MQL5-Google-Onedrive/scripts/review_pull_requests.py

#!/usr/bin/env python3
"""
Pull Request Review Script
Reviews all pull requests and creates a comprehensive summary
"""

import subprocess
import sys
import json
from pathlib import Path
from datetime import datetime
from collections import defaultdict

REPO_ROOT = Path(__file__).resolve().parents[1]

# ⚡ Bolt: Global cache for branch metadata to avoid redundant git calls.
BRANCH_METADATA_CACHE = {}


def get_git_version():
    """⚡ Bolt: Get git version to check for ahead-behind support (2.41+)."""
    result = run_command(["git", "--version"])
    if result and result.returncode == 0:
        try:
            # "git version 2.45.1" -> [2, 45, 1]
            version_str = result.stdout.strip().split()[-1]
            return [int(x) for x in version_str.split('.')[:3]]
        except (ValueError, IndexError):
            pass
    return [0, 0, 0]


def fetch_branch_metadata(base_branch="main"):
    """⚡ Bolt: Fetch metadata for all remote branches in bulk (O(1) subprocess calls)."""
    global BRANCH_METADATA_CACHE
    if BRANCH_METADATA_CACHE:
        return BRANCH_METADATA_CACHE

    version = get_git_version()
    # ahead-behind atom was added in git 2.41
    supports_ahead_behind = version >= [2, 41, 0]

    # ⚡ Bolt: Use objectname:short to match 'git log --oneline' format (hash + subject)
    fmt = "%(refname:short)|%(committerdate:iso8601)|%(objectname:short) %(subject)"
    if supports_ahead_behind:
        fmt = f"%(ahead-behind:{base_branch})|{fmt}"

    # Scan all remote branches under refs/remotes/
    result = run_command(["git", "for-each-ref", f"--format={fmt}", "refs/remotes/"])
    if not result or result.returncode != 0:
        return {}

    for line in result.stdout.strip().split('\n'):
        if not line: continue
        parts = line.split('|', 3 if supports_ahead_behind else 2)

        if supports_ahead_behind:
            # format: "ahead behind|ref|date|subject"
            counts, ref, date, subject = parts
            ahead, behind = map(int, counts.split())
        else:
            # fallback: "ref|date|subject"
            ref, date, subject = parts
            ahead, behind = None, None

        # Filter out HEAD, main, and the remote name itself
        if "/HEAD" in ref or ref.endswith("/main") or ref == "origin":
            continue

        BRANCH_METADATA_CACHE[ref] = {
            "ahead": ahead,
            "behind": behind,
            "date": date,
            "subject": subject
        }

    return BRANCH_METADATA_CACHE


def run_command(cmd, capture_output=True):
    """Run a command and return the result."""
    try:
        result = subprocess.run(
            cmd,
            cwd=REPO_ROOT,
            capture_output=capture_output,
            text=True,
            timeout=30,
            encoding='utf-8',
            errors='replace'
        )
        return result
    except Exception as e:
        print(f"Error running command: {e}", file=sys.stderr)
        return None


def get_prs_via_gh_cli():
    """Get PRs using GitHub CLI."""
    result = run_command(["gh", "pr", "list", "--state", "all", "--json", "number,title,state,author,createdAt,updatedAt,headRefName,baseRefName,isDraft,labels"])
    if result and result.returncode == 0:
        try:
            return json.loads(result.stdout)
        except json.JSONDecodeError:
            return []
    return None


def get_prs_via_git():
    """⚡ Bolt: Get PR information via git branches using cached metadata with fallback."""
    metadata = fetch_branch_metadata()
    
