | [Global and wiki configurations](https://pr-agent-docs.codium.ai/usage-guide/configuration_options/) 💎 | ✅ | ✅ | ✅ | ✅ |
| | [PR interactive actions](https://www.codium.ai/images/pr_agent/pr-actions.mp4) 💎 | ✅ | | | |
- 💎 means this feature is available only in [PR-Agent Pro](https://www.codium.ai/pricing/)
[//]: # (- Support for additional git providers is described in [here]&#40;./docs/Full_environments.md&#41;)
@ -120,13 +122,13 @@ ___
\
‣ **Analyze 💎 ([`/analyze`](https://pr-agent-docs.codium.ai/tools/analyze/))**: Identify code components that changed in the PR, and enables to interactively generate tests, docs, and code suggestions for each component.
\
‣ **Custom Suggestions 💎 ([`/custom_suggestions`](https://pr-agent-docs.codium.ai/tools/custom_suggestions/))**: Automatically generates custom suggestions for improving the PR code, based on specific guidelines defined by the user.
‣ **Custom Prompt 💎 ([`/custom_prompt`](https://pr-agent-docs.codium.ai/tools/custom_prompt/))**: Automatically generates custom suggestions for improving the PR code, based on specific guidelines defined by the user.
\
‣ **Generate Tests 💎 ([`/test component_name`](https://pr-agent-docs.codium.ai/tools/test/))**: Generates unit tests for a selected component, based on the PR code changes.
\
‣ **CI Feedback 💎 ([`/checks ci_job`](https://pr-agent-docs.codium.ai/tools/ci_feedback/))**: Automatically generates feedback and analysis for a failed CI job.
\
‣ **Similar Code 💎 ([`/find_similar_component`](https://pr-agent-docs.codium.ai/tools/similar_code//))**: Retrieves the most similar code components from inside the organization's codebase, or from open-source code.
‣ **Similar Code 💎 ([`/find_similar_component`](https://pr-agent-docs.codium.ai/tools/similar_code/))**: Retrieves the most similar code components from inside the organization's codebase, or from open-source code.
___
## Example results
@ -225,31 +227,47 @@ Note that when you set your own PR-Agent or use CodiumAI hosted PR-Agent, there
---
## Installation
To use your own version of PR-Agent, you first need to acquire two tokens:
[//]: # (## Installation)
1. An OpenAI key from [here](https://platform.openai.com/), with access to GPT-4.
2. A GitHub personal access token (classic) with the repo scope.
[//]: # (To use your own version of PR-Agent, you first need to acquire two tokens:)
There are several ways to use PR-Agent:
[//]: # ()
[//]: # (1. An OpenAI key from [here]&#40;https://platform.openai.com/&#41;, with access to GPT-4.)
**Locally**
- [Use Docker image (no installation required)](https://pr-agent-docs.codium.ai/installation/locally/#use-docker-image-no-installation-required)
- [Run from source](https://pr-agent-docs.codium.ai/installation/locally/#run-from-source)
[//]: # (2. A GitHub personal access token &#40;classic&#41; with the repo scope.)
**GitHub specific methods**
- [Run as a GitHub Action](https://pr-agent-docs.codium.ai/installation/github/#run-as-a-github-action)
- [Run as a GitHub App](https://pr-agent-docs.codium.ai/installation/github/#run-as-a-github-app)
[//]: # ()
[//]: # (There are several ways to use PR-Agent:)
**GitLab specific methods**
- [Run a GitLab webhook server](https://pr-agent-docs.codium.ai/installation/gitlab/)
[//]: # ()
[//]: # (**Locally**)
**BitBucket specific methods**
- [Run as a Bitbucket Pipeline](https://pr-agent-docs.codium.ai/installation/bitbucket/)
[//]: # (- [Using pip package]&#40;https://pr-agent-docs.codium.ai/installation/locally/#using-pip-package&#41;)
[//]: # (- [Using Docker image]&#40;https://pr-agent-docs.codium.ai/installation/locally/#using-docker-image&#41;)
[//]: # (- [Run from source]&#40;https://pr-agent-docs.codium.ai/installation/locally/#run-from-source&#41;)
[//]: # ()
[//]: # (**GitHub specific methods**)
[//]: # (- [Run as a GitHub Action]&#40;https://pr-agent-docs.codium.ai/installation/github/#run-as-a-github-action&#41;)
[//]: # (- [Run as a GitHub App]&#40;https://pr-agent-docs.codium.ai/installation/github/#run-as-a-github-app&#41;)
[//]: # ()
[//]: # (**GitLab specific methods**)
[//]: # (- [Run a GitLab webhook server]&#40;https://pr-agent-docs.codium.ai/installation/gitlab/&#41;)
[//]: # ()
[//]: # (**BitBucket specific methods**)
[//]: # (- [Run as a Bitbucket Pipeline]&#40;https://pr-agent-docs.codium.ai/installation/bitbucket/&#41;)
## PR-Agent Pro 💎
[PR-Agent Pro](https://www.codium.ai/pricing/) is a hosted version of PR-Agent, provided by CodiumAI. It is available for a monthly fee, and provides the following benefits:
1. **Fully managed** - We take care of everything for you - hosting, models, regular updates, and more. Installation is as simple as signing up and adding the PR-Agent app to your GitHub\BitBucket repo.
1. **Fully managed** - We take care of everything for you - hosting, models, regular updates, and more. Installation is as simple as signing up and adding the PR-Agent app to your GitHub\GitLab\BitBucket repo.
2. **Improved privacy** - No data will be stored or used to train models. PR-Agent Pro will employ zero data retention, and will use an OpenAI account with zero data retention.
3. **Improved support** - PR-Agent Pro users will receive priority support, and will be able to request new features and capabilities.
4. **Extra features** -In addition to the benefits listed above, PR-Agent Pro will emphasize more customization, and the usage of static code analysis, in addition to LLM logic, to improve results.
@ -279,11 +297,22 @@ Here are some advantages of PR-Agent:
## Data privacy
If you host PR-Agent with your OpenAI API key, it is between you and OpenAI. You can read their API data privacy policy here:
### Self-hosted PR-Agent
- If you host PR-Agent with your OpenAI API key, it is between you and OpenAI. You can read their API data privacy policy here:
https://openai.com/enterprise-privacy
When using PR-Agent Pro 💎, hosted by CodiumAI, we will not store any of your data, nor will we use it for training.
You will also benefit from an OpenAI account with zero data retention.
### CodiumAI-hosted PR-Agent Pro 💎
- When using PR-Agent Pro 💎, hosted by CodiumAI, we will not store any of your data, nor will we use it for training. You will also benefit from an OpenAI account with zero data retention.
- For certain clients, CodiumAI-hosted PR-Agent Pro will use CodiumAI’s proprietary models — if this is the case, you will be notified.
- No passive collection of Code and Pull Requests’ data — PR-Agent will be active only when you invoke it, and it will then extract and analyze only data relevant to the executed command and queried pull request.
### PR-Agent Chrome extension
- The [PR-Agent Chrome extension](https://chromewebstore.google.com/detail/pr-agent-chrome-extension/ephlnjeghhogofkifjloamocljapahnl) serves solely to modify the visual appearance of a GitHub PR screen. It does not transmit any user's repo or pull request code. Code is only sent for processing when a user submits a GitHub comment that activates a PR-Agent tool, in accordance with the standard privacy policy of PR-Agent.
## Links

4
docker/Dockerfile.lambda
View File

@ -1,10 +1,10 @@
FROM public.ecr.aws/lambda/python:3.10
RUN yum update -y && \
yum install -y gcc python3-devel && \
yum install -y gcc python3-devel git && \
yum clean all
ADD pyproject.toml .
ADD pyproject.toml requirements.txt .
RUN pip install . && rm pyproject.toml
RUN pip install mangum==0.17.0
COPY pr_agent/ ${LAMBDA_TASK_ROOT}/pr_agent/

16
docs/README.md
View File

@ -1,15 +1 @@
# To install:
pip install mkdocs
pip install mkdocs-material
pip install mkdocs-material-extensions
pip install "mkdocs-material[imaging]"
# docs
To run localy: `mkdocs serve`
To expand and customize the theme: [Material MKDocs](https://squidfunk.github.io/mkdocs-material/)
The deployment is managed on the gh-pages branches.
After each merge to main the deplloyment will be taken care of by GH action automatically and the new version will be available at: [Docs](https://codium-ai.github.io/docs/)
Github action is located in `.github/workflows/ci.yml` file.
# [Visit Our Docs Portal](https://pr-agent-docs.codium.ai/)

140
docs/docs/assets/logo.svg Normal file
View File

@ -0,0 +1,140 @@
<?xml version="1.0" encoding="utf-8"?>
<!-- Generator: Adobe Illustrator 28.1.0, SVG Export Plug-In . SVG Version: 6.00 Build 0) -->
<svg version="1.1" id="Layer_1" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" x="0px" y="0px"
width="64px" height="64px" viewBox="0 0 64 64" enable-background="new 0 0 64 64" xml:space="preserve">
<g>
<defs>
<rect id="SVGID_1_" x="0.4" y="0.1" width="63.4" height="63.4"/>
</defs>
<clipPath id="SVGID_00000008836131916906499950000015813697852011234749_">
<use xlink:href="#SVGID_1_" overflow="visible"/>
</clipPath>
<g clip-path="url(#SVGID_00000008836131916906499950000015813697852011234749_)">
<path fill="#05E5AD" d="M21.4,9.8c3,0,5.9,0.7,8.5,1.9c-5.7,3.4-9.8,11.1-9.8,20.1c0,9,4,16.7,9.8,20.1c-2.6,1.2-5.5,1.9-8.5,1.9
c-11.6,0-21-9.8-21-22S9.8,9.8,21.4,9.8z"/>
<radialGradient id="SVGID_00000150822754378345238340000008985053211526864828_" cx="-140.0905" cy="350.1757" r="4.8781" gradientTransform="matrix(-4.7708 -6.961580e-02 -0.1061 7.2704 -601.3099 -2523.8489)" gradientUnits="userSpaceOnUse">
<stop offset="0" style="stop-color:#6447FF"/>
<stop offset="6.666670e-02" style="stop-color:#6348FE"/>
<stop offset="0.1333" style="stop-color:#614DFC"/>
<stop offset="0.2" style="stop-color:#5C54F8"/>
<stop offset="0.2667" style="stop-color:#565EF3"/>
<stop offset="0.3333" style="stop-color:#4E6CEC"/>
<stop offset="0.4" style="stop-color:#447BE4"/>
<stop offset="0.4667" style="stop-color:#3A8DDB"/>
<stop offset="0.5333" style="stop-color:#2F9FD1"/>
<stop offset="0.6" style="stop-color:#25B1C8"/>
<stop offset="0.6667" style="stop-color:#1BC0C0"/>
<stop offset="0.7333" style="stop-color:#13CEB9"/>
<stop offset="0.8" style="stop-color:#0DD8B4"/>
<stop offset="0.8667" style="stop-color:#08DFB0"/>
<stop offset="0.9333" style="stop-color:#06E4AE"/>
<stop offset="1" style="stop-color:#05E5AD"/>
</radialGradient>
<path fill="url(#SVGID_00000150822754378345238340000008985053211526864828_)" d="M21.4,9.8c3,0,5.9,0.7,8.5,1.9
c-5.7,3.4-9.8,11.1-9.8,20.1c0,9,4,16.7,9.8,20.1c-2.6,1.2-5.5,1.9-8.5,1.9c-11.6,0-21-9.8-21-22S9.8,9.8,21.4,9.8z"/>
<radialGradient id="SVGID_00000022560571240417802950000012439139323268113305_" cx="-191.7649" cy="385.7387" r="4.8781" gradientTransform="matrix(-2.5514 -0.7616 -0.8125 2.7217 -130.733 -1180.2209)" gradientUnits="userSpaceOnUse">
<stop offset="0" style="stop-color:#6447FF"/>
<stop offset="6.666670e-02" style="stop-color:#6348FE"/>
<stop offset="0.1333" style="stop-color:#614DFC"/>
<stop offset="0.2" style="stop-color:#5C54F8"/>
<stop offset="0.2667" style="stop-color:#565EF3"/>
<stop offset="0.3333" style="stop-color:#4E6CEC"/>
<stop offset="0.4" style="stop-color:#447BE4"/>
<stop offset="0.4667" style="stop-color:#3A8DDB"/>
<stop offset="0.5333" style="stop-color:#2F9FD1"/>
<stop offset="0.6" style="stop-color:#25B1C8"/>
<stop offset="0.6667" style="stop-color:#1BC0C0"/>
<stop offset="0.7333" style="stop-color:#13CEB9"/>
<stop offset="0.8" style="stop-color:#0DD8B4"/>
<stop offset="0.8667" style="stop-color:#08DFB0"/>
<stop offset="0.9333" style="stop-color:#06E4AE"/>
<stop offset="1" style="stop-color:#05E5AD"/>
</radialGradient>
<path fill="url(#SVGID_00000022560571240417802950000012439139323268113305_)" d="M38,18.3c-2.1-2.8-4.9-5.1-8.1-6.6
c2-1.2,4.2-1.9,6.6-1.9c2.2,0,4.3,0.6,6.2,1.7C40.8,12.9,39.2,15.3,38,18.3L38,18.3z"/>
<radialGradient id="SVGID_00000143611122169386473660000017673587931016751800_" cx="-194.7918" cy="395.2442" r="4.8781" gradientTransform="matrix(-2.5514 -0.7616 -0.8125 2.7217 -130.733 -1172.9556)" gradientUnits="userSpaceOnUse">
<stop offset="0" style="stop-color:#6447FF"/>
<stop offset="6.666670e-02" style="stop-color:#6348FE"/>
<stop offset="0.1333" style="stop-color:#614DFC"/>
<stop offset="0.2" style="stop-color:#5C54F8"/>
<stop offset="0.2667" style="stop-color:#565EF3"/>
<stop offset="0.3333" style="stop-color:#4E6CEC"/>
<stop offset="0.4" style="stop-color:#447BE4"/>
<stop offset="0.4667" style="stop-color:#3A8DDB"/>
<stop offset="0.5333" style="stop-color:#2F9FD1"/>
<stop offset="0.6" style="stop-color:#25B1C8"/>
<stop offset="0.6667" style="stop-color:#1BC0C0"/>
<stop offset="0.7333" style="stop-color:#13CEB9"/>
<stop offset="0.8" style="stop-color:#0DD8B4"/>
<stop offset="0.8667" style="stop-color:#08DFB0"/>
<stop offset="0.9333" style="stop-color:#06E4AE"/>
<stop offset="1" style="stop-color:#05E5AD"/>
</radialGradient>
<path fill="url(#SVGID_00000143611122169386473660000017673587931016751800_)" d="M38,45.2c1.2,3,2.9,5.3,4.7,6.8
c-1.9,1.1-4,1.7-6.2,1.7c-2.3,0-4.6-0.7-6.6-1.9C33.1,50.4,35.8,48.1,38,45.2L38,45.2z"/>
<path fill="#684BFE" d="M20.1,31.8c0-9,4-16.7,9.8-20.1c3.2,1.5,6,3.8,8.1,6.6c-1.5,3.7-2.5,8.4-2.5,13.5s0.9,9.8,2.5,13.5
c-2.1,2.8-4.9,5.1-8.1,6.6C24.1,48.4,20.1,40.7,20.1,31.8z"/>
<radialGradient id="SVGID_00000147942998054305738810000004710078864578628519_" cx="-212.7358" cy="363.2475" r="4.8781" gradientTransform="matrix(-2.3342 -1.063 -1.623 3.5638 149.3813 -1470.1027)" gradientUnits="userSpaceOnUse">
<stop offset="0" style="stop-color:#6447FF"/>
<stop offset="6.666670e-02" style="stop-color:#6348FE"/>
<stop offset="0.1333" style="stop-color:#614DFC"/>
<stop offset="0.2" style="stop-color:#5C54F8"/>
<stop offset="0.2667" style="stop-color:#565EF3"/>
<stop offset="0.3333" style="stop-color:#4E6CEC"/>
<stop offset="0.4" style="stop-color:#447BE4"/>
<stop offset="0.4667" style="stop-color:#3A8DDB"/>
<stop offset="0.5333" style="stop-color:#2F9FD1"/>
<stop offset="0.6" style="stop-color:#25B1C8"/>
<stop offset="0.6667" style="stop-color:#1BC0C0"/>
<stop offset="0.7333" style="stop-color:#13CEB9"/>
<stop offset="0.8" style="stop-color:#0DD8B4"/>
<stop offset="0.8667" style="stop-color:#08DFB0"/>
<stop offset="0.9333" style="stop-color:#06E4AE"/>
<stop offset="1" style="stop-color:#05E5AD"/>
</radialGradient>
<path fill="url(#SVGID_00000147942998054305738810000004710078864578628519_)" d="M50.7,42.5c0.6,3.3,1.5,6.1,2.5,8
c-1.8,2-3.8,3.1-6,3.1c-1.6,0-3.1-0.6-4.5-1.7C46.1,50.2,48.9,46.8,50.7,42.5L50.7,42.5z"/>
<radialGradient id="SVGID_00000083770737908230256670000016126156495859285174_" cx="-208.5327" cy="357.2025" r="4.8781" gradientTransform="matrix(-2.3342 -1.063 -1.623 3.5638 149.3813 -1476.8097)" gradientUnits="userSpaceOnUse">
<stop offset="0" style="stop-color:#6447FF"/>
<stop offset="6.666670e-02" style="stop-color:#6348FE"/>
<stop offset="0.1333" style="stop-color:#614DFC"/>
<stop offset="0.2" style="stop-color:#5C54F8"/>
<stop offset="0.2667" style="stop-color:#565EF3"/>
<stop offset="0.3333" style="stop-color:#4E6CEC"/>
<stop offset="0.4" style="stop-color:#447BE4"/>
<stop offset="0.4667" style="stop-color:#3A8DDB"/>
<stop offset="0.5333" style="stop-color:#2F9FD1"/>
<stop offset="0.6" style="stop-color:#25B1C8"/>
<stop offset="0.6667" style="stop-color:#1BC0C0"/>
<stop offset="0.7333" style="stop-color:#13CEB9"/>
<stop offset="0.8" style="stop-color:#0DD8B4"/>
<stop offset="0.8667" style="stop-color:#08DFB0"/>
<stop offset="0.9333" style="stop-color:#06E4AE"/>
<stop offset="1" style="stop-color:#05E5AD"/>
</radialGradient>
<path fill="url(#SVGID_00000083770737908230256670000016126156495859285174_)" d="M42.7,11.5c1.4-1.1,2.9-1.7,4.5-1.7
c2.2,0,4.3,1.1,6,3.1c-1,2-1.9,4.7-2.5,8C48.9,16.7,46.1,13.4,42.7,11.5L42.7,11.5z"/>
<path fill="#684BFE" d="M38,45.2c2.8-3.7,4.4-8.4,4.4-13.5c0-5.1-1.7-9.8-4.4-13.5c1.2-3,2.9-5.3,4.7-6.8c3.4,1.9,6.2,5.3,8,9.5
c-0.6,3.2-0.9,6.9-0.9,10.8s0.3,7.6,0.9,10.8c-1.8,4.3-4.6,7.6-8,9.5C40.8,50.6,39.2,48.2,38,45.2L38,45.2z"/>
<path fill="#321BB2" d="M38,45.2c-1.5-3.7-2.5-8.4-2.5-13.5S36.4,22,38,18.3c2.8,3.7,4.4,8.4,4.4,13.5S40.8,41.5,38,45.2z"/>
<path fill="#05E6AD" d="M53.2,12.9c1.1-2,2.3-3.1,3.6-3.1c3.9,0,7,9.8,7,22s-3.1,22-7,22c-1.3,0-2.6-1.1-3.6-3.1
c3.4-3.8,5.7-10.8,5.7-18.8C58.8,23.8,56.6,16.8,53.2,12.9z"/>
<radialGradient id="SVGID_00000009565123575973598080000009335550354766300606_" cx="-7.8671" cy="278.2442" r="4.8781" gradientTransform="matrix(1.5187 0 0 -7.8271 69.237 2209.3281)" gradientUnits="userSpaceOnUse">
<stop offset="0" style="stop-color:#05E5AD"/>
<stop offset="0.32" style="stop-color:#05E5AD;stop-opacity:0"/>
<stop offset="0.9028" style="stop-color:#6447FF"/>
</radialGradient>
<path fill="url(#SVGID_00000009565123575973598080000009335550354766300606_)" d="M53.2,12.9c1.1-2,2.3-3.1,3.6-3.1
c3.9,0,7,9.8,7,22s-3.1,22-7,22c-1.3,0-2.6-1.1-3.6-3.1c3.4-3.8,5.7-10.8,5.7-18.8C58.8,23.8,56.6,16.8,53.2,12.9z"/>
<path fill="#684BFE" d="M52.8,31.8c0-3.9-0.8-7.6-2.1-10.8c0.6-3.3,1.5-6.1,2.5-8c3.4,3.8,5.7,10.8,5.7,18.8c0,8-2.3,15-5.7,18.8
c-1-2-1.9-4.7-2.5-8C52,39.3,52.8,35.7,52.8,31.8z"/>
<path fill="#321BB2" d="M50.7,42.5c-0.6-3.2-0.9-6.9-0.9-10.8s0.3-7.6,0.9-10.8c1.3,3.2,2.1,6.9,2.1,10.8S52,39.3,50.7,42.5z"/>
</g>
</g>
</svg>
Side by Side

After

Width:  |  Height:  |  Size: 9.0 KiB

49
docs/docs/chrome-extension/index.md Normal file
View File

@ -0,0 +1,49 @@
## PR-Agent chrome extension
PR-Agent Chrome extension is a collection of tools that integrates seamlessly with your GitHub environment, aiming to enhance your PR-Agent usage experience, and providing additional features.
## Features
### Toolbar extension
With PR-Agent Chrome extension, it's [easier than ever](https://www.youtube.com/watch?v=gT5tli7X4H4) to interactively configure and experiment with the different tools and configuration options.
After you found the setup that works for you, you can also easily export it as a persistent configuration file, and use it for automatic commands.
<img src="https://codium.ai/images/pr_agent/toolbar1.png" width="512">
<img src="https://codium.ai/images/pr_agent/toolbar2.png" width="512">
### PR-Agent filters
PR-Agent filters is a sidepanel option. that allows you to filter different message in the conversation tab.
For example, you can choose to present only message from PR-Agent, or filter those messages, focusing only on user's comments.
<img src="https://codium.ai/images/pr_agent/pr_agent_filters1.png" width="256">
<img src="https://codium.ai/images/pr_agent/pr_agent_filters2.png" width="256">
### Enhanced code suggestions
PR-Agent Chrome extension adds the following capabilities to code suggestions tool's comments:
- Auto-expand the table when you are viewing a code block, to avoid clipping.
- Adding a "quote-and-reply" button, that enables to address and comment on a specific suggestion (for example, asking the author to fix the issue)
<img src="https://codium.ai/images/pr_agent/chrome_extension_code_suggestion1.png" width="512">
<img src="https://codium.ai/images/pr_agent/chrome_extension_code_suggestion2.png" width="512">
## Installation
Go to the marketplace and install the extension:
[PR-Agent Chrome Extension](https://chromewebstore.google.com/detail/pr-agent-chrome-extension/ephlnjeghhogofkifjloamocljapahnl)
## Pre-requisites
The PR-Agent Chrome extension will work on any repo where you have previously [installed PR-Agent](https://pr-agent-docs.codium.ai/installation/).
## Data privacy and security
The PR-Agent Chrome extension only modifies the visual appearance of a GitHub PR screen. It does not transmit any user's repo or pull request code. Code is only sent for processing when a user submits a GitHub comment that activates a PR-Agent tool, in accordance with the standard privacy policy of PR-Agent.

2
docs/docs/core-abilities/index.md
View File

@ -43,7 +43,7 @@ We use [tiktoken](https://github.com/openai/tiktoken) to tokenize the patches af
#### Example
<kbd><img src=https://codium.ai/images/git_patch_logic.png width="768"></kbd>
![Core Abilities](https://codium.ai/images/git_patch_logic.png){width=768}
## YAML Prompting
TBD

93
docs/docs/css/custom.css
View File

@ -5,94 +5,23 @@
--md-accent-fg-color: #AEA1F1;
}
.md-nav__title, .md-nav__link {
font-size: 16px; /* Adjust the font size as needed */
font-size: 16px;
}
.md-tabs__link {
font-size: 16px; /* Adjust the font size as needed */
font-size: 16px;
}
.md-header__title {
font-size: 20px; /* Adjust the font size as needed */
font-size: 20px;
margin-left: 0px !important;
}
/*
@media (prefers-color-scheme: light) {
body {
--md-primary-fg-color: #00ffee !important;
--md-primary-bg-color: #ff0000 !important;
.md-content img {
border-width: 1px;
border-style: solid;
border-color: black;
outline-width: 1px;
outline-style: solid;
outline-color: darkgray;
}
body, .md-main, .md-content {
background-color: #4312f5 !important;
}
}
@media (prefers-color-scheme: dark) {
body {
--md-primary-fg-color: #171518 !important;
--md-primary-bg-color: #171518 !important;
}
body, .md-main, .md-content {
background-color: #171518 !important;
}
.md-header__title {
color: #ffffff !important;
}
.md-tabs .md-tabs__link {
color: #ffffff !important;
}
.md-tabs .md-tabs__link:hover,
.md-tabs .md-tabs__link:focus {
color: #ffffff !important;
}
.md-header__button {
color: #ffffff !important;
}
.md-header__button svg {
fill: currentColor !important;
}
.md-header__button:hover,
.md-header__button:focus {
color: #ffffff !important;
}
.md-header__button:hover svg,
.md-header__button:focus svg {
fill: currentColor !important;
}
.md-search__icon svg {
fill: #ffffff !important;
}
.md-search__input {
color: #ffffff !important;
}
.md-nav__item--active > .md-nav__link--active,
.md-nav__link--active {
color: #AEA1F1 !important;
}
.md-nav--secondary .md-nav__title {
background: #171518;
box-shadow: 0 0 0.4rem 0.4rem #171518;
}
.md-nav--lifted>.md-nav__list>.md-nav__item--active>.md-nav__link {
background: #171518;
box-shadow: 0 0 0.4rem 0.4rem #171518;
}
.md-content a {
color: #AEA1F1 !important;
}
} */

92
docs/docs/finetuning_benchmark/index.md Normal file
View File

@ -0,0 +1,92 @@
# PR-Agent Code Fine-tuning Benchmark
On coding tasks, the gap between open-source models and top closed-source models such as GPT4 is significant.
<br>
In practice, open-source models are unsuitable for most real-world code tasks, and require further fine-tuning to produce acceptable results.
_PR-Agent fine-tuning benchmark_ aims to benchmark open-source models on their ability to be fine-tuned for a coding task.
Specifically, we chose to fine-tune open-source models on the task of analyzing a pull request, and providing useful feedback and code suggestions.
Here are the results:
<br>
<br>
**Model performance:**
| Model name | Model size [B] | Better than gpt-4 rate, after fine-tuning [%] |
|-----------------------------|----------------|----------------------------------------------|
| **DeepSeek 34B-instruct** | **34** | **40.7** |
| DeepSeek 34B-base | 34 | 38.2 |
| Phind-34b | 34 | 38 |
| Granite-34B | 34 | 37.6 |
| Codestral-22B-v0.1 | 22 | 32.7 |
| QWEN-1.5-32B | 32 | 29 |
| | | |
| **CodeQwen1.5-7B** | **7** | **35.4** |
| Granite-8b-code-instruct | 8 | 34.2 |
| CodeLlama-7b-hf | 7 | 31.8 |
| Gemma-7B | 7 | 27.2 |
| DeepSeek coder-7b-instruct | 7 | 26.8 |
| Llama-3-8B-Instruct | 8 | 26.8 |
| Mistral-7B-v0.1 | 7 | 16.1 |
<br>
**Fine-tuning impact:**
| Model name | Model size [B] | Fine-tuned | Better than gpt-4 rate [%] |
|---------------------------|----------------|------------|----------------------------|
| DeepSeek 34B-instruct | 34 | yes | 40.7 |
| DeepSeek 34B-instruct | 34 | no | 3.6 |
## Results analysis
- **Fine-tuning is a must** - without fine-tuning, open-source models provide poor results on most real-world code tasks, which include complicated prompt and lengthy context. We clearly see that without fine-tuning, deepseek model was 96.4% of the time inferior to GPT-4, while after fine-tuning, it is better 40.7% of the time.
- **Always start from a code-dedicated model** — When fine-tuning, always start from a code-dedicated model, and not from a general-usage model. The gaps in downstream results are very big.
- **Don't believe the hype** —newer models, or models from big-tech companies (Llama3, Gemma, Mistral), are not always better for fine-tuning.
- **The best large model** - For large 34B code-dedicated models, the gaps when doing proper fine-tuning are small. The current top model is **DeepSeek 34B-instruct**
- **The best small model** - For small 7B code-dedicated models, the gaps when fine-tuning are much larger. **CodeQWEN 1.5-7B** is by far the best model for fine-tuning.
- **Base vs. instruct** - For the top model (deepseek), we saw small advantage when starting from the instruct version. However, we recommend testing both versions on each specific task, as the base model is generally considered more suitable for fine-tuning.
## The dataset
### Training dataset
Our training dataset comprises 25,000 pull requests, aggregated from permissive license repos. For each pull request, we generated responses for the three main tools of PR-Agent:
[Describe](https://pr-agent-docs.codium.ai/tools/describe/), [Review](https://pr-agent-docs.codium.ai/tools/improve/) and [Improve](https://pr-agent-docs.codium.ai/tools/improve/).
On the raw data collected, we employed various automatic and manual cleaning techniques to ensure the outputs were of the highest quality, and suitable for instruct-tuning.
Here are the prompts, and example outputs, used as input-output pairs to fine-tune the models:
| Tool | Prompt | Example output |
|----------|------------------------------------------------------------------------------------------------------------|----------------|
| Describe | [link](https://github.com/Codium-ai/pr-agent/blob/main/pr_agent/settings/pr_description_prompts.toml) | [link](https://github.com/Codium-ai/pr-agent/pull/910#issue-2303989601) |
| Review | [link](https://github.com/Codium-ai/pr-agent/blob/main/pr_agent/settings/pr_reviewer_prompts.toml) | [link](https://github.com/Codium-ai/pr-agent/pull/910#issuecomment-2118761219) |
| Improve | [link](https://github.com/Codium-ai/pr-agent/blob/main/pr_agent/settings/pr_code_suggestions_prompts.toml) | [link](https://github.com/Codium-ai/pr-agent/pull/910#issuecomment-2118761309) |
### Evaluation dataset
- For each tool, we aggregated 100 additional examples to be used for evaluation. These examples were not used in the training dataset, and were manually selected to represent diverse real-world use-cases.
- For each test example, we generated two responses: one from the fine-tuned model, and one from the best code model in the world, `gpt-4-turbo-2024-04-09`.
- We used a third LLM to judge which response better answers the prompt, and will likely be perceived by a human as better response.
<br>
We experimented with three model as judges: `gpt-4-turbo-2024-04-09`, `gpt-4o`, and `claude-3-opus-20240229`. All three produced similar results, with the same ranking order. This strengthens the validity of our testing protocol.
The evaluation prompt can be found [here](https://github.com/Codium-ai/pr-agent/blob/main/pr_agent/settings/pr_evaluate_prompt_response.toml)
Here is an example of a judge model feedback:
```
command: improve
model1_score: 9,
model2_score: 6,
why: |
Response 1 is better because it provides more actionable and specific suggestions that directly
enhance the code's maintainability, performance, and best practices. For example, it suggests
using a variable for reusable widget instances and using named routes for navigation, which
are practical improvements. In contrast, Response 2 focuses more on general advice and less
actionable suggestions, such as changing variable names and adding comments, which are less
critical for immediate code improvement."
```

