pull_mr.py

import requests
import datetime
import csv

# Function to get the release date for a given tag (e.g., "3.4.0")
def get_release_date(tag="3.4.0"):
    releases_url = "https://gitlab.com/api/v4/projects/libeigen%2Feigen/releases"
    response = requests.get(releases_url)
    if response.status_code == 200:
        releases = response.json()
        for release in releases:
            if release.get("tag_name") == tag:
                release_date = release.get("released_at")
                if release_date:
                    # Remove the trailing 'Z' if present and convert to datetime
                    return datetime.datetime.fromisoformat(release_date.rstrip("Z"))
    return None

# Function to fetch all merged merge requests from the repository using pagination
def fetch_merge_requests():
    merge_requests = []
    api_url = "https://gitlab.com/api/v4/projects/libeigen%2Feigen/merge_requests"
    page = 1
    per_page = 100
    while True:
        params = {
            'state': 'merged',
            'per_page': per_page,
            'page': page,
            'order_by': 'merged_at',
            'sort': 'asc'  # Sorting in ascending order so older MRs come first
        }
        response = requests.get(api_url, params=params)
        if response.status_code != 200:
            print("Error fetching merge requests:", response.status_code)
            break
        page_data = response.json()
        if not page_data:
            break  # No more pages
        merge_requests.extend(page_data)
        page += 1
    return merge_requests

def main():
    release_tag = "3.4.0"
    release_date = get_release_date(release_tag)
    if release_date is None:
        print(f"Could not determine release date for tag {release_tag}. Exiting.")
        return

    print(f"Release {release_tag} date: {release_date.isoformat()}")

    mrs = fetch_merge_requests()
    print(f"Total merged MRs: {len(mrs)}")
    filtered_mrs = []

    for mr in mrs:
        merged_at_str = mr.get("merged_at")
        if merged_at_str:
            merged_at = datetime.datetime.fromisoformat(merged_at_str.rstrip("Z"))
            # Only include merge requests merged after the release date
            if merged_at > release_date:
                # Extract key pieces of information; default to "NA" if missing.
                mr_id = mr.get("iid", "NA")
                link = mr.get("web_url", "NA")
                title = mr.get("title", "NA")
                description = mr.get("description", "NA")
                if isinstance(description, str):
                    description = description.strip() or "NA"
                author = mr.get("author", {}).get("name", "NA")
                merged_date = merged_at_str or "NA"
                labels = mr.get("labels", [])
                labels_str = ", ".join(labels) if labels else "NA"
                # Impacted Areas/Components is not directly available via the API, so we default to "NA"
                impacted_areas = "NA"

                filtered_mrs.append({
                    "mr_id_link": f"{mr_id} ({link})",
                    "title_subject": title,
                    "description_summary": description,
                    "author": author,
                    "merge_date": merged_date,
                    "category_labels": labels_str,
                    "impacted_areas_components": impacted_areas
                })

    # Write the collected data to a CSV file.
    output_file = "eigen_release_notes.csv"
    fieldnames = ["mr_id_link", "title_subject", "description_summary", "author", "merge_date", "category_labels", "impacted_areas_components"]
    try:
        with open(output_file, "w", newline='', encoding="utf-8") as csvfile:
            writer = csv.DictWriter(csvfile, fieldnames=fieldnames)
            writer.writeheader()
            for row in filtered_mrs:
                writer.writerow(row)
        print(f"Release notes saved to {output_file}")
    except Exception as e:
        print("Error writing CSV file:", e)

if __name__ == "__main__":
    main()

write_mr_notes.py

import openai
import csv
import os
# Set your API key for the model provider
api_key = os.environ['OPENAI_API_KEY']
client = openai.OpenAI(api_key=api_key)

def summarize_merge_request(mr):
    # Construct the instructions (prompt) for the summarization.
    instructions = (
        "You are a summarization assistant. Summarize the following merge request "
        "for the Eigen C++ library. Focus on the key changes, improvements, and their impact. "
        "If any field is marked as NA, do not include it in the summary.\n\n"
        "The summary should be in the form and have the sections as follows:\n"
        "## Title: \n"
        "## Authors: \n"
        "## Summary:\n"
        "### Key Changes:\n"
        "### Improvements:\n"
        "### Impact:\n")
    input = (f"Title: {mr.get('title_subject', 'NA')}\n"
        f"Description: {mr.get('description_summary', 'NA')}\n"
        f"Author: {mr.get('author', 'NA')}\n"
        f"Merge Date: {mr.get('merge_date', 'NA')}\n"
        f"Labels: {mr.get('category_labels', 'NA')}\n"
        f"Impacted Areas: {mr.get('impacted_areas_components', 'NA')}\n\n")

    # Use the new API call with the gpt-4o-mini model.
    response = client.responses.create(
        model="gpt-4o-mini",
        instructions=instructions,
        input=input  # No additional input is needed.
    )
    summary = response.output_text.strip()
    return summary

def main():
    input_csv = "eigen_release_notes.csv"
    output_csv = "eigen_release_notes_with_summary.csv"

    # Read the merge request data from CSV
    with open(input_csv, newline='', encoding='utf-8') as csvfile:
        reader = csv.DictReader(csvfile)
        mrs = list(reader)

    # Process each merge request to get its summary
    check_first = True
    for mr in mrs:
        if check_first:
            import pdb; pdb.set_trace()
        mr["summary"] = summarize_merge_request(mr)
        print(f"Summary for MR {mr.get('mr_id_link', 'NA')}:\n{mr['summary']}\n")

    # Write the updated data (with summaries) to a new CSV file
    fieldnames = list(mrs[0].keys())
    with open(output_csv, "w", newline='', encoding='utf-8') as csvfile:
        writer = csv.DictWriter(csvfile, fieldnames=fieldnames)
        writer.writeheader()
        writer.writerows(mrs)
    print(f"Summaries saved to {output_csv}")

if __name__ == "__main__":
    main()