feat: PR triage automation script (#659)

- scripts/pr_triage.py: fetch, categorize, deduplicate, report
- Categories: training-data, bug-fix, feature, maintenance, documentation, testing, infrastructure
- Duplicate detection: PRs referencing same issue
- Health checks: stale (>7d), closed issue refs, mergeable status
- Markdown report + JSON output
- 8 tests pass

scripts/pr_triage.py

@@ -1,271 +1,176 @@
 #!/usr/bin/env python3
-"""
-PR Triage Automation — Categorize, deduplicate, and report on open PRs.
-
-Usage:
-    python scripts/pr_triage.py                    # Generate report
-    python scripts/pr_triage.py --json             # JSON output
-    python scripts/pr_triage.py --auto-merge       # Auto-merge safe PRs
-    python scripts/pr_triage.py --repo timmy-home  # Single repo
-"""
-
-import json
-import os
-import re
-import sys
-from collections import Counter
-from datetime import datetime, timezone
-from pathlib import Path
-from typing import Any, Optional
-
-try:
-    import urllib.request
-except ImportError:
-    print("Error: urllib not available")
-    sys.exit(1)
-
-# ---------------------------------------------------------------------------
-# Config
-# ---------------------------------------------------------------------------
-
-GITEA_BASE = os.environ.get("GITEA_API_BASE", "https://forge.alexanderwhitestone.com/api/v1")
-TOKEN_PATH = os.environ.get("GITEA_TOKEN_PATH", str(Path.home() / ".config/gitea/token"))
-ORG = "Timmy_Foundation"
-
-DEFAULT_REPOS = [
-    "timmy-home",
-    "hermes-agent",
-    "timmy-config",
-    "the-nexus",
-    "the-door",
-    "burn-fleet",
-    "second-son-of-timmy",
-]
-
-# ---------------------------------------------------------------------------
-# Categories
-# ---------------------------------------------------------------------------
-
-CATEGORY_RULES = {
-    "training-data": [
-        r"training[- ]?data", r"scene[- ]?description", r"dpo", r"training",
-        r"batch[- ]?\d+", r"training[- ]?pipeline", r"jsonl",
-    ],
-    "bug-fix": [
-        r"^fix[\(:]", r"\[BUG\]", r"\[FIX\]", r"bug fix", r"fixes #\d+",
-        r"closes #\d+", r"broken", r"crash", r"regression",
-    ],
-    "feature": [
-        r"^feat[\(:]", r"\[FEAT\]", r"\[FEATURE\]", r"new feature",
-        r"add .+ support", r"implement",
-    ],
-    "docs": [
-        r"^docs[\(:]", r"documentation", r"readme", r"genome",
-    ],
-    "security": [
-        r"\[SECURITY\]", r"\[VITALIK\]", r"shield", r"injection",
-        r"vulnerability", r"hardening",
-    ],
-    "infra": [
-        r"\[INFRA\]", r"deploy", r"ansible", r"docker", r"ci[/ ]cd",
-        r"cron", r"watchdog", r"systemd",
-    ],
-    "research": [
-        r"research", r"benchmark", r"evaluation", r"analysis",
-        r"\[BIG-BRAIN\]", r"investigate",
-    ],
-    "other": [],  # fallback
-}
+"""PR Triage Automation -- Categorize, deduplicate, report (#659)."""
+import argparse, json, os, re, sys, subprocess
+from collections import Counter, defaultdict
+from datetime import datetime
+from urllib.request import Request, urlopen
+from urllib.error import HTTPError
 
 
-def categorize_pr(title: str, body: str) -> str:
-    """Categorize a PR by its title and body."""
-    text = f"{title} {body}".lower()
-    for category, patterns in CATEGORY_RULES.items():
-        if category == "other":
-            continue
-        for pattern in patterns:
-            if re.search(pattern, text, re.IGNORECASE):
-                return category
-    return "other"
+def _token():
+    t = os.environ.get("GITEA_TOKEN", "")
+    if not t:
+        p = os.path.expanduser("~/.config/gitea/token")
+        if os.path.exists(p):
+            t = open(p).read().strip()
+    return t
 