184
docs/docs/index.md
View File

@ -13,119 +13,66 @@ CodiumAI PR-Agent is an open-source tool to help efficiently review and handle p
PR-Agent offers extensive pull request functionalities across various git providers.
| | | GitHub | Gitlab | Bitbucket | Azure DevOps |
|-------|---------------------------------------------------------------------------------------------------------------------|:------:|:------:|:---------:|:------------:|
| TOOLS | Review | ✔️ | ✔️ | ✔️ | ✔️ |
| | ⮑ Incremental | ✔️ | | | |
| | ⮑ [SOC2 Compliance](https://pr-agent-docs.codium.ai/tools/review/#soc2-ticket-compliance){:target="_blank"} 💎 | ✔️ | ✔️ | ✔️ | ✔️ |
| | Ask | ✔️ | ✔️ | ✔️ | ✔️ |
| | Describe | ✔️ | ✔️ | ✔️ | ✔️ |
| | ⮑ [Inline file summary](https://pr-agent-docs.codium.ai/tools/describe/#inline-file-summary){:target="_blank"} 💎 | ✔️ | ✔️ | ✔️ | ✔️ |
| | Improve | ✔️ | ✔️ | ✔️ | ✔️ |
| | ⮑ Extended | ✔️ | ✔️ | ✔️ | ✔️ |
| | [Custom Suggestions](./tools/custom_suggestions.md){:target="_blank"} 💎 | ✔️ | ✔️ | ✔️ | ✔️ |
| | Reflect and Review | ✔️ | ✔️ | ✔️ | ✔️ |
| | Update CHANGELOG.md | ✔️ | ✔️ | ✔️ | ️ |
| | Find Similar Issue | ✔️ | | | ️ |
| | [Add PR Documentation](./tools/documentation.md){:target="_blank"} 💎 | ✔️ | ✔️ | ✔️ | ✔️ |
| | [Generate Custom Labels](./tools/describe.md#handle-custom-labels-from-the-repos-labels-page-💎){:target="_blank"} 💎 | ✔️ | ✔️ | | ✔️ |
| | [Analyze PR Components](./tools/analyze.md){:target="_blank"} 💎 | ✔️ | ✔️ | ✔️ | ✔️ |
|-------|-----------------------------------------------------------------------------------------------------------------------|:------:|:------:|:---------:|:------------:|
| TOOLS | Review | ✅ | ✅ | ✅ | ✅ |
| | ⮑ Incremental | ✅ | | | |
| | ⮑ [SOC2 Compliance](https://pr-agent-docs.codium.ai/tools/review/#soc2-ticket-compliance){:target="_blank"} 💎 | ✅ | ✅ | ✅ | ✅ |
| | Ask | ✅ | ✅ | ✅ | ✅ |
| | Describe | ✅ | ✅ | ✅ | ✅ |
| | ⮑ [Inline file summary](https://pr-agent-docs.codium.ai/tools/describe/#inline-file-summary){:target="_blank"} 💎 | ✅ | ✅ | | ✅ |
| | Improve | ✅ | ✅ | ✅ | ✅ |
| | ⮑ Extended | ✅ | ✅ | ✅ | ✅ |
| | [Custom Prompt](./tools/custom_prompt.md){:target="_blank"} 💎 | ✅ | ✅ | ✅ | ✅ |
| | Reflect and Review | ✅ | ✅ | ✅ | ✅ |
| | Update CHANGELOG.md | ✅ | ✅ | ✅ | ️ |
| | Find Similar Issue | ✅ | | | ️ |
| | [Add PR Documentation](./tools/documentation.md){:target="_blank"} 💎 | ✅ | ✅ | | ✅ |
| | [Generate Custom Labels](./tools/describe.md#handle-custom-labels-from-the-repos-labels-page-💎){:target="_blank"} 💎 | ✅ | ✅ | | ✅ |
| | [Analyze PR Components](./tools/analyze.md){:target="_blank"} 💎 | ✅ | ✅ | | ✅ |
| | | | | | ️ |
| USAGE | CLI | ✔️ | ✔️ | ✔️ | ✔️ |
| | App / webhook | ✔️ | ✔️ | | ✔️ |
| | Tagging bot | ✔️ | | | ✔️ |
| | Actions | ✔️ | | | ️ |
| USAGE | CLI | ✅ | ✅ | ✅ | ✅ |
| | App / webhook | ✅ | ✅ | ✅ | ✅ |
| | Actions | ✅ | | | ️ |
| | | | | |
| CORE | PR compression | ✔️ | ✔️ | ✔️ | ✔️ |
| | Repo language prioritization | ✔️ | ✔️ | ✔️ | ✔️ |
| | Adaptive and token-aware file patch fitting | ✔️ | ✔️ | ✔️ | ✔️ |
| | Multiple models support | ✔️ | ✔️ | ✔️ | ✔️ |
| | Incremental PR review | ✔️ | | | |
| | [Static code analysis](./tools/analyze.md/){:target="_blank"} 💎 | ✔️ | ✔️ | ✔️ | ✔️ |
| | [Multiple configuration options](./usage-guide/configuration_options.md){:target="_blank"} 💎 | ✔️ | ✔️ | ✔️ | ✔️ |
| CORE | PR compression | ✅ | ✅ | ✅ | ✅ |
| | Repo language prioritization | ✅ | ✅ | ✅ | ✅ |
| | Adaptive and token-aware file patch fitting | ✅ | ✅ | ✅ | ✅ |
| | Multiple models support | ✅ | ✅ | ✅ | ✅ |
| | Incremental PR review | ✅ | | | |
| | [Static code analysis](./tools/analyze.md/){:target="_blank"} 💎 | ✅ | ✅ | ✅ | ✅ |
| | [Multiple configuration options](./usage-guide/configuration_options.md){:target="_blank"} 💎 | ✅ | ✅ | ✅ | ✅ |
💎 marks a feature available only in [PR-Agent Pro](https://www.codium.ai/pricing/){:target="_blank"}
## Example results
<hr>
<h4><a href="https://github.com/Codium-ai/pr-agent/pull/530">/describe</a></h4>
<div align="center">
<p float="center">
<img src="https://www.codium.ai/images/pr_agent/describe_new_short_main.png" width="512">
</p>
</div>
## Example Results
<hr>
<h4><a href="https://github.com/Codium-ai/pr-agent/pull/732#issuecomment-1975099151">/review</a></h4>
<div align="center">
<p float="center">
<kbd>
<img src="https://www.codium.ai/images/pr_agent/review_new_short_main.png" width="512">
</kbd>
</p>
</div>
#### [/describe](https://github.com/Codium-ai/pr-agent/pull/530)
<figure markdown="1">
![/describe](https://www.codium.ai/images/pr_agent/describe_new_short_main.png){width=512}
</figure>
<hr>
<h4><a href="https://github.com/Codium-ai/pr-agent/pull/732#issuecomment-1975099159">/improve</a></h4>
<div align="center">
<p float="center">
<kbd>
<img src="https://www.codium.ai/images/pr_agent/improve_new_short_main.png" width="512">
</kbd>
</p>
</div>
#### [/review](https://github.com/Codium-ai/pr-agent/pull/732#issuecomment-1975099151)
<figure markdown="1">
![/review](https://www.codium.ai/images/pr_agent/review_new_short_main.png){width=512}
</figure>
<hr>
<h4><a href="https://github.com/Codium-ai/pr-agent/pull/530">/generate_labels</a></h4>
<div align="center">
<p float="center">
<kbd><img src="https://www.codium.ai/images/pr_agent/geneare_custom_labels_main_short.png" width="300"></kbd>
</p>
</div>
[//]: # (<h4><a href="https://github.com/Codium-ai/pr-agent/pull/78#issuecomment-1639739496">/reflect_and_review:</a></h4>)
[//]: # (<div align="center">)
[//]: # (<p float="center">)
[//]: # (<img src="https://www.codium.ai/images/reflect_and_review.gif" width="800">)
[//]: # (</p>)
[//]: # (</div>)
[//]: # (<h4><a href="https://github.com/Codium-ai/pr-agent/pull/229#issuecomment-1695020538">/ask:</a></h4>)
[//]: # (<div align="center">)
[//]: # (<p float="center">)
[//]: # (<img src="https://www.codium.ai/images/ask-2.gif" width="800">)
[//]: # (</p>)
[//]: # (</div>)
[//]: # (<h4><a href="https://github.com/Codium-ai/pr-agent/pull/229#issuecomment-1695024952">/improve:</a></h4>)
[//]: # (<div align="center">)
[//]: # (<p float="center">)
[//]: # (<img src="https://www.codium.ai/images/improve-2.gif" width="800">)
[//]: # (</p>)
[//]: # (</div>)
<div align="left">
</div>
#### [/improve](https://github.com/Codium-ai/pr-agent/pull/732#issuecomment-1975099159)
<figure markdown="1">
![/improve](https://www.codium.ai/images/pr_agent/improve_new_short_main.png){width=512}
</figure>
<hr>
#### [/generate_labels](https://github.com/Codium-ai/pr-agent/pull/530)
<figure markdown="1">
![/generate_labels](https://www.codium.ai/images/pr_agent/geneare_custom_labels_main_short.png){width=300}
</figure>
<hr>
## How it works
## How it Works
The following diagram illustrates PR-Agent tools and their flow:
@ -139,24 +86,39 @@ Check out the [PR Compression strategy](core-abilities/index.md) page for more d
[PR-Agent Pro](https://www.codium.ai/pricing/) is a hosted version of PR-Agent, provided by CodiumAI. It is available for a monthly fee, and provides the following benefits:
1. **Fully managed** - We take care of everything for you - hosting, models, regular updates, and more. Installation is as simple as signing up and adding the PR-Agent app to your GitHub\BitBucket repo.
1. **Fully managed** - We take care of everything for you - hosting, models, regular updates, and more. Installation is as simple as signing up and adding the PR-Agent app to your GitHub\GitLab\BitBucket repo.
2. **Improved privacy** - No data will be stored or used to train models. PR-Agent Pro will employ zero data retention, and will use an OpenAI account with zero data retention.
3. **Improved support** - PR-Agent Pro users will receive priority support, and will be able to request new features and capabilities.
4. **Extra features** -In addition to the benefits listed above, PR-Agent Pro will emphasize more customization, and the usage of static code analysis, in addition to LLM logic, to improve results. It has the following additional tools and features:
- [**Analyze PR components**](./tools/analyze.md/)
- [**Custom Code Suggestions**](./tools/custom_suggestions.md/)
- [**Tests**](./tools/test.md/)
- [**PR documentation**](./tools/documentation.md/)
- [**CI feedback**](./tools/ci_feedback.md/)
- [**SOC2 compliance check**](./tools/review.md/#soc2-ticket-compliance)
- [**Custom labels**](./tools/describe.md/#handle-custom-labels-from-the-repos-labels-page)
- [**Global and wiki configuration**](./usage-guide/configuration_options.md/#wiki-configuration-file)
- (Tool): [**Analyze PR components**](./tools/analyze.md/)
- (Tool): [**Custom Prompt Suggestions**](./tools/custom_prompt.md/)
- (Tool): [**Tests**](./tools/test.md/)
- (Tool): [**PR documentation**](./tools/documentation.md/)
- (Tool): [**Improve Component**](https://pr-agent-docs.codium.ai/tools/improve_component/)
- (Tool): [**Similar code search**](https://pr-agent-docs.codium.ai/tools/similar_code/)
- (Tool): [**CI feedback**](./tools/ci_feedback.md/)
- (Feature): [**Interactive triggering**](./usage-guide/automations_and_usage.md/#interactive-triggering)
- (Feature): [**SOC2 compliance check**](./tools/review.md/#soc2-ticket-compliance)
- (Feature): [**Custom labels**](./tools/describe.md/#handle-custom-labels-from-the-repos-labels-page)
- (Feature): [**Global and wiki configuration**](./usage-guide/configuration_options.md/#wiki-configuration-file)
- (Feature): [**Inline file summary**](https://pr-agent-docs.codium.ai/tools/describe/#inline-file-summary)
## Data privacy
## Data Privacy
If you host PR-Agent with your OpenAI API key, it is between you and OpenAI. You can read their API data privacy policy here:
### Self-hosted PR-Agent
- If you host PR-Agent with your OpenAI API key, it is between you and OpenAI. You can read their API data privacy policy here:
https://openai.com/enterprise-privacy
When using PR-Agent Pro 💎, hosted by CodiumAI, we will not store any of your data, nor will we use it for training.
You will also benefit from an OpenAI account with zero data retention.
### CodiumAI-hosted PR-Agent Pro 💎
- When using PR-Agent Pro 💎, hosted by CodiumAI, we will not store any of your data, nor will we use it for training. You will also benefit from an OpenAI account with zero data retention.
- For certain clients, CodiumAI-hosted PR-Agent Pro will use CodiumAI’s proprietary models — if this is the case, you will be notified.
- No passive collection of Code and Pull Requests’ data — PR-Agent will be active only when you invoke it, and it will then extract and analyze only data relevant to the executed command and queried pull request.
### PR-Agent Chrome extension
- The [PR-Agent Chrome extension](https://chromewebstore.google.com/detail/pr-agent-chrome-extension/ephlnjeghhogofkifjloamocljapahnl) serves solely to modify the visual appearance of a GitHub PR screen. It does not transmit any user's repo or pull request code. Code is only sent for processing when a user submits a GitHub comment that activates a PR-Agent tool, in accordance with the standard privacy policy of PR-Agent.

11
docs/docs/installation/azure.md
View File

@ -4,16 +4,15 @@ To use Azure DevOps provider use the following settings in configuration.toml:
```
[config]
git_provider="azure"
use_repo_settings_file=false
```
Azure DevOps provider supports [PAT token](https://learn.microsoft.com/en-us/azure/devops/organizations/accounts/use-personal-access-tokens-to-authenticate?view=azure-devops&tabs=Windows) or [DefaultAzureCredential](https://learn.microsoft.com/en-us/azure/developer/python/sdk/authentication-overview#authentication-in-server-environments) authentication.
PAT is faster to create, but has build in experation date, and will use the user identity for API calls.
Using DefaultAzureCredential you can use managed identity or Service principle, which are more secure and will create seperate ADO user identity (via AAD) to the agent.
PAT is faster to create, but has build in expiration date, and will use the user identity for API calls.
Using DefaultAzureCredential you can use managed identity or Service principle, which are more secure and will create separate ADO user identity (via AAD) to the agent.
If PAT was choosen, you can assign the value in .secrets.toml.
If DefaultAzureCredential was choosen, you can assigned the additional env vars like AZURE_CLIENT_SECRET directly,
or use managed identity/az cli (for local develpment) without any additional configuration.
If PAT was chosen, you can assign the value in .secrets.toml.
If DefaultAzureCredential was chosen, you can assigned the additional env vars like AZURE_CLIENT_SECRET directly,
or use managed identity/az cli (for local development) without any additional configuration.
in any case, 'org' value must be assigned in .secrets.toml:
```
[azure_devops]

27
docs/docs/installation/github.md
View File

@ -7,9 +7,11 @@ You can use our pre-built Github Action Docker image to run PR-Agent as a Github
```yaml
on:
pull_request:
types: [opened, reopened, ready_for_review]
issue_comment:
jobs:
pr_agent_job:
if: ${{ github.event.sender.type != 'Bot' }}
runs-on: ubuntu-latest
permissions:
issues: write
@ -24,27 +26,14 @@ jobs:
OPENAI_KEY: ${{ secrets.OPENAI_KEY }}
GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
```
** if you want to pin your action to a specific release (v0.7 for example) for stability reasons, use:
** if you want to pin your action to a specific release (v2.0 for example) for stability reasons, use:
```yaml
on:
pull_request:
issue_comment:
jobs:
pr_agent_job:
runs-on: ubuntu-latest
permissions:
issues: write
pull-requests: write
contents: write
name: Run pr agent on every pull request, respond to user comments
...
steps:
- name: PR Agent action step
id: pragent
uses: Codium-ai/pr-agent@v0.7
env:
OPENAI_KEY: ${{ secrets.OPENAI_KEY }}
GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
uses: Codium-ai/pr-agent@v2.0
...
```
2) Add the following secret to your repository under `Settings > Secrets and variables > Actions > New repository secret > Add secret`:
@ -66,6 +55,7 @@ When you open your next PR, you should see a comment from `github-actions` bot w
PR_REVIEWER.REQUIRE_TESTS_REVIEW: "false" # Disable tests review
PR_CODE_SUGGESTIONS.NUM_CODE_SUGGESTIONS: 6 # Increase number of code suggestions
```
See detailed usage instructions in the [USAGE GUIDE](https://pr-agent-docs.codium.ai/usage-guide/automations_and_usage/#github-action)
---
@ -159,6 +149,9 @@ cp pr_agent/settings/.secrets_template.toml pr_agent/settings/.secrets.toml
## Deploy as a Lambda Function
Note that since AWS Lambda env vars cannot have "." in the name, you can replace each "." in an env variable with "__".<br>
For example: `GITHUB.WEBHOOK_SECRET` --> `GITHUB__WEBHOOK_SECRET`
1. Follow steps 1-5 from [here](#run-as-a-github-app).
2. Build a docker image that can be used as a lambda function
```shell

37
docs/docs/installation/gitlab_pro.md
View File

@ -1,37 +0,0 @@
## Install Hosted PR-Agent Pro for GitLab (Teams & Enterprise)
### Step 1
Acquire a personal, project or group level access token. Enable the “api” scope in order to allow PR-Agent to read pull requests, comment and respond to requests.
<kbd><img src=https://www.codium.ai/images/pr_agent/gitlab_pro_pat.png></kbd>
Store the token in a safe place, you won’t be able to access it again after it was generated.
### Step 2
Generate a shared secret and link it to the access token. Browse to [https://register.gitlab.pr-agent.codium.ai](https://register.gitlab.pr-agent.codium.ai).
Fill in your generated GitLab token and your company or personal name in the appropriate fields and click "Submit".
You should see "Success!" displayed above the Submit button, and a shared secret will be generated. Store it in a safe place, you won’t be able to access it again after it was generated.
### Step 3
Install a webhook for your repository or groups, by clicking “webhooks” on the settings menu. Click the “Add new webhook” button.
<kbd><img src=https://www.codium.ai/images/pr_agent/gitlab_pro_add_webhook.png></kbd>
In the webhook definition form, fill in the following fields:
URL: https://pro.gitlab.pr-agent.codium.ai/webhook
Secret token: Your CodiumAI key
Trigger: Check the ‘comments’ and ‘merge request events’ boxes.
Enable SSL verification: Check the box.
<kbd><img src=https://www.codium.ai/images/pr_agent/gitlab_pro_webhooks.png></kbd>
### Step 4
You’re all set!
Open a new merge request or add a MR comment with one of PR-Agent’s commands such as /review, /describe or /improve.

7
docs/docs/installation/index.md
View File

@ -1,5 +1,6 @@
# Installation
## Self-hosted PR-Agent
If you choose to host you own PR-Agent, you first need to acquire two tokens:
1. An OpenAI key from [here](https://platform.openai.com/api-keys), with access to GPT-4 (or a key for [other models](../usage-guide/additional_configurations.md/#changing-a-model), if you prefer).
@ -13,7 +14,9 @@ There are several ways to use self-hosted PR-Agent:
- [BitBucket](./bitbucket.md)
- [Azure DevOps](./azure.md)
___
Note that [PR-Agent Pro 💎](https://app.codium.ai/), an app for GitHub\GitLab\BitBucket hosted by CodiumAI, is also available.
## PR-Agent Pro 💎
PR-Agent Pro, an app for GitHub\GitLab\BitBucket hosted by CodiumAI, is also available.
<br>
With PR-Agent Pro Installation is as simple as signing up and adding the PR-Agent app to your relevant repo.
<br>
See [here](./pr_agent_pro.md) for more details.

39
docs/docs/installation/locally.md
View File

@ -1,4 +1,41 @@
## Use Docker image (no installation required)
## Using pip package
Install the package:
```
pip install pr-agent
```
Then run the relevant tool with the script below.
<br>
Make sure to fill in the required parameters (`user_token`, `openai_key`, `pr_url`, `command`):
```python
from pr_agent import cli
from pr_agent.config_loader import get_settings
def main():
# Fill in the following values
provider = "github" # GitHub provider
user_token = "..." # GitHub user token
openai_key = "..." # OpenAI key
pr_url = "..." # PR URL, for example 'https://github.com/Codium-ai/pr-agent/pull/809'
command = "/review" # Command to run (e.g. '/review', '/describe', '/ask="What is the purpose of this PR?"', ...)
# Setting the configurations
get_settings().set("CONFIG.git_provider", provider)
get_settings().set("openai.key", openai_key)
get_settings().set("github.user_token", user_token)
# Run the command. Feedback will appear in GitHub PR comments
cli.run_command(pr_url, command)
if __name__ == '__main__':
main()
```
## Using Docker image
A list of the relevant tools can be found in the [tools guide](../tools/ask.md).

64
docs/docs/installation/pr_agent_pro.md Normal file
View File

@ -0,0 +1,64 @@
## Getting Started with PR-Agent Pro
PR-Agent Pro is a versatile application compatible with GitHub, GitLab, and BitBucket, hosted by CodiumAI.
See [here](https://pr-agent-docs.codium.ai/#pr-agent-pro) for more details about the benefits of using PR-Agent Pro.
Interested parties can subscribe to PR-Agent Pro through the following [link](https://www.codium.ai/pricing/).
After subscribing, you are granted the ability to easily install the application across any of your repositories.
![PR Agent Pro](https://codium.ai/images/pr_agent/pr_agent_pro_install.png){width=468}
Each user who wants to use PR-Agent pro needs to buy a seat.
Initially, CodiumAI offers a two-week trial period at no cost, after which continued access requires each user to secure a personal seat.
Once a user acquires a seat, they gain the flexibility to use PR-Agent Pro across any repository where it was enabled.
Users without a purchased seat who interact with a repository featuring PR-Agent Pro are entitled to receive up to five complimentary feedbacks.
Beyond this limit, PR-Agent Pro will cease to respond to their inquiries unless a seat is purchased.
## Install PR-Agent Pro for GitLab (Teams & Enterprise)
Since GitLab platform does not support apps, installing PR-Agent Pro for GitLab is a bit more involved, and requires the following steps:
### Step 1
Acquire a personal, project or group level access token. Enable the “api” scope in order to allow PR-Agent to read pull requests, comment and respond to requests.
<figure markdown="1">
![Step 1](https://www.codium.ai/images/pr_agent/gitlab_pro_pat.png){width=750}
</figure>
Store the token in a safe place, you won’t be able to access it again after it was generated.
### Step 2
Generate a shared secret and link it to the access token. Browse to [https://register.gitlab.pr-agent.codium.ai](https://register.gitlab.pr-agent.codium.ai).
Fill in your generated GitLab token and your company or personal name in the appropriate fields and click "Submit".
You should see "Success!" displayed above the Submit button, and a shared secret will be generated. Store it in a safe place, you won’t be able to access it again after it was generated.
### Step 3
Install a webhook for your repository or groups, by clicking “webhooks” on the settings menu. Click the “Add new webhook” button.
<figure markdown="1">
![Step 3.1](https://www.codium.ai/images/pr_agent/gitlab_pro_add_webhook.png)
</figure>
In the webhook definition form, fill in the following fields:
URL: https://pro.gitlab.pr-agent.codium.ai/webhook
Secret token: Your CodiumAI key
Trigger: Check the ‘comments’ and ‘merge request events’ boxes.
Enable SSL verification: Check the box.
<figure markdown="1">
![Step 3.2](https://www.codium.ai/images/pr_agent/gitlab_pro_webhooks.png){width=750}
</figure>
### Step 4
You’re all set!
Open a new merge request or add a MR comment with one of PR-Agent’s commands such as /review, /describe or /improve.

15
docs/docs/tools/analyze.md
View File

@ -1,7 +1,7 @@
## Overview
The `analyze` tool combines static code analysis with LLM capabilities to provide a comprehensive analysis of the PR code changes.
The `analyze` tool combines advanced static code analysis with LLM capabilities to provide a comprehensive analysis of the PR code changes.
The tool scans the PR code changes, find the code components (methods, functions, classes) that changed, and summarizes the changes in each component.
The tool scans the PR code changes, find the code components (methods, functions, classes) that changed, and enables to interactively generate tests, docs, code suggestions and similar code search for each component.
It can be invoked manually by commenting on any PR:
```
@ -9,16 +9,11 @@ It can be invoked manually by commenting on any PR:
```
## Example usage
An example [result](https://github.com/Codium-ai/pr-agent/pull/546#issuecomment-1868524805):
<kbd><img src=https://codium.ai/images/pr_agent/analyze_1.png width="768"></kbd>
<kbd><img src=https://codium.ai/images/pr_agent/analyze_2.png width="768"></kbd>
<kbd><img src=https://codium.ai/images/pr_agent/analyze_3.png width="768"></kbd>
An example result:
![Analyze 1](https://codium.ai/images/pr_agent/analyze_1.png){width=750}
**Notes**
- Language that are currently supported: Python, Java, C++, JavaScript, TypeScript.
- Language that are currently supported: Python, Java, C++, JavaScript, TypeScript, C#.

44
docs/docs/tools/ask.md
View File

@ -5,11 +5,12 @@ It can be invoked manually by commenting on any PR:
```
/ask "..."
```
For example:
<kbd><img src="https://codium.ai/images/pr_agent/ask_comment.png" width="768"></kbd>
## Example usage
<kbd><img src="https://codium.ai/images/pr_agent/ask.png" width="768"></kbd>
![Ask Comment](https://codium.ai/images/pr_agent/ask_comment.png){width=512}
![Ask](https://codium.ai/images/pr_agent/ask.png){width=512}
## Ask lines
@ -18,6 +19,41 @@ You can run `/ask` on specific lines of code in the PR from the PR's diff view.
- To select multiple lines, click on the '+' sign of the first line and then hold and drag to select the rest of the lines.
- write `/ask "..."` in the comment box and press `Add single comment` button.
<kbd><img src="https://codium.ai/images/pr_agent/Ask_line.png" width="768"></kbd>
![Ask Line](https://codium.ai/images/pr_agent/Ask_line.png){width=512}
Note that the tool does not have "memory" of previous questions, and answers each question independently.
## Ask on images
You can also ask questions about images that appear in the comment, where the entire PR code will be used as context.
<br>
The basic syntax is:
```
/ask "..."
[Image](https://real_link_to_image)
```
where `https://real_link_to_image` is the direct link to the image.
Note that GitHub has a built-in mechanism of pasting images in comments. However, pasted image does not provide a direct link.
To get a direct link to an image, we recommend using the following scheme:
1) First, post a comment that contains **only** the image:
![Ask image1](https://codium.ai/images/pr_agent/ask_images1.png){width=512}
2) Quote reply to that comment:
![Ask image2](https://codium.ai/images/pr_agent/ask_images2.png){width=512}
3) In the screen opened, type the question below the image:
![Ask image3](https://codium.ai/images/pr_agent/ask_images3.png){width=512}
![Ask image4](https://codium.ai/images/pr_agent/ask_images4.png){width=512}
4) Post the comment, and receive the answer:
![Ask image5](https://codium.ai/images/pr_agent/ask_images5.png){width=512}
See a full video tutorial [here](https://codium.ai/images/pr_agent/ask_image_video.mov)

19
docs/docs/tools/ci_feedback.md
View File

@ -8,13 +8,12 @@ The tool analyzes the failed checks and provides several feedbacks:
- Failure summary
- Relevant error logs
<kbd>
<img src="https://www.codium.ai/images/pr_agent/failed_check1.png" width="768">
</kbd>
## Example usage
![Failed Check 1](https://www.codium.ai/images/pr_agent/failed_check1.png){width=768}
&rarr;
<kbd>
<img src="https://www.codium.ai/images/pr_agent/failed_check2.png" width="768">
</kbd>
![Failed Check 2](https://www.codium.ai/images/pr_agent/failed_check2.png){width=768}
___
@ -24,6 +23,14 @@ In addition to being automatically triggered, the tool can also be invoked manua
```
where `{repo_name}` is the name of the repository, `{run_number}` is the run number of the failed check, and `{job_number}` is the job number of the failed check.
## Disabling the tool from running automatically
If you wish to disable the tool from running automatically, you can do so by adding the following configuration to the configuration file:
```
[checks]
enable_auto_checks_feedback = false
```
## Configuration options
- `enable_auto_checks_feedback` - if set to true, the tool will automatically provide feedback when a check is failed. Default is true.
- `excluded_checks_list` - a list of checks to exclude from the feedback, for example: ["check1", "check2"]. Default is an empty list.

10
docs/docs/tools/custom_labels.md
View File

@ -5,15 +5,16 @@ It can be invoked manually by commenting on any PR:
```
/generate_labels
```
For example:
## Example usage
If we wish to add detect changes to SQL queries in a given PR, we can add the following custom label along with its description:
<kbd><img src=https://codium.ai/images/pr_agent/custom_labels_list.png width="768"></kbd>
![Custom labels list](https://codium.ai/images/pr_agent/custom_labels_list.png){width=768}
When running the `generate_labels` tool on a PR that includes changes in SQL queries, it will automatically suggest the custom label:
<kbd><img src=https://codium.ai/images/pr_agent/custom_label_published.png width="768"></kbd>
![Custom labels published](https://codium.ai/images/pr_agent/custom_label_published.png){width=768}
Note that in addition to the dedicated tool `generate_labels`, the custom labels will also be used by the `describe` tool.
@ -36,7 +37,8 @@ b. Add/edit the custom labels. It should be formatted as follows:
* Label name: The name of the custom label.
* Description: Start the description of with prefix `pr_agent:`, for example: `pr_agent: Description of when AI should suggest this label`.<br>
The description should be comprehensive and detailed, indicating when to add the desired label.
<kbd><img src=https://codium.ai/images/pr_agent/add_native_custom_labels.png width="880"></kbd>
![Add native custom labels](https://codium.ai/images/pr_agent/add_native_custom_labels.png){width=880}
c. Now the custom labels will be included in the `generate_labels` tool.

58
docs/docs/tools/custom_prompt.md Normal file
View File

@ -0,0 +1,58 @@
## Overview
The `custom_prompt` tool scans the PR code changes, and automatically generates suggestions for improving the PR code.
It shares similarities with the `improve` tool, but with one main difference: the `custom_prompt` tool will **only propose suggestions that follow specific guidelines defined by the prompt** in: `pr_custom_prompt.prompt` configuration.
The tool can be triggered [automatically](../usage-guide/automations_and_usage.md#github-app-automatic-tools-when-a-new-pr-is-opened) every time a new PR is opened, or can be invoked manually by commenting on a PR.
When commenting, use the following template:
```
/custom_prompt --pr_custom_prompt.prompt="
The code suggestions should focus only on the following:
- ...
- ...
"
```
With a [configuration file](../usage-guide/automations_and_usage.md#github-app), use the following template:
```
[pr_custom_prompt]
prompt="""\
The suggestions should focus only on the following:
-...
-...
"""
```
Remember - with this tool, you are the prompter. Be specific, clear, and concise in the instructions. Specify relevant aspects that you want the model to focus on. \
You might benefit from several trial-and-error iterations, until you get the correct prompt for your use case.
## Example usage
Here is an example of a possible prompt, defined in the configuration file:
```
[pr_custom_prompt]
prompt="""\
The code suggestions should focus only on the following:
- look for edge cases when implementing a new function
- make sure every variable has a meaningful name
- make sure the code is efficient
"""
```
(The instructions above are just an example. We want to emphasize that the prompt should be specific and clear, and be tailored to the needs of your project)
Results obtained with the prompt above:
![Custom prompt results](https://codium.ai/images/pr_agent/custom_suggestions_result.png){width=768}
## Configuration options
`prompt`: the prompt for the tool. It should be a multi-line string.
`num_code_suggestions`: number of code suggestions provided by the 'custom_prompt' tool. Default is 4.
`enable_help_text`: if set to true, the tool will display a help text in the comment. Default is true.

57
docs/docs/tools/custom_suggestions.md
View File

@ -1,57 +0,0 @@
## Overview
The `custom_suggestions` tool scans the PR code changes, and automatically generates custom suggestions for improving the PR code.
It shares similarities with the `improve` tool, but with one main difference: the `custom_suggestions` tool will only propose suggestions that follow specific guidelines defined by the prompt in: `pr_custom_suggestions.prompt` configuration.
The tool can be triggered [automatically](../usage-guide/automations_and_usage.md#github-app-automatic-tools-when-a-new-pr-is-opened) every time a new PR is opened, or can be invoked manually by commenting on a PR.
When commenting, use the following template:
```
/custom_suggestions --pr_custom_suggestions.prompt="The suggestions should focus only on the following:\n-...\n-...\n-..."
```
With a [configuration file](../usage-guide/automations_and_usage.md#github-app), use the following template:
```
[pr_custom_suggestions]
prompt="""\
The suggestions should focus only on the following:
-...
-...
-...
"""
```
Using a configuration file is recommended, since it allows to use multi-line instructions.
Don't forget - with this tool, you are the prompter. Be specific, clear, and concise in the instructions. Specify relevant aspects that you want the model to focus on. \
You might benefit from several trial-and-error iterations, until you get the correct prompt for your use case.
## Example usage
Here is an example of a possible prompt:
```
[pr_custom_suggestions]
prompt="""\
The suggestions should focus only on the following:
- look for edge cases when implementing a new function
- make sure every variable has a meaningful name
- make sure the code is efficient
"""
```
The instructions above are just an example. We want to emphasize that the prompt should be specific and clear, and be tailored to the needs of your project.
Results obtained with the prompt above:
<kbd><img src=https://codium.ai/images/pr_agent/custom_suggestions_prompt.png width="512"></kbd>
<kbd><img src=https://codium.ai/images/pr_agent/custom_suggestions_result.png width="768"></kbd>
## Configuration options
`prompt`: the prompt for the tool. It should be a multi-line string.
`num_code_suggestions`: number of code suggestions provided by the 'custom_suggestions' tool. Default is 4.
`enable_help_text`: if set to true, the tool will display a help text in the comment. Default is true.

