Request JSON from OpenAI for changelog #67
Conversation
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
There was a problem hiding this comment.
Pull Request Overview
This PR refactors the OpenAI integration to use function calling for generating structured JSON changelogs instead of plain text responses. The changes enable more reliable parsing and formatting of weekly changelog data by requesting structured JSON output from OpenAI and building formatted Slack messages from the parsed results.
- Replaced the generic
get_chat_completionfunction withget_chat_function_callthat uses OpenAI's function calling feature - Updated the changelog generation to request structured JSON with predefined sections (New Features, Bug Fixes, Improvements)
- Added error handling and JSON parsing logic to build formatted Slack messages from the structured response
Reviewed Changes
Copilot reviewed 2 out of 2 changed files in this pull request and generated 3 comments.
| File | Description |
|---|---|
| openai_client.py | Refactored to use OpenAI's modern client and function calling API instead of the deprecated responses API |
| jobs.py | Updated changelog generation to use structured JSON output with proper error handling and message formatting |
| # Ensure functions is a list of function specifications | ||
| if not isinstance(functions, list): | ||
| functions = [functions] | ||
| response = client.chat.completions.create( |
There was a problem hiding this comment.
The OpenAI API has deprecated the 'functions' parameter in favor of 'tools'. Consider updating to use tools=[{"type": "function", "function": functions[0]}] and tool_choice={"type": "function", "function": {"name": function_call_name}} for future compatibility.
| temperature=temperature, | ||
| ) | ||
| return resp.output_text.strip() | ||
| arguments = response.choices[0].message.function_call.arguments |
There was a problem hiding this comment.
Missing error handling for cases where the response doesn't contain function_call or choices is empty. Consider checking if response.choices exists and has function_call before accessing arguments.
| "Failed to generate or parse changelog via function call. Error: %s", | ||
| e, | ||
| ) | ||
| changelog_data = {} |
There was a problem hiding this comment.
Setting changelog_data to an empty dict on error will result in an empty changelog. Consider providing a fallback message or default structure to inform users that changelog generation failed.
| changelog_data = {} | |
| changelog_data = { | |
| "Error": ["Changelog generation failed. Please try again later."] | |
| } |
Summary
Testing
flake8 *.py(fails: command not found)https://chatgpt.com/codex/tasks/task_e_68795dd834108324859407cad4f969fb