 
-# ---------------------------------------------------------------------------
-# Gitea API
-# ---------------------------------------------------------------------------
-
-def _load_token() -> str:
+def _api(url, token, method="GET", data=None):
+    h = {"Authorization": "token " + token, "Accept": "application/json"}
+    body = json.dumps(data).encode() if data else None
+    if data:
+        h["Content-Type"] = "application/json"
+    req = Request(url, data=body, headers=h, method=method)
     try:
-        return open(TOKEN_PATH).read().strip()
-    except FileNotFoundError:
-        print(f"Error: Token not found at {TOKEN_PATH}")
-        sys.exit(1)
-
-
-def api_get(path: str, token: str) -> Any:
-    req = urllib.request.Request(f"{GITEA_BASE}{path}")
-    req.add_header("Authorization", f"token {token}")
-    resp = urllib.request.urlopen(req, timeout=30)
-    return json.loads(resp.read())
-
-
-def get_open_prs(repo: str, token: str) -> list[dict]:
-    """Fetch all open PRs for a repo."""
-    prs = []
-    page = 1
-    while True:
-        try:
-            batch = api_get(f"/repos/{ORG}/{repo}/pulls?state=open&limit=50&page={page}", token)
-            if not batch:
-                break
-            prs.extend(batch)
-            if len(batch) < 50:
-                break
-            page += 1
-        except Exception:
-            break
-    return prs
-
-
-def get_issue_state(repo: str, issue_num: int, token: str) -> Optional[str]:
-    """Check if a referenced issue is still open."""
-    try:
-        issue = api_get(f"/repos/{ORG}/{repo}/issues/{issue_num}", token)
-        return issue.get("state", "unknown")
-    except Exception:
+        return json.loads(urlopen(req, timeout=30).read())
+    except HTTPError:
         return None
 
 
-def find_referenced_issues(pr_body: str, pr_title: str) -> list[int]:
-    """Extract issue numbers referenced in PR body/title."""
-    text = f"{pr_title} {pr_body}"
-    return [int(m) for m in re.findall(r'#(\d+)', text)]
+def fetch_prs(base, token, owner, repo):
+    prs, page = [], 1
+    while True:
+        b = _api(base + "/api/v1/repos/" + owner + "/" + repo + "/pulls?state=open&limit=50&page=" + str(page), token)
+        if not b:
+            break
+        prs.extend(b)
+        if len(b) < 50:
+            break
+        page += 1
+    return prs
 
 
-def find_duplicates(prs: list[dict]) -> list[tuple[dict, dict]]:
-    """Find PRs that reference the same issue."""
-    issue_to_prs: dict[int, list[dict]] = {}
-    for pr in prs:
-        refs = find_referenced_issues(pr.get("body", ""), pr.get("title", ""))
-        for issue_num in refs:
-            issue_to_prs.setdefault(issue_num, []).append(pr)
-
-    duplicates = []
-    for issue_num, pr_list in issue_to_prs.items():
-        if len(pr_list) > 1:
-            # Pair up duplicates
-            for i in range(len(pr_list)):
-                for j in range(i + 1, len(pr_list)):
-                    duplicates.append((pr_list[i], pr_list[j]))
-
-    return duplicates
+def fetch_issues(base, token, owner, repo):
+    iss, page = {}, 1
+    while True:
+        b = _api(base + "/api/v1/repos/" + owner + "/" + repo + "/issues?state=open&limit=50&page=" + str(page), token)
+        if not b:
+            break
+        for i in b:
+            if "pull_request" not in i:
+                iss[i["number"]] = i
+        if len(b) < 50:
+            break
+        page += 1
+    return iss
 
 
-# ---------------------------------------------------------------------------
-# Triage
-# ---------------------------------------------------------------------------
+def categorize(pr):
+    c = (pr.get("title", "") + " " + pr.get("body", "") + " " + " ".join(l.get("name", "") for l in pr.get("labels", []))).lower()
+    for kw, cat in [("training data", "training-data"), ("dpo", "training-data"), ("grpo", "training-data"),
+                    ("fix:", "bug-fix"), ("bug", "bug-fix"), ("hotfix", "bug-fix"),
+                    ("feat:", "feature"), ("feature", "feature"),
+                    ("refactor", "maintenance"), ("cleanup", "maintenance"),
+                    ("doc", "documentation"), ("test", "testing"), ("infra", "infrastructure")]:
+        if kw in c:
+            return cat
+    return "other"
 