222
docs/docs/tools/describe.md
View File

@ -5,80 +5,119 @@ The tool can be triggered automatically every time a new PR is [opened](../usage
```
/describe
```
For example:
<kbd><img src=https://codium.ai/images/pr_agent/describe_comment.png width="768"></kbd>
## Example usage
<kbd><img src=https://codium.ai/images/pr_agent/describe_new.png width="768"></kbd>
### Manual triggering
Invoke the tool manually by commenting `/describe` on any PR:
![Describe comment](https://codium.ai/images/pr_agent/describe_comment.png){width=512}
## Configuration options
To edit [configurations](https://github.com/Codium-ai/pr-agent/blob/main/pr_agent/settings/configuration.toml#L46) related to the describe tool (`pr_description` section), use the following template:
After ~30 seconds, the tool will generate a description for the PR:
![Describe New](https://codium.ai/images/pr_agent/describe_new.png){width=512}
If you want to edit [configurations](#configuration-options), add the relevant ones to the command:
```
/describe --pr_description.some_config1=... --pr_description.some_config2=...
```
### Possible configurations:
### Automatic triggering
- `publish_labels`: if set to true, the tool will publish the labels to the PR. Default is true.
To run the `describe` automatically when a PR is opened, define in a [configuration file](https://pr-agent-docs.codium.ai/usage-guide/configuration_options/#wiki-configuration-file):
```
[github_app]
pr_commands = [
"/describe",
...
]
- `publish_description_as_comment`: if set to true, the tool will publish the description as a comment to the PR. If false, it will overwrite the origianl description. Default is false.
[pr_description]
publish_labels = ...
...
```
- `add_original_user_description`: if set to true, the tool will add the original user description to the generated description. Default is true.
- The `pr_commands` lists commands that will be executed automatically when a PR is opened.
- The `[pr_description]` section contains the configurations for the `describe` tool you want to edit (if any).
- `keep_original_user_title`: if set to true, the tool will keep the original PR title, and won't change it. Default is true.
- `extra_instructions`: Optional extra instructions to the tool. For example: "focus on the changes in the file X. Ignore change in ...".
## Configuration options
- To enable `custom labels`, apply the configuration changes described [here](./custom_labels.md#configuration-options)
!!! example "Possible configurations"
- `enable_pr_type`: if set to false, it will not show the `PR type` as a text value in the description content. Default is true.
<table>
<tr>
<td><b>publish_labels</b></td>
<td>If set to true, the tool will publish the labels to the PR. Default is true.</td>
</tr>
<tr>
<td><b>publish_description_as_comment</b></td>
<td>If set to true, the tool will publish the description as a comment to the PR. If false, it will overwrite the original description. Default is false.</td>
</tr>
<tr>
<td><b>publish_description_as_comment_persistent</b></td>
<td>If set to true and `publish_description_as_comment` is true, the tool will publish the description as a persistent comment to the PR. Default is true.</td>
</tr>
<tr>
<td><b>add_original_user_description</b></td>
<td>If set to true, the tool will add the original user description to the generated description. Default is true.</td>
</tr>
<tr>
<td><b>generate_ai_title</b></td>
<td>If set to true, the tool will also generate an AI title for the PR. Default is false.</td>
</tr>
<tr>
<td><b>extra_instructions</b></td>
<td>Optional extra instructions to the tool. For example: "focus on the changes in the file X. Ignore change in ..."</td>
</tr>
<tr>
<td><b>enable_pr_type</b></td>
<td>If set to false, it will not show the `PR type` as a text value in the description content. Default is true.</td>
</tr>
<tr>
<td><b>final_update_message</b></td>
<td>If set to true, it will add a comment message [`PR Description updated to latest commit...`](https://github.com/Codium-ai/pr-agent/pull/499#issuecomment-1837412176) after finishing calling `/describe`. Default is false.</td>
</tr>
<tr>
<td><b>enable_semantic_files_types</b></td>
<td>If set to true, "Changes walkthrough" section will be generated. Default is true.</td>
</tr>
<tr>
<td><b>collapsible_file_list</b></td>
<td>If set to true, the file list in the "Changes walkthrough" section will be collapsible. If set to "adaptive", the file list will be collapsible only if there are more than 8 files. Default is "adaptive".</td>
</tr>
<tr>
<td><b>enable_help_text</b></td>
<td>If set to true, the tool will display a help text in the comment. Default is false.</td>
</tr>
</table>
- `final_update_message`: if set to true, it will add a comment message [`PR Description updated to latest commit...`](https://github.com/Codium-ai/pr-agent/pull/499#issuecomment-1837412176) after finishing calling `/describe`. Default is true.
- `enable_semantic_files_types`: if set to true, "Changes walkthrough" section will be generated. Default is true.
- `collapsible_file_list`: if set to true, the file list in the "Changes walkthrough" section will be collapsible. If set to "adaptive", the file list will be collapsible only if there are more than 8 files. Default is "adaptive".
## Inline file summary 💎
### Inline file summary 💎
> This feature is available only in PR-Agent Pro
This feature enables you to copy the `changes walkthrough` table to the "Files changed" tab, so you can quickly understand the changes in each file while reviewing the code changes (diff view).
This feature will enable you to quickly understand the changes in each file while reviewing the code changes (diff view).
To copy the `changes walkthrough` table to the "Files changed" tab, you can click on the checkbox that appears PR Description status message below the main PR Description:
To add the walkthrough table to the "Files changed" tab, you can click on the checkbox that appears PR Description status message below the main PR Description:
<kbd><img src=https://codium.ai/images/pr_agent/add_table_checkbox.png width="512"></kbd>
![Add table checkbox](https://codium.ai/images/pr_agent/add_table_checkbox.png){width=512}
If you prefer to have the file summaries appear in the "Files changed" tab on every PR, change the `pr_description.inline_file_summary` parameter in the configuration file, possible values are:
- `'table'`: File changes walkthrough table will be displayed on the top of the "Files changed" tab, in addition to the "Conversation" tab.
<kbd><img src=https://codium.ai/images/pr_agent/diffview-table.png width="768"></kbd>
![Diffview table](https://codium.ai/images/pr_agent/diffview-table.png){width=512}
- `true`: A collapsable file comment with changes title and a changes summary for each file in the PR.
<kbd><img src=https://codium.ai/images/pr_agent/diffview_changes.png width="768"></kbd>
![Diffview changes](https://codium.ai/images/pr_agent/diffview_changes.png){width=512}
- `false` (`default`): File changes walkthrough will be added only to the "Conversation" tab.
**Note** that this feature is currently available only for GitHub.
**Note**: that this feature is currently available only for GitHub.
### Handle custom labels from the Repo's labels page 💎
> This feature is available only in PR-Agent Pro
You can control the custom labels that will be suggested by the `describe` tool, from the repo's labels page:
* GitHub : go to `https://github.com/{owner}/{repo}/labels` (or click on the "Labels" tab in the issues or PRs page)
* GitLab : go to `https://gitlab.com/{owner}/{repo}/-/labels` (or click on "Manage" -> "Labels" on the left menu)
Now add/edit the custom labels. they should be formatted as follows:
* Label name: The name of the custom label.
* Description: Start the description of with prefix `pr_agent:`, for example: `pr_agent: Description of when AI should suggest this label`.<br>
The description should be comprehensive and detailed, indicating when to add the desired label. For example:
<kbd><img src=https://codium.ai/images/pr_agent/add_native_custom_labels.png width="880"></kbd>
### Markers template
## Markers template
To enable markers, set `pr_description.use_description_markers=true`.
Markers enable to easily integrate user's content and auto-generated content, with a template-like mechanism.
@ -98,56 +137,73 @@ pr_agent:walkthrough
```
The marker `pr_agent:type` will be replaced with the PR type, `pr_agent:summary` will be replaced with the PR summary, and `pr_agent:walkthrough` will be replaced with the PR walkthrough.
<kbd><img src=https://codium.ai/images/pr_agent/describe_markers_before.png width="768"></kbd>
&rarr;
<kbd><img src=https://codium.ai/images/pr_agent/describe_markers_after.png width="768"></kbd>
![Describe markers before](https://codium.ai/images/pr_agent/describe_markers_before.png){width=512}
### Configuration params:
&rarr;
![Describe markers after](https://codium.ai/images/pr_agent/describe_markers_after.png){width=512}
**Configuration params**:
- `use_description_markers`: if set to true, the tool will use markers template. It replaces every marker of the form `pr_agent:marker_name` with the relevant content. Default is false.
- `include_generated_by_header`: if set to true, the tool will add a dedicated header: 'Generated by PR Agent at ...' to any automatic content. Default is true.
## Usage Tips
### Automation
- When you first install the app, the [default mode](../usage-guide/automations_and_usage.md#github-app) for the describe tool is:
```
pr_commands = ["/describe --pr_description.add_original_user_description=true"
"--pr_description.keep_original_user_title=true", ...]
```
meaning the `describe` tool will run automatically on every PR, will keep the original title, and will add the original user description above the generated description.
<br> This default settings aim to strike a good balance between automation and control:
If you want more automation, just give the PR a title, and the tool will auto-write a full description; If you want more control, you can add a detailed description, and the tool will add the complementary description below it.
- For maximal automation, you can change the default mode to:
```
pr_commands = ["/describe --pr_description.add_original_user_description=false"
"--pr_description.keep_original_user_title=true", ...]
```
so the title will be auto-generated as well.
- Markers are an alternative way to control the generated description, to give maximal control to the user. If you set:
```
pr_commands = ["/describe --pr_description.use_description_markers=true", ...]
```
the tool will replace every marker of the form `pr_agent:marker_name` in the PR description with the relevant content, where `marker_name` is one of the following:
- `type`: the PR type.
- `summary`: the PR summary.
- `walkthrough`: the PR walkthrough.
Note that when markers are enabled, if the original PR description does not contain any markers, the tool will not alter the description at all.
### Custom labels
## Custom labels
The default labels of the describe tool are quite generic, since they are meant to be used in any repo: [`Bug fix`, `Tests`, `Enhancement`, `Documentation`, `Other`].
If you specify [custom labels](#handle-custom-labels-from-the-repos-labels-page) in the repo's labels page, you can get tailored labels for your use cases.
You can define custom labels that are relevant for your repo and use cases.
Custom labels can be defined in a [configuration file](https://pr-agent-docs.codium.ai/tools/custom_labels/#configuration-options), or directly in the repo's [labels page](#handle-custom-labels-from-the-repos-labels-page).
Examples for custom labels:
- `Main topic:performance` - pr_agent:The main topic of this PR is performance
- `New endpoint` - pr_agent:A new endpoint was added in this PR
- `SQL query` - pr_agent:A new SQL query was added in this PR
- `Dockerfile changes` - pr_agent:The PR contains changes in the Dockerfile
- ...
- `Main topic:performance` - pr_agent:The main topic of this PR is performance
- `New endpoint` - pr_agent:A new endpoint was added in this PR
- `SQL query` - pr_agent:A new SQL query was added in this PR
- `Dockerfile changes` - pr_agent:The PR contains changes in the Dockerfile
- ...
The list above is eclectic, and aims to give an idea of different possibilities. Define custom labels that are relevant for your repo and use cases.
Note that Labels are not mutually exclusive, so you can add multiple label categories.
<br>Make sure to provide proper title, and a detailed and well-phrased description for each label, so the tool will know when to suggest it.
<br>
Make sure to provide proper title, and a detailed and well-phrased description for each label, so the tool will know when to suggest it.
Each label description should be a **conditional statement**, that indicates if to add the label to the PR or not, according to the PR content.
### Handle custom labels from the Repo's labels page 💎
You can control the custom labels that will be suggested by the `describe` tool, from the repo's labels page:
* GitHub : go to `https://github.com/{owner}/{repo}/labels` (or click on the "Labels" tab in the issues or PRs page)
* GitLab : go to `https://gitlab.com/{owner}/{repo}/-/labels` (or click on "Manage" -> "Labels" on the left menu)
Now add/edit the custom labels. they should be formatted as follows:
* Label name: The name of the custom label.
* Description: Start the description of with prefix `pr_agent:`, for example: `pr_agent: Description of when AI should suggest this label`.<br>
The description should be comprehensive and detailed, indicating when to add the desired label. For example:
![Add native custom labels](https://codium.ai/images/pr_agent/add_native_custom_labels.png){width=768}
## Usage Tips
!!! tip "Automation"
- When you first install PR-Agent app, the [default mode](../usage-guide/automations_and_usage.md#github-app) for the describe tool is:
```
pr_commands = ["/describe", ...]
```
meaning the `describe` tool will run automatically on every PR, with the default configurations.
- Markers are an alternative way to control the generated description, to give maximal control to the user. If you set:
```
pr_commands = ["/describe --pr_description.use_description_markers=true", ...]
```
the tool will replace every marker of the form `pr_agent:marker_name` in the PR description with the relevant content, where `marker_name` is one of the following:
* `type`: the PR type.
* `summary`: the PR summary.
* `walkthrough`: the PR walkthrough.
- Note that when markers are enabled, if the original PR description does not contain any markers, the tool will not alter the description at all.

23
docs/docs/tools/documentation.md
View File

@ -5,13 +5,23 @@ It can be invoked manually by commenting on any PR:
```
/add_docs
```
For example:
<kbd><img src=https://codium.ai/images/pr_agent/docs_command.png width="768"></kbd>
## Example usage
<kbd><img src=https://codium.ai/images/pr_agent/docs_components.png width="768"></kbd>
Invoke the tool manually by commenting `/add_docs` on any PR:
<kbd><img src=https://codium.ai/images/pr_agent/docs_single_component.png width="768"></kbd>
![Docs command](https://codium.ai/images/pr_agent/docs_command.png){width=768}
The tool will generate documentation for all the components that changed in the PR:
![Docs component](https://codium.ai/images/pr_agent/docs_components.png){width=768}
![Docs single component](https://codium.ai/images/pr_agent/docs_single_component.png){width=768}
You can state a name of a specific component in the PR to get documentation only for that component:
```
/add_docs component_name
```
## Configuration options
- `docs_style`: The exact style of the documentation (for python docstring). you can choose between: `google`, `numpy`, `sphinx`, `restructuredtext`, `plain`. Default is `sphinx`.
@ -19,6 +29,5 @@ For example:
**Notes**
- Language that are currently fully supported: Python, Java, C++, JavaScript, TypeScript.
- For languages that are not fully supported, the tool will suggest documentation only for new components in the PR.
- A previous version of the tool, that offered support only for new components, was deprecated.
- Language that are currently fully supported: Python, Java, C++, JavaScript, TypeScript, C#.
- This tool can also be triggered interactively by using the [`analyze`](./analyze.md) tool.

17
docs/docs/tools/help.md Normal file
View File

@ -0,0 +1,17 @@
## Overview
The `help` tool provides a list of all the available tools and their descriptions.
For PR-Agent Pro users, it also enables to trigger each tool by checking the relevant box.
It can be invoked manually by commenting on any PR:
```
/help
```
## Example usage
An example [result](https://github.com/Codium-ai/pr-agent/pull/546#issuecomment-1868524805):
![Help 1](https://codium.ai/images/pr_agent/help1.png){width=750}
&rarr;
![Analyze 2](https://codium.ai/images/pr_agent/help2.png){width=750}

202
docs/docs/tools/improve.md
View File

@ -5,85 +5,163 @@ The tool can be triggered automatically every time a new PR is [opened](../usage
/improve
```
### Summarized vs committable code suggestions
## Example usage
The code suggestions can be presented as a single comment (via `pr_code_suggestions.summarize=true`):
### Manual triggering
<kbd><img src=https://codium.ai/images/pr_agent/code_suggestions_as_comment.png width="768"></kbd>
Invoke the tool manually by commenting `/improve` on any PR. The code suggestions by default are presented as a single comment:
![code suggestions as comment](https://codium.ai/images/pr_agent/code_suggestions_as_comment.png){width=512}
Or as a separate commitable code comment for each suggestion:
<kbd><img src=https://codium.ai/images/pr_agent/improve.png width="768"></kbd>
Note that a single comment has a significantly smaller PR footprint. We recommend this mode for most cases.
Also note that collapsible are not supported in _Bitbucket_. Hence, the suggestions are presented there as code comments.
### Extended mode
An extended mode, which does not involve PR Compression and provides more comprehensive suggestions, can be invoked by commenting on any PR:
```
/improve --extended
```
or by setting:
```
[pr_code_suggestions]
auto_extended_mode=true
```
(True by default).
Note that the extended mode divides the PR code changes into chunks, up to the token limits, where each chunk is handled separately (might use multiple calls to GPT-4 for large PRs).
Hence, the total number of suggestions is proportional to the number of chunks, i.e., the size of the PR.
### Configuration options
To edit [configurations](https://github.com/Codium-ai/pr-agent/blob/main/pr_agent/settings/configuration.toml#L66) related to the improve tool (`pr_code_suggestions` section), use the following template:
To edit [configurations](#configuration-options) related to the improve tool, use the following template:
```
/improve --pr_code_suggestions.some_config1=... --pr_code_suggestions.some_config2=...
```
### General options
For example, you can choose to present the suggestions as commitable code comments, by running the following command:
```
/improve --pr_code_suggestions.commitable_code_suggestions=true
```
- `num_code_suggestions`: number of code suggestions provided by the 'improve' tool. Default is 4.
- `extra_instructions`: Optional extra instructions to the tool. For example: "focus on the changes in the file X. Ignore change in ...".
- `rank_suggestions`: if set to true, the tool will rank the suggestions, based on importance. Default is false.
- `summarize`: if set to true, the tool will display the suggestions in a single comment. Default is false.
- `enable_help_text`: if set to true, the tool will display a help text in the comment. Default is true.
![improve](https://codium.ai/images/pr_agent/improve.png){width=512}
### params for '/improve --extended' mode
- `auto_extended_mode`: enable extended mode automatically (no need for the `--extended` option). Default is true.
- `num_code_suggestions_per_chunk`: number of code suggestions provided by the 'improve' tool, per chunk. Default is 8.
- `rank_extended_suggestions`: if set to true, the tool will rank the suggestions, based on importance. Default is true.
- `max_number_of_calls`: maximum number of chunks. Default is 5.
- `final_clip_factor`: factor to remove suggestions with low confidence. Default is 0.9.
Note that a single comment has a significantly smaller PR footprint. We recommend this mode for most cases.
Also note that collapsible are not supported in _Bitbucket_. Hence, the suggestions are presented there as code comments.
### Automatic triggering
To run the `improve` automatically when a PR is opened, define in a [configuration file](https://pr-agent-docs.codium.ai/usage-guide/configuration_options/#wiki-configuration-file):
```
[github_app]
pr_commands = [
"/improve",
...
]
[pr_code_suggestions]
num_code_suggestions_per_chunk = ...
...
```
- The `pr_commands` lists commands that will be executed automatically when a PR is opened.
- The `[pr_code_suggestions]` section contains the configurations for the `improve` tool you want to edit (if any)
### Extended mode
An extended mode, which does not involve PR Compression and provides more comprehensive suggestions, can be invoked by commenting on any PR by setting:
```
[pr_code_suggestions]
auto_extended_mode=true
```
(This mode is true by default).
Note that the extended mode divides the PR code changes into chunks, up to the token limits, where each chunk is handled separately (might use multiple calls to GPT-4 for large PRs).
Hence, the total number of suggestions is proportional to the number of chunks, i.e., the size of the PR.
## Configuration options
!!! example "General options"
<table>
<tr>
<td><b>num_code_suggestions</b></td>
<td>Number of code suggestions provided by the 'improve' tool. Default is 4 for CLI, 0 for auto tools.</td>
</tr>
<tr>
<td><b>extra_instructions</b></td>
<td>Optional extra instructions to the tool. For example: "focus on the changes in the file X. Ignore change in ...".</td>
</tr>
<tr>
<td><b>rank_suggestions</b></td>
<td>If set to true, the tool will rank the suggestions, based on importance. Default is false.</td>
</tr>
<tr>
<td><b>commitable_code_suggestions</b></td>
<td>If set to true, the tool will display the suggestions as commitable code comments. Default is false.</td>
</tr>
<tr>
<td><b>persistent_comment</b></td>
<td>If set to true, the improve comment will be persistent, meaning that every new improve request will edit the previous one. Default is false.</td>
</tr>
<tr>
<td><b>self_reflect_on_suggestions</b></td>
<td>If set to true, the improve tool will calculate an importance score for each suggestion [1-10], and sort the suggestion labels group based on this score. Default is true.</td>
</tr>
<tr>
<td><b>suggestions_score_threshold</b></td>
<td> Any suggestion with importance score less than this threshold will be removed. Default is 0. Highly recommend not to set this value above 7-8, since above it may clip relevant suggestions that can be useful. </td>
</tr>
<tr>
<td><b>enable_help_text</b></td>
<td>If set to true, the tool will display a help text in the comment. Default is true.</td>
</tr>
</table>
!!! example "params for 'extended' mode"
<table>
<tr>
<td><b>auto_extended_mode</b></td>
<td>Enable extended mode automatically (no need for the --extended option). Default is true.</td>
</tr>
<tr>
<td><b>num_code_suggestions_per_chunk</b></td>
<td>Number of code suggestions provided by the 'improve' tool, per chunk. Default is 5.</td>
</tr>
<tr>
<td><b>rank_extended_suggestions</b></td>
<td>If set to true, the tool will rank the suggestions, based on importance. Default is true.</td>
</tr>
<tr>
<td><b>max_number_of_calls</b></td>
<td>Maximum number of chunks. Default is 5.</td>
</tr>
<tr>
<td><b>final_clip_factor</b></td>
<td>Factor to remove suggestions with low confidence. Default is 0.9.</td>
</tr>
</table>
## Usage Tips
### Extra instructions
!!! tip "Extra instructions"
Extra instructions are very important for the `imrpove` tool, since they enable you to guide the model to suggestions that are more relevant to the specific needs of the project.
Extra instructions are very important for the `imrpove` tool, since they enable you to guide the model to suggestions that are more relevant to the specific needs of the project.
Be specific, clear, and concise in the instructions. With extra instructions, you are the prompter. Specify relevant aspects that you want the model to focus on.
Be specific, clear, and concise in the instructions. With extra instructions, you are the prompter. Specify relevant aspects that you want the model to focus on.
Examples for extra instructions:
```
[pr_code_suggestions] # /improve #
extra_instructions="""
Emphasize the following aspects:
- Does the code logic cover relevant edge cases?
- Is the code logic clear and easy to understand?
- Is the code logic efficient?
...
"""
```
Use triple quotes to write multi-line instructions. Use bullet points to make the instructions more readable.
Examples for extra instructions:
```
[pr_code_suggestions] # /improve #
extra_instructions="""\
Emphasize the following aspects:
- Does the code logic cover relevant edge cases?
- Is the code logic clear and easy to understand?
- Is the code logic efficient?
...
"""
```
Use triple quotes to write multi-line instructions. Use bullet points to make the instructions more readable.
### A note on code suggestions quality
!!! tip "Review vs. Improve tools comparison"
- While the current AI for code is getting better and better (GPT-4), it's not flawless. Not all the suggestions will be perfect, and a user should not accept all of them automatically.
- Suggestions are not meant to be [simplistic](https://github.com/Codium-ai/pr-agent/blob/main/pr_agent/settings/pr_code_suggestions_prompts.toml#L34). Instead, they aim to give deep feedback and raise questions, ideas and thoughts to the user, who can then use his judgment, experience, and understanding of the code base.
- Recommended to use the `exra_instructions` field to guide the model to suggestions that are more relevant to the specific needs of the project.
- Consider also trying the [Custom Suggestions Tool](./custom_suggestions.md) 💎, that will **only** propose suggestions that follow specific guidelines defined by user.
- The [review](https://pr-agent-docs.codium.ai/tools/review/) tool includes a section called 'Possible issues', that also provide feedback on the PR Code.
In this section, the model is instructed to focus **only** on [major bugs and issues](https://github.com/Codium-ai/pr-agent/blob/main/pr_agent/settings/pr_reviewer_prompts.toml#L71).
- The `improve` tool, on the other hand, has a broader mandate, and in addition to bugs and issues, it can also give suggestions for improving code quality and making the code more efficient, readable, and maintainable (see [here](https://github.com/Codium-ai/pr-agent/blob/main/pr_agent/settings/pr_code_suggestions_prompts.toml#L34)).
- Hence, if you are interested only in feedback about clear bugs, the `review` tool might suffice. If you want a more detailed feedback, including broader suggestions for improving the PR code, also enable the `improve` tool to run on each PR.
## A note on code suggestions quality
- While the current AI for code is getting better and better (GPT-4), it's not flawless. Not all the suggestions will be perfect, and a user should not accept all of them automatically. Critical reading and judgment are required.
- While mistakes of the AI are rare but can happen, a real benefit from the suggestions of the `improve` (and [`review`](https://pr-agent-docs.codium.ai/tools/review/)) tool is to catch, with high probability, **mistakes or bugs done by the PR author**, when they happen. So, it's a good practice to spend the needed ~30-60 seconds to review the suggestions, even if not all of them are always relevant.
- The hierarchical structure of the suggestions is designed to help the user to _quickly_ understand them, and to decide which ones are relevant and which are not:
- Only if the `Category` header is relevant, the user should move to the summarized suggestion description
- Only if the summarized suggestion description is relevant, the user should click on the collapsible, to read the full suggestion description with a code preview example.
In addition, we recommend to use the `exra_instructions` field to guide the model to suggestions that are more relevant to the specific needs of the project.
<br>
Consider also trying the [Custom Prompt Tool](./custom_prompt.md) 💎, that will **only** propose code suggestions that follow specific guidelines defined by user.

29
docs/docs/tools/improve_component.md Normal file
View File

@ -0,0 +1,29 @@
## Overview
The `improve_component` tool generates code suggestions for a specific code component that changed in the PR.
it can be invoked manually by commenting on any PR:
```
/improve_component component_name
```
To get a list of the components that changed in the PR and choose the relevant component interactively, use the [`analyze`](./analyze.md) tool.
## Example usage
Invoke the tool manually by commenting `/improve_component` on any PR:
![improve_component1](https://codium.ai/images/pr_agent/improve_component1.png){width=768}
The tool will generate code suggestions for the selected component (if no component is stated, it will generate code suggestions for the largest component):
![improve_component2](https://codium.ai/images/pr_agent/improve_component2.png){width=768}
**Notes**
- Language that are currently supported by the tool: Python, Java, C++, JavaScript, TypeScript, C#.
- This tool can also be triggered interactively by using the [`analyze`](./analyze.md) tool.
## Configuration options
- `num_code_suggestions`: number of code suggestions to provide. Default is 4
- `extra_instructions`: Optional extra instructions to the tool. For example: "focus on ...".
- `file`: in case there are several components with the same name, you can specify the relevant file.
- `class_name`: in case there are several methods with the same name in the same file, you can specify the relevant class name.

6
docs/docs/tools/index.md
View File

@ -3,18 +3,20 @@
Here is a list of PR-Agent tools, each with a dedicated page that explains how to use it:
| Tool | Description |
|-------------------------------------------------------------------------------|---------------------------------------------------------------------------------------------------------------------------------------------|
|------------------------------------------------------------------------------------------|---------------------------------------------------------------------------------------------------------------------------------------------|
| **[PR Description (`/describe`](./describe.md))** | Automatically generating PR description - title, type, summary, code walkthrough and labels |
| **[PR Review (`/review`](./review.md))** | Adjustable feedback about the PR, possible issues, security concerns, review effort and more |
| **[Code Suggestions (`/improve`](./improve.md))** | Code suggestions for improving the PR |
| **[Question Answering (`/ask ...`](./ask.md))** | Answering free-text questions about the PR, or on specific code lines |
| **[Update Changelog (`/update_changelog`](./update_changelog.md))** | Automatically updating the CHANGELOG.md file with the PR changes |
| **[Find Similar Issue (`/similar_issue`](./similar_issues.md))** | Automatically retrieves and presents similar issues |
| **[Help (`/help`](./help.md))** | Provides a list of all the available tools. Also enables to trigger them interactively (💎) |
| **💎 [Add Documentation (`/add_docs`](./documentation.md))** | Generates documentation to methods/functions/classes that changed in the PR |
| **💎 [Generate Custom Labels (`/generate_labels`](./custom_labels.md))** | Generates custom labels for the PR, based on specific guidelines defined by the user |
| **💎 [Analyze (`/analyze`](./analyze.md))** | Identify code components that changed in the PR, and enables to interactively generate tests, docs, and code suggestions for each component |
| **💎 [Custom Suggestions (`/custom_suggestions`](./custom_suggestions.md))** | Automatically generates custom suggestions for improving the PR code, based on specific guidelines defined by the user |
| **💎 [Custom Prompt (`/custom_prompt`](./custom_prompt.md))** | Automatically generates custom suggestions for improving the PR code, based on specific guidelines defined by the user |
| **💎 [Generate Tests (`/test component_name`](./test.md))** | Automatically generates unit tests for a selected component, based on the PR code changes |
| **💎 [Improve Component (`/improve_component component_name`](./improve_component.md))** | Generates code suggestions for a specific code component that changed in the PR |
| **💎 [CI Feedback (`/checks ci_job`](./ci_feedback.md))** | Automatically generates feedback and analysis for a failed CI job |
Note that the tools marked with 💎 are available only for PR-Agent Pro users.

282
docs/docs/tools/review.md
View File

@ -4,48 +4,43 @@ The tool can be triggered automatically every time a new PR is [opened](../usage
```
/review
```
For example:
<kbd><img src=https://codium.ai/images/pr_agent/review_comment.png width="768"></kbd>
## Example usage
<kbd><img src=https://codium.ai/images/pr_agent/review.png width="768"></kbd>
### Manual triggering
Invoke the tool manually by commenting `/review` on any PR:
## Configuration options
![review comment](https://codium.ai/images/pr_agent/review_comment.png){width=512}
To edit [configurations](https://github.com/Codium-ai/pr-agent/blob/main/pr_agent/settings/configuration.toml#L19) related to the review tool (`pr_reviewer` section), use the following template:
After ~30 seconds, the tool will generate a review for the PR:
![review](https://codium.ai/images/pr_agent/review3.png){width=512}
If you want to edit [configurations](#configuration-options), add the relevant ones to the command:
```
/review --pr_reviewer.some_config1=... --pr_reviewer.some_config2=...
```
#### General options
- `num_code_suggestions`: number of code suggestions provided by the 'review' tool. For manual comments, default is 4. For [PR-Agent app](https://github.com/Codium-ai/pr-agent/blob/main/pr_agent/settings/configuration.toml#L142) auto tools, default is 0, meaning no code suggestions will be provided by the review tool, unless you manually edit `pr_commands`.
- `inline_code_comments`: if set to true, the tool will publish the code suggestions as comments on the code diff. Default is false.
- `persistent_comment`: if set to true, the review comment will be persistent, meaning that every new review request will edit the previous one. Default is true.
- `extra_instructions`: Optional extra instructions to the tool. For example: "focus on the changes in the file X. Ignore change in ...".
### Automatic triggering
#### Enable\\disable features
- `require_focused_review`: if set to true, the tool will add a section - 'is the PR a focused one'. Default is false.
- `require_score_review`: if set to true, the tool will add a section that scores the PR. Default is false.
- `require_tests_review`: if set to true, the tool will add a section that checks if the PR contains tests. Default is true.
- `require_estimate_effort_to_review`: if set to true, the tool will add a section that estimates the effort needed to review the PR. Default is true.
To run the `review` automatically when a PR is opened, define in a [configuration file](https://pr-agent-docs.codium.ai/usage-guide/configuration_options/#wiki-configuration-file):
```
[github_app]
pr_commands = [
"/review",
...
]
#### SOC2 ticket compliance 💎
> This feature is available only in PR-Agent Pro
[pr_reviewer]
num_code_suggestions = ...
...
```
This sub-tool checks if the PR description properly contains a ticket to a project management system (e.g., Jira, Asana, Trello, etc.), as required by SOC2 compliance. If not, it will add a label to the PR: "Missing SOC2 ticket".
- `require_soc2_ticket`: If set to true, the SOC2 ticket checker sub-tool will be enabled. Default is false.
- `soc2_ticket_prompt`: The prompt for the SOC2 ticket review. Default is: `Does the PR description include a link to ticket in a project management system (e.g., Jira, Asana, Trello, etc.) ?`. Edit this field if your compliance requirements are different.
- The `pr_commands` lists commands that will be executed automatically when a PR is opened.
- The `[pr_reviewer]` section contains the configurations for the `review` tool you want to edit (if any).
#### Adding PR labels
- `enable_review_labels_security`: if set to true, the tool will publish a 'possible security issue' label if it detects a security issue. Default is true.
- `enable_review_labels_effort`: if set to true, the tool will publish a 'Review effort [1-5]: x' label. Default is true.
#### Auto-approval
- `enable_auto_approval`: if set to true, the tool will approve the PR when invoked with the 'auto_approve' command. Default is false. This flag can be changed only from configuration file.
- `maximal_review_effort`: maximal effort level for auto-approval. If the PR's estimated review effort is above this threshold, the auto-approval will not run. Default is 5.
#### Incremental Mode
### Incremental Mode
Incremental review only considers changes since the last PR-Agent review. This can be useful when working on the PR in an iterative manner, and you want to focus on the changes since the last review instead of reviewing the entire PR again.
For invoking the incremental mode, the following command can be used:
```
@ -53,23 +48,9 @@ For invoking the incremental mode, the following command can be used:
```
Note that the incremental mode is only available for GitHub.
<kbd><img src=https://codium.ai/images/pr_agent/incremental_review.png width="768"></kbd>
![incremental review](https://codium.ai/images/pr_agent/incremental_review_2.png){width=512}
Under the section 'pr_reviewer', the [configuration file](https://github.com/Codium-ai/pr-agent/blob/main/pr_agent/settings/configuration.toml#L19) contains options to customize the 'review -i' tool.
These configurations can be used to control the rate at which the incremental review tool will create new review comments when invoked automatically, to prevent making too much noise in the PR.
- `minimal_commits_for_incremental_review`: Minimal number of commits since the last review that are required to create incremental review.
If there are less than the specified number of commits since the last review, the tool will not perform any action.
Default is 0 - the tool will always run, no matter how many commits since the last review.
- `minimal_minutes_for_incremental_review`: Minimal number of minutes that need to pass since the last reviewed commit to create incremental review.
If less than the specified number of minutes have passed between the last reviewed commit and running this command, the tool will not perform any action.
Default is 0 - the tool will always run, no matter how much time have passed since the last reviewed commit.
- `require_all_thresholds_for_incremental_review`: If set to true, all the previous thresholds must be met for incremental review to run. If false, only one is enough to run the tool.
For example, if `minimal_commits_for_incremental_review=2` and `minimal_minutes_for_incremental_review=2`, and we have 3 commits since the last review, but the last reviewed commit is from 1 minute ago:
When `require_all_thresholds_for_incremental_review=true` the incremental review __will not__ run, because only 1 out of 2 conditions were met (we have enough commits but the last review is too recent),
but when `require_all_thresholds_for_incremental_review=false` the incremental review __will__ run, because one condition is enough (we have 3 commits which is more than the configured 2).
Default is false - the tool will run as long as at least once conditions is met.
#### PR Reflection
### PR Reflection
By invoking:
```
@ -77,92 +58,185 @@ By invoking:
```
The tool will first ask the author questions about the PR, and will guide the review based on their answers.
<kbd><img src=https://codium.ai/images/pr_agent/reflection_questions.png width="768"></kbd>
![reflection questions](https://codium.ai/images/pr_agent/reflection_questions.png){width=512}
<kbd><img src=https://codium.ai/images/pr_agent/reflection_answers.png width="768"></kbd>
![reflection answers](https://codium.ai/images/pr_agent/reflection_answers.png){width=512}
<kbd><img src=https://codium.ai/images/pr_agent/reflection_insights.png width="768"></kbd>
![reflection insights](https://codium.ai/images/pr_agent/reflection_insights.png){width=512}
## Configuration options
!!! example "General options"
<table>
<tr>
<td><b>num_code_suggestions</b></td>
<td>Number of code suggestions provided by the 'review' tool. For manual comments, default is 4. For PR-Agent app auto tools, default is 0, meaning no code suggestions will be provided by the review tool, unless you manually edit pr_commands.</td>
</tr>
<tr>
<td><b>inline_code_comments</b></td>
<td>If set to true, the tool will publish the code suggestions as comments on the code diff. Default is false.</td>
</tr>
<tr>
<td><b>persistent_comment</b></td>
<td>If set to true, the review comment will be persistent, meaning that every new review request will edit the previous one. Default is true.</td>
</tr>
<tr>
<td><b>extra_instructions</b></td>
<td>Optional extra instructions to the tool. For example: "focus on the changes in the file X. Ignore change in ...".</td>
</tr>
<tr>
<td><b>enable_help_text</b></td>
<td>If set to true, the tool will display a help text in the comment. Default is true.</td>
</tr>
</table>
!!! example "Enable\\disable specific sub-sections"
<table>
<tr>
<td><b>require_score_review</b></td>
<td>If set to true, the tool will add a section that scores the PR. Default is false.</td>
</tr>
<tr>
<td><b>require_tests_review</b></td>
<td>If set to true, the tool will add a section that checks if the PR contains tests. Default is true.</td>
</tr>
<tr>
<td><b>require_estimate_effort_to_review</b></td>
<td>If set to true, the tool will add a section that estimates the effort needed to review the PR. Default is true.</td>
</tr>
<tr>
<td><b>require_can_be_split_review</b></td>
<td>If set to true, the tool will add a section that checks if the PR contains several themes, and can be split into smaller PRs. Default is false.</td>
</tr>
</table>
!!! example "SOC2 ticket compliance 💎"
This sub-tool checks if the PR description properly contains a ticket to a project management system (e.g., Jira, Asana, Trello, etc.), as required by SOC2 compliance. If not, it will add a label to the PR: "Missing SOC2 ticket".
<table>
<tr>
<td><b>require_soc2_ticket</b></td>
<td>If set to true, the SOC2 ticket checker sub-tool will be enabled. Default is false.</td>
</tr>
<tr>
<td><b>soc2_ticket_prompt</b></td>
<td>The prompt for the SOC2 ticket review. Default is: `Does the PR description include a link to ticket in a project management system (e.g., Jira, Asana, Trello, etc.) ?`. Edit this field if your compliance requirements are different.</td>
</tr>
</table>
!!! example "Adding PR labels"
You can enable\disable the `review` tool to add specific labels to the PR:
<table>
<tr>
<td><b>enable_review_labels_security</b></td>
<td>If set to true, the tool will publish a 'possible security issue' label if it detects a security issue. Default is true.</td>
</tr>
<tr>
<td><b>enable_review_labels_effort</b></td>
<td>If set to true, the tool will publish a 'Review effort [1-5]: x' label. Default is true.</td>
</tr>
</table>
!!! example "Auto-approval"
If enabled, the `review` tool can approve a PR when a specific comment, `/review auto_approve`, is invoked.
<table>
<tr>
<td><b>enable_auto_approval</b></td>
<td>If set to true, the tool will approve the PR when invoked with the 'auto_approve' command. Default is false. This flag can be changed only from configuration file.</td>
</tr>
<tr>
<td><b>maximal_review_effort</b></td>
<td>Maximal effort level for auto-approval. If the PR's estimated review effort is above this threshold, the auto-approval will not run. Default is 5.</td>
</tr>
</table>
## Usage Tips
### General guidelines
!!! tip "General guidelines"
The `review` tool provides a collection of possible feedbacks about a PR.
It is recommended to review the [Configuration options](#configuration-options) section, and choose the relevant options for your use case.
The `review` tool provides a collection of possible feedbacks about a PR.
It is recommended to review the [Configuration options](#configuration-options) section, and choose the relevant options for your use case.
Some of the features that are disabled by default are quite useful, and should be considered for enabling. For example:
`require_score_review`, `require_soc2_ticket`, and more.
Some of the features that are disabled by default are quite useful, and should be considered for enabling. For example:
`require_score_review`, `require_soc2_ticket`, and more.
On the other hand, if you find one of the enabled features to be irrelevant for your use case, disable it. No default configuration can fit all use cases.
On the other hand, if you find one of the enabled features to be irrelevant for your use case, disable it. No default configuration can fit all use cases.
### Code suggestions
!!! tip "Automation"
When you first install PR-Agent app, the [default mode](../usage-guide/automations_and_usage.md#github-app-automatic-tools-when-a-new-pr-is-opened) for the `review` tool is:
```
pr_commands = ["/review --pr_reviewer.num_code_suggestions=0", ...]
```
Meaning the `review` tool will run automatically on every PR, without providing code suggestions.
Edit this field to enable/disable the tool, or to change the used configurations.
If you set `num_code_suggestions`>0 , the `review` tool will also provide code suggestions.
!!! tip "Code suggestions"
Notice If you are interested **only** in the code suggestions, it is recommended to use the [`improve`](./improve.md) feature instead, since it is a dedicated only to code suggestions, and usually gives better results.
Use the `review` tool if you want to get more comprehensive feedback, which includes code suggestions as well.
If you set `num_code_suggestions`>0 , the `review` tool will also provide code suggestions.
### Automation
- When you first install the app, the [default mode](../usage-guide/automations_and_usage.md#github-app-automatic-tools-when-a-new-pr-is-opened) for the `review` tool is:
```
pr_commands = ["/review", ...]
```
Meaning the `review` tool will run automatically on every PR, with the default configuration.
Edit this field to enable/disable the tool, or to change the used configurations.
Notice If you are interested **only** in the code suggestions, it is recommended to use the [`improve`](./improve.md) feature instead, since it is a dedicated only to code suggestions, and usually gives better results.
Use the `review` tool if you want to get more comprehensive feedback, which includes code suggestions as well.
### Auto-labels
!!! tip "Possible labels from the review tool"
The `review` tool can auto-generate two specific types of labels for a PR:
The `review` tool can auto-generate two specific types of labels for a PR:
- a `possible security issue` label that detects a possible [security issue](https://github.com/Codium-ai/pr-agent/blob/tr/user_description/pr_agent/settings/pr_reviewer_prompts.toml#L136) (`enable_review_labels_security` flag)
- a `Review effort [1-5]: x` label, where x is the estimated effort to review the PR (`enable_review_labels_effort` flag)
- a `possible security issue` label that detects if a possible [security issue](https://github.com/Codium-ai/pr-agent/blob/tr/user_description/pr_agent/settings/pr_reviewer_prompts.toml#L136) exists in the PR code (`enable_review_labels_security` flag)
- a `Review effort [1-5]: x` label, where x is the estimated effort to review the PR (`enable_review_labels_effort` flag)
Both modes are useful, and we recommended to enable them.
Both modes are useful, and we recommended to enable them.
### Extra instructions
!!! tip "Extra instructions"
Extra instructions are important.
The `review` tool can be configured with extra instructions, which can be used to guide the model to a feedback tailored to the needs of your project.
Extra instructions are important.
The `review` tool can be configured with extra instructions, which can be used to guide the model to a feedback tailored to the needs of your project.
Be specific, clear, and concise in the instructions. With extra instructions, you are the prompter. Specify the relevant sub-tool, and the relevant aspects of the PR that you want to emphasize.
Be specific, clear, and concise in the instructions. With extra instructions, you are the prompter. Specify the relevant sub-tool, and the relevant aspects of the PR that you want to emphasize.
Examples for extra instructions:
```
[pr_reviewer] # /review #
extra_instructions="""
In the code feedback section, emphasize the following:
- Does the code logic cover relevant edge cases?
- Is the code logic clear and easy to understand?
- Is the code logic efficient?
...
"""
```
Use triple quotes to write multi-line instructions. Use bullet points to make the instructions more readable.
Examples for extra instructions:
```
[pr_reviewer]
extra_instructions="""\
In the code feedback section, emphasize the following:
- Does the code logic cover relevant edge cases?
- Is the code logic clear and easy to understand?
- Is the code logic efficient?
...
"""
```
Use triple quotes to write multi-line instructions. Use bullet points to make the instructions more readable.
### Auto-approval
!!! tip "Auto-approval"
PR-Agent can approve a PR when a specific comment is invoked.
PR-Agent can approve a PR when a specific comment is invoked.
To ensure safety, the auto-approval feature is disabled by default. To enable auto-approval, you need to actively set in a pre-defined configuration file the following:
```
[pr_reviewer]
enable_auto_approval = true
```
(this specific flag cannot be set with a command line argument, only in the configuration file, committed to the repository)
To ensure safety, the auto-approval feature is disabled by default. To enable auto-approval, you need to actively set in a pre-defined configuration file the following:
```
[pr_reviewer]
enable_auto_approval = true
```
(this specific flag cannot be set with a command line argument, only in the configuration file, committed to the repository)
After enabling, by commenting on a PR:
```
/review auto_approve
```
PR-Agent will automatically approve the PR, and add a comment with the approval.
After enabling, by commenting on a PR:
```
/review auto_approve
```
PR-Agent will automatically approve the PR, and add a comment with the approval.
You can also enable auto-approval only if the PR meets certain requirements, such as that the `estimated_review_effort` label is equal or below a certain threshold, by adjusting the flag:
```
[pr_reviewer]
maximal_review_effort = 5
```
You can also enable auto-approval only if the PR meets certain requirements, such as that the `estimated_review_effort` label is equal or below a certain threshold, by adjusting the flag:
```
[pr_reviewer]
maximal_review_effort = 5
```