    # Check if we have ahead/behind data (Git 2.41+)
    has_ahead_behind = any(d["ahead"] is not None for d in metadata.values())

    if not has_ahead_behind:
        # ⚡ Bolt: Fallback to original logic for old git versions
        result = run_command(["git", "branch", "-r", "--no-merged", "main"])
        open_branches = []
        if result and result.returncode == 0:
            open_branches = [b.strip() for b in result.stdout.strip().split("\n")
                        if b.strip() and "origin/main" not in b and "HEAD" not in b]

        result_merged = run_command(["git", "branch", "-r", "--merged", "main"])
        merged_branches = []
        if result_merged and result_merged.returncode == 0:
            merged_branches = [b.strip() for b in result_merged.stdout.strip().split("\n")
                              if b.strip() and "origin/main" not in b and "HEAD" not in b]

        return {
            "open": sorted(open_branches),
            "merged": sorted(merged_branches)
        }

    open_branches = []
    merged_branches = []
    
    for ref, data in metadata.items():
        if data["ahead"] > 0:
            open_branches.append(ref)
        else:
            merged_branches.append(ref)

    return {
        "open": sorted(open_branches),
        "merged": sorted(merged_branches)
    }


def analyze_branch_name(branch_name):
    """Analyze branch name to extract PR information."""
    branch = branch_name.replace("origin/", "")
    
    info = {
        "type": "unknown",
        "category": "other",
        "description": branch
    }
    
    # Categorize branches
    if branch.startswith("Cursor/"):
        info["type"] = "cursor"
        info["category"] = "ai-generated"
        info["description"] = branch.replace("Cursor/A6-9V/", "")
    elif branch.startswith("copilot/"):
        info["type"] = "copilot"
        info["category"] = "ai-generated"
        info["description"] = branch.replace("copilot/", "")
    elif branch.startswith("bolt-"):
        info["type"] = "bolt"
        info["category"] = "optimization"
        info["description"] = branch.replace("bolt-", "")
    elif branch.startswith("feat/"):
        info["type"] = "feature"
        info["category"] = "feature"
        info["description"] = branch.replace("feat/", "")
    elif branch.startswith("feature/"):
        info["type"] = "feature"
        info["category"] = "feature"
        info["description"] = branch.replace("feature/", "")
    
    return info


def get_branch_info(branch_name):
    """⚡ Bolt: Get detailed information about a branch using cached metadata."""
    metadata = fetch_branch_metadata()
    data = metadata.get(branch_name, {})
    
    branch = branch_name.replace("origin/", "")
    ahead = data.get("ahead")
    
    # ⚡ Bolt: Check if we have the optimized metadata (ahead count is not None)
    if ahead is None:
        # Fallback for old git or missing metadata: use original slow path
        result = run_command(["git", "log", "--oneline", f"main..{branch_name}"])
        commit_count = 0
        commits = []
        if result and result.returncode == 0:
            commits = [c.strip() for c in result.stdout.strip().split("\n") if c.strip()]
            commit_count = len(commits)

        result_date = run_command(["git", "log", "-1", "--format=%ci", branch_name])
        last_commit = None
        if result_date and result_date.returncode == 0 and result_date.stdout.strip():
            last_commit = result_date.stdout.strip()
    else:
        # Optimized path using cached metadata
        commit_count = ahead
        last_commit = data.get("date")

        commits = []
        if commit_count > 0:
            # If we only have 1 commit ahead, we can use the cached subject (which includes hash)
            if commit_count == 1 and data.get("subject"):
                commits.append(data["subject"])
            else:
                # ⚡ Bolt: Still need git log for multiple commits to match original behavior
                result = run_command(["git", "log", "--oneline", "-n", "5", f"main..{branch_name}"])
                if result and result.returncode == 0:
                    commits = [c.strip() for c in result.stdout.strip().split("\n") if c.strip()]

    return {
        "branch": branch,
        "full_name": branch_name,
        "commit_count": commit_count,
        "commits": commits[:5],
        "last_commit_date": last_commit
    }


def main():
    """Main review function."""
    print("=" * 80)
    print("PULL REQUEST REVIEW")
    print(f"Generated: {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}")
    print("=" * 80)
    print()
    