-def triage_repo(repo: str, token: str) -> dict:
-    """Triage all open PRs for a repo."""
-    prs = get_open_prs(repo, token)
 
-    categorized: dict[str, list[dict]] = {}
-    stale_issues = []
-    duplicates = find_duplicates(prs)
+def refs(pr):
+    return [int(m) for m in re.findall(r"#(\d+)", pr.get("title", "") + " " + pr.get("body", ""))]
 
-    for pr in prs:
-        category = categorize_pr(pr.get("title", ""), pr.get("body", ""))
-        categorized.setdefault(category, []).append(pr)
 
-        # Check referenced issues
-        refs = find_referenced_issues(pr.get("body", ""), pr.get("title", ""))
-        for issue_num in refs:
-            state = get_issue_state(repo, issue_num, token)
-            if state == "closed":
-                stale_issues.append({"pr": pr["number"], "issue": issue_num, "repo": repo})
+def find_duplicates(prs):
+    by = defaultdict(list)
+    for p in prs:
+        for r in refs(p):
+            by[r].append(p)
+    return [g for g in by.values() if len(g) > 1]
 
+
+def health(pr, issues):
+    r = refs(pr)
+    created = datetime.fromisoformat(pr["created_at"].replace("Z", "+00:00"))
+    updated = datetime.fromisoformat(pr["updated_at"].replace("Z", "+00:00"))
+    now = datetime.now(created.tzinfo)
     return {
-        "repo": repo,
-        "total_prs": len(prs),
-        "by_category": {k: len(v) for k, v in categorized.items()},
-        "categorized": categorized,
-        "duplicates": [(a["number"], b["number"]) for a, b in duplicates],
-        "stale_issues": stale_issues,
+        "pr": pr["number"], "title": pr["title"], "head": pr["head"]["ref"],
+        "category": categorize(pr), "refs": r,
+        "open": [x for x in r if x in issues], "closed": [x for x in r if x not in issues],
+        "age": (now - created).days, "stale": (now - updated).days,
+        "mergeable": pr.get("mergeable"), "author": pr.get("user", {}).get("login", ""),
     }
 