12
docs/docs/tools/similar_code.md
View File

@ -5,7 +5,8 @@ For example:
`Global Search` for a method called `chat_completion`:
<kbd><img src=https://codium.ai/images/pr_agent/similar_code_global2.png width="768"></kbd>
![similar code global](https://codium.ai/images/pr_agent/similar_code_global2.png){width=768}
PR-Agent will examine the code component and will extract the most relevant keywords to search for similar code:
@ -16,11 +17,12 @@ PR-Agent will examine the code component and will extract the most relevant keyw
Search result link example:
<kbd><img src=https://codium.ai/images/pr_agent/code_search_result_single.png width="768"></kbd>
![code search result single](https://codium.ai/images/pr_agent/code_search_result_single.png){width=768}
`Organization Search`:
<kbd><img src=https://codium.ai/images/pr_agent/similar_code_org.png width="768"></kbd>
![similar code org](https://codium.ai/images/pr_agent/similar_code_org.png){width=768}
## How to use
@ -47,11 +49,11 @@ It can be invoked automatically from the analyze table, can be accessed by:
/analyze
```
Choose the components you want to find similar code for, and click on the `similar` checkbox.
<kbd><img src=https://codium.ai/images/pr_agent/analyze_similar.png width="768"></kbd>
![analyze similar](https://codium.ai/images/pr_agent/analyze_similar.png){width=768}
If you are looking to search for similar code in the organization's codebase, you can click on the `Organization` checkbox, and it will invoke a new search command just for the organization's codebase.
<kbd><img src=https://codium.ai/images/pr_agent/similar_code_global.png width="768"></kbd>
![similar code global](https://codium.ai/images/pr_agent/similar_code_global.png){width=768}
## Configuration options

10
docs/docs/tools/similar_issues.md
View File

@ -4,13 +4,15 @@ It can be invoked manually by commenting on any PR:
```
/similar_issue
```
For example:
<kbd><img src=https://codium.ai/images/pr_agent/similar_issue_original_issue.png width="768"></kbd>
<kbd><img src=https://codium.ai/images/pr_agent/similar_issue_comment.png width="768"></kbd>
## Example usage
<kbd><img src=https://codium.ai/images/pr_agent/similar_issue.png width="768"></kbd>
![similar_issue_original_issue](https://codium.ai/images/pr_agent/similar_issue_original_issue.png){width=768}
![similar_issue_comment](https://codium.ai/images/pr_agent/similar_issue_comment.png){width=768}
![similar_issue](https://codium.ai/images/pr_agent/similar_issue.png){width=768}
Note that to perform retrieval, the `similar_issue` tool indexes all the repo previous issues (once).

18
docs/docs/tools/test.md
View File

@ -5,19 +5,25 @@ It can be invoked manually by commenting on any PR:
/test component_name
```
where 'component_name' is the name of a specific component in the PR.
To get a list of the components that changed in the PR, use the [`analyze`](./analyze.md) tool.
To get a list of the components that changed in the PR and choose the relevant component interactively, use the [`analyze`](./analyze.md) tool.
## Example usage
An example [result](https://github.com/Codium-ai/pr-agent/pull/598#issuecomment-1913679429):
Invoke the tool manually by commenting `/test` on any PR:
<kbd><img src=https://codium.ai/images/pr_agent/test1.png width="704"></kbd>
![test1](https://codium.ai/images/pr_agent/test1.png){width=704}
<kbd><img src=https://codium.ai/images/pr_agent/test2.png width="768"></kbd>
The tool will generate tests for the selected component (if no component is stated, it will generate tests for largest component):
<kbd><img src=https://codium.ai/images/pr_agent/test3.png width="768"></kbd>
![test2](https://codium.ai/images/pr_agent/test2.png){width=768}
![test3](https://codium.ai/images/pr_agent/test3.png){width=768}
(Example taken from [here](https://github.com/Codium-ai/pr-agent/pull/598#issuecomment-1913679429)):
**Notes**
- Language that are currently supported by the tool: Python, Java, C++, JavaScript, TypeScript.
- Language that are currently supported by the tool: Python, Java, C++, JavaScript, TypeScript, C#.
- This tool can also be triggered interactively by using the [`analyze`](./analyze.md) tool.
## Configuration options

8
docs/docs/tools/update_changelog.md
View File

@ -4,16 +4,16 @@ It can be invoked manually by commenting on any PR:
```
/update_changelog
```
For example:
<kbd><img src=https://codium.ai/images/pr_agent/update_changelog_comment.png width="768"></kbd>
## Example usage
<kbd><img src=https://codium.ai/images/pr_agent/update_changelog.png width="768"></kbd>
![update_changelog_comment](https://codium.ai/images/pr_agent/update_changelog_comment.png){width=768}
![update_changelog](https://codium.ai/images/pr_agent/update_changelog.png){width=768}
## Configuration options
Under the section 'pr_update_changelog', the [configuration file](https://github.com/Codium-ai/pr-agent/blob/main/pr_agent/settings/configuration.toml#L50) contains options to customize the 'update changelog' tool:
Under the section `pr_update_changelog`, the [configuration file](https://github.com/Codium-ai/pr-agent/blob/main/pr_agent/settings/configuration.toml#L50) contains options to customize the 'update changelog' tool:
- `push_changelog_changes`: whether to push the changes to CHANGELOG.md, or just print them. Default is false (print only).
- `extra_instructions`: Optional extra instructions to the tool. For example: "focus on the changes in the file X. Ignore change in ...