    # Try GitHub CLI first
    prs = get_prs_via_gh_cli()
    
    if prs is not None:
        print(f"Found {len(prs)} pull requests via GitHub CLI")
        print()
        
        # Group by state
        by_state = defaultdict(list)
        for pr in prs:
            by_state[pr.get("state", "unknown")].append(pr)
        
        print("Pull Requests by State:")
        for state, pr_list in sorted(by_state.items()):
            print(f"  {state.upper()}: {len(pr_list)}")
        print()
        
        # Show open PRs
        open_prs = by_state.get("OPEN", [])
        if open_prs:
            print("=" * 80)
            print("OPEN PULL REQUESTS")
            print("=" * 80)
            for pr in open_prs:
                print(f"\nPR #{pr.get('number', 'N/A')}: {pr.get('title', 'No title')}")
                print(f"  Author: {pr.get('author', {}).get('login', 'Unknown')}")
                print(f"  Branch: {pr.get('headRefName', 'N/A')} -> {pr.get('baseRefName', 'main')}")
                print(f"  Created: {pr.get('createdAt', 'N/A')}")
                print(f"  Updated: {pr.get('updatedAt', 'N/A')}")
                print(f"  Draft: {'Yes' if pr.get('isDraft') else 'No'}")
                labels = [l.get('name') for l in pr.get('labels', [])]
                if labels:
                    print(f"  Labels: {', '.join(labels)}")
        
        # Show merged PRs
        merged_prs = by_state.get("MERGED", [])
        if merged_prs:
            print("\n" + "=" * 80)
            print(f"MERGED PULL REQUESTS ({len(merged_prs)} total)")
            print("=" * 80)
            print(f"\nShowing last 10 merged PRs:")
            for pr in merged_prs[-10:]:
                print(f"  PR #{pr.get('number', 'N/A')}: {pr.get('title', 'No title')}")
    
    else:
        # Fallback to git branch analysis
        print("GitHub CLI not available, analyzing branches...")
        print()
        
        branch_info = get_prs_via_git()
        
        open_branches = branch_info["open"]
        merged_branches = branch_info["merged"]
        
        print(f"Open branches (potential PRs): {len(open_branches)}")
        print(f"Merged branches (completed PRs): {len(merged_branches)}")
        print()
        
        # Categorize open branches
        categories = defaultdict(list)
        for branch in open_branches:
            info = analyze_branch_name(branch)
            categories[info["category"]].append((branch, info))
        
        print("=" * 80)
        print("OPEN BRANCHES (Potential Pull Requests)")
        print("=" * 80)
        print()
        
        for category, branches in sorted(categories.items()):
            print(f"{category.upper()}: {len(branches)} branches")
            for branch, info in branches[:10]:  # Show first 10
                branch_details = get_branch_info(branch)
                print(f"  - {info['description']}")
                print(f"    Branch: {branch_details['branch']}")
                print(f"    Commits: {branch_details['commit_count']}")
                if branch_details['last_commit_date']:
                    print(f"    Last commit: {branch_details['last_commit_date']}")
            if len(branches) > 10:
                print(f"    ... and {len(branches) - 10} more")
            print()
        
        print("=" * 80)
        print("MERGED BRANCHES (Completed Pull Requests)")
        print("=" * 80)
        print(f"\nTotal merged: {len(merged_branches)}")
        print("\nRecent merged branches:")
        for branch in merged_branches[:20]:
            info = analyze_branch_name(branch)
            print(f"  - {info['description']}")
    
    print("\n" + "=" * 80)
    print("REVIEW COMPLETE")
    print("=" * 80)
    print("\nNote: GitHub doesn't support 'pinning' pull requests directly.")
    print("Consider:")
    print("1. Creating a tracking issue for important PRs")
    print("2. Using labels to categorize PRs")
    print("3. Adding PRs to project boards")
    print("4. Creating a PR summary document")


if __name__ == "__main__":
    main()