xiaxinhang/pr-agent
1
0
Fork 0
mirror of https://github.com/qodo-ai/pr-agent.git synced 2025-07-06 05:40:38 +08:00
Code Issues Packages Projects Releases Wiki Activity

s

Browse Source
This commit is contained in:
mrT23
2024-06-27 09:07:19 +03:00
parent 056eb3a954
commit 15f854336a
1 changed files with 17 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files

30
pr_agent/tools/pr_description.py
View File

@ -168,7 +168,8 @@ class PRDescription:
return None return None
large_pr_handling = get_settings().pr_description.enable_large_pr_handling and "pr_description_only_files_prompts" in get_settings() large_pr_handling = get_settings().pr_description.enable_large_pr_handling and "pr_description_only_files_prompts" in get_settings()
patches_diff = get_pr_diff(self.git_provider, self.token_handler, model, large_pr_handling=large_pr_handling) patches_diff = get_pr_diff(self.git_provider, self.token_handler, model,
large_pr_handling=large_pr_handling)
if not large_pr_handling or patches_diff: if not large_pr_handling or patches_diff:
self.patches_diff = patches_diff self.patches_diff = patches_diff
if patches_diff: if patches_diff:
@ -211,29 +212,32 @@ class PRDescription:
get_settings().pr_description_only_description_prompts.system, get_settings().pr_description_only_description_prompts.system,
get_settings().pr_description_only_description_prompts.user) get_settings().pr_description_only_description_prompts.user)
files_walkthrough = "\n".join(file_description_str_list) files_walkthrough = "\n".join(file_description_str_list)
files_walkthrough_prompt = copy.deepcopy(files_walkthrough)
if remaining_files_list: if remaining_files_list:
files_walkthrough += "\n\nNo more token budget. Additional unprocessed files:" files_walkthrough_prompt += "\n\nNo more token budget. Additional unprocessed files:"
for file in remaining_files_list: for file in remaining_files_list:
files_walkthrough += f"\n- {file}" files_walkthrough_prompt += f"\n- {file}"
if deleted_files_list: if deleted_files_list:
files_walkthrough += "\n\nAdditional deleted files:" files_walkthrough_prompt += "\n\nAdditional deleted files:"
for file in deleted_files_list: for file in deleted_files_list:
files_walkthrough += f"\n- {file}" files_walkthrough_prompt += f"\n- {file}"
tokens_files_walkthrough = len(token_handler_only_description_prompt.encoder.encode(files_walkthrough)) tokens_files_walkthrough = len(
token_handler_only_description_prompt.encoder.encode(files_walkthrough_prompt))
total_tokens = token_handler_only_description_prompt.prompt_tokens + tokens_files_walkthrough total_tokens = token_handler_only_description_prompt.prompt_tokens + tokens_files_walkthrough
max_tokens_model = get_max_tokens(model) max_tokens_model = get_max_tokens(model)
if total_tokens > max_tokens_model - OUTPUT_BUFFER_TOKENS_HARD_THRESHOLD: if total_tokens > max_tokens_model - OUTPUT_BUFFER_TOKENS_HARD_THRESHOLD:
# clip files_walkthrough to git the tokens within the limit # clip files_walkthrough to git the tokens within the limit
files_walkthrough = clip_tokens(files_walkthrough, files_walkthrough_prompt = clip_tokens(files_walkthrough_prompt,
max_tokens_model - OUTPUT_BUFFER_TOKENS_HARD_THRESHOLD - token_handler_only_description_prompt.prompt_tokens, max_tokens_model - OUTPUT_BUFFER_TOKENS_HARD_THRESHOLD - token_handler_only_description_prompt.prompt_tokens,
num_input_tokens=tokens_files_walkthrough) num_input_tokens=tokens_files_walkthrough)
# PR header inference # PR header inference
# toDo - add deleted and unprocessed files to the prompt ('files_walkthrough'), as extra data get_logger().debug(f"PR diff only description", artifact=files_walkthrough_prompt)
get_logger().debug(f"PR diff only description", artifact=files_walkthrough) prediction_headers = await self._get_prediction(model, patches_diff=files_walkthrough_prompt,
prediction_headers = await self._get_prediction(model, patches_diff=files_walkthrough,
prompt="pr_description_only_description_prompts") prompt="pr_description_only_description_prompts")
prediction_headers = prediction_headers.strip().removeprefix('```yaml').strip('`').strip() prediction_headers = prediction_headers.strip().removeprefix('```yaml').strip('`').strip()
# manually add extra files to final prediction
if get_settings().pr_description.mention_extra_files: if get_settings().pr_description.mention_extra_files:
for file in remaining_files_list: for file in remaining_files_list:
extra_file_yaml = f"""\ extra_file_yaml = f"""\
@ -247,7 +251,6 @@ class PRDescription:
not processed (token-limit) not processed (token-limit)
""" """
files_walkthrough = files_walkthrough.strip() + "\n" + extra_file_yaml.strip() files_walkthrough = files_walkthrough.strip() + "\n" + extra_file_yaml.strip()
# final processing # final processing
self.prediction = prediction_headers + "\n" + "pr_files:\n" + files_walkthrough self.prediction = prediction_headers + "\n" + "pr_files:\n" + files_walkthrough
if not load_yaml(self.prediction): if not load_yaml(self.prediction):
@ -256,6 +259,7 @@ class PRDescription:
get_logger().debug(f"Using only headers for describe {self.pr_id}") get_logger().debug(f"Using only headers for describe {self.pr_id}")
self.prediction = prediction_headers self.prediction = prediction_headers
async def _get_prediction(self, model: str, patches_diff: str, prompt="pr_description_prompt") -> str: async def _get_prediction(self, model: str, patches_diff: str, prompt="pr_description_prompt") -> str:
variables = copy.deepcopy(self.vars) variables = copy.deepcopy(self.vars)
variables["diff"] = patches_diff # update diff variables["diff"] = patches_diff # update diff