56
docs/docs/usage-guide/additional_configurations.md
View File

@ -7,10 +7,10 @@ To ignore files or directories, edit the **[ignore.toml](https://github.com/Codi
- `IGNORE.GLOB`
- `IGNORE.REGEX`
For example, to ignore python files in a PR with online usage, comment on a PR:
For example, to ignore Python files in a PR with online usage, comment on a PR:
`/review --ignore.glob=['*.py']`
To ignore python files in all PRs, set in a configuration file:
To ignore Python files in all PRs, set in a configuration file:
```
[ignore]
glob = ['*.py']
@ -26,13 +26,13 @@ All PR-Agent tools have a parameter called `extra_instructions`, that enables to
## Working with large PRs
The default mode of CodiumAI is to have a single call per tool, using GPT-4, which has a token limit of 8000 tokens.
This mode provide a very good speed-quality-cost tradeoff, and can handle most PRs successfully.
This mode provides a very good speed-quality-cost tradeoff, and can handle most PRs successfully.
When the PR is above the token limit, it employs a [PR Compression strategy](../core-abilities/index.md).
However, for very large PRs, or in case you want to emphasize quality over speed and cost, there are 2 possible solutions:
However, for very large PRs, or in case you want to emphasize quality over speed and cost, there are two possible solutions:
1) [Use a model](https://codium-ai.github.io/Docs-PR-Agent/usage-guide/#changing-a-model) with larger context, like GPT-32K, or claude-100K. This solution will be applicable for all the tools.
2) For the `/improve` tool, there is an ['extended' mode](https://codium-ai.github.io/Docs-PR-Agent/tools/#improve) (`/improve --extended`),
which divides the PR to chunks, and process each chunk separately. With this mode, regardless of the model, no compression will be done (but for large PRs, multiple model calls may occur)
which divides the PR to chunks, and processes each chunk separately. With this mode, regardless of the model, no compression will be done (but for large PRs, multiple model calls may occur)
## Changing a model
@ -59,12 +59,12 @@ and set in your configuration file:
model="" # the OpenAI model you've deployed on Azure (e.g. gpt-3.5-turbo)
```
### Huggingface
### Hugging Face
**Local**
You can run Huggingface models locally through either [VLLM](https://docs.litellm.ai/docs/providers/vllm) or [Ollama](https://docs.litellm.ai/docs/providers/ollama)
You can run Hugging Face models locally through either [VLLM](https://docs.litellm.ai/docs/providers/vllm) or [Ollama](https://docs.litellm.ai/docs/providers/ollama)
E.g. to use a new Huggingface model locally via Ollama, set:
E.g. to use a new Hugging Face model locally via Ollama, set:
```
[__init__.py]
MAX_TOKENS = {
@ -79,16 +79,17 @@ MAX_TOKENS={
[config] # in configuration.toml
model = "ollama/llama2"
model_turbo = "ollama/llama2"
[ollama] # in .secrets.toml
api_base = ... # the base url for your huggingface inference endpoint
api_base = ... # the base url for your Hugging Face inference endpoint
# e.g. if running Ollama locally, you may use:
api_base = "http://localhost:11434/"
```
### Inference Endpoints
To use a new model with Huggingface Inference Endpoints, for example, set:
To use a new model with Hugging Face Inference Endpoints, for example, set:
```
[__init__.py]
MAX_TOKENS = {
@ -101,10 +102,11 @@ MAX_TOKENS={
}
[config] # in configuration.toml
model = "huggingface/meta-llama/Llama-2-7b-chat-hf"
model_turbo = "huggingface/meta-llama/Llama-2-7b-chat-hf"
[huggingface] # in .secrets.toml
key = ... # your huggingface api key
api_base = ... # the base url for your huggingface inference endpoint
key = ... # your Hugging Face api key
api_base = ... # the base url for your Hugging Face inference endpoint
```
(you can obtain a Llama2 key from [here](https://replicate.com/replicate/llama-2-70b-chat/api))
@ -114,13 +116,27 @@ To use Llama2 model with Replicate, for example, set:
```
[config] # in configuration.toml
model = "replicate/llama-2-70b-chat:2c1608e18606fad2812020dc541930f2d0495ce32eee50074220b87300bc16e1"
model_turbo = "replicate/llama-2-70b-chat:2c1608e18606fad2812020dc541930f2d0495ce32eee50074220b87300bc16e1"
[replicate] # in .secrets.toml
key = ...
```
(you can obtain a Llama2 key from [here](https://replicate.com/replicate/llama-2-70b-chat/api))
Also review the [AiHandler](https://github.com/Codium-ai/pr-agent/blob/main/pr_agent/algo/ai_handler.py) file for instruction how to set keys for other models.
Also, review the [AiHandler](https://github.com/Codium-ai/pr-agent/blob/main/pr_agent/algo/ai_handler.py) file for instructions on how to set keys for other models.
### Groq
To use Llama3 model with Groq, for example, set:
```
[config] # in configuration.toml
model = "llama3-70b-8192"
model_turbo = "llama3-70b-8192"
fallback_models = ["groq/llama3-70b-8192"]
[groq] # in .secrets.toml
key = ... # your Groq api key
```
(you can obtain a Groq key from [here](https://console.groq.com/keys))
### Vertex AI
@ -129,6 +145,7 @@ To use Google's Vertex AI platform and its associated models (chat-bison/codecha
```
[config] # in configuration.toml
model = "vertex_ai/codechat-bison"
model_turbo = "vertex_ai/codechat-bison"
fallback_models="vertex_ai/codechat-bison"
[vertexai] # in .secrets.toml
@ -162,8 +179,9 @@ To use Amazon Bedrock and its foundational models, add the below configuration:
```
[config] # in configuration.toml
model = "anthropic.claude-v2"
fallback_models="anthropic.claude-instant-v1"
model="bedrock/anthropic.claude-3-sonnet-20240229-v1:0"
model_turbo="bedrock/anthropic.claude-3-sonnet-20240229-v1:0"
fallback_models=["bedrock/anthropic.claude-v2:1"]
[aws] # in .secrets.toml
bedrock_region = "us-east-1"
@ -171,12 +189,18 @@ bedrock_region = "us-east-1"
Note that you have to add access to foundational models before using them. Please refer to [this document](https://docs.aws.amazon.com/bedrock/latest/userguide/setting-up.html) for more details.
If you are using the claude-3 model, please configure the following settings as there are parameters incompatible with claude-3.
```
[litellm]
drop_params = true
```
AWS session is automatically authenticated from your environment, but you can also explicitly set `AWS_ACCESS_KEY_ID` and `AWS_SECRET_ACCESS_KEY` environment variables.
## Patch Extra Lines
By default, around any change in your PR, git patch provides 3 lines of context above and below the change.
By default, around any change in your PR, git patch provides three lines of context above and below the change.
```
@@ -12,5 +12,5 @@ def func1():
code line that already existed in the file...

55
docs/docs/usage-guide/automations_and_usage.md
View File

@ -1,5 +1,5 @@
## Local repo (CLI)
When running from your local repo (CLI), your local configuration file will be used.
When running from your locally cloned PR-Agent repo (CLI), your local configuration file will be used.
Examples of invoking the different tools via the CLI:
- **Review**: `python -m pr_agent.cli --pr_url=<pr_url> review`
@ -50,6 +50,10 @@ Any configuration value in [configuration file](https://github.com/Codium-ai/pr-
## GitHub App
!!! note "Configurations for PR-Agent Pro"
PR-Agent Pro for GitHub is an App, hosted by CodiumAI. So all the instructions below are relevant also for PR-Agent Pro users.
Same goes for [GitLab webhook](#gitlab-webhook) and [BitBucket App](#bitbucket-app) sections.
### GitHub app automatic tools when a new PR is opened
The [github_app](https://github.com/Codium-ai/pr-agent/blob/main/pr_agent/settings/configuration.toml#L108) section defines GitHub app specific configurations.
@ -58,22 +62,21 @@ The configuration parameter `pr_commands` defines the list of tools that will be
```
[github_app]
pr_commands = [
"/describe --pr_description.add_original_user_description=true --pr_description.keep_original_user_title=true --pr_description.final_update_message=false",
"/review --pr_reviewer.num_code_suggestions=0 --pr_reviewer.final_update_message=false",
"/describe --pr_description.final_update_message=false",
"/review --pr_reviewer.num_code_suggestions=0",
"/improve",
]
```
This means that when a new PR is opened/reopened or marked as ready for review, PR-Agent will run the `describe`, `review` and `improve` tools.
For the `describe` tool, for example, the `add_original_user_description` and `keep_original_user_title` parameters will be set to true.
For the `review` tool, for example, the `num_code_suggestions` parameter will be set to 0.
You can override the default tool parameters by using one the three options for a [configuration file](https://codium-ai.github.io/Docs-PR-Agent/usage-guide/#configuration-options): **wiki**, **local**, or **global**.
For example, if your local `.pr_agent.toml` file contains:
```
[pr_description]
add_original_user_description = false
keep_original_user_title = false
generate_ai_title = true
```
When a new PR is opened, PR-Agent will run the `describe` tool with the above parameters.
Every time you run the `describe` tool, including automatic runs, the PR title will be generated by the AI.
To cancel the automatic run of all the tools, set:
```
@ -98,14 +101,14 @@ The configuration parameter `push_commands` defines the list of tools that will
[github_app]
handle_push_trigger = true
push_commands = [
"/describe --pr_description.add_original_user_description=true --pr_description.keep_original_user_title=true",
"/review --pr_reviewer.num_code_suggestions=0",
"/describe",
"/review --pr_reviewer.num_code_suggestions=0 --pr_reviewer.final_update_message=false",
]
```
This means that when new code is pushed to the PR, the PR-Agent will run the `describe` and `review` tools, with the specified parameters.
## GitHub Action
`GitHub Action` is a different way to trigger PR-Agent tools, and uses a different configuration mechanism than `GitHub App`.
`GitHub Action` is a different way to trigger PR-Agent tools, and uses a different configuration mechanism than `GitHub App`.<br>
You can configure settings for `GitHub Action` by adding environment variables under the env section in `.github/workflows/pr_agent.yml` file.
Specifically, start by setting the following environment variables:
```yaml
@ -119,20 +122,25 @@ Specifically, start by setting the following environment variables:
`github_action_config.auto_review`, `github_action_config.auto_describe` and `github_action_config.auto_improve` are used to enable/disable automatic tools that run when a new PR is opened.
If not set, the default configuration is for all three tools to run automatically when a new PR is opened.
Note that you can give additional config parameters by adding environment variables to `.github/workflows/pr_agent.yml`, or by using a `.pr_agent.toml` file in the root of your repo, similar to the GitHub App usage.
`github_action_config.enable_output` are used to enable/disable github actions [output parameter](https://docs.github.com/en/actions/creating-actions/metadata-syntax-for-github-actions#outputs-for-docker-container-and-javascript-actions) (default is `true`).
Review result is output as JSON to `steps.{step-id}.outputs.review` property.
The JSON structure is equivalent to the yaml data structure defined in [pr_reviewer_prompts.toml](https://github.com/idubnori/pr-agent/blob/main/pr_agent/settings/pr_reviewer_prompts.toml).
For example, you can set an environment variable: `pr_description.add_original_user_description=false`, or add a `.pr_agent.toml` file with the following content:
Note that you can give additional config parameters by adding environment variables to `.github/workflows/pr_agent.yml`, or by using a `.pr_agent.toml` [configuration file](https://pr-agent-docs.codium.ai/usage-guide/configuration_options/#global-configuration-file) in the root of your repo
For example, you can set an environment variable: `pr_description.publish_labels=false`, or add a `.pr_agent.toml` file with the following content:
```
[pr_description]
add_original_user_description = false
publish_labels = false
```
to prevent PR-Agent from publishing labels when running the `describe` tool.
## GitLab Webhook
After setting up a GitLab webhook, to control which commands will run automatically when a new PR is opened, you can set the `pr_commands` parameter in the configuration file, similar to the GitHub App:
```
[gitlab]
pr_commands = [
"/describe --pr_description.add_original_user_description=true --pr_description.keep_original_user_title=true",
"/describe",
"/review --pr_reviewer.num_code_suggestions=0",
"/improve",
]
@ -153,14 +161,14 @@ Each time you invoke a `/review` tool, it will use inline code comments.
### BitBucket Self-Hosted App automatic tools
to control which commands will run automatically when a new PR is opened, you can set the `pr_commands` parameter in the configuration file:
To control which commands will run automatically when a new PR is opened, you can set the `pr_commands` parameter in the configuration file:
Specifically, set the following values:
[bitbucket_app]
```
[bitbucket_app]
pr_commands = [
"/review --pr_reviewer.num_code_suggestions=0",
"/improve --pr_code_suggestions.summarize=false",
"/improve --pr_code_suggestions.commitable_code_suggestions=true",
]
```
@ -170,16 +178,15 @@ To use Azure DevOps provider use the following settings in configuration.toml:
```
[config]
git_provider="azure"
use_repo_settings_file=false
```
Azure DevOps provider supports [PAT token](https://learn.microsoft.com/en-us/azure/devops/organizations/accounts/use-personal-access-tokens-to-authenticate?view=azure-devops&tabs=Windows) or [DefaultAzureCredential](https://learn.microsoft.com/en-us/azure/developer/python/sdk/authentication-overview#authentication-in-server-environments) authentication.
PAT is faster to create, but has build in experation date, and will use the user identity for API calls.
Using DefaultAzureCredential you can use managed identity or Service principle, which are more secure and will create seperate ADO user identity (via AAD) to the agent.
PAT is faster to create, but has build in expiration date, and will use the user identity for API calls.
Using DefaultAzureCredential you can use managed identity or Service principle, which are more secure and will create separate ADO user identity (via AAD) to the agent.
If PAT was choosen, you can assign the value in .secrets.toml.
If DefaultAzureCredential was choosen, you can assigned the additional env vars like AZURE_CLIENT_SECRET directly,
or use managed identity/az cli (for local develpment) without any additional configuration.
If PAT was chosen, you can assign the value in .secrets.toml.
If DefaultAzureCredential was chosen, you can assigned the additional env vars like AZURE_CLIENT_SECRET directly,
or use managed identity/az cli (for local development) without any additional configuration.
in any case, 'org' value must be assigned in .secrets.toml:
```
[azure_devops]
@ -193,7 +200,7 @@ To control which commands will run automatically when a new PR is opened, you ca
```
[azure_devops_server]
pr_commands = [
"/describe --pr_description.add_original_user_description=true --pr_description.keep_original_user_title=true",
"/describe",
"/review --pr_reviewer.num_code_suggestions=0",
"/improve",
]

11
docs/docs/usage-guide/configuration_options.md
View File

@ -11,19 +11,24 @@ There are three ways to set persistent configurations:
In terms of precedence, wiki configurations will override local configurations, and local configurations will override global configurations.
!!! tip "Tip1: edit only what you need"
Your configuration file should be minimal, and edit only the relevant values. Don't copy the entire configuration options, since it can lead to legacy problems when something changes.
!!! tip "Tip2: show relevant configurations"
If you set `config.output_relevant_configurations=true`, each tool will also output in a collapsible section its relevant configurations. This can be useful for debugging, or getting to know the configurations better.
## Wiki configuration file 💎
Specifically for GitHub, with PR-Agent-Pro you can set configurations by creating a page called `.pr_agent.toml` in the [wiki](https://github.com/Codium-ai/pr-agent/wiki/pr_agent.toml) of the repo.
The advantage of this method is that it allows to set configurations without needing to commit new content to the repo - just edit the wiki page and **save**.
<kbd><img src="https://codium.ai/images/pr_agent/wiki_configuration.png" width="512"></kbd>
![wiki_configuration](https://codium.ai/images/pr_agent/wiki_configuration.png){width=512}
Click [here](https://codium.ai/images/pr_agent/wiki_configuration_pr_agent.mp4) to see a short instructional video. We recommend surrounding the configuration content with triple-quotes, to allow better presentation when displayed in the wiki as markdown.
An example content:
```
[pr_description] # /describe #
keep_original_user_title=false
[pr_description]
generate_ai_title=true
```
PR-Agent will know to remove the triple-quotes when reading the configuration content.

5
docs/docs/usage-guide/index.md
View File

@ -1,8 +1,10 @@
# Usage guide
This page provides a detailed guide on how to use PR-Agent. It includes information on how to adjust PR-Agent configurations, define which tools will run automatically, manage mail notifications, and other advanced configurations.
- [Introduction](./introduction.md)
- [Configuration Options](./configuration_options.md)
- [Managing Mail Notifications](./mail_notifications.md)
- [Usage and Automation](./automations_and_usage.md)
- [Local Repo (CLI)](./automations_and_usage.md#local-repo-cli)
- [Online Usage](./automations_and_usage.md#online-usage)
@ -11,6 +13,7 @@
- [GitLab Webhook](./automations_and_usage.md#gitlab-webhook)
- [BitBucket App](./automations_and_usage.md#bitbucket-app)
- [Azure DevOps Provider](./automations_and_usage.md#azure-devops-provider)
- [Managing Mail Notifications](./mail_notifications.md)
- [Additional Configurations Walkthrough](./additional_configurations.md)
- [Ignoring files from analysis](./additional_configurations.md#ignoring-files-from-analysis)
- [Extra instructions](./additional_configurations.md#extra-instructions)

13
docs/docs/usage-guide/mail_notifications.md
View File

@ -2,8 +2,17 @@
Unfortunately, it is not possible in GitHub to disable mail notifications from a specific user.
If you are subscribed to notifications for a repo with PR-Agent, we recommend turning off notifications for PR comments, to avoid lengthy emails:
<kbd><img src="https://codium.ai/images/pr_agent/notifications.png" width="512"></kbd>
![notifications](https://codium.ai/images/pr_agent/notifications.png){width=512}
As an alternative, you can filter in your mail provider the notifications specifically from the PR-Agent bot, [see how](https://www.quora.com/How-can-you-filter-emails-for-specific-people-in-Gmail#:~:text=On%20the%20Filters%20and%20Blocked,the%20body%20of%20the%20email).
<kbd><img src="https://codium.ai/images/pr_agent/filter_mail_notifications.png" width="512"></kbd>
![filter_mail_notifications](https://codium.ai/images/pr_agent/filter_mail_notifications.png){width=512}
Another option to reduce the mail overload, yet still receive notifications on PR-Agent tools, is to disable the help collapsible section in PR-Agent bot comments.
This can done by setting `enable_help_text=false` for the relevant tool in the configuration file.
For example, to disable the help text for the `pr_reviewer` tool, set:
```
[pr_reviewer]
enable_help_text = false
```

25
docs/mkdocs.yml
View File

@ -1,4 +1,6 @@
site_name: PR-Agent
site_name: PR-Agent Documentation
repo_url: https://github.com/Codium-ai/pr-agent
repo_name: Codium-ai/pr-agent
nav:
- Overview: 'index.md'
@ -7,14 +9,14 @@ nav:
- Locally: 'installation/locally.md'
- GitHub: 'installation/github.md'
- GitLab: 'installation/gitlab.md'
- 💎 GitLab Pro: 'installation/gitlab_pro.md'
- BitBucket: 'installation/bitbucket.md'
- Azure DevOps: 'installation/azure.md'
- 💎 PR-Agent Pro: 'installation/pr_agent_pro.md'
- Usage Guide:
- 'usage-guide/index.md'
- Introduction: 'usage-guide/introduction.md'
- Configuration Options: 'usage-guide/configuration_options.md'
- Managing email notifications: 'usage-guide/mail_notifications.md'
- Managing Mail Notifications: 'usage-guide/mail_notifications.md'
- Usage and Automation: 'usage-guide/automations_and_usage.md'
- Additional Configurations: 'usage-guide/additional_configurations.md'
- Tools:
@ -25,19 +27,25 @@ nav:
- Ask: 'tools/ask.md'
- Update Changelog: 'tools/update_changelog.md'
- Similar Issues: 'tools/similar_issues.md'
- Help: 'tools/help.md'
- 💎 Analyze: 'tools/analyze.md'
- 💎 Test: 'tools/test.md'
- 💎 Improve Component: 'tools/improve_component.md'
- 💎 Documentation: 'tools/documentation.md'
- 💎 Custom Labels: 'tools/custom_labels.md'
- 💎 Custom Suggestions: 'tools/custom_suggestions.md'
- 💎 Custom Prompt: 'tools/custom_prompt.md'
- 💎 CI Feedback: 'tools/ci_feedback.md'
- 💎 Similar Code: 'tools/similar_code.md'
- Core Abilities: 'core-abilities/index.md'
- Chrome Extension: 'chrome-extension/index.md'
- Code Fine-tuning Benchmark: 'finetuning_benchmark/index.md'
theme:
logo: assets/logo.png
logo: assets/logo.svg
favicon: assets/favicon.ico
name: material
icon:
repo: fontawesome/brands/github
features:
- navigation.tabs
- navigation.expand
@ -77,6 +85,7 @@ theme:
plugins:
- social
- search
- glightbox
extra:
generator: false
@ -93,6 +102,9 @@ extra:
link: https://twitter.com/CodiumAI
- icon: fontawesome/brands/instagram
link: https://www.instagram.com/codiumai/
analytics:
provider: custom
property: ${{ secrets.GOOGLE_ANALYTICS_ID }}
extra_css:
- css/custom.css
@ -109,10 +121,11 @@ markdown_extensions:
- pymdownx.details
- pymdownx.superfences
- pymdownx.mark
- md_in_html
- attr_list
- pymdownx.emoji:
emoji_index: !!python/name:material.extensions.emoji.twemoji
emoji_generator: !!python/name:materialx.emoji.to_svg
emoji_generator: !!python/name:material.extensions.emoji.to_svg
- toc:
title: On this page
toc_depth: 3

10
docs/overrides/main.html Normal file
View File

@ -0,0 +1,10 @@
{% extends "base.html" %}
{% block scripts %}
{{ super() }}
<!-- Google Tag Manager (noscript) -->
<noscript><iframe src="https://www.googletagmanager.com/ns.html?id=GTM-5C9KZBM3"
height="0" width="0" style="display:none;visibility:hidden"></iframe></noscript>
<!-- End Google Tag Manager (noscript) -->
{% endblock %}

7
docs/overrides/partials/integrations/analytics/custom.html Normal file
View File

@ -0,0 +1,7 @@
<!-- Google Tag Manager -->
<script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':
new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],
j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src=
'https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);
})(window,document,'script','dataLayer','GTM-5C9KZBM3');</script>
<!-- End Google Tag Manager -->

3
pr_agent/agent/pr_agent.py
View File

@ -73,6 +73,9 @@ class PRAgent:
args = update_settings_from_args(args)
action = action.lstrip("/").lower()
if action not in command2class:
get_logger().debug(f"Unknown command: {action}")
return False
with get_logger().contextualize(command=action):
get_logger().info("PR-Agent request handler started", analytics=True)
if action == "reflect_and_review":

17
pr_agent/algo/__init__.py
View File

@ -1,8 +1,9 @@
MAX_TOKENS = {
'text-embedding-ada-002': 8000,
'gpt-3.5-turbo': 4000,
'gpt-3.5-turbo': 16000,
'gpt-3.5-turbo-0125': 16000,
'gpt-3.5-turbo-0613': 4000,
'gpt-3.5-turbo-0301': 4000,
'gpt-3.5-turbo-1106': 16000,
'gpt-3.5-turbo-16k': 16000,
'gpt-3.5-turbo-16k-0613': 16000,
'gpt-4': 8000,
@ -10,6 +11,11 @@ MAX_TOKENS = {
'gpt-4-32k': 32000,
'gpt-4-1106-preview': 128000, # 128K, but may be limited by config.max_model_tokens
'gpt-4-0125-preview': 128000, # 128K, but may be limited by config.max_model_tokens
'gpt-4o': 128000, # 128K, but may be limited by config.max_model_tokens
'gpt-4o-2024-05-13': 128000, # 128K, but may be limited by config.max_model_tokens
'gpt-4-turbo-preview': 128000, # 128K, but may be limited by config.max_model_tokens
'gpt-4-turbo-2024-04-09': 128000, # 128K, but may be limited by config.max_model_tokens
'gpt-4-turbo': 128000, # 128K, but may be limited by config.max_model_tokens
'claude-instant-1': 100000,
'claude-2': 100000,
'command-nightly': 4096,
@ -23,4 +29,11 @@ MAX_TOKENS = {
'anthropic.claude-v1': 100000,
'anthropic.claude-v2': 100000,
'anthropic/claude-3-opus-20240229': 100000,
'bedrock/anthropic.claude-instant-v1': 100000,
'bedrock/anthropic.claude-v2': 100000,
'bedrock/anthropic.claude-v2:1': 100000,
'bedrock/anthropic.claude-3-sonnet-20240229-v1:0': 100000,
'bedrock/anthropic.claude-3-haiku-20240307-v1:0': 100000,
'groq/llama3-8b-8192': 8192,
'groq/llama3-70b-8192': 8192,
}

2
pr_agent/algo/ai_handlers/base_ai_handler.py
View File

@ -15,7 +15,7 @@ class BaseAiHandler(ABC):
pass
@abstractmethod
async def chat_completion(self, model: str, system: str, user: str, temperature: float = 0.2):
async def chat_completion(self, model: str, system: str, user: str, temperature: float = 0.2, img_path: str = None):
"""
This method should be implemented to return a chat completion from the AI model.
Args:

47
pr_agent/algo/ai_handlers/litellm_ai_handler.py
View File

@ -1,5 +1,5 @@
import os
import requests
import boto3
import litellm
import openai
@ -36,6 +36,8 @@ class LiteLLMAIHandler(BaseAiHandler):
assert litellm_token, "LITELLM_TOKEN is required"
os.environ["LITELLM_TOKEN"] = litellm_token
litellm.use_client = True
if get_settings().get("LITELLM.DROP_PARAMS", None):
litellm.drop_params = get_settings().litellm.drop_params
if get_settings().get("OPENAI.ORG", None):
litellm.organization = get_settings().openai.org
if get_settings().get("OPENAI.API_TYPE", None):
@ -50,8 +52,8 @@ class LiteLLMAIHandler(BaseAiHandler):
litellm.anthropic_key = get_settings().anthropic.key
if get_settings().get("COHERE.KEY", None):
litellm.cohere_key = get_settings().cohere.key
if get_settings().get("REPLICATE.KEY", None):
litellm.replicate_key = get_settings().replicate.key
if get_settings().get("GROQ.KEY", None):
litellm.api_key = get_settings().groq.key
if get_settings().get("REPLICATE.KEY", None):
litellm.replicate_key = get_settings().replicate.key
if get_settings().get("HUGGINGFACE.KEY", None):
@ -59,6 +61,9 @@ class LiteLLMAIHandler(BaseAiHandler):
if get_settings().get("HUGGINGFACE.API_BASE", None) and 'huggingface' in get_settings().config.model:
litellm.api_base = get_settings().huggingface.api_base
self.api_base = get_settings().huggingface.api_base
if get_settings().get("OLLAMA.API_BASE", None) :
litellm.api_base = get_settings().ollama.api_base
self.api_base = get_settings().ollama.api_base
if get_settings().get("HUGGINGFACE.REPITITION_PENALTY", None):
self.repetition_penalty = float(get_settings().huggingface.repetition_penalty)
if get_settings().get("VERTEXAI.VERTEX_PROJECT", None):
@ -68,11 +73,24 @@ class LiteLLMAIHandler(BaseAiHandler):
)
if get_settings().get("AWS.BEDROCK_REGION", None):
litellm.AmazonAnthropicConfig.max_tokens_to_sample = 2000
litellm.AmazonAnthropicClaude3Config.max_tokens = 2000
self.aws_bedrock_client = boto3.client(
service_name="bedrock-runtime",
region_name=get_settings().aws.bedrock_region,
)
def prepare_logs(self, response, system, user, resp, finish_reason):
response_log = response.dict().copy()
response_log['system'] = system
response_log['user'] = user
response_log['output'] = resp
response_log['finish_reason'] = finish_reason
if hasattr(self, 'main_pr_language'):
response_log['main_pr_language'] = self.main_pr_language
else:
response_log['main_pr_language'] = 'unknown'
return response_log
@property
def deployment_id(self):
"""
@ -84,13 +102,27 @@ class LiteLLMAIHandler(BaseAiHandler):
retry=retry_if_exception_type((openai.APIError, openai.APIConnectionError, openai.Timeout)), # No retry on RateLimitError
stop=stop_after_attempt(OPENAI_RETRIES)
)
async def chat_completion(self, model: str, system: str, user: str, temperature: float = 0.2):
async def chat_completion(self, model: str, system: str, user: str, temperature: float = 0.2, img_path: str = None):
try:
resp, finish_reason = None, None
deployment_id = self.deployment_id
if self.azure:
model = 'azure/' + model
messages = [{"role": "system", "content": system}, {"role": "user", "content": user}]
if img_path:
try:
# check if the image link is alive
r = requests.head(img_path, allow_redirects=True)
if r.status_code == 404:
error_msg = f"The image link is not [alive](img_path).\nPlease repost the original image as a comment, and send the question again with 'quote reply' (see [instructions](https://pr-agent-docs.codium.ai/tools/ask/#ask-on-images-using-the-pr-code-as-context))."
get_logger().error(error_msg)
return f"{error_msg}", "error"
except Exception as e:
get_logger().error(f"Error fetching image: {img_path}", e)
return f"Error fetching image: {img_path}", "error"
messages[1]["content"] = [{"type": "text", "text": messages[1]["content"]},
{"type": "image_url", "image_url": {"url": img_path}}]
kwargs = {
"model": model,
"deployment_id": deployment_id,
@ -125,10 +157,13 @@ class LiteLLMAIHandler(BaseAiHandler):
else:
resp = response["choices"][0]['message']['content']
finish_reason = response["choices"][0]["finish_reason"]
# usage = response.get("usage")
get_logger().debug(f"\nAI response:\n{resp}")
get_logger().debug("Full_response", artifact=response)
# log the full response for debugging
response_log = self.prepare_logs(response, system, user, resp, finish_reason)
get_logger().debug("Full_response", artifact=response_log)
# for CLI debugging
if get_settings().config.verbosity_level >= 2:
get_logger().info(f"\nAI response:\n{resp}")

3
pr_agent/algo/git_patch_processing.py
View File

@ -23,7 +23,10 @@ def extend_patch(original_file_str, patch_str, num_lines) -> str:
return patch_str
if type(original_file_str) == bytes:
try:
original_file_str = original_file_str.decode('utf-8')
except UnicodeDecodeError:
return ""
original_lines = original_file_str.splitlines()
patch_lines = patch_str.splitlines()

67
pr_agent/algo/pr_processing.py
View File

@ -9,7 +9,7 @@ from pr_agent.algo.git_patch_processing import convert_to_hunks_with_lines_numbe
from pr_agent.algo.language_handler import sort_files_by_main_languages
from pr_agent.algo.file_filter import filter_ignored
from pr_agent.algo.token_handler import TokenHandler
from pr_agent.algo.utils import get_max_tokens, ModelType
from pr_agent.algo.utils import get_max_tokens, clip_tokens, ModelType
from pr_agent.config_loader import get_settings
from pr_agent.git_providers.git_provider import GitProvider
from pr_agent.algo.types import EDIT_TYPE, FilePatchInfo
@ -87,18 +87,30 @@ def get_pr_diff(git_provider: GitProvider, token_handler: TokenHandler, model: s
# if we are over the limit, start pruning
get_logger().info(f"Tokens: {total_tokens}, total tokens over limit: {get_max_tokens(model)}, "
f"pruning diff.")
patches_compressed, modified_file_names, deleted_file_names, added_file_names = \
patches_compressed, modified_file_names, deleted_file_names, added_file_names, total_tokens_new = \
pr_generate_compressed_diff(pr_languages, token_handler, model, add_line_numbers_to_hunks)
# Insert additional information about added, modified, and deleted files if there is enough space
max_tokens = get_max_tokens(model) - OUTPUT_BUFFER_TOKENS_HARD_THRESHOLD
curr_token = total_tokens_new # == token_handler.count_tokens(final_diff)+token_handler.prompt_tokens
final_diff = "\n".join(patches_compressed)
if added_file_names:
delta_tokens = 10
if added_file_names and (max_tokens - curr_token) > delta_tokens:
added_list_str = ADDED_FILES_ + "\n".join(added_file_names)
added_list_str = clip_tokens(added_list_str, max_tokens - curr_token)
if added_list_str:
final_diff = final_diff + "\n\n" + added_list_str
if modified_file_names:
curr_token += token_handler.count_tokens(added_list_str) + 2
if modified_file_names and (max_tokens - curr_token) > delta_tokens:
modified_list_str = MORE_MODIFIED_FILES_ + "\n".join(modified_file_names)
modified_list_str = clip_tokens(modified_list_str, max_tokens - curr_token)
if modified_list_str:
final_diff = final_diff + "\n\n" + modified_list_str
if deleted_file_names:
curr_token += token_handler.count_tokens(modified_list_str) + 2
if deleted_file_names and (max_tokens - curr_token) > delta_tokens:
deleted_list_str = DELETED_FILES_ + "\n".join(deleted_file_names)
deleted_list_str = clip_tokens(deleted_list_str, max_tokens - curr_token)
if deleted_list_str:
final_diff = final_diff + "\n\n" + deleted_list_str
try:
get_logger().debug(f"After pruning, added_list_str: {added_list_str}, modified_list_str: {modified_list_str}, "
@ -134,6 +146,9 @@ def pr_generate_extended_diff(pr_languages: list,
# extend each patch with extra lines of context
extended_patch = extend_patch(original_file_content_str, patch, num_lines=patch_extra_lines)
if not extended_patch:
get_logger().warning(f"Failed to extend patch for file: {file.filename}")
continue
full_extended_patch = f"\n\n## {file.filename}\n\n{extended_patch}\n"
if add_line_numbers_to_hunks:
@ -149,7 +164,7 @@ def pr_generate_extended_diff(pr_languages: list,
def pr_generate_compressed_diff(top_langs: list, token_handler: TokenHandler, model: str,
convert_hunks_to_line_numbers: bool) -> Tuple[list, list, list, list]:
convert_hunks_to_line_numbers: bool) -> Tuple[list, list, list, list, int]:
"""
Generate a compressed diff string for a pull request, using diff minimization techniques to reduce the number of
tokens used.
@ -195,10 +210,11 @@ def pr_generate_compressed_diff(top_langs: list, token_handler: TokenHandler, mo
patch = handle_patch_deletions(patch, original_file_content_str,
new_file_content_str, file.filename, file.edit_type)
if patch is None:
if not deleted_files_list:
total_tokens += token_handler.count_tokens(DELETED_FILES_)
# if not deleted_files_list:
# total_tokens += token_handler.count_tokens(DELETED_FILES_)
if file.filename not in deleted_files_list:
deleted_files_list.append(file.filename)
total_tokens += token_handler.count_tokens(file.filename) + 1
# total_tokens += token_handler.count_tokens(file.filename) + 1
continue
if convert_hunks_to_line_numbers:
@ -219,14 +235,17 @@ def pr_generate_compressed_diff(top_langs: list, token_handler: TokenHandler, mo
if get_settings().config.verbosity_level >= 2:
get_logger().warning(f"Patch too large, minimizing it, {file.filename}")
if file.edit_type == EDIT_TYPE.ADDED:
if not added_files_list:
total_tokens += token_handler.count_tokens(ADDED_FILES_)
# if not added_files_list:
# total_tokens += token_handler.count_tokens(ADDED_FILES_)
if file.filename not in added_files_list:
added_files_list.append(file.filename)
# total_tokens += token_handler.count_tokens(file.filename) + 1
else:
if not modified_files_list:
total_tokens += token_handler.count_tokens(MORE_MODIFIED_FILES_)
# if not modified_files_list:
# total_tokens += token_handler.count_tokens(MORE_MODIFIED_FILES_)
if file.filename not in modified_files_list:
modified_files_list.append(file.filename)
total_tokens += token_handler.count_tokens(file.filename) + 1
# total_tokens += token_handler.count_tokens(file.filename) + 1
continue
if patch:
@ -239,7 +258,7 @@ def pr_generate_compressed_diff(top_langs: list, token_handler: TokenHandler, mo
if get_settings().config.verbosity_level >= 2:
get_logger().info(f"Tokens: {total_tokens}, last filename: {file.filename}")
return patches, modified_files_list, deleted_files_list, added_files_list
return patches, modified_files_list, deleted_files_list, added_files_list, total_tokens
async def retry_with_fallback_models(f: Callable, model_type: ModelType = ModelType.REGULAR):
@ -358,7 +377,23 @@ def get_pr_multi_diffs(git_provider: GitProvider,
patch = convert_to_hunks_with_lines_numbers(patch, file)
new_patch_tokens = token_handler.count_tokens(patch)
if patch and (token_handler.prompt_tokens + new_patch_tokens) > get_max_tokens(model) - OUTPUT_BUFFER_TOKENS_SOFT_THRESHOLD:
if patch and (token_handler.prompt_tokens + new_patch_tokens) > get_max_tokens(
model) - OUTPUT_BUFFER_TOKENS_SOFT_THRESHOLD:
if get_settings().config.get('large_patch_policy', 'skip') == 'skip':
get_logger().warning(f"Patch too large, skipping: {file.filename}")
continue
elif get_settings().config.get('large_patch_policy') == 'clip':
delta_tokens = get_max_tokens(model) - OUTPUT_BUFFER_TOKENS_SOFT_THRESHOLD - token_handler.prompt_tokens
patch_clipped = clip_tokens(patch, delta_tokens, delete_last_line=True, num_input_tokens=new_patch_tokens)
new_patch_tokens = token_handler.count_tokens(patch_clipped)
if patch_clipped and (token_handler.prompt_tokens + new_patch_tokens) > get_max_tokens(
model) - OUTPUT_BUFFER_TOKENS_SOFT_THRESHOLD:
get_logger().warning(f"Patch too large, skipping: {file.filename}")
continue
else:
get_logger().info(f"Clipped large patch for file: {file.filename}")
patch = patch_clipped
else:
get_logger().warning(f"Patch too large, skipping: {file.filename}")
continue

21
pr_agent/algo/token_handler.py
View File

@ -1,12 +1,25 @@
from jinja2 import Environment, StrictUndefined
from tiktoken import encoding_for_model, get_encoding
from pr_agent.config_loader import get_settings
from threading import Lock
def get_token_encoder():
return encoding_for_model(get_settings().config.model) if "gpt" in get_settings().config.model else get_encoding(
class TokenEncoder:
_encoder_instance = None
_model = None
_lock = Lock() # Create a lock object
@classmethod
def get_token_encoder(cls):
model = get_settings().config.model
if cls._encoder_instance is None or model != cls._model: # Check without acquiring the lock for performance
with cls._lock: # Lock acquisition to ensure thread safety
if cls._encoder_instance is None or model != cls._model:
cls._model = model
cls._encoder_instance = encoding_for_model(cls._model) if "gpt" in cls._model else get_encoding(
"cl100k_base")
return cls._encoder_instance
class TokenHandler:
"""
@ -31,7 +44,7 @@ class TokenHandler:
- system: The system string.
- user: The user string.
"""
self.encoder = get_token_encoder()
self.encoder = TokenEncoder.get_token_encoder()
if pr is not None:
self.prompt_tokens = self._get_system_user_tokens(pr, self.encoder, vars, system, user)

160
pr_agent/algo/utils.py
View File

@ -2,8 +2,10 @@ from __future__ import annotations
import difflib
import json
import os
import re
import textwrap
import time
from datetime import datetime
from enum import Enum
from typing import Any, List, Tuple
@ -12,7 +14,7 @@ import yaml
from starlette_context import context
from pr_agent.algo import MAX_TOKENS
from pr_agent.algo.token_handler import get_token_encoder
from pr_agent.algo.token_handler import TokenEncoder
from pr_agent.config_loader import get_settings, global_settings
from pr_agent.algo.types import FilePatchInfo
from pr_agent.log import get_logger
@ -70,10 +72,12 @@ def convert_to_markdown(output_data: dict, gfm_supported: bool = True, increment
"""
emojis = {
"Possible issues": "🔍",
"Can be split": "🔀",
"Possible issues": "⚡",
"Score": "🏅",
"Relevant tests": "🧪",
"Focused PR": "✨",
"Relevant ticket": "🎫",
"Security concerns": "🔒",
"Insights from user's answers": "📝",
"Code feedback": "🤖",
@ -81,9 +85,9 @@ def convert_to_markdown(output_data: dict, gfm_supported: bool = True, increment
}
markdown_text = ""
if not incremental_review:
markdown_text += f"## PR Review\n\n"
markdown_text += f"## PR Review 🔍\n\n"
else:
markdown_text += f"## Incremental PR Review\n\n"
markdown_text += f"## Incremental PR Review 🔍\n\n"
markdown_text += f"⏮️ Review for commits since previous PR-Agent review {incremental_review}.\n\n"
if gfm_supported:
markdown_text += "<table>\n<tr>\n"
@ -94,6 +98,7 @@ def convert_to_markdown(output_data: dict, gfm_supported: bool = True, increment
for key, value in output_data['review'].items():
if value is None or value == '' or value == {} or value == []:
if key.lower() != 'can_be_split':
continue
key_nice = key.replace('_', ' ').capitalize()
emoji = emojis.get(key_nice, "")
@ -103,6 +108,8 @@ def convert_to_markdown(output_data: dict, gfm_supported: bool = True, increment
if 'security concerns' in key_nice.lower():
value = emphasize_header(value.strip())
markdown_text += f"<tr><td> {emoji}&nbsp;<strong>{key_nice}</strong></td><td>\n\n{value}\n\n</td></tr>\n"
elif 'can be split' in key_nice.lower():
markdown_text += process_can_be_split(emoji, value)
elif 'possible issues' in key_nice.lower():
value = value.strip()
issues = value.split('\n- ')
@ -154,6 +161,38 @@ def convert_to_markdown(output_data: dict, gfm_supported: bool = True, increment
return markdown_text
def process_can_be_split(emoji, value):
# key_nice = "Can this PR be split?"
key_nice = "Multiple PR themes"
markdown_text = ""
if not value or isinstance(value, list) and len(value) == 1:
value = "No"
markdown_text += f"<tr><td> {emoji}&nbsp;<strong>{key_nice}</strong></td><td>\n\n{value}\n\n</td></tr>\n"
else:
number_of_splits = len(value)
markdown_text += f"<tr><td rowspan={number_of_splits}> {emoji}&nbsp;<strong>{key_nice}</strong></td>\n"
for i, split in enumerate(value):
title = split.get('title', '')
relevant_files = split.get('relevant_files', [])
if i == 0:
markdown_text += f"<td><details><summary>\nSub-PR theme: <strong>{title}</strong></summary>\n\n"
markdown_text += f"<hr>\n"
markdown_text += f"Relevant files:\n"
markdown_text += f"<ul>\n"
for file in relevant_files:
markdown_text += f"<li>{file}</li>\n"
markdown_text += f"</ul>\n\n</details></td></tr>\n"
else:
markdown_text += f"<tr>\n<td><details><summary>\nSub-PR theme: <strong>{title}</strong></summary>\n\n"
markdown_text += f"<hr>\n"
markdown_text += f"Relevant files:\n"
markdown_text += f"<ul>\n"
for file in relevant_files:
markdown_text += f"<li>{file}</li>\n"
markdown_text += f"</ul>\n\n</details></td></tr>\n"
return markdown_text
def parse_code_suggestion(code_suggestion: dict, i: int = 0, gfm_supported: bool = True) -> str:
"""
Convert a dictionary of data into markdown format.
@ -319,7 +358,7 @@ def convert_str_to_datetime(date_str):
return datetime.strptime(date_str, datetime_format)
def load_large_diff(filename, new_file_content_str: str, original_file_content_str: str) -> str:
def load_large_diff(filename, new_file_content_str: str, original_file_content_str: str, show_warning: bool = True) -> str:
"""
Generate a patch for a modified file by comparing the original content of the file with the new content provided as
input.
@ -338,7 +377,7 @@ def load_large_diff(filename, new_file_content_str: str, original_file_content_s
try:
diff = difflib.unified_diff(original_file_content_str.splitlines(keepends=True),
new_file_content_str.splitlines(keepends=True))
if get_settings().config.verbosity_level >= 2:
if get_settings().config.verbosity_level >= 2 and show_warning:
get_logger().warning(f"File was modified, but no patch was found. Manually creating patch: {filename}.")
patch = ''.join(diff)
except Exception:
@ -433,6 +472,7 @@ def try_fix_yaml(response_text: str, keys_fix_yaml: List[str] = []) -> dict:
except:
pass
# third fallback - try to remove leading and trailing curly brackets
response_text_copy = response_text.strip().rstrip().removeprefix('{').removesuffix('}').rstrip(':\n')
try:
@ -515,7 +555,7 @@ def get_max_tokens(model):
return max_tokens_model
def clip_tokens(text: str, max_tokens: int, add_three_dots=True) -> str:
def clip_tokens(text: str, max_tokens: int, add_three_dots=True, num_input_tokens=None, delete_last_line=False) -> str:
"""
Clip the number of tokens in a string to a maximum number of tokens.
@ -530,16 +570,30 @@ def clip_tokens(text: str, max_tokens: int, add_three_dots=True) -> str:
return text
try:
encoder = get_token_encoder()
if num_input_tokens is None:
encoder = TokenEncoder.get_token_encoder()
num_input_tokens = len(encoder.encode(text))
if num_input_tokens <= max_tokens:
return text
if max_tokens < 0:
return ""
# calculate the number of characters to keep
num_chars = len(text)
chars_per_token = num_chars / num_input_tokens
num_output_chars = int(chars_per_token * max_tokens)
factor = 0.9 # reduce by 10% to be safe
num_output_chars = int(factor * chars_per_token * max_tokens)
# clip the text
if num_output_chars > 0:
clipped_text = text[:num_output_chars]
if delete_last_line:
clipped_text = clipped_text.rsplit('\n', 1)[0]
if add_three_dots:
clipped_text += "...(truncated)"
clipped_text += "\n...(truncated)"
else: # if the text is empty
clipped_text = ""
return clipped_text
except Exception as e:
get_logger().warning(f"Failed to clip tokens: {e}")
@ -625,3 +679,89 @@ def find_line_number_of_relevant_line_in_file(diff_files: List[FilePatchInfo],
absolute_position = start2 + delta - 1
break
return position, absolute_position
def validate_and_await_rate_limit(rate_limit_status=None, git_provider=None, get_rate_limit_status_func=None):
if git_provider and not rate_limit_status:
rate_limit_status = {'resources': git_provider.github_client.get_rate_limit().raw_data}
if not rate_limit_status:
rate_limit_status = get_rate_limit_status_func()
# validate that the rate limit is not exceeded
is_rate_limit = False
for key, value in rate_limit_status['resources'].items():
if value['remaining'] == 0:
print(f"key: {key}, value: {value}")
is_rate_limit = True
sleep_time_sec = value['reset'] - datetime.now().timestamp()
sleep_time_hour = sleep_time_sec / 3600.0
print(f"Rate limit exceeded. Sleeping for {sleep_time_hour} hours")
if sleep_time_sec > 0:
time.sleep(sleep_time_sec+1)
if git_provider:
rate_limit_status = {'resources': git_provider.github_client.get_rate_limit().raw_data}
else:
rate_limit_status = get_rate_limit_status_func()
return is_rate_limit
def get_largest_component(pr_url):
from pr_agent.tools.pr_analyzer import PRAnalyzer
publish_output = get_settings().config.publish_output
get_settings().config.publish_output = False # disable publish output
analyzer = PRAnalyzer(pr_url)
methods_dict_files = analyzer.run_sync()
get_settings().config.publish_output = publish_output
max_lines_changed = 0
file_b = ""
component_name_b = ""
for file in methods_dict_files:
for method in methods_dict_files[file]:
try:
if methods_dict_files[file][method]['num_plus_lines'] > max_lines_changed:
max_lines_changed = methods_dict_files[file][method]['num_plus_lines']
file_b = file
component_name_b = method
except:
pass
if component_name_b:
get_logger().info(f"Using the largest changed component: '{component_name_b}'")
return component_name_b, file_b
else:
return None, None
def github_action_output(output_data: dict, key_name: str):
try:
if not get_settings().get('github_action_config.enable_output', False):
return
key_data = output_data.get(key_name, {})
with open(os.environ['GITHUB_OUTPUT'], 'a') as fh:
print(f"{key_name}={json.dumps(key_data, indent=None, ensure_ascii=False)}", file=fh)
except Exception as e:
get_logger().error(f"Failed to write to GitHub Action output: {e}")
return
def show_relevant_configurations(relevant_section: str) -> str:
forbidden_keys = ['ai_disclaimer', 'ai_disclaimer_title', 'ANALYTICS_FOLDER', 'secret_provider',
'trial_prefix_message', 'no_eligible_message', 'identity_provider', 'ALLOWED_REPOS','APP_NAME']
markdown_text = ""
markdown_text += "\n<hr>\n<details> <summary><strong>🛠️ Relevant configurations:</strong></summary> \n\n"
markdown_text +="<br>These are the relevant [configurations](https://github.com/Codium-ai/pr-agent/blob/main/pr_agent/settings/configuration.toml) for this tool:\n\n"
markdown_text += f"**[config**]\n```yaml\n\n"
for key, value in get_settings().config.items():
if key in forbidden_keys:
continue
markdown_text += f"{key}: {value}\n"
markdown_text += "\n```\n"
markdown_text += f"\n**[{relevant_section}]**\n```yaml\n\n"
for key, value in get_settings().get(relevant_section, {}).items():
if key in forbidden_keys:
continue
markdown_text += f"{key}: {value}\n"
markdown_text += "\n```"
markdown_text += "\n</details>\n"
return markdown_text

63
pr_agent/cli.py
View File

@ -9,47 +9,58 @@ from pr_agent.log import setup_logger
log_level = os.environ.get("LOG_LEVEL", "INFO")
setup_logger(log_level)
def run(inargs=None):
def set_parser():
parser = argparse.ArgumentParser(description='AI based pull request analyzer', usage=
"""\
Usage: cli.py --pr-url=<URL on supported git hosting service> <command> [<args>].
For example:
- cli.py --pr_url=... review
- cli.py --pr_url=... describe
- cli.py --pr_url=... improve
- cli.py --pr_url=... ask "write me a poem about this PR"
- cli.py --pr_url=... reflect
- cli.py --issue_url=... similar_issue
"""\
Usage: cli.py --pr-url=<URL on supported git hosting service> <command> [<args>].
For example:
- cli.py --pr_url=... review
- cli.py --pr_url=... describe
- cli.py --pr_url=... improve
- cli.py --pr_url=... ask "write me a poem about this PR"
- cli.py --pr_url=... reflect
- cli.py --issue_url=... similar_issue
Supported commands:
- review / review_pr - Add a review that includes a summary of the PR and specific suggestions for improvement.
Supported commands:
- review / review_pr - Add a review that includes a summary of the PR and specific suggestions for improvement.
- ask / ask_question [question] - Ask a question about the PR.
- ask / ask_question [question] - Ask a question about the PR.
- describe / describe_pr - Modify the PR title and description based on the PR's contents.
- describe / describe_pr - Modify the PR title and description based on the PR's contents.
- improve / improve_code - Suggest improvements to the code in the PR as pull request comments ready to commit.
Extended mode ('improve --extended') employs several calls, and provides a more thorough feedback
- improve / improve_code - Suggest improvements to the code in the PR as pull request comments ready to commit.
Extended mode ('improve --extended') employs several calls, and provides a more thorough feedback
- reflect - Ask the PR author questions about the PR.
- reflect - Ask the PR author questions about the PR.
- update_changelog - Update the changelog based on the PR's contents.
- update_changelog - Update the changelog based on the PR's contents.
- add_docs
- add_docs
- generate_labels
- generate_labels
Configuration:
To edit any configuration parameter from 'configuration.toml', just add -config_path=<value>.
For example: 'python cli.py --pr_url=... review --pr_reviewer.extra_instructions="focus on the file: ..."'
""")
Configuration:
To edit any configuration parameter from 'configuration.toml', just add -config_path=<value>.
For example: 'python cli.py --pr_url=... review --pr_reviewer.extra_instructions="focus on the file: ..."'
""")
parser.add_argument('--pr_url', type=str, help='The URL of the PR to review', default=None)
parser.add_argument('--issue_url', type=str, help='The URL of the Issue to review', default=None)
parser.add_argument('command', type=str, help='The', choices=commands, default='review')
parser.add_argument('rest', nargs=argparse.REMAINDER, default=[])
return parser
def run_command(pr_url, command):
# Preparing the command
run_command_str = f"--pr_url={pr_url} {command.lstrip('/')}"
args = set_parser().parse_args(run_command_str.split())
# Run the command. Feedback will appear in GitHub PR comments
run(args=args)
def run(inargs=None, args=None):
parser = set_parser()
if not args:
args = parser.parse_args(inargs)
if not args.pr_url and not args.issue_url:
parser.print_help()

23
pr_agent/cli_pip.py Normal file
View File

@ -0,0 +1,23 @@
from pr_agent import cli
from pr_agent.config_loader import get_settings
def main():
# Fill in the following values
provider = "github" # GitHub provider
user_token = "..." # GitHub user token
openai_key = "..." # OpenAI key
pr_url = "..." # PR URL, for example 'https://github.com/Codium-ai/pr-agent/pull/809'
command = "/review" # Command to run (e.g. '/review', '/describe', '/ask="What is the purpose of this PR?"')
# Setting the configurations
get_settings().set("CONFIG.git_provider", provider)
get_settings().set("openai.key", openai_key)
get_settings().set("github.user_token", user_token)
# Run the command. Feedback will appear in GitHub PR comments
cli.run_command(pr_url, command)
if __name__ == '__main__':
main()

1
pr_agent/config_loader.py
View File

@ -21,6 +21,7 @@ global_settings = Dynaconf(
"settings/pr_line_questions_prompts.toml",
"settings/pr_description_prompts.toml",
"settings/pr_code_suggestions_prompts.toml",
"settings/pr_code_suggestions_reflect_prompts.toml",
"settings/pr_sort_code_suggestions_prompts.toml",
"settings/pr_information_from_user_prompts.toml",
"settings/pr_update_changelog_prompts.toml",

82
pr_agent/git_providers/azuredevops_provider.py
View File

@ -26,6 +26,7 @@ try:
CommentThread,
GitVersionDescriptor,
GitPullRequest,
GitPullRequestIterationChanges,
)
except ImportError:
AZURE_DEVOPS_AVAILABLE = False
@ -230,29 +231,58 @@ class AzureDevopsProvider(GitProvider):
base_sha = self.pr.last_merge_target_commit
head_sha = self.pr.last_merge_source_commit
commits = self.azure_devops_client.get_pull_request_commits(
project=self.workspace_slug,
# Get PR iterations
iterations = self.azure_devops_client.get_pull_request_iterations(
repository_id=self.repo_slug,
pull_request_id=self.pr_num,
project=self.workspace_slug
)
changes = None
if iterations:
iteration_id = iterations[-1].id # Get the last iteration (most recent changes)
# Get changes for the iteration
changes = self.azure_devops_client.get_pull_request_iteration_changes(
repository_id=self.repo_slug,
pull_request_id=self.pr_num,
iteration_id=iteration_id,
project=self.workspace_slug
)
diff_files = []
diffs = []
diff_types = {}
if changes:
for change in changes.change_entries:
item = change.additional_properties.get('item', {})
path = item.get('path', None)
if path:
diffs.append(path)
diff_types[path] = change.additional_properties.get('changeType', 'Unknown')
for c in commits:
changes_obj = self.azure_devops_client.get_changes(
project=self.workspace_slug,
repository_id=self.repo_slug,
commit_id=c.commit_id,
)
for i in changes_obj.changes:
if i["item"]["gitObjectType"] == "tree":
continue
diffs.append(i["item"]["path"])
diff_types[i["item"]["path"]] = i["changeType"]
# wrong implementation - gets all the files that were changed in any commit in the PR
# commits = self.azure_devops_client.get_pull_request_commits(
# project=self.workspace_slug,
# repository_id=self.repo_slug,
# pull_request_id=self.pr_num,
# )
#
# diff_files = []
# diffs = []
# diff_types = {}
diffs = list(set(diffs))
# for c in commits:
# changes_obj = self.azure_devops_client.get_changes(
# project=self.workspace_slug,
# repository_id=self.repo_slug,
# commit_id=c.commit_id,
# )
# for i in changes_obj.changes:
# if i["item"]["gitObjectType"] == "tree":
# continue
# diffs.append(i["item"]["path"])
# diff_types[i["item"]["path"]] = i["changeType"]
#
# diffs = list(set(diffs))
for file in diffs:
if not is_valid_file(file):
@ -273,12 +303,13 @@ class AzureDevopsProvider(GitProvider):
new_file_content_str = new_file_content_str.content
except Exception as error:
get_logger().error(
"Failed to retrieve new file content of %s at version %s. Error: %s",
file,
version,
str(error),
)
get_logger().error(f"Failed to retrieve new file content of {file} at version {version}. Error: {str(error)}")
# get_logger().error(
# "Failed to retrieve new file content of %s at version %s. Error: %s",
# file,
# version,
# str(error),
# )
new_file_content_str = ""
edit_type = EDIT_TYPE.MODIFIED
@ -303,17 +334,12 @@ class AzureDevopsProvider(GitProvider):
)
original_file_content_str = original_file_content_str.content
except Exception as error:
get_logger().error(
"Failed to retrieve original file content of %s at version %s. Error: %s",
file,
version,
str(error),
)
get_logger().error(f"Failed to retrieve original file content of {file} at version {version}. Error: {str(error)}")
original_file_content_str = ""
patch = load_large_diff(
file, new_file_content_str, original_file_content_str
)
file, new_file_content_str, original_file_content_str, show_warning=False
).rstrip()
diff_files.append(
FilePatchInfo(

74
pr_agent/git_providers/bitbucket_server_provider.py
View File

@ -1,13 +1,14 @@
import json
from typing import Optional, Tuple
from urllib.parse import urlparse
from urllib.parse import quote_plus, urlparse
import requests
from atlassian.bitbucket import Bitbucket
from starlette_context import context
from .git_provider import GitProvider
from pr_agent.algo.types import FilePatchInfo
from ..algo.types import EDIT_TYPE, FilePatchInfo
from ..algo.language_handler import is_valid_file
from ..algo.utils import load_large_diff, find_line_number_of_relevant_line_in_file
from ..config_loader import get_settings
from ..log import get_logger
@ -59,6 +60,9 @@ class BitbucketServerProvider(GitProvider):
except Exception:
return ""
def get_pr_id(self):
return self.pr_num
def publish_code_suggestions(self, code_suggestions: list) -> bool:
"""
Publishes code suggestions as comments on the PR.
@ -140,14 +144,8 @@ class BitbucketServerProvider(GitProvider):
if self.diff_files:
return self.diff_files
commits_in_pr = self.bitbucket_client.get_pull_requests_commits(
self.workspace_slug,
self.repo_slug,
self.pr_num
)
commit_list = list(commits_in_pr)
base_sha, head_sha = commit_list[0]['parents'][0]['id'], commit_list[-1]['id']
base_sha = self.pr.toRef['latestCommit']
head_sha = self.pr.fromRef['latestCommit']
diff_files = []
original_file_content_str = ""
@ -156,6 +154,10 @@ class BitbucketServerProvider(GitProvider):
changes = self.bitbucket_client.get_pull_requests_changes(self.workspace_slug, self.repo_slug, self.pr_num)
for change in changes:
file_path = change['path']['toString']
if not is_valid_file(file_path.split("/")[-1]):
get_logger().info(f"Skipping a non-code file: {file_path}")
continue
match change['type']:
case 'ADD':
edit_type = EDIT_TYPE.ADDED
@ -241,8 +243,11 @@ class BitbucketServerProvider(GitProvider):
}
}
response = requests.post(url=self._get_pr_comments_url(), json=payload, headers=self.headers)
return response
try:
requests.post(url=self._get_pr_comments_url(), json=payload, headers=self.headers).raise_for_status()
except Exception as e:
get_logger().error(f"Failed to publish inline comment to '{file}' at line {from_line}, error: {e}")
raise e
def generate_link_to_relevant_line_number(self, suggestion) -> str:
try:
@ -255,18 +260,37 @@ class BitbucketServerProvider(GitProvider):
position, absolute_position = find_line_number_of_relevant_line_in_file \
(diff_files, relevant_file, relevant_line_str)
if absolute_position != -1:
if self.pr:
link = f"{self.pr_url}/diff#{quote_plus(relevant_file)}?t={absolute_position}"
return link
else:
if get_settings().config.verbosity_level >= 2:
get_logger().info(f"Failed adding line link to '{relevant_file}' since PR not set")
else:
if get_settings().config.verbosity_level >= 2:
get_logger().info(f"Failed adding line link to '{relevant_file}' since position not found")
if absolute_position != -1 and self.pr_url:
link = f"{self.pr_url}/#L{relevant_file}T{absolute_position}"
link = f"{self.pr_url}/diff#{quote_plus(relevant_file)}?t={absolute_position}"
return link
except Exception as e:
if get_settings().config.verbosity_level >= 2:
get_logger().info(f"Failed adding line link, error: {e}")
get_logger().info(f"Failed adding line link to '{relevant_file}', error: {e}")
return ""
def publish_inline_comments(self, comments: list[dict]):
for comment in comments:
if 'position' in comment:
self.publish_inline_comment(comment['body'], comment['position'], comment['path'])
elif 'start_line' in comment: # multi-line comment
# note that bitbucket does not seem to support range - only a comment on a single line - https://community.developer.atlassian.com/t/api-post-endpoint-for-inline-pull-request-comments/60452
self.publish_inline_comment(comment['body'], comment['start_line'], comment['path'])
elif 'line' in comment: # single-line comment
self.publish_inline_comment(comment['body'], comment['line'], comment['path'])
else:
get_logger().error(f"Could not publish inline comment: {comment}")
def get_title(self):
return self.pr.title
@ -278,7 +302,10 @@ class BitbucketServerProvider(GitProvider):
return self.pr.fromRef['displayId']
def get_pr_description_full(self):
if hasattr(self.pr, "description"):
return self.pr.description
else:
return None
def get_user_id(self):
return 0
@ -305,7 +332,7 @@ class BitbucketServerProvider(GitProvider):
path_parts = parsed_url.path.strip("/").split("/")
if len(path_parts) < 6 or path_parts[4] != "pull-requests":
raise ValueError(
"The provided URL does not appear to be a Bitbucket PR URL"
f"The provided URL '{pr_url}' does not appear to be a Bitbucket PR URL"
)
workspace_slug = path_parts[1]
@ -313,7 +340,7 @@ class BitbucketServerProvider(GitProvider):
try:
pr_number = int(path_parts[5])
except ValueError as e:
raise ValueError("Unable to convert PR number to integer") from e
raise ValueError(f"Unable to convert PR number '{path_parts[5]}' to integer") from e
return workspace_slug, repo_slug, pr_number
@ -334,13 +361,18 @@ class BitbucketServerProvider(GitProvider):
raise NotImplementedError("Get commit messages function not implemented yet.")
# bitbucket does not support labels
def publish_description(self, pr_title: str, description: str):
payload = json.dumps({
payload = {
"version": self.pr.version,
"description": description,
"title": pr_title
})
"title": pr_title,
"reviewers": self.pr.reviewers # needs to be sent otherwise gets wiped
}
try:
self.bitbucket_client.update_pull_request(self.workspace_slug, self.repo_slug, str(self.pr_num), payload)
except Exception as e:
get_logger().error(f"Failed to update pull request, error: {e}")
raise e
response = requests.put(url=self.bitbucket_pull_request_api_url, headers=self.headers, data=payload)
return response
# bitbucket does not support labels
def publish_labels(self, pr_types: list):

2
pr_agent/git_providers/codecommit_provider.py
View File

@ -10,7 +10,7 @@ from ..algo.utils import load_large_diff
from .git_provider import GitProvider
from ..config_loader import get_settings
from ..log import get_logger
from pr_agent.algo.language_handler import is_valid_file
class PullRequestCCMimic:
"""

13
pr_agent/git_providers/git_provider.py
View File

@ -71,7 +71,7 @@ class GitProvider(ABC):
# if the existing description was generated by the pr-agent, but it doesn't contain a user description,
# return nothing (empty string) because it means there is no user description
user_description_header = "## **user description**"
user_description_header = "### **user description**"
if user_description_header not in description_lowercase:
get_logger().info(f"Existing description was generated by the pr-agent, but it doesn't contain a user description")
return ""
@ -102,8 +102,8 @@ class GitProvider(ABC):
return original_user_description
def _possible_headers(self):
return ("## **user description**", "## **pr type**", "## **pr description**", "## **pr labels**", "## **type**", "## **description**",
"## **labels**", "### 🤖 generated by pr agent")
return ("### **user description**", "### **pr type**", "### **pr description**", "### **pr labels**", "### **type**", "### **description**",
"### **labels**", "### 🤖 generated by pr agent")
def _is_generated_by_pr_agent(self, description_lowercase: str) -> bool:
possible_headers = self._possible_headers()
@ -197,6 +197,12 @@ class GitProvider(ABC):
def calc_pr_statistics(self, pull_request_data: dict):
return {}
def get_num_of_files(self):
try:
return len(self.get_diff_files())
except Exception as e:
return -1
def get_main_pr_language(languages, files) -> str:
"""
@ -266,6 +272,7 @@ def get_main_pr_language(languages, files) -> str:
return main_language_str
class IncrementalPR:
def __init__(self, is_incremental: bool = False):
self.is_incremental = is_incremental

24
pr_agent/git_providers/github_provider.py
View File

@ -114,6 +114,11 @@ class GithubProvider(GitProvider):
self.git_files = self.pr.get_files()
return self.git_files
def get_num_of_files(self):
if self.git_files:
return self.git_files.totalCount
else:
return -1
@retry(exceptions=RateLimitExceeded,
tries=get_settings().github.ratelimit_retries, delay=2, backoff=2, jitter=(1, 3))
@ -142,6 +147,7 @@ class GithubProvider(GitProvider):
for file in files:
if not is_valid_file(file.filename):
get_logger().info(f"Skipping a non-code file: {file.filename}")
continue
new_file_content_str = self._get_pr_file_content(file, self.pr.head.sha) # communication with GitHub
@ -740,22 +746,4 @@ class GithubProvider(GitProvider):
return False
def calc_pr_statistics(self, pull_request_data: dict):
try:
out = {}
from datetime import datetime
created_at = pull_request_data['created_at']
closed_at = pull_request_data['closed_at']
closed_at_datetime = datetime.strptime(closed_at, "%Y-%m-%dT%H:%M:%SZ")
created_at_datetime = datetime.strptime(created_at, "%Y-%m-%dT%H:%M:%SZ")
difference = closed_at_datetime - created_at_datetime
out['hours'] = difference.total_seconds() / 3600
out['commits'] = pull_request_data['commits']
out['comments'] = pull_request_data['comments']
out['review_comments'] = pull_request_data['review_comments']
out['changed_files'] = pull_request_data['changed_files']
out['additions'] = pull_request_data['additions']
out['deletions'] = pull_request_data['deletions']
except Exception as e:
get_logger().exception(f"Failed to calculate PR statistics, error: {e}")
return {}
return out

0
pr_agent/servers/__init__.py Normal file
View File

12
pr_agent/servers/azuredevops_server_webhook.py
View File

@ -26,8 +26,9 @@ from pr_agent.git_providers.utils import apply_repo_settings
from pr_agent.log import get_logger
from fastapi import Request, Depends
from fastapi.security import HTTPBasic, HTTPBasicCredentials
from pr_agent.log import get_logger
from pr_agent.log import LoggingFormat, get_logger, setup_logger
setup_logger(fmt=LoggingFormat.JSON, level="DEBUG")
security = HTTPBasic()
router = APIRouter()
available_commands_rgx = re.compile(r"^\/(" + "|".join(command2class.keys()) + r")\s*")
@ -40,8 +41,15 @@ def handle_request(
):
log_context["action"] = body
log_context["api_url"] = url
async def inner():
try:
with get_logger().contextualize(**log_context):
background_tasks.add_task(PRAgent().handle_request, url, body)
await PRAgent().handle_request(url, body)
except Exception as e:
get_logger().error(f"Failed to handle webhook: {e}")
background_tasks.add_task(inner)
# currently only basic auth is supported with azure webhooks

12
pr_agent/servers/bitbucket_server_webhook.py
View File

@ -13,9 +13,10 @@ from starlette_context.middleware import RawContextMiddleware
from pr_agent.agent.pr_agent import PRAgent
from pr_agent.config_loader import get_settings
from pr_agent.log import get_logger
from pr_agent.log import LoggingFormat, get_logger, setup_logger
from pr_agent.servers.utils import verify_signature
setup_logger(fmt=LoggingFormat.JSON, level="DEBUG")
router = APIRouter()
@ -24,8 +25,15 @@ def handle_request(
):
log_context["action"] = body
log_context["api_url"] = url
async def inner():
try:
with get_logger().contextualize(**log_context):
background_tasks.add_task(PRAgent().handle_request, url, body)
await PRAgent().handle_request(url, body)
except Exception as e:
get_logger().error(f"Failed to handle webhook: {e}")
background_tasks.add_task(inner)
@router.post("/")

12
pr_agent/servers/github_action_runner.py
View File

@ -46,19 +46,22 @@ async def run_action():
if not GITHUB_EVENT_PATH:
print("GITHUB_EVENT_PATH not set")
return
if not OPENAI_KEY:
print("OPENAI_KEY not set")
return
if not GITHUB_TOKEN:
print("GITHUB_TOKEN not set")
return
# Set the environment variables in the settings
if OPENAI_KEY:
get_settings().set("OPENAI.KEY", OPENAI_KEY)
else:
# Might not be set if the user is using models not from OpenAI
print("OPENAI_KEY not set")
if OPENAI_ORG:
get_settings().set("OPENAI.ORG", OPENAI_ORG)
get_settings().set("GITHUB.USER_TOKEN", GITHUB_TOKEN)
get_settings().set("GITHUB.DEPLOYMENT_TYPE", "user")
enable_output = get_setting_or_env("GITHUB_ACTION_CONFIG.ENABLE_OUTPUT", True)
get_settings().set("GITHUB_ACTION_CONFIG.ENABLE_OUTPUT", enable_output)
# Load the event payload
try:
@ -96,11 +99,14 @@ async def run_action():
# invoke by default all three tools
if auto_describe is None or is_true(auto_describe):
get_settings().pr_description.final_update_message = False # No final update message when auto_describe is enabled
await PRDescription(pr_url).run()
if auto_review is None or is_true(auto_review):
await PRReviewer(pr_url).run()
if auto_improve is None or is_true(auto_improve):
await PRCodeSuggestions(pr_url).run()
else:
get_logger().info(f"Skipping action: {action}")
# Handle issue comment event
elif GITHUB_EVENT_NAME == "issue_comment" or GITHUB_EVENT_NAME == "pull_request_review_comment":

22
pr_agent/servers/github_app.py
View File

@ -86,6 +86,11 @@ async def handle_comments_on_pr(body: Dict[str, Any],
return {}
comment_body = body.get("comment", {}).get("body")
if comment_body and isinstance(comment_body, str) and not comment_body.lstrip().startswith("/"):
if '/ask' in comment_body and comment_body.strip().startswith('> ![image]'):
comment_body_split = comment_body.split('/ask')
comment_body = '/ask' + comment_body_split[1] +' \n' +comment_body_split[0].strip().lstrip('>')
get_logger().info(f"Reformatting comment_body so command is at the beginning: {comment_body}")
else:
get_logger().info("Ignoring comment not starting with /")
return {}
disable_eyes = False
@ -135,7 +140,7 @@ async def handle_new_pr_opened(body: Dict[str, Any],
if not (pull_request and api_url):
get_logger().info(f"Invalid PR event: {action=} {api_url=}")
return {}
if action in get_settings().github_app.handle_pr_actions: # ['opened', 'reopened', 'ready_for_review', 'review_requested']
if action in get_settings().github_app.handle_pr_actions: # ['opened', 'reopened', 'ready_for_review']
if get_identity_provider().verify_eligibility("github", sender_id, api_url) is not Eligibility.NOT_ELIGIBLE:
await _perform_auto_commands_github("pr_commands", agent, body, api_url, log_context)
else:
@ -224,19 +229,22 @@ def handle_closed_pr(body, event, action, log_context):
def get_log_context(body, event, action, build_number):
sender = ""
sender_id = ""
sender_type = ""
try:
sender = body.get("sender", {}).get("login")
sender_id = body.get("sender", {}).get("id")
sender_type = body.get("sender", {}).get("type")
repo = body.get("repository", {}).get("full_name", "")
git_org = body.get("organization", {}).get("login", "")
installation_id = body.get("installation", {}).get("id", "")
app_name = get_settings().get("CONFIG.APP_NAME", "Unknown")
log_context = {"action": action, "event": event, "sender": sender, "server_type": "github_app",
"request_id": uuid.uuid4().hex, "build_number": build_number, "app_name": app_name,
"repo": repo, "git_org": git_org}
"repo": repo, "git_org": git_org, "installation_id": installation_id}
except Exception as e:
get_logger().error("Failed to get log context", e)
log_context = {}
return log_context, sender, sender_id
return log_context, sender, sender_id, sender_type
async def handle_request(body: Dict[str, Any], event: str):
@ -251,7 +259,13 @@ async def handle_request(body: Dict[str, Any], event: str):
if not action:
return {}
agent = PRAgent()
log_context, sender, sender_id = get_log_context(body, event, action, build_number)
log_context, sender, sender_id, sender_type = get_log_context(body, event, action, build_number)
# logic to ignore PRs opened by bot
if get_settings().get("GITHUB_APP.IGNORE_BOT_PR", False) and sender_type == "Bot":
if 'pr-agent' not in sender:
get_logger().info(f"Ignoring PR from '{sender=}' because it is a bot")
return {}
# handle comments on PRs
if action == 'created':

193
pr_agent/servers/help.py
View File

@ -22,116 +22,22 @@ class HelpMessage:
@staticmethod
def get_review_usage_guide():
output ="**Overview:**\n"
output +="The `review` tool scans the PR code changes, and generates a PR review. The tool can be triggered [automatically](https://pr-agent-docs.codium.ai/usage-guide/automations_and_usage/#github-app-automatic-tools-when-a-new-pr-is-opened) every time a new PR is opened, or can be invoked manually by commenting on any PR.\n"
output +=("The `review` tool scans the PR code changes, and generates a PR review which includes several types of feedbacks, such as possible PR issues, security threats and relevant test in the PR. More feedbacks can be [added](https://pr-agent-docs.codium.ai/tools/review/#general-configurations) by configuring the tool.\n\n"
"The tool can be triggered [automatically](https://pr-agent-docs.codium.ai/usage-guide/automations_and_usage/#github-app-automatic-tools-when-a-new-pr-is-opened) every time a new PR is opened, or can be invoked manually by commenting on any PR.\n")
output +="""\
When commenting, to edit [configurations](https://github.com/Codium-ai/pr-agent/blob/main/pr_agent/settings/configuration.toml#L19) related to the review tool (`pr_reviewer` section), use the following template:
- When commenting, to edit [configurations](https://github.com/Codium-ai/pr-agent/blob/main/pr_agent/settings/configuration.toml#L23) related to the review tool (`pr_reviewer` section), use the following template:
```
/review --pr_reviewer.some_config1=... --pr_reviewer.some_config2=...
```
With a [configuration file](https://pr-agent-docs.codium.ai/usage-guide/configuration_options/), use the following template:
- With a [configuration file](https://pr-agent-docs.codium.ai/usage-guide/configuration_options/), use the following template:
```
[pr_reviewer]
some_config1=...
some_config2=...
```
"""
output +="\n\n<table>"
# extra instructions
output += "<tr><td><details> <summary><strong> Utilizing extra instructions</strong></summary><hr>\n\n"
output += '''\
The `review` tool can be configured with extra instructions, which can be used to guide the model to a feedback tailored to the needs of your project.
Be specific, clear, and concise in the instructions. With extra instructions, you are the prompter. Specify the relevant sub-tool, and the relevant aspects of the PR that you want to emphasize.
Examples for extra instructions:
```
[pr_reviewer] # /review #
extra_instructions="""
In the 'possible issues' section, emphasize the following:
- Does the code logic cover relevant edge cases?
- Is the code logic clear and easy to understand?
- Is the code logic efficient?
...
"""
```
Use triple quotes to write multi-line instructions. Use bullet points to make the instructions more readable.
'''
output += "\n\n</details></td></tr>\n\n"
# automation
output += "<tr><td><details> <summary><strong> How to enable\\disable automation</strong></summary><hr>\n\n"
output += """\
- When you first install PR-Agent app, the [default mode](https://pr-agent-docs.codium.ai/usage-guide/automations_and_usage/#github-app-automatic-tools-when-a-new-pr-is-opened) for the `review` tool is:
```
pr_commands = ["/review", ...]
```
meaning the `review` tool will run automatically on every PR, with the default configuration.
Edit this field to enable/disable the tool, or to change the used configurations
"""
output += "\n\n</details></td></tr>\n\n"
# # code feedback
# output += "<tr><td><details> <summary><strong> About the 'Code feedback' section</strong></summary><hr>\n\n"
# output+="""\
# The `review` tool provides several type of feedbacks, one of them is code suggestions.
# If you are interested **only** in the code suggestions, it is recommended to use the [`improve`](https://github.com/Codium-ai/pr-agent/blob/main/docs/IMPROVE.md) feature instead, since it dedicated only to code suggestions, and usually gives better results.
# Use the `review` tool if you want to get a more comprehensive feedback, which includes code suggestions as well.
# """
# output += "\n\n</details></td></tr>\n\n"
# auto-labels
output += "<tr><td><details> <summary><strong> Auto-labels</strong></summary><hr>\n\n"
output+="""\
The `review` tool can auto-generate two specific types of labels for a PR:
- a `possible security issue` label, that detects possible [security issues](https://github.com/Codium-ai/pr-agent/blob/tr/user_description/pr_agent/settings/pr_reviewer_prompts.toml#L136) (`enable_review_labels_security` flag)
- a `Review effort [1-5]: x` label, where x is the estimated effort to review the PR (`enable_review_labels_effort` flag)
"""
output += "\n\n</details></td></tr>\n\n"
# extra sub-tools
output += "<tr><td><details> <summary><strong> Extra sub-tools</strong></summary><hr>\n\n"
output += """\
The `review` tool provides a collection of possible feedbacks about a PR.
It is recommended to review the [possible options](https://pr-agent-docs.codium.ai/tools/review/#enabledisable-features), and choose the ones relevant for your use case.
Some of the feature that are disabled by default are quite useful, and should be considered for enabling. For example:
`require_score_review`, `require_soc2_ticket`, and more.
"""
output += "\n\n</details></td></tr>\n\n"
output += "<tr><td><details> <summary><strong> Auto-approve PRs</strong></summary><hr>\n\n"
output += '''\
By invoking:
```
/review auto_approve
```
The tool will automatically approve the PR, and add a comment with the approval.
To ensure safety, the auto-approval feature is disabled by default. To enable auto-approval, you need to actively set in a pre-defined configuration file the following:
```
[pr_reviewer]
enable_auto_approval = true
```
(this specific flag cannot be set with a command line argument, only in the configuration file, committed to the repository)
You can also enable auto-approval only if the PR meets certain requirements, such as that the `estimated_review_effort` is equal or below a certain threshold, by adjusting the flag:
```
[pr_reviewer]
maximal_review_effort = 5
```
'''
output += "\n\n</details></td></tr>\n\n"
# general
output += "\n\n<tr><td><details> <summary><strong> More PR-Agent commands</strong></summary><hr> \n\n"
output += HelpMessage.get_general_bot_help_text()
output += "\n\n</details></td></tr>\n\n"
output += "</table>"
output += f"\n\nSee the [review usage](https://pr-agent-docs.codium.ai/tools/review/) page for a comprehensive guide on using this tool.\n\n"
output += f"\n\nSee the review [usage page](https://pr-agent-docs.codium.ai/tools/review/) for a comprehensive guide on using this tool.\n\n"
return output
@ -162,10 +68,9 @@ some_config2=...
output += """\
- When you first install the app, the [default mode](https://pr-agent-docs.codium.ai/usage-guide/automations_and_usage/#github-app-automatic-tools-when-a-new-pr-is-opened) for the describe tool is:
```
pr_commands = ["/describe --pr_description.add_original_user_description=true"
"--pr_description.keep_original_user_title=true", ...]
pr_commands = ["/describe", ...]
```
meaning the `describe` tool will run automatically on every PR, will keep the original title, and will add the original user description above the generated description.
meaning the `describe` tool will run automatically on every PR.
- Markers are an alternative way to control the generated description, to give maximal control to the user. If you set:
```
@ -222,7 +127,7 @@ Be specific, clear, and concise in the instructions. With extra instructions, yo
Examples for extra instructions:
```
[pr_description]
extra_instructions="""
extra_instructions="""\
- The PR title should be in the format: '<PR type>: <title>'
- The title should be short and concise (up to 10 words)
- ...
@ -255,15 +160,16 @@ It can be invoked manually by commenting on any PR:
```
Note that the tool does not have "memory" of previous questions, and answers each question independently.
You can ask questions about the entire PR, about specific code lines, or about an image related to the PR code changes.
"""
output += "\n\n<table>"
# general
output += "\n\n<tr><td><details> <summary><strong> More PR-Agent commands</strong></summary><hr> \n\n"
output += HelpMessage.get_general_bot_help_text()
output += "\n\n</details></td></tr>\n\n"
output += "</table>"
# output += "\n\n<table>"
#
# # # general
# # output += "\n\n<tr><td><details> <summary><strong> More PR-Agent commands</strong></summary><hr> \n\n"
# # output += HelpMessage.get_general_bot_help_text()
# # output += "\n\n</details></td></tr>\n\n"
#
# output += "</table>"
output += f"\n\nSee the [ask usage](https://pr-agent-docs.codium.ai/tools/ask/) page for a comprehensive guide on using this tool.\n\n"
@ -273,16 +179,16 @@ Note that the tool does not have "memory" of previous questions, and answers eac
@staticmethod
def get_improve_usage_guide():
output = "**Overview:**\n"
output += "The `improve` tool scans the PR code changes, and automatically generates suggestions for improving the PR code. "
output += "The code suggestions tool, named `improve`, scans the PR code changes, and automatically generates code suggestions for improving the PR."
output += "The tool can be triggered [automatically](https://pr-agent-docs.codium.ai/usage-guide/automations_and_usage/#github-app-automatic-tools-when-a-new-pr-is-opened) every time a new PR is opened, or can be invoked manually by commenting on a PR.\n"
output += """\
When commenting, to edit [configurations](https://github.com/Codium-ai/pr-agent/blob/main/pr_agent/settings/configuration.toml#L69) related to the improve tool (`pr_code_suggestions` section), use the following template:
- When commenting, to edit [configurations](https://github.com/Codium-ai/pr-agent/blob/main/pr_agent/settings/configuration.toml#L78) related to the improve tool (`pr_code_suggestions` section), use the following template:
```
/improve --pr_code_suggestions.some_config1=... --pr_code_suggestions.some_config2=...
```
With a [configuration file](https://pr-agent-docs.codium.ai/usage-guide/configuration_options/), use the following template:
- With a [configuration file](https://pr-agent-docs.codium.ai/usage-guide/configuration_options/), use the following template:
```
[pr_code_suggestions]
@ -291,64 +197,7 @@ some_config2=...
```
"""
output += "\n\n<table>"
# automation
output += "<tr><td><details> <summary><strong> Enabling\\disabling automation </strong></summary><hr>\n\n"
output += """\
When you first install the app, the [default mode](https://pr-agent-docs.codium.ai/usage-guide/automations_and_usage/#github-app-automatic-tools-when-a-new-pr-is-opened) for the improve tool is:
```
pr_commands = ["/improve --pr_code_suggestions.summarize=true", ...]
```
meaning the `improve` tool will run automatically on every PR, with summarization enabled. Delete this line to disable the tool from running automatically.
"""
output += "\n\n</details></td></tr>\n\n"
# extra instructions
output += "<tr><td><details> <summary><strong> Utilizing extra instructions</strong></summary><hr>\n\n"
output += '''\
Extra instructions are very important for the `improve` tool, since they enable to guide the model to suggestions that are more relevant to the specific needs of the project.
Be specific, clear, and concise in the instructions. With extra instructions, you are the prompter. Specify relevant aspects that you want the model to focus on.
Examples for extra instructions:
```
[pr_code_suggestions] # /improve #
extra_instructions="""
Emphasize the following aspects:
- Does the code logic cover relevant edge cases?
- Is the code logic clear and easy to understand?
- Is the code logic efficient?
...
"""
```
Use triple quotes to write multi-line instructions. Use bullet points to make the instructions more readable.
'''
output += "\n\n</details></td></tr>\n\n"
# suggestions quality
output += "\n\n<tr><td><details> <summary><strong> A note on code suggestions quality</strong></summary><hr> \n\n"
output += """\
- While the current AI for code is getting better and better (GPT-4), it's not flawless. Not all the suggestions will be perfect, and a user should not accept all of them automatically.
- Suggestions are not meant to be simplistic. Instead, they aim to give deep feedback and raise questions, ideas and thoughts to the user, who can then use his judgment, experience, and understanding of the code base.
- Recommended to use the 'extra_instructions' field to guide the model to suggestions that are more relevant to the specific needs of the project, or use the [custom suggestions :gem:](https://pr-agent-docs.codium.ai/tools/custom_suggestions/) tool
- With large PRs, best quality will be obtained by using 'improve --extended' mode.
"""
output += "\n\n</details></td></tr>\n\n"\
# general
output += "\n\n<tr><td><details> <summary><strong> More PR-Agent commands</strong></summary><hr> \n\n"
output += HelpMessage.get_general_bot_help_text()
output += "\n\n</details></td></tr>\n\n"
output += "</table>"
output += f"\n\nSee the [improve usage](https://pr-agent-docs.codium.ai/tools/improve/) page for a more comprehensive guide on using this tool.\n\n"
output += f"\n\nSee the improve [usage page](https://pr-agent-docs.codium.ai/tools/improve/) for a comprehensive guide on using this tool.\n\n"
return output

3
pr_agent/settings/.secrets_template.toml
View File

@ -29,6 +29,9 @@ key = "" # Optional, uncomment if you want to use Cohere. Acquire through https:
[replicate]
key = "" # Optional, uncomment if you want to use Replicate. Acquire through https://replicate.com/
[groq]
key = "" # Acquire through https://console.groq.com/keys
[huggingface]
key = "" # Optional, uncomment if you want to use Huggingface Inference API. Acquire through https://huggingface.co/docs/api-inference/quicktour
api_base = "" # the base url for your huggingface inference endpoint

54
pr_agent/settings/configuration.toml
View File

@ -1,7 +1,7 @@
[config]
model="gpt-4" # "gpt-4-0125-preview"
model_turbo="gpt-4-0125-preview"
fallback_models=["gpt-3.5-turbo-16k"]
model="gpt-4-turbo-2024-04-09"
model_turbo="gpt-4o"
fallback_models=["gpt-4-0125-preview"]
git_provider="github"
publish_output=true
publish_output_progress=true
@ -19,13 +19,15 @@ secret_provider="google_cloud_storage"
cli_mode=false
ai_disclaimer_title="" # Pro feature, title for a collapsible disclaimer to AI outputs
ai_disclaimer="" # Pro feature, full text for the AI disclaimer
output_relevant_configurations=false
large_patch_policy = "clip" # "clip", "skip"
[pr_reviewer] # /review #
# enable/disable features
require_focused_review=false
require_score_review=false
require_tests_review=true
require_estimate_effort_to_review=true
require_can_be_split_review=false
# soc2
require_soc2_ticket=false
soc2_ticket_prompt="Does the PR description include a link to ticket in a project management system (e.g., Jira, Asana, Trello, etc.) ?"
@ -44,7 +46,7 @@ enable_review_labels_effort=true
require_all_thresholds_for_incremental_review=false
minimal_commits_for_incremental_review=0
minimal_minutes_for_incremental_review=0
enable_help_text=true # Determines whether to include help text in the PR review. Enabled by default.
enable_help_text=false # Determines whether to include help text in the PR review. Enabled by default.
# auto approval
enable_auto_approval=false
maximal_review_effort=5
@ -52,15 +54,17 @@ maximal_review_effort=5
[pr_description] # /describe #
publish_labels=true
publish_description_as_comment=false
add_original_user_description=true
keep_original_user_title=true
generate_ai_title=false
use_bullet_points=true
extra_instructions = ""
enable_pr_type=true
final_update_message = true
enable_help_text=false
enable_help_comment=true
# describe as comment
publish_description_as_comment=false
publish_description_as_comment_persistent=true
## changes walkthrough section
enable_semantic_files_types=true
collapsible_file_list='adaptive' # true, false, 'adaptive'
@ -72,19 +76,23 @@ include_generated_by_header=true
#custom_labels = ['Bug fix', 'Tests', 'Bug fix with tests', 'Enhancement', 'Documentation', 'Other']
[pr_questions] # /ask #
enable_help_text=true
enable_help_text=false
[pr_code_suggestions] # /improve #
max_context_tokens=8000
num_code_suggestions=4
summarize = true
commitable_code_suggestions = false
extra_instructions = ""
rank_suggestions = false
enable_help_text=true
enable_help_text=false
persistent_comment=false
# suggestions scoring
self_reflect_on_suggestions=true
suggestions_score_threshold=0 # [0-10]. highly recommend not to set this value above 8, since above it may clip highly relevant suggestions
# params for '/improve --extended' mode
auto_extended_mode=true
num_code_suggestions_per_chunk=5
num_code_suggestions_per_chunk=4
max_number_of_calls = 3
parallel_calls = true
rank_extended_suggestions = false
@ -107,8 +115,13 @@ num_tests=3 # number of tests to generate. max 5.
avoid_mocks=true # if true, the generated tests will prefer to use real objects instead of mocks
file = "" # in case there are several components with the same name, you can specify the relevant file
class_name = "" # in case there are several methods with the same name in the same file, you can specify the relevant class name
enable_help_text=true
enable_help_text=false
[pr_improve_component] # /improve_component #
num_code_suggestions=4
extra_instructions = ""
file = "" # in case there are several components with the same name, you can specify the relevant file
class_name = ""
[checks] # /checks (pro feature) #
enable_auto_checks_feedback=true
@ -132,6 +145,7 @@ try_fix_invalid_inline_comments = true
# auto_review = true # set as env var in .github/workflows/pr-agent.yaml
# auto_describe = true # set as env var in .github/workflows/pr-agent.yaml
# auto_improve = true # set as env var in .github/workflows/pr-agent.yaml
# enable_output = true # set as env var in .github/workflows/pr-agent.yaml
[github_app]
# these toggles allows running the github app from custom deployments
@ -139,9 +153,9 @@ override_deployment_type = true
# settings for "pull_request" event
handle_pr_actions = ['opened', 'reopened', 'ready_for_review']
pr_commands = [
"/describe --pr_description.add_original_user_description=true --pr_description.keep_original_user_title=true",
"/describe --pr_description.final_update_message=false",
"/review --pr_reviewer.num_code_suggestions=0",
"/improve --pr_code_suggestions.summarize=true",
"/improve",
]
# settings for "pull_request" event with "synchronize" action - used to detect and handle push triggers for new commits
handle_push_trigger = false
@ -151,23 +165,24 @@ push_trigger_wait_for_initial_review = true
push_trigger_pending_tasks_backlog = true
push_trigger_pending_tasks_ttl = 300
push_commands = [
"/describe --pr_description.add_original_user_description=true --pr_description.keep_original_user_title=true",
"/describe",
"/review --pr_reviewer.num_code_suggestions=0",
]
ignore_pr_title = []
ignore_bot_pr = true
[gitlab]
url = "https://gitlab.com" # URL to the gitlab service
pr_commands = [
"/describe --pr_description.add_original_user_description=true --pr_description.keep_original_user_title=true",
"/describe",
"/review --pr_reviewer.num_code_suggestions=0",
"/improve --pr_code_suggestions.summarize=true",
"/improve",
]
[bitbucket_app]
pr_commands = [
"/review --pr_reviewer.num_code_suggestions=0",
"/improve --pr_code_suggestions.summarize=false",
"/improve --pr_code_suggestions.commitable_code_suggestions=true",
]
@ -192,7 +207,8 @@ pr_commands = [
url = ""
[litellm]
#use_client = false
# use_client = false
# drop_params = false
[pr_similar_issue]
skip_comments = false

1
pr_agent/settings/language_extensions.toml
View File

@ -44,6 +44,7 @@ default = [
'ss',
'svg',
'tar',
'tgz',
'tsv',
'ttf',
'war',

43
pr_agent/settings/pr_code_suggestions_prompts.toml
View File

@ -1,8 +1,9 @@
[pr_code_suggestions_prompt]
system="""You are PR-Reviewer, a language model that specializes in suggesting code improvements for a Pull Request (PR).
Your task is to provide meaningful and actionable code suggestions, to improve the new code presented in a PR diff (lines starting with '+').
system="""You are PR-Reviewer, a language model that specializes in suggesting ways to improve for a Pull Request (PR) code.
Your task is to provide meaningful and actionable code suggestions, to improve the new code presented in a PR diff.
Example for the PR Diff format:
The format we will use to present the PR code diff:
======
## file: 'src/file1.py'
@ -26,22 +27,26 @@ __old hunk__
## file: 'src/file2.py'
...
======
- In this format, we separated each hunk of code to '__new hunk__' and '__old hunk__' sections. The '__new hunk__' section contains the new code of the chunk, and the '__old hunk__' section contains the old code that was removed.
- Code lines are prefixed symbols ('+', '-', ' '). The '+' symbol indicates new code added in the PR, the '-' symbol indicates code removed in the PR, and the ' ' symbol indicates unchanged code.
- We also added line numbers for the '__new hunk__' sections, to help you refer to the code lines in your suggestions. These line numbers are not part of the actual code, and are only used for reference.
Specific instructions:
Specific instructions for generating code suggestions:
- Provide up to {{ num_code_suggestions }} code suggestions. The suggestions should be diverse and insightful.
- The suggestions should refer only to code from the '__new hunk__' sections, and focus on new lines of code (lines starting with '+').
- Prioritize suggestions that address major problems, issues and bugs in the PR code. As a second priority, suggestions should focus on enhancement, best practice, performance, maintainability, and other aspects.
- The suggestions should focus on ways to improve the new code in the PR, meaning focusing on lines from '__new hunk__' sections, starting with '+'. Use the '__old hunk__' sections to understand the context of the code changes.
- Prioritize suggestions that address possible issues, major problems, and bugs in the PR code.
- Don't suggest to add docstring, type hints, or comments, or to remove unused imports.
- Suggestions should not repeat code already present in the '__new hunk__' sections.
- Provide the exact line numbers range (inclusive) for each suggestion.
- Provide the exact line numbers range (inclusive) for each suggestion. Use the line numbers from the '__new hunk__' sections.
- When quoting variables or names from the code, use backticks (`) instead of single quote (').
- Take into account that you are reviewing a PR code diff, and that the entire codebase is not available for you as context. Hence, avoid suggestions that might conflict with unseen parts of the codebase.
{%- if extra_instructions %}
Extra instructions from the user:
Extra instructions from the user, that should be taken into account with high priority:
======
{{ extra_instructions }}
======
@ -54,17 +59,12 @@ class CodeSuggestion(BaseModel):
relevant_file: str = Field(description="the relevant file full path")
language: str = Field(description="the code language of the relevant file")
suggestion_content: str = Field(description="an actionable suggestion for meaningfully improving the new code introduced in the PR")
{%- if summarize_mode %}
existing_code: str = Field(description="a short code snippet from a '__new hunk__' section to illustrate the relevant existing code. Don't show the line numbers.")
improved_code: str = Field(description="a short code snippet to illustrate the improved code, after applying the suggestion.")
one_sentence_summary:str = Field(description="a short summary of the suggestion action, in a single sentence. Focus on the 'what'. Be general, and avoid method or variable names.")
{%- else %}
existing_code: str = Field(description="a code snippet, demonstrating the relevant code lines from a '__new hunk__' section. It must be contiguous, correctly formatted and indented, and without line numbers")
improved_code: str = Field(description="a new code snippet, that can be used to replace the relevant lines in '__new hunk__' code. Replacement suggestions should be complete, correctly formatted and indented, and without line numbers")
{%- endif %}
existing_code: str = Field(description="a short code snippet, demonstrating the relevant code lines from a '__new hunk__' section. It must be without line numbers. Use abbreviations if needed")
improved_code: str = Field(description="a new code snippet, that can be used to replace the relevant 'existing_code' lines in '__new hunk__' code after applying the suggestion")
one_sentence_summary: str = Field(description="a short summary of the suggestion action, in a single sentence. Focus on the 'what'. Be general, and avoid method or variable names.")
relevant_lines_start: int = Field(description="The relevant line number, from a '__new hunk__' section, where the suggestion starts (inclusive). Should be derived from the hunk line numbers, and correspond to the 'existing code' snippet above")
relevant_lines_end: int = Field(description="The relevant line number, from a '__new hunk__' section, where the suggestion ends (inclusive). Should be derived from the hunk line numbers, and correspond to the 'existing code' snippet above")
label: str = Field(description="a single label for the suggestion, to help the user understand the suggestion type. For example: 'security', 'bug', 'performance', 'enhancement', 'possible issue', 'best practice', 'maintainability', etc. Other labels are also allowed")
label: str = Field(description="a single label for the suggestion, to help the user understand the suggestion type. For example: 'security', 'possible bug', 'possible issue', 'performance', 'enhancement', 'best practice', 'maintainability', etc. Other labels are also allowed")
class PRCodeSuggestions(BaseModel):
code_suggestions: List[CodeSuggestion]
@ -80,7 +80,6 @@ code_suggestions:
python
suggestion_content: |
...
{%- if summarize_mode %}
existing_code: |
...
improved_code: |
@ -89,14 +88,6 @@ code_suggestions:
...
relevant_lines_start: 12
relevant_lines_end: 13
{%- else %}
existing_code: |
...
relevant_lines_start: 12
relevant_lines_end: 13
improved_code: |
...
{%- endif %}
label: |
...
```

88
pr_agent/settings/pr_code_suggestions_reflect_prompts.toml Normal file
View File

@ -0,0 +1,88 @@
[pr_code_suggestions_reflect_prompt]
system="""You are a language model that specializes in reviewing and evaluating suggestions for a Pull Request (PR) code.
Your input is a PR code, and a list of code suggestions that were generated for the PR.
Your goal is to inspect, review and score the suggestsions.
Be aware - the suggestions may not always be correct or accurate, and you should evaluate them in relation to the actual PR code diff presented. Sometimes the suggestion may ignore parts of the actual code diff, and in that case, you should give it a score of 0.
Specific instructions:
- Carefully review both the suggestion content, and the related PR code diff. Mistakes in the suggestions can occur. Make sure the suggestions are correct, and properly derived from the PR code diff.
- In addition to the exact code lines mentioned in each suggestion, review the code around them, to ensure that the suggestions are contextually accurate.
- Also check that the 'existing_code' and 'improved_code' fields correctly reflect the suggested changes.
- Make sure the suggestions focus on new code introduced in the PR, and not on existing code that was not changed.
- High scores (8 to 10) should be given to correct suggestions that address major bugs and issues, or security concerns. Lower scores (3 to 7) should be for correct suggestions addressing minor issues, code style, code readability, maintainability, etc. Don't give high scores to suggestions that are not crucial, and bring only small improvement or optimization.
- Order the feedback the same way the suggestions are ordered in the input.
The format that is used to present the PR code diff is as follows:
======
## file: 'src/file1.py'
@@ ... @@ def func1():
__new hunk__
12 code line1 that remained unchanged in the PR
13 +new hunk code line2 added in the PR
14 code line3 that remained unchanged in the PR
__old hunk__
code line1 that remained unchanged in the PR
-old hunk code line2 that was removed in the PR
code line3 that remained unchanged in the PR
@@ ... @@ def func2():
__new hunk__
...
__old hunk__
...
## file: 'src/file2.py'
...
======
- In this format, we separated each hunk of code to '__new hunk__' and '__old hunk__' sections. The '__new hunk__' section contains the new code of the chunk, and the '__old hunk__' section contains the old code that was removed.
- Code lines are prefixed symbols ('+', '-', ' '). The '+' symbol indicates new code added in the PR, the '-' symbol indicates code removed in the PR, and the ' ' symbol indicates unchanged code.
- We also added line numbers for the '__new hunk__' sections, to help you refer to the code lines in your suggestions. These line numbers are not part of the actual code, and are only used for reference.
The output must be a YAML object equivalent to type $PRCodeSuggestionsFeedback, according to the following Pydantic definitions:
=====
class CodeSuggestionFeedback(BaseModel):
suggestion_summary: str = Field(description="repeated from the input")
relevant_file: str = Field(description="repeated from the input")
suggestion_score: int = Field(description="The actual output - the score of the suggestion, from 0 to 10. Give 0 if the suggestion is plain wrong. Otherwise, give a score from 1 to 10 (inclusive), where 1 is the lowest and 10 is the highest.")
why: str = Field(description="Short and concise explanation of why the suggestion received the score (one to two sentences).")
class PRCodeSuggestionsFeedback(BaseModel):
code_suggestions: List[CodeSuggestionFeedback]
=====
Example output:
```yaml
code_suggestions:
- suggestion_content: |
Use a more descriptive variable name here
relevant_file: "src/file1.py"
suggestion_score: 6
why: |
The variable name 't' is not descriptive enough
```
Each YAML output MUST be after a newline, indented, with block scalar indicator ('|').
"""
user="""You are given a Pull Request (PR) code diff:
======
{{ diff|trim }}
======
And here is a list of corresponding {{ num_code_suggestions }} code suggestions to improve this Pull Request code:
======
{{ suggestion_str|trim }}
======
Response (should be a valid YAML, and nothing else):
```yaml
"""

68
pr_agent/settings/pr_evaluate_prompt_response.toml Normal file
View File

@ -0,0 +1,68 @@
[pr_evaluate_prompt]
prompt="""\
You are the PR-task-evaluator, a language model that compares and ranks the quality of two responses provided in response to a lengthy task regarding a Pull Request (PR) code diff.
The task to be evaluated is:
***** Start of Task *****
{{pr_task|trim}}
***** End of Task *****
Response 1 to the task is:
***** Start of Response 1 *****
{{pr_response1|trim}}
***** End of Response 1 *****
Response 2 to the task is:
***** Start of Response 2 *****
{{pr_response2|trim}}
***** End of Response 2 *****
Guidelines to evaluate the responses:
- Thoroughly read the 'Task' part. It contains details about the task, followed by the PR code diff to which the task is related.
- Thoroughly read 'Response1' and 'Response2' parts. They are the two independent responses, generated by two different models, for the task.
After that, rank each response. Criterions to rank each response:
- How well does the response follow the specific task instructions and requirements?
- How well does the response analyze and understand the PR code diff?
- How well will a person perceive it as a good response that correctly addresses the task?
- How well does the reponse prioritize key feedback, related to the task instructions, that a human reader seeing that feedback would also consider as important?
- Don't neccessarily rank higher a response that is longer. A shorter response might be better if it is more concise, and still addresses the task better.
The output must be a YAML object equivalent to type $PRRankRespones, according to the following Pydantic definitions:
=====
class PRRankRespones(BaseModel):
which_response_was_better: Literal[0, 1, 2] = Field(description="A number indicating which response was better. 0 means both responses are equally good.")
why: str = Field(description="In a short and concise manner, explain why the chosen response is better than the other. Be specific and give examples if relevant.")
score_response1: int = Field(description="A score between 1 and 10, indicating the quality of the response1, based on the criterions mentioned in the prompt.")
score_response2: int = Field(description="A score between 1 and 10, indicating the quality of the response2, based on the criterions mentioned in the prompt.")
=====
Example output:
```yaml
which_response_was_better: "X"
why: "Response X is better because it is more practical, and addresses the task requirements better since ..."
score_response1: ...
score_response2: ...
```
Response (should be a valid YAML, and nothing else):
```yaml
"""

24
pr_agent/settings/pr_reviewer_prompts.toml
View File

@ -49,6 +49,12 @@ Extra instructions from the user:
The output must be a YAML object equivalent to type $PRReview, according to the following Pydantic definitions:
=====
{%- if require_can_be_split_review %}
class SubPR(BaseModel):
relevant_files: List[str] = Field(description="The relevant files of the sub-PR")
title: str = Field(description="Short and concise title for an independent and meaningful sub-PR, composed only from the relevant files")
{%- endif %}
class Review(BaseModel):
{%- if require_estimate_effort_to_review %}
estimated_effort_to_review_[1-5]: str = Field(description="Estimate, on a scale of 1-5 (inclusive), the time and effort required to review this PR by an experienced and knowledgeable developer. 1 means short and easy review , 5 means long and hard review. Take into account the size, complexity, quality, and the needed changes of the PR code diff. Explain your answer in a short and concise manner.")
@ -61,12 +67,12 @@ class Review(BaseModel):
{%- endif %}
{%- if question_str %}
insights_from_user_answers: str = Field(description="shortly summarize the insights you gained from the user's answers to the questions")
{%- endif %}
{%- if require_focused %}
focused_pr: str = Field(description="Is this a focused PR, in the sense that all the PR code diff changes are united under a single focused theme ? If the theme is too broad, or the PR code diff changes are too scattered, then the PR is not focused. Explain your answer shortly.")
{%- endif %}
possible_issues: str = Field(description="Does this PR code introduce clear issues, bugs, or major performance concerns? If there are no apparent issues, respond with 'No'. If there are any issues, describe them briefly. Use bullet points if more than one issue. Be specific, and provide examples if possible. Start each bullet point with a short specific header, such as: "- Possible Bug: ...", etc.")
security_concerns: str = Field(description="does this PR code introduce possible vulnerabilities such as exposure of sensitive information (e.g., API keys, secrets, passwords), or security concerns like SQL injection, XSS, CSRF, and others ? Answer 'No' if there are no possible issues. If there are security concerns or issues, start your answer with a short header, such as: 'Sensitive information exposure: ...', 'SQL injection: ...' etc. Explain your answer. Be specific and give examples if possible")
{%- if require_can_be_split_review %}
can_be_split: List[SubPR] = Field(min_items=0, max_items=3, description="Can this PR, which contains {{ num_pr_files }} changed files in total, be divided into smaller sub-PRs with distinct tasks that can be reviewed and merged independently, regardless of the order ? Make sure that the sub-PRs are indeed independent, with no code dependencies between them, and that each sub-PR represent a meaningfull independent task. Output an empty list if the PR code does not needd to be split.")
{%- endif %}
{%- if num_code_suggestions > 0 %}
class CodeSuggestion(BaseModel):
@ -100,14 +106,18 @@ review:
{%- endif %}
relevant_tests: |
No
{%- if require_focused %}
focused_pr: |
no, because ...
{%- endif %}
possible_issues: |
No
security_concerns: |
No
{%- if require_can_be_split_review %}
can_be_split: |
- relevant_files:
- ...
- ...
title: ...
- ...
{%- endif %}
{%- if num_code_suggestions > 0 %}
code_feedback
- relevant_file: |

2
pr_agent/tools/pr_add_docs.py
View File

@ -26,6 +26,8 @@ class PRAddDocs:
)
self.ai_handler = ai_handler()
self.ai_handler.main_pr_language = self.main_language
self.patches_diff = None
self.prediction = None
self.cli_mode = cli_mode

208
pr_agent/tools/pr_code_suggestions.py
View File

@ -9,7 +9,7 @@ from pr_agent.algo.ai_handlers.base_ai_handler import BaseAiHandler
from pr_agent.algo.ai_handlers.litellm_ai_handler import LiteLLMAIHandler
from pr_agent.algo.pr_processing import get_pr_diff, get_pr_multi_diffs, retry_with_fallback_models
from pr_agent.algo.token_handler import TokenHandler
from pr_agent.algo.utils import load_yaml, replace_code_tags, ModelType
from pr_agent.algo.utils import load_yaml, replace_code_tags, ModelType, show_relevant_configurations
from pr_agent.config_loader import get_settings
from pr_agent.git_providers import get_git_provider
from pr_agent.git_providers.git_provider import get_main_pr_language
@ -46,6 +46,7 @@ class PRCodeSuggestions:
num_code_suggestions = get_settings().pr_code_suggestions.num_code_suggestions
self.ai_handler = ai_handler()
self.ai_handler.main_pr_language = self.main_language
self.patches_diff = None
self.prediction = None
self.cli_mode = cli_mode
@ -56,7 +57,6 @@ class PRCodeSuggestions:
"language": self.main_language,
"diff": "", # empty diff for initial calculation
"num_code_suggestions": num_code_suggestions,
"summarize_mode": get_settings().pr_code_suggestions.summarize,
"extra_instructions": get_settings().pr_code_suggestions.extra_instructions,
"commit_messages_str": self.git_provider.get_commit_messages(),
}
@ -75,21 +75,22 @@ class PRCodeSuggestions:
relevant_configs = {'pr_code_suggestions': dict(get_settings().pr_code_suggestions),
'config': dict(get_settings().config)}
get_logger().debug("Relevant configs", artifacts=relevant_configs)
if get_settings().config.publish_output:
if get_settings().config.publish_output and get_settings().config.publish_output_progress:
if self.git_provider.is_supported("gfm_markdown"):
self.progress_response = self.git_provider.publish_comment(self.progress)
else:
self.git_provider.publish_comment("Preparing suggestions...", is_temporary=True)
if not self.is_extended:
await retry_with_fallback_models(self._prepare_prediction, ModelType.TURBO)
data = self._prepare_pr_code_suggestions()
data = await retry_with_fallback_models(self._prepare_prediction)
else:
data = await retry_with_fallback_models(self._prepare_prediction_extended, ModelType.TURBO)
data = await retry_with_fallback_models(self._prepare_prediction_extended)
if not data:
data = {"code_suggestions": []}
if (not data) or (not 'code_suggestions' in data) or (not data['code_suggestions']):
if data is None or 'code_suggestions' not in data or not data['code_suggestions']:
get_logger().error('No code suggestions found for PR.')
pr_body = "## PR Code Suggestions\n\nNo code suggestions found for PR."
pr_body = "## PR Code Suggestions ✨\n\nNo code suggestions found for PR."
get_logger().debug(f"PR output", artifact=pr_body)
if self.progress_response:
self.git_provider.edit_comment(self.progress_response, body=pr_body)
@ -104,7 +105,8 @@ class PRCodeSuggestions:
if get_settings().config.publish_output:
self.git_provider.remove_initial_comment()
if get_settings().pr_code_suggestions.summarize and self.git_provider.is_supported("gfm_markdown"):
if ((not get_settings().pr_code_suggestions.commitable_code_suggestions) and
self.git_provider.is_supported("gfm_markdown")):
# generate summarized suggestions
pr_body = self.generate_summarized_suggestions(data)
@ -112,10 +114,25 @@ class PRCodeSuggestions:
# add usage guide
if get_settings().pr_code_suggestions.enable_help_text:
pr_body += "<hr>\n\n<details> <summary><strong>✨ Improve tool usage guide:</strong></summary><hr> \n\n"
pr_body += "<hr>\n\n<details> <summary><strong>💡 Tool usage guide:</strong></summary><hr> \n\n"
pr_body += HelpMessage.get_improve_usage_guide()
pr_body += "\n</details>\n"
# Output the relevant configurations if enabled
if get_settings().get('config', {}).get('output_relevant_configurations', False):
pr_body += show_relevant_configurations(relevant_section='pr_code_suggestions')
if get_settings().pr_code_suggestions.persistent_comment:
final_update_message = False
self.git_provider.publish_persistent_comment(pr_body,
initial_header="## PR Code Suggestions ✨",
update_header=True,
name="suggestions",
final_update_message=final_update_message, )
if self.progress_response:
self.progress_response.delete()
else:
if self.progress_response:
self.git_provider.edit_comment(self.progress_response, body=pr_body)
else:
@ -136,7 +153,7 @@ class PRCodeSuggestions:
except Exception as e:
pass
async def _prepare_prediction(self, model: str):
async def _prepare_prediction(self, model: str) -> dict:
self.patches_diff = get_pr_diff(self.git_provider,
self.token_handler,
model,
@ -150,7 +167,10 @@ class PRCodeSuggestions:
get_logger().error(f"Error getting PR diff")
self.prediction = None
async def _get_prediction(self, model: str, patches_diff: str):
data = self.prediction
return data
async def _get_prediction(self, model: str, patches_diff: str) -> dict:
variables = copy.deepcopy(self.vars)
variables["diff"] = patches_diff # update diff
environment = Environment(undefined=StrictUndefined)
@ -159,7 +179,34 @@ class PRCodeSuggestions:
response, finish_reason = await self.ai_handler.chat_completion(model=model, temperature=0.2,
system=system_prompt, user=user_prompt)
return response
# load suggestions from the AI response
data = self._prepare_pr_code_suggestions(response)
# self-reflect on suggestions
if get_settings().pr_code_suggestions.self_reflect_on_suggestions:
model = get_settings().config.model_turbo # use turbo model for self-reflection, since it is an easier task
response_reflect = await self.self_reflect_on_suggestions(data["code_suggestions"], patches_diff, model=model)
if response_reflect:
response_reflect_yaml = load_yaml(response_reflect)
code_suggestions_feedback = response_reflect_yaml["code_suggestions"]
if len(code_suggestions_feedback) == len(data["code_suggestions"]):
for i, suggestion in enumerate(data["code_suggestions"]):
try:
suggestion["score"] = code_suggestions_feedback[i]["suggestion_score"]
suggestion["score_why"] = code_suggestions_feedback[i]["why"]
except Exception as e: #
get_logger().error(f"Error processing suggestion score {i}",
artifact={"suggestion": suggestion,
"code_suggestions_feedback": code_suggestions_feedback[i]})
suggestion["score"] = 7
suggestion["score_why"] = ""
else:
# get_logger().error(f"Could not self-reflect on suggestions. using default score 7")
for i, suggestion in enumerate(data["code_suggestions"]):
suggestion["score"] = 7
suggestion["score_why"] = ""
return data
@staticmethod
def _truncate_if_needed(suggestion):
@ -173,19 +220,19 @@ class PRCodeSuggestions:
f"characters to {max_code_suggestion_length} characters")
return suggestion
def _prepare_pr_code_suggestions(self) -> Dict:
review = self.prediction.strip()
data = load_yaml(review,
def _prepare_pr_code_suggestions(self, predictions: str) -> Dict:
data = load_yaml(predictions.strip(),
keys_fix_yaml=["relevant_file", "suggestion_content", "existing_code", "improved_code"])
if isinstance(data, list):
data = {'code_suggestions': data}
# remove invalid suggestions
# remove or edit invalid suggestions
suggestion_list = []
one_sentence_summary_list = []
for i, suggestion in enumerate(data['code_suggestions']):
if get_settings().pr_code_suggestions.summarize:
if not suggestion or 'one_sentence_summary' not in suggestion or 'label' not in suggestion or 'relevant_file' not in suggestion:
try:
if (not suggestion or 'one_sentence_summary' not in suggestion or
'label' not in suggestion or 'relevant_file' not in suggestion):
get_logger().debug(f"Skipping suggestion {i + 1}, because it is invalid: {suggestion}")
continue
@ -193,15 +240,26 @@ class PRCodeSuggestions:
get_logger().debug(f"Skipping suggestion {i + 1}, because it is a duplicate: {suggestion}")
continue
if ('existing_code' in suggestion) and ('improved_code' in suggestion) and (
suggestion['existing_code'] != suggestion['improved_code']):
if 'const' in suggestion['suggestion_content'] and 'instead' in suggestion['suggestion_content'] and 'let' in suggestion['suggestion_content']:
get_logger().debug(f"Skipping suggestion {i + 1}, because it uses 'const instead let': {suggestion}")
continue
if ('existing_code' in suggestion) and ('improved_code' in suggestion):
if suggestion['existing_code'] == suggestion['improved_code']:
get_logger().debug(
f"edited improved suggestion {i + 1}, because equal to existing code: {suggestion['existing_code']}")
if get_settings().pr_code_suggestions.commitable_code_suggestions:
suggestion['improved_code'] = "" # we need 'existing_code' to locate the code in the PR
else:
suggestion['existing_code'] = ""
suggestion = self._truncate_if_needed(suggestion)
if get_settings().pr_code_suggestions.summarize:
one_sentence_summary_list.append(suggestion['one_sentence_summary'])
suggestion_list.append(suggestion)
else:
get_logger().debug(
f"Skipping suggestion {i + 1}, because existing code is equal to improved code {suggestion['existing_code']}")
get_logger().info(
f"Skipping suggestion {i + 1}, because it does not contain 'existing_code' or 'improved_code': {suggestion}")
except Exception as e:
get_logger().error(f"Error processing suggestion {i + 1}: {suggestion}, error: {e}")
data['code_suggestions'] = suggestion_list
return data
@ -228,6 +286,9 @@ class PRCodeSuggestions:
if new_code_snippet:
new_code_snippet = self.dedent_code(relevant_file, relevant_lines_start, new_code_snippet)
if d.get('score'):
body = f"**Suggestion:** {content} [{label}, importance: {d.get('score')}]\n```suggestion\n" + new_code_snippet + "\n```"
else:
body = f"**Suggestion:** {content} [{label}]\n```suggestion\n" + new_code_snippet + "\n```"
code_suggestions.append({'body': body, 'relevant_file': relevant_file,
'relevant_lines_start': relevant_lines_start,
@ -277,7 +338,8 @@ class PRCodeSuggestions:
self.patches_diff_list = get_pr_multi_diffs(self.git_provider, self.token_handler, model,
max_calls=get_settings().pr_code_suggestions.max_number_of_calls)
if self.patches_diff_list:
get_logger().debug(f"PR diff", artifact=self.patches_diff_list)
get_logger().info(f"Number of PR chunk calls: {len(self.patches_diff_list)}")
get_logger().debug(f"PR diff:", artifact=self.patches_diff_list)
# parallelize calls to AI:
if get_settings().pr_code_suggestions.parallel_calls:
@ -290,14 +352,24 @@ class PRCodeSuggestions:
prediction = await self._get_prediction(model, patches_diff)
prediction_list.append(prediction)
data = {}
for prediction in prediction_list:
self.prediction = prediction
data_per_chunk = self._prepare_pr_code_suggestions()
if "code_suggestions" in data:
data["code_suggestions"].extend(data_per_chunk["code_suggestions"])
data = {"code_suggestions": []}
for j, predictions in enumerate(prediction_list): # each call adds an element to the list
if "code_suggestions" in predictions:
score_threshold = max(1, get_settings().pr_code_suggestions.suggestions_score_threshold)
for i, prediction in enumerate(predictions["code_suggestions"]):
try:
if get_settings().pr_code_suggestions.self_reflect_on_suggestions:
score = int(prediction["score"])
if score >= score_threshold:
data["code_suggestions"].append(prediction)
else:
data.update(data_per_chunk)
get_logger().info(
f"Removing suggestions {i} from call {j}, because score is {score}, and score_threshold is {score_threshold}",
artifact=prediction)
else:
data["code_suggestions"].append(prediction)
except Exception as e:
get_logger().error(f"Error getting PR diff for suggestion {i} in call {j}, error: {e}")
self.data = data
else:
get_logger().error(f"Error getting PR diff")
@ -363,7 +435,7 @@ class PRCodeSuggestions:
def generate_summarized_suggestions(self, data: Dict) -> str:
try:
pr_body = "## PR Code Suggestions\n\n"
pr_body = "## PR Code Suggestions ✨\n\n"
if len(data.get('code_suggestions', [])) == 0:
pr_body += "No suggestions found to improve this PR."
@ -375,13 +447,16 @@ class PRCodeSuggestions:
for ext in extensions:
extension_to_language[ext] = language
pr_body = "## PR Code Suggestions\n\n"
pr_body = "## PR Code Suggestions ✨\n\n"
pr_body += "<table>"
header = f"Suggestions"
delta = 76
header = f"Suggestion"
delta = 66
header += "&nbsp; " * delta
pr_body += f"""<thead><tr><td>Category</td><td align=left>{header}</td></tr></thead>"""
if get_settings().pr_code_suggestions.self_reflect_on_suggestions:
pr_body += f"""<thead><tr><td>Category</td><td align=left>{header}</td><td align=center>Score</td></tr>"""
else:
pr_body += f"""<thead><tr><td>Category</td><td align=left>{header}</td></tr>"""
pr_body += """<tbody>"""
suggestions_labels = dict()
# add all suggestions related to each label
@ -391,13 +466,17 @@ class PRCodeSuggestions:
suggestions_labels[label] = []
suggestions_labels[label].append(suggestion)
# sort suggestions_labels by the suggestion with the highest score
if get_settings().pr_code_suggestions.self_reflect_on_suggestions:
suggestions_labels = dict(sorted(suggestions_labels.items(), key=lambda x: max([s['score'] for s in x[1]]), reverse=True))
# sort the suggestions inside each label group by score
for label, suggestions in suggestions_labels.items():
suggestions_labels[label] = sorted(suggestions, key=lambda x: x['score'], reverse=True)
for label, suggestions in suggestions_labels.items():
num_suggestions=len(suggestions)
# pr_body += f"""<tr><td><strong>{label}</strong></td>"""
pr_body += f"""<tr><td rowspan={num_suggestions}><strong>{label.capitalize()}</strong></td>\n"""
# pr_body += f"""<td>"""
# pr_body += f"""<details><summary>{len(suggestions)} suggestions</summary>"""
# pr_body += f"""<table>"""
for i, suggestion in enumerate(suggestions):
relevant_file = suggestion['relevant_file'].strip()
@ -408,8 +487,12 @@ class PRCodeSuggestions:
range_str = f"[{relevant_lines_start}]"
else:
range_str = f"[{relevant_lines_start}-{relevant_lines_end}]"
try:
code_snippet_link = self.git_provider.get_line_link(relevant_file, relevant_lines_start,
relevant_lines_end)
except:
code_snippet_link = ""
# add html table for each suggestion
suggestion_content = suggestion['suggestion_content'].rstrip().rstrip()
@ -430,12 +513,11 @@ class PRCodeSuggestions:
pr_body += f"""<td>\n\n"""
else:
pr_body += f"""<tr><td>\n\n"""
suggestion_summary = suggestion['one_sentence_summary'].strip()
suggestion_summary = suggestion['one_sentence_summary'].strip().rstrip('.')
if '`' in suggestion_summary:
suggestion_summary = replace_code_tags(suggestion_summary)
# suggestion_summary = suggestion_summary + max((77-len(suggestion_summary)), 0)*"&nbsp;"
pr_body += f"""\n\n<details><summary>{suggestion_summary}</summary>\n\n___\n\n"""
pr_body += f"""\n\n<details><summary>{suggestion_summary}</summary>\n\n___\n\n"""
pr_body += f"""
**{suggestion_content}**
@ -443,14 +525,50 @@ class PRCodeSuggestions:
{example_code}
"""
if get_settings().pr_code_suggestions.self_reflect_on_suggestions:
pr_body +=f"\n\n<details><summary><b>Suggestion importance[1-10]: {suggestion['score']}</b></summary>\n\n"
pr_body += f"Why: {suggestion['score_why']}\n\n"
pr_body += f"</details>"
pr_body += f"</details>"
# # add another column for 'score'
if get_settings().pr_code_suggestions.self_reflect_on_suggestions:
pr_body += f"</td><td align=center>{suggestion['score']}\n\n"
pr_body += f"</td></tr>"
# pr_body += "</details>"
pr_body += """</td></tr>"""
# pr_body += """</td></tr>"""
pr_body += """</tr></tbody></table>"""
return pr_body
except Exception as e:
get_logger().info(f"Failed to publish summarized code suggestions, error: {e}")
return ""
async def self_reflect_on_suggestions(self, suggestion_list: List, patches_diff: str, model: str) -> str:
if not suggestion_list:
return ""
try:
suggestion_str = ""
for i, suggestion in enumerate(suggestion_list):
suggestion_str += f"suggestion {i + 1}: " + str(suggestion) + '\n\n'
variables = {'suggestion_list': suggestion_list,
'suggestion_str': suggestion_str,
"diff": patches_diff,
'num_code_suggestions': len(suggestion_list)}
environment = Environment(undefined=StrictUndefined)
system_prompt_reflect = environment.from_string(get_settings().pr_code_suggestions_reflect_prompt.system).render(
variables)
user_prompt_reflect = environment.from_string(get_settings().pr_code_suggestions_reflect_prompt.user).render(variables)
with get_logger().contextualize(command="self_reflect_on_suggestions"):
response_reflect, finish_reason_reflect = await self.ai_handler.chat_completion(model=model,
system=system_prompt_reflect,
user=user_prompt_reflect)
except Exception as e:
get_logger().info(f"Could not reflect on suggestions, error: {e}")
return ""
return response_reflect

28
pr_agent/tools/pr_description.py
View File

@ -9,7 +9,7 @@ from pr_agent.algo.ai_handlers.base_ai_handler import BaseAiHandler
from pr_agent.algo.ai_handlers.litellm_ai_handler import LiteLLMAIHandler
from pr_agent.algo.pr_processing import get_pr_diff, retry_with_fallback_models
from pr_agent.algo.token_handler import TokenHandler
from pr_agent.algo.utils import load_yaml, set_custom_labels, get_user_labels, ModelType
from pr_agent.algo.utils import load_yaml, set_custom_labels, get_user_labels, ModelType, show_relevant_configurations
from pr_agent.config_loader import get_settings
from pr_agent.git_providers import get_git_provider
from pr_agent.git_providers.git_provider import get_main_pr_language
@ -41,6 +41,7 @@ class PRDescription:
# Initialize the AI handler
self.ai_handler = ai_handler()
self.ai_handler.main_pr_language = self.main_pr_language
# Initialize the variables dictionary
@ -81,7 +82,7 @@ class PRDescription:
if get_settings().config.publish_output:
self.git_provider.publish_comment("Preparing PR description...", is_temporary=True)
await retry_with_fallback_models(self._prepare_prediction, ModelType.TURBO) # turbo model because larger context
await retry_with_fallback_models(self._prepare_prediction, ModelType.TURBO)
if self.prediction:
self._prepare_data()
@ -112,9 +113,13 @@ class PRDescription:
pr_body += HelpMessage.get_describe_usage_guide()
pr_body += "\n</details>\n"
elif get_settings().pr_description.enable_help_comment:
pr_body += "\n\n___\n\n> ✨ **PR-Agent usage**:"
pr_body += "\n\n___\n\n> 💡 **PR-Agent usage**:"
pr_body += "\n>Comment `/help` on the PR to get a list of all available PR-Agent tools and their descriptions\n\n"
# Output the relevant configurations if enabled
if get_settings().get('config', {}).get('output_relevant_configurations', False):
pr_body += show_relevant_configurations(relevant_section='pr_description')
if get_settings().config.publish_output:
# publish labels
if get_settings().pr_description.publish_labels and self.git_provider.is_supported("get_labels"):
@ -131,6 +136,13 @@ class PRDescription:
# publish description
if get_settings().pr_description.publish_description_as_comment:
full_markdown_description = f"## Title\n\n{pr_title}\n\n___\n{pr_body}"
if get_settings().pr_description.publish_description_as_comment_persistent:
self.git_provider.publish_persistent_comment(full_markdown_description,
initial_header="## Title",
update_header=True,
name="describe",
final_update_message=False, )
else:
self.git_provider.publish_comment(full_markdown_description)
else:
self.git_provider.publish_description(pr_title, pr_body)
@ -293,7 +305,7 @@ class PRDescription:
# Remove the 'PR Title' key from the dictionary
ai_title = self.data.pop('title', self.vars["title"])
if get_settings().pr_description.keep_original_user_title:
if (not get_settings().pr_description.generate_ai_title):
# Assign the original PR title to the 'title' variable
title = self.vars["title"]
else:
@ -309,7 +321,11 @@ class PRDescription:
value = self.file_label_dict
else:
key_publish = key.rstrip(':').replace("_", " ").capitalize()
pr_body += f"## **{key_publish}**\n"
if key_publish== "Type":
key_publish = "PR Type"
# elif key_publish == "Description":
# key_publish = "PR Description"
pr_body += f"### **{key_publish}**\n"
if 'walkthrough' in key.lower():
if self.git_provider.is_supported("gfm_markdown"):
pr_body += "<details> <summary>files:</summary>\n\n"
@ -321,7 +337,7 @@ class PRDescription:
pr_body += "</details>\n"
elif 'pr_files' in key.lower():
changes_walkthrough, pr_file_changes = self.process_pr_files_prediction(changes_walkthrough, value)
changes_walkthrough = f"## **Changes walkthrough**\n{changes_walkthrough}"
changes_walkthrough = f"### **Changes walkthrough** 📝\n{changes_walkthrough}"
else:
# if the value is a list, join its items by comma
if isinstance(value, list):

1
pr_agent/tools/pr_generate_labels.py
View File

@ -35,6 +35,7 @@ class PRGenerateLabels:
# Initialize the AI handler
self.ai_handler = ai_handler()
self.ai_handler.main_pr_language = self.main_pr_language
# Initialize the variables dictionary
self.vars = {

57
pr_agent/tools/pr_help_message.py
View File

@ -18,8 +18,8 @@ class PRHelpMessage:
relevant_configs = {'pr_help': dict(get_settings().pr_help),
'config': dict(get_settings().config)}
get_logger().debug("Relevant configs", artifacts=relevant_configs)
pr_comment = "## PR Agent Walkthrough\n\n"
pr_comment += "🤖 Welcome to the PR Agent, an AI-powered tool for automated pull request analysis, feedback, suggestions and more."""
pr_comment = "## PR Agent Walkthrough 🤖\n\n"
pr_comment += "Welcome to the PR Agent, an AI-powered tool for automated pull request analysis, feedback, suggestions and more."""
pr_comment += "\n\nHere is a list of tools you can use to interact with the PR Agent:\n"
base_path = "https://pr-agent-docs.codium.ai/tools"
@ -27,51 +27,58 @@ class PRHelpMessage:
tool_names.append(f"[DESCRIBE]({base_path}/describe/)")
tool_names.append(f"[REVIEW]({base_path}/review/)")
tool_names.append(f"[IMPROVE]({base_path}/improve/)")
tool_names.append(f"[ANALYZE]({base_path}/analyze/) 💎")
tool_names.append(f"[UPDATE CHANGELOG]({base_path}/update_changelog/)")
tool_names.append(f"[ADD DOCUMENTATION]({base_path}/documentation/) 💎")
tool_names.append(f"[ASK]({base_path}/ask/)")
tool_names.append(f"[GENERATE CUSTOM LABELS]({base_path}/custom_labels/)")
tool_names.append(f"[ADD DOCS]({base_path}/documentation/) 💎")
tool_names.append(f"[TEST]({base_path}/test/) 💎")
tool_names.append(f"[IMPROVE COMPONENT]({base_path}/improve_component/) 💎")
tool_names.append(f"[ANALYZE]({base_path}/analyze/) 💎")
tool_names.append(f"[ASK]({base_path}/ask/)")
tool_names.append(f"[GENERATE CUSTOM LABELS]({base_path}/custom_labels/) 💎")
tool_names.append(f"[CI FEEDBACK]({base_path}/ci_feedback/) 💎")
tool_names.append(f"[CUSTOM SUGGESTIONS]({base_path}/custom_suggestions/) 💎")
tool_names.append(f"[CUSTOM PROMPT]({base_path}/custom_prompt/) 💎")
tool_names.append(f"[SIMILAR ISSUE]({base_path}/similar_issues/)")
descriptions = []
descriptions.append("Generates PR description - title, type, summary, code walkthrough and labels")
descriptions.append("Adjustable feedback about the PR, possible issues, security concerns, review effort and more")
descriptions.append("Code suggestions for improving the PR.")
descriptions.append("Identifies code components that changed in the PR, and enables to interactively generate tests, docs, and code suggestions for each component.")
descriptions.append("Automatically updates the changelog.")
descriptions.append("Generates documentation to methods/functions/classes that changed in the PR.")
descriptions.append("Answering free-text questions about the PR.")
descriptions.append("Code suggestions for improving the PR")
descriptions.append("Automatically updates the changelog")
descriptions.append("Generates documentation to methods/functions/classes that changed in the PR")
descriptions.append("Generates unit tests for a specific component, based on the PR code change")
descriptions.append("Code suggestions for a specific component that changed in the PR")
descriptions.append("Identifies code components that changed in the PR, and enables to interactively generate tests, docs, and code suggestions for each component")
descriptions.append("Answering free-text questions about the PR")
descriptions.append("Generates custom labels for the PR, based on specific guidelines defined by the user")
descriptions.append("Generates unit tests for a specific component, based on the PR code change.")
descriptions.append("Generates feedback and analysis for a failed CI job.")
descriptions.append("Generates custom suggestions for improving the PR code, based on specific guidelines defined by the user.")
descriptions.append("Automatically retrieves and presents similar issues.")
descriptions.append("Generates feedback and analysis for a failed CI job")
descriptions.append("Generates custom suggestions for improving the PR code, derived only from a specific guidelines prompt defined by the user")
descriptions.append("Automatically retrieves and presents similar issues")
commands =[]
commands.append("`/describe`")
commands.append("`/review`")
commands.append("`/improve`")
commands.append("`/analyze`")
commands.append("`/update_changelog`")
commands.append("`/add_docs`")
commands.append("`/test`")
commands.append("`/improve_component`")
commands.append("`/analyze`")
commands.append("`/ask`")
commands.append("`/generate_labels`")
commands.append("`/test`")
commands.append("`/checks`")
commands.append("`/custom_suggestions`")
commands.append("`/custom_prompt`")
commands.append("`/similar_issue`")
checkbox_list = []
checkbox_list.append(" - [ ] Run <!-- /describe -->")
checkbox_list.append(" - [ ] Run <!-- /review -->")
checkbox_list.append(" - [ ] Run <!-- /improve -->")
checkbox_list.append(" - [ ] Run <!-- /analyze -->")
checkbox_list.append(" - [ ] Run <!-- /update_changelog -->")
checkbox_list.append(" - [ ] Run <!-- /add_docs -->")
checkbox_list.append(" - [ ] Run <!-- /test -->")
checkbox_list.append(" - [ ] Run <!-- /improve_component -->")
checkbox_list.append(" - [ ] Run <!-- /analyze -->")
checkbox_list.append("[*]")
checkbox_list.append("[*]")
checkbox_list.append("[*]")
checkbox_list.append("[*]")
checkbox_list.append("[*]")
@ -79,17 +86,17 @@ class PRHelpMessage:
checkbox_list.append("[*]")
checkbox_list.append("[*]")
if isinstance(self.git_provider, GithubProvider):
pr_comment += f"<table><tr align='center'><th align='center'>Tool</th><th align='center'>Description</th><th align='center'>Invoke Interactively :gem:</th></tr>"
if isinstance(self.git_provider, GithubProvider) and not get_settings().config.get('disable_checkboxes', False):
pr_comment += f"<table><tr align='left'><th align='left'>Tool</th><th align='left'>Description</th><th align='left'>Trigger Interactively :gem:</th></tr>"
for i in range(len(tool_names)):
pr_comment += f"\n<tr><td align='center'>\n\n<strong>{tool_names[i]}</strong></td>\n<td>{descriptions[i]}</td>\n<td>\n\n{checkbox_list[i]}\n</td></tr>"
pr_comment += f"\n<tr><td align='left'>\n\n<strong>{tool_names[i]}</strong></td>\n<td>{descriptions[i]}</td>\n<td>\n\n{checkbox_list[i]}\n</td></tr>"
pr_comment += "</table>\n\n"
pr_comment += f"""\n\n(1) Note that each tool be [triggered automatically](https://github.com/Codium-ai/pr-agent/blob/main/Usage.md#github-app-automatic-tools-for-pr-actions) when a new PR is opened, or called manually by [commenting on a PR](https://github.com/Codium-ai/pr-agent/blob/main/Usage.md#online-usage)."""
pr_comment += f"""\n\n(2) Tools marked with [*] require additional parameters to be passed. For example, to invoke the `/ask` tool, you need to comment on a PR: `/ask "<question content>"`. See the relevant documentation for each tool for more details."""
else:
pr_comment += f"<table><tr align='center'><th align='center'>Tool</th><th align='left'>Command</th><th align='left'>Description</th></tr>"
pr_comment += f"<table><tr align='left'><th align='left'>Tool</th><th align='left'>Command</th><th align='left'>Description</th></tr>"
for i in range(len(tool_names)):
pr_comment += f"\n<tr><td align='center'>\n\n<strong>{tool_names[i]}</strong></td><td>{commands[i]}</td><td>{descriptions[i]}</td></tr>"
pr_comment += f"\n<tr><td align='left'>\n\n<strong>{tool_names[i]}</strong></td><td>{commands[i]}</td><td>{descriptions[i]}</td></tr>"
pr_comment += "</table>\n\n"
pr_comment += f"""\n\nNote that each tool be [invoked automatically](https://pr-agent-docs.codium.ai/usage-guide/automations_and_usage/) when a new PR is opened, or called manually by [commenting on a PR](https://pr-agent-docs.codium.ai/usage-guide/automations_and_usage/#online-usage)."""
if get_settings().config.publish_output:

2
pr_agent/tools/pr_information_from_user.py
View File

@ -21,6 +21,8 @@ class PRInformationFromUser:
self.git_provider.get_languages(), self.git_provider.get_files()
)
self.ai_handler = ai_handler()
self.ai_handler.main_pr_language = self.main_pr_language
self.vars = {
"title": self.git_provider.pr.title,
"branch": self.git_provider.get_pr_branch(),

5
pr_agent/tools/pr_line_questions.py
View File

@ -22,8 +22,11 @@ class PR_LineQuestions:
def __init__(self, pr_url: str, args=None, ai_handler: partial[BaseAiHandler,] = LiteLLMAIHandler):
self.question_str = self.parse_args(args)
self.git_provider = get_git_provider()(pr_url)
self.main_pr_language = get_main_pr_language(
self.git_provider.get_languages(), self.git_provider.get_files()
)
self.ai_handler = ai_handler()
self.ai_handler.main_pr_language = self.main_pr_language
self.vars = {
"title": self.git_provider.pr.title,

39
pr_agent/tools/pr_questions.py
View File

@ -7,6 +7,7 @@ from pr_agent.algo.ai_handlers.base_ai_handler import BaseAiHandler
from pr_agent.algo.ai_handlers.litellm_ai_handler import LiteLLMAIHandler
from pr_agent.algo.pr_processing import get_pr_diff, retry_with_fallback_models
from pr_agent.algo.token_handler import TokenHandler
from pr_agent.algo.utils import ModelType
from pr_agent.config_loader import get_settings
from pr_agent.git_providers import get_git_provider
from pr_agent.git_providers.git_provider import get_main_pr_language
@ -17,11 +18,14 @@ from pr_agent.servers.help import HelpMessage
class PRQuestions:
def __init__(self, pr_url: str, args=None, ai_handler: partial[BaseAiHandler,] = LiteLLMAIHandler):
question_str = self.parse_args(args)
self.pr_url = pr_url
self.git_provider = get_git_provider()(pr_url)
self.main_pr_language = get_main_pr_language(
self.git_provider.get_languages(), self.git_provider.get_files()
)
self.ai_handler = ai_handler()
self.ai_handler.main_pr_language = self.main_pr_language
self.question_str = question_str
self.vars = {
"title": self.git_provider.pr.title,
@ -47,19 +51,25 @@ class PRQuestions:
return question_str
async def run(self):
get_logger().info('Answering a PR question...')
get_logger().info(f'Answering a PR question about the PR {self.pr_url} ')
relevant_configs = {'pr_questions': dict(get_settings().pr_questions),
'config': dict(get_settings().config)}
get_logger().debug("Relevant configs", artifacts=relevant_configs)
if get_settings().config.publish_output:
self.git_provider.publish_comment("Preparing answer...", is_temporary=True)
await retry_with_fallback_models(self._prepare_prediction)
# identify image
img_path = self.idenfity_image_in_comment()
if img_path:
get_logger().debug(f"Image path identified", artifact=img_path)
await retry_with_fallback_models(self._prepare_prediction, model_type=ModelType.TURBO)
pr_comment = self._prepare_pr_answer()
get_logger().debug(f"PR output", artifact=pr_comment)
if self.git_provider.is_supported("gfm_markdown") and get_settings().pr_questions.enable_help_text:
pr_comment += "<hr>\n\n<details> <summary><strong>✨ Ask tool usage guide:</strong></summary><hr> \n\n"
pr_comment += "<hr>\n\n<details> <summary><strong>💡 Tool usage guide:</strong></summary><hr> \n\n"
pr_comment += HelpMessage.get_ask_usage_guide()
pr_comment += "\n</details>\n"
@ -68,6 +78,19 @@ class PRQuestions:
self.git_provider.remove_initial_comment()
return ""
def idenfity_image_in_comment(self):
img_path = ''
if '![image]' in self.question_str:
# assuming structure:
# /ask question ... > ![image](img_path)
img_path = self.question_str.split('![image]')[1].strip().strip('()')
self.vars['img_path'] = img_path
elif 'https://' in self.question_str and ('.png' in self.question_str or 'jpg' in self.question_str): # direct image link
# include https:// in the image path
img_path = 'https://' + self.question_str.split('https://')[1]
self.vars['img_path'] = img_path
return img_path
async def _prepare_prediction(self, model: str):
self.patches_diff = get_pr_diff(self.git_provider, self.token_handler, model)
if self.patches_diff:
@ -83,11 +106,17 @@ class PRQuestions:
environment = Environment(undefined=StrictUndefined)
system_prompt = environment.from_string(get_settings().pr_questions_prompt.system).render(variables)
user_prompt = environment.from_string(get_settings().pr_questions_prompt.user).render(variables)
if 'img_path' in variables:
img_path = self.vars['img_path']
response, finish_reason = await self.ai_handler.chat_completion(model=model, temperature=0.2,
system=system_prompt, user=user_prompt,
img_path=img_path)
else:
response, finish_reason = await self.ai_handler.chat_completion(model=model, temperature=0.2,
system=system_prompt, user=user_prompt)
return response
def _prepare_pr_answer(self) -> str:
answer_str = f"Question: {self.question_str}\n\n"
answer_str += f"Answer:\n{self.prediction.strip()}\n\n"
answer_str = f"### **Ask**❓\n{self.question_str}\n\n"
answer_str += f"### **Answer:**\n{self.prediction.strip()}\n\n"
return answer_str

24
pr_agent/tools/pr_reviewer.py
View File

@ -8,7 +8,8 @@ from pr_agent.algo.ai_handlers.base_ai_handler import BaseAiHandler
from pr_agent.algo.ai_handlers.litellm_ai_handler import LiteLLMAIHandler
from pr_agent.algo.pr_processing import get_pr_diff, retry_with_fallback_models
from pr_agent.algo.token_handler import TokenHandler
from pr_agent.algo.utils import convert_to_markdown, load_yaml, ModelType
from pr_agent.algo.utils import convert_to_markdown, github_action_output, load_yaml, ModelType, \
show_relevant_configurations
from pr_agent.config_loader import get_settings
from pr_agent.git_providers import get_git_provider
from pr_agent.git_providers.git_provider import IncrementalPR, get_main_pr_language
@ -46,6 +47,8 @@ class PRReviewer:
if self.is_answer and not self.git_provider.is_supported("get_issue_comments"):
raise Exception(f"Answer mode is not supported for {get_settings().config.git_provider} for now")
self.ai_handler = ai_handler()
self.ai_handler.main_pr_language = self.main_language
self.patches_diff = None
self.prediction = None
@ -56,10 +59,11 @@ class PRReviewer:
"description": self.git_provider.get_pr_description(),
"language": self.main_language,
"diff": "", # empty diff for initial calculation
"num_pr_files": self.git_provider.get_num_of_files(),
"require_score": get_settings().pr_reviewer.require_score_review,
"require_tests": get_settings().pr_reviewer.require_tests_review,
"require_focused": get_settings().pr_reviewer.require_focused_review,
"require_estimate_effort_to_review": get_settings().pr_reviewer.require_estimate_effort_to_review,
'require_can_be_split_review': get_settings().pr_reviewer.require_can_be_split_review,
'num_code_suggestions': get_settings().pr_reviewer.num_code_suggestions,
'question_str': question_str,
'answer_str': answer_str,
@ -121,7 +125,7 @@ class PRReviewer:
if get_settings().config.publish_output:
self.git_provider.publish_comment("Preparing review...", is_temporary=True)
await retry_with_fallback_models(self._prepare_prediction, model_type=ModelType.TURBO)
await retry_with_fallback_models(self._prepare_prediction)
if not self.prediction:
self.git_provider.remove_initial_comment()
return None
@ -134,7 +138,7 @@ class PRReviewer:
if get_settings().pr_reviewer.persistent_comment and not self.incremental.is_incremental:
final_update_message = get_settings().pr_reviewer.final_update_message
self.git_provider.publish_persistent_comment(pr_review,
initial_header="## PR Review",
initial_header="## PR Review 🔍",
update_header=True,
final_update_message=final_update_message, )
else:
@ -189,6 +193,7 @@ class PRReviewer:
data = load_yaml(self.prediction.strip(),
keys_fix_yaml=["estimated_effort_to_review_[1-5]:", "security_concerns:", "possible_issues:",
"relevant_file:", "relevant_line:", "suggestion:"])
github_action_output(data, 'review')
if 'code_feedback' in data:
code_feedback = data['code_feedback']
@ -230,10 +235,14 @@ class PRReviewer:
# Add help text if gfm_markdown is supported
if self.git_provider.is_supported("gfm_markdown") and get_settings().pr_reviewer.enable_help_text:
markdown_text += "<hr>\n\n<details> <summary><strong>✨ Review tool usage guide:</strong></summary><hr> \n\n"
markdown_text += "<hr>\n\n<details> <summary><strong>💡 Tool usage guide:</strong></summary><hr> \n\n"
markdown_text += HelpMessage.get_review_usage_guide()
markdown_text += "\n</details>\n"
# Output the relevant configurations if enabled
if get_settings().get('config', {}).get('output_relevant_configurations', False):
markdown_text += show_relevant_configurations(relevant_section='pr_reviewer')
# Add custom labels from the review prediction (effort, security)
self.set_review_labels(data)
@ -354,6 +363,9 @@ class PRReviewer:
return True
def set_review_labels(self, data):
if not get_settings().config.publish_output:
return
if (get_settings().pr_reviewer.enable_review_labels_security or
get_settings().pr_reviewer.enable_review_labels_effort):
try:
@ -370,6 +382,8 @@ class PRReviewer:
review_labels.append('Possible security concern')
current_labels = self.git_provider.get_pr_labels(update=True)
if not current_labels:
current_labels = []
get_logger().debug(f"Current labels:\n{current_labels}")
if current_labels:
current_labels_filtered = [label for label in current_labels if

12
pr_agent/tools/pr_similar_issue.py
View File

@ -3,9 +3,6 @@ from enum import Enum
from typing import List
import openai
import pandas as pd
import pinecone
from pinecone_datasets import Dataset, DatasetMetadata
from pydantic import BaseModel, Field
from pr_agent.algo import MAX_TOKENS
@ -36,6 +33,12 @@ class PRSimilarIssue:
index_name = self.index_name = "codium-ai-pr-agent-issues"
if get_settings().pr_similar_issue.vectordb == "pinecone":
try:
import pinecone
from pinecone_datasets import Dataset, DatasetMetadata
import pandas as pd
except:
raise Exception("Please install 'pinecone' and 'pinecone_datasets' to use pinecone as vectordb")
# assuming pinecone api key and environment are set in secrets file
try:
api_key = get_settings().pinecone.api_key
@ -107,7 +110,10 @@ class PRSimilarIssue:
get_logger().info('No new issues to update')
elif get_settings().pr_similar_issue.vectordb == "lancedb":
try:
import lancedb # import lancedb only if needed
except:
raise Exception("Please install lancedb to use lancedb as vectordb")
self.db = lancedb.connect(get_settings().lancedb.uri)
self.table = None

14
pr_agent/tools/pr_update_changelog.py
View File

@ -8,7 +8,7 @@ from pr_agent.algo.ai_handlers.base_ai_handler import BaseAiHandler
from pr_agent.algo.ai_handlers.litellm_ai_handler import LiteLLMAIHandler
from pr_agent.algo.pr_processing import get_pr_diff, retry_with_fallback_models
from pr_agent.algo.token_handler import TokenHandler
from pr_agent.algo.utils import ModelType
from pr_agent.algo.utils import ModelType, show_relevant_configurations
from pr_agent.config_loader import get_settings
from pr_agent.git_providers import get_git_provider, GithubProvider
from pr_agent.git_providers.git_provider import get_main_pr_language
@ -26,7 +26,10 @@ class PRUpdateChangelog:
)
self.commit_changelog = get_settings().pr_update_changelog.push_changelog_changes
self._get_changlog_file() # self.changelog_file_str
self.ai_handler = ai_handler()
self.ai_handler.main_pr_language = self.main_language
self.patches_diff = None
self.prediction = None
self.cli_mode = cli_mode
@ -71,6 +74,11 @@ class PRUpdateChangelog:
await retry_with_fallback_models(self._prepare_prediction, model_type=ModelType.TURBO)
new_file_content, answer = self._prepare_changelog_update()
# Output the relevant configurations if enabled
if get_settings().get('config', {}).get('output_relevant_configurations', False):
answer += show_relevant_configurations(relevant_section='pr_update_changelog')
get_logger().debug(f"PR output", artifact=answer)
if get_settings().config.publish_output:
@ -78,7 +86,7 @@ class PRUpdateChangelog:
if self.commit_changelog:
self._push_changelog_update(new_file_content, answer)
else:
self.git_provider.publish_comment(f"**Changelog updates:**\n\n{answer}")
self.git_provider.publish_comment(f"**Changelog updates:** 🔄\n\n{answer}")
async def _prepare_prediction(self, model: str):
self.patches_diff = get_pr_diff(self.git_provider, self.token_handler, model)
@ -138,7 +146,7 @@ class PRUpdateChangelog:
self.git_provider.pr.create_review(commit=last_commit_id, comments=[d])
except Exception:
# we can't create a review for some reason, let's just publish a comment
self.git_provider.publish_comment(f"**Changelog updates:**\n\n{answer}")
self.git_provider.publish_comment(f"**Changelog updates: 🔄**\n\n{answer}")
def _get_default_changelog(self):
example_changelog = \

28
pyproject.toml
View File

@ -3,44 +3,44 @@ requires = ["setuptools>=61.0"]
build-backend = "setuptools.build_meta"
[project]
name = "pr_agent"
version = "0.0.1"
name = "pr-agent"
version = "0.2.2"
authors = [{name= "CodiumAI", email = "tal.r@codium.ai"}]
authors = [
{name = "Itamar Friedman", email = "itamar.f@codium.ai"},
]
maintainers = [
{name = "Ori Kotek", email = "ori.k@codium.ai"},
{name = "Tal Ridnik", email = "tal.r@codium.ai"},
{name = "Ori Kotek", email = "ori.k@codium.ai"},
{name = "Hussam Lawen", email = "hussam.l@codium.ai"},
{name = "Sagi Medina", email = "sagi.m@codium.ai"}
]
description = "CodiumAI PR-Agent is an open-source tool to automatically analyze a pull request and provide several types of feedback"
description = "CodiumAI PR-Agent aims to help efficiently review and handle pull requests, by providing AI feedbacks and suggestions."
readme = "README.md"
requires-python = ">=3.10"
keywords = ["ai", "tool", "developer", "review", "agent"]
license = {file = "LICENSE", name = "Apache 2.0 License"}
keywords = ["AI", "Agents", "Pull Request", "Automation", "Code Review"]
license = {name = "Apache 2.0", file = "LICENSE"}
classifiers = [
"Development Status :: 3 - Alpha",
"Intended Audience :: Developers",
"Operating System :: Independent",
"Programming Language :: Python :: 3",
]
dynamic = ["dependencies"]
[tool.setuptools.dynamic]
dependencies = {file = ["requirements.txt"]}
[project.urls]
"Homepage" = "https://github.com/Codium-ai/pr-agent"
"Documentation" = "https://pr-agent-docs.codium.ai/"
[tool.setuptools]
include-package-data = false
include-package-data = true
license-files = ["LICENSE"]
[tool.setuptools.packages.find]
where = ["."]
include = ["pr_agent"]
include = ["pr_agent*"] # include pr_agent and any sub-packages it finds under it.
[project.scripts]
pr-agent = "pr_agent.cli:run"

2
requirements-dev.txt
View File

@ -1 +1,3 @@
pytest==7.4.0
poetry
twine

14
requirements.txt
View File

@ -9,21 +9,23 @@ GitPython==3.1.32
google-cloud-aiplatform==1.35.0
google-cloud-storage==2.10.0
Jinja2==3.1.2
litellm==1.29.1
litellm==1.31.10
loguru==0.7.2
msrest==0.7.1
openai==1.13.3
pinecone-client
pinecone-datasets @ git+https://github.com/mrT23/pinecone-datasets.git@main
lancedb==0.5.1
pytest==7.4.0
PyGithub==1.59.*
PyYAML==6.0.1
python-gitlab==3.15.0
retry==0.9.2
starlette-context==0.3.6
tiktoken==0.5.2
tiktoken==0.7.0
ujson==5.8.0
uvicorn==0.22.0
tenacity==8.2.3
# langchain==0.0.349 # uncomment this to support language LangChainOpenAIHandler
# Uncomment the following lines to enable the 'similar issue' tool
# pinecone-client
# pinecone-datasets @ git+https://github.com/mrT23/pinecone-datasets.git@main
# lancedb==0.5.1
# uncomment this to support language LangChainOpenAIHandler
# langchain==0.0.349

2
tests/unittest/test_clip_tokens.py
View File

@ -15,5 +15,5 @@ class TestClipTokens:
max_tokens = 10
result = clip_tokens(text, max_tokens)
expected_results = 'line1\nline2\nline3\nli...(truncated)'
expected_results = 'line1\nline2\nline3\n\n...(truncated)'
assert result == expected_results

2
tests/unittest/test_convert_to_markdown.py
View File

@ -52,7 +52,7 @@ class TestConvertToMarkdown:
'suggestion': "Consider raising an exception or logging a warning when 'pr_url' attribute is not found. This can help in debugging issues related to the absence of 'pr_url' in instances where it's expected. [important]\n",
'relevant_line': '[return ""](https://github.com/Codium-ai/pr-agent-pro/pull/102/files#diff-52d45f12b836f77ed1aef86e972e65404634ea4e2a6083fb71a9b0f9bb9e062fR199)'}]}
expected_output = '## PR Review\n\n<table>\n<tr>\n<tr><td> ⏱️&nbsp;<strong>Estimated&nbsp;effort&nbsp;to&nbsp;review [1-5]</strong></td><td>\n\n1, because the changes are minimal and straightforward, focusing on a single functionality addition.\n\n\n</td></tr>\n<tr><td> 🧪&nbsp;<strong>Relevant tests</strong></td><td>\n\nNo\n\n\n</td></tr>\n<tr><td> 🔍&nbsp;<strong>Possible issues</strong></td><td>\n\nNo\n\n</td></tr>\n<tr><td> 🔒&nbsp;<strong>Security concerns</strong></td><td>\n\nNo\n\n</td></tr>\n</table>\n\n\n<details><summary> <strong>Code feedback:</strong></summary>\n\n<hr><table><tr><td>relevant file</td><td>pr_agent/git_providers/git_provider.py\n</td></tr><tr><td>suggestion &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</td><td>\n\n<strong>\n\nConsider raising an exception or logging a warning when \'pr_url\' attribute is not found. This can help in debugging issues related to the absence of \'pr_url\' in instances where it\'s expected. [important]\n\n</strong>\n</td></tr><tr><td>relevant line</td><td><a href=\'https://github.com/Codium-ai/pr-agent-pro/pull/102/files#diff-52d45f12b836f77ed1aef86e972e65404634ea4e2a6083fb71a9b0f9bb9e062fR199\'>return ""</a></td></tr></table><hr>\n\n</details>'
expected_output = '## PR Review 🔍\n\n<table>\n<tr>\n<tr><td> ⏱️&nbsp;<strong>Estimated&nbsp;effort&nbsp;to&nbsp;review [1-5]</strong></td><td>\n\n1, because the changes are minimal and straightforward, focusing on a single functionality addition.\n\n\n</td></tr>\n<tr><td> 🧪&nbsp;<strong>Relevant tests</strong></td><td>\n\nNo\n\n\n</td></tr>\n<tr><td> ⚡&nbsp;<strong>Possible issues</strong></td><td>\n\nNo\n\n</td></tr>\n<tr><td> 🔒&nbsp;<strong>Security concerns</strong></td><td>\n\nNo\n\n</td></tr>\n</table>\n\n\n<details><summary> <strong>Code feedback:</strong></summary>\n\n<hr><table><tr><td>relevant file</td><td>pr_agent/git_providers/git_provider.py\n</td></tr><tr><td>suggestion &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</td><td>\n\n<strong>\n\nConsider raising an exception or logging a warning when \'pr_url\' attribute is not found. This can help in debugging issues related to the absence of \'pr_url\' in instances where it\'s expected. [important]\n\n</strong>\n</td></tr><tr><td>relevant line</td><td><a href=\'https://github.com/Codium-ai/pr-agent-pro/pull/102/files#diff-52d45f12b836f77ed1aef86e972e65404634ea4e2a6083fb71a9b0f9bb9e062fR199\'>return ""</a></td></tr></table><hr>\n\n</details>'
assert convert_to_markdown(input_data).strip() == expected_output.strip()

50
tests/unittest/test_github_action_output.py Normal file
View File

@ -0,0 +1,50 @@
import os
import json
from pr_agent.algo.utils import get_settings, github_action_output
class TestGitHubOutput:
def test_github_action_output_enabled(self, monkeypatch, tmp_path):
get_settings().set('GITHUB_ACTION_CONFIG.ENABLE_OUTPUT', True)
monkeypatch.setenv('GITHUB_OUTPUT', str(tmp_path / 'output'))
output_data = {'key1': {'value1': 1, 'value2': 2}}
key_name = 'key1'
github_action_output(output_data, key_name)
with open(str(tmp_path / 'output'), 'r') as f:
env_value = f.read()
actual_key = env_value.split('=')[0]
actual_data = json.loads(env_value.split('=')[1])
assert actual_key == key_name
assert actual_data == output_data[key_name]
def test_github_action_output_disabled(self, monkeypatch, tmp_path):
get_settings().set('GITHUB_ACTION_CONFIG.ENABLE_OUTPUT', False)
monkeypatch.setenv('GITHUB_OUTPUT', str(tmp_path / 'output'))
output_data = {'key1': {'value1': 1, 'value2': 2}}
key_name = 'key1'
github_action_output(output_data, key_name)
assert not os.path.exists(str(tmp_path / 'output'))
def test_github_action_output_notset(self, monkeypatch, tmp_path):
# not set config
monkeypatch.setenv('GITHUB_OUTPUT', str(tmp_path / 'output'))
output_data = {'key1': {'value1': 1, 'value2': 2}}
key_name = 'key1'
github_action_output(output_data, key_name)
assert not os.path.exists(str(tmp_path / 'output'))
def test_github_action_output_error_case(self, monkeypatch, tmp_path):
monkeypatch.setenv('GITHUB_OUTPUT', str(tmp_path / 'output'))
output_data = None # invalid data
key_name = 'key1'
github_action_output(output_data, key_name)
assert not os.path.exists(str(tmp_path / 'output'))