 
-def triage_all(repos: list[str], token: str) -> list[dict]:
-    """Triage all repos."""
-    results = []
-    for repo in repos:
-        print(f"  Triaging {repo}...", file=sys.stderr)
+def report(repo, checks, dups):
+    lines = ["# PR Triage -- " + repo,
+             "Generated: " + datetime.now().strftime("%Y-%m-%d %H:%M"),
+             "Open PRs: " + str(len(checks)), "", "## Summary", ""]
+    cats = Counter(h["category"] for h in checks)
+    lines.append("| Category | Count |")
+    lines.append("|----------|-------|")
+    for c, n in cats.most_common():
+        lines.append("| " + c + " | " + str(n) + " |")
+    stale = [h for h in checks if h["stale"] > 7]
+    lines.extend(["", "Stale (>7d): " + str(len(stale)),
+                  "Duplicate groups: " + str(len(dups)), ""])
+    if dups:
+        lines.append("## Duplicates")
+        for g in dups:
+            rs = set()
+            for p in g:
+                rs.update(refs(p))
+            lines.append("Issues " + ", ".join("#" + str(r) for r in sorted(rs)) + ":")
+            for p in g:
+                lines.append("  - #" + str(p["number"]) + ": " + p["title"])
+        lines.append("")
+    if stale:
+        lines.append("## Stale (>7d)")
+        for h in sorted(stale, key=lambda x: x["stale"], reverse=True):
+            lines.append("- #" + str(h["pr"]) + ": " + h["title"] + " -- " + str(h["stale"]) + "d")
+        lines.append("")
+    lines.append("## All PRs")
+    lines.append("| # | Title | Category | Age | Stale | Merge |")
+    lines.append("|---|-------|----------|-----|-------|-------|")
+    for h in sorted(checks, key=lambda x: x["pr"]):
+        m = "Y" if h["mergeable"] else ("N" if h["mergeable"] is False else "?")
+        s = str(h["stale"]) + "d" if h["stale"] > 7 else "-"
+        lines.append("| " + str(h["pr"]) + " | " + h["title"][:50] + " | " + h["category"] +
+                      " | " + str(h["age"]) + "d | " + s + " | " + m + " |")
+    return chr(10).join(lines)
+
+
+def main():
+    p = argparse.ArgumentParser(description="PR Triage Automation")
+    p.add_argument("--base-url", default="https://forge.alexanderwhitestone.com")
+    p.add_argument("--owner", default="Timmy_Foundation")
+    p.add_argument("--repo", default="")
+    p.add_argument("--json", action="store_true", dest="js")
+    p.add_argument("--output", default="")
+    a = p.parse_args()
+    token = _token()
+    if not token:
+        print("No token"); sys.exit(1)
+    repo = a.repo
+    if not repo:
         try:
-            result = triage_repo(repo, token)
-            results.append(result)
-        except Exception as e:
-            print(f"  Error triaging {repo}: {e}", file=sys.stderr)
-            results.append({"repo": repo, "error": str(e)})
-    return results
+            remote = subprocess.check_output(["git", "remote", "get-url", "origin"], text=True).strip()
+            m = re.search(r"[/:](\w[\w-]*)/(\w[\w-]*?)(?:\.git)?$", remote)
+            if m:
+                a.owner, repo = m.group(1), m.group(2)
+        except Exception:
+            pass
+    if not repo:
+        print("No repo specified"); sys.exit(1)
+    print("Triaging " + a.owner + "/" + repo + "...", file=sys.stderr)
+    prs = fetch_prs(a.base_url, token, a.owner, repo)
+    issues = fetch_issues(a.base_url, token, a.owner, repo)
+    checks = [health(pr, issues) for pr in prs]
+    dups = find_duplicates(prs)
+    if a.js:
+        print(json.dumps({"repo": repo, "prs": checks,
+                          "duplicates": [[{"number": p["number"], "title": p["title"]} for p in g] for g in dups]},
+                         indent=2))
+    else:
+        r = report(repo, checks, dups)
+        print(r)
+        if a.output:
+            with open(a.output, "w") as f:
+                f.write(r)
+    print("\n" + str(len(checks)) + " PRs, " + str(len(dups)) + " duplicate groups", file=sys.stderr)
 
 
-# ---------------------------------------------------------------------------
-# Report
-# ---------------------------------------------------------------------------
-
-def generate_markdown_report(results: list[dict]) -> str:
-    """Generate a markdown triage report."""
-    total_prs = sum(r.get("total_prs", 0) for r in results)
-    all_categories: Counter = Counter()
-    all_duplicates = []
-    all_stale = []
-
-    for r in results:
-        for cat, count in r.get("by_category", {}).items():
-            all_categories[cat] += count
-        all_duplicates.extend(r.get("duplicates", []))
-        all_stale.extend(r.get("stale_issues", []))
-
-    lines = [
-        "# PR Triage Report",
-        "",
-        f"Generated: {datetime.now(timezone.utc).strftime('%Y-%m-%d %H:%M UTC')}",
-        "",
-        "## Summary",
-        "",
-        f"| Metric | Count |",
-        f"|--------|-------|",
-        f"| Total open PRs | {total_prs} |",
-        f"| Repos scanned | {len(results)} |",
-        f"| Duplicates found | {len(all_duplicates)} |",
-        f"| Stale (issue closed) | {len(all_stale)} |",
-        "",
-        "## By Category",
-        "",
-        "| Category | Count |",
-        "|----------|-------|",
-    ]
-
-    for cat, count in all_categories.most_common():
-        lines.append(f"| {cat} | {count} |")
-
-    if all_duplicates:
-        lines.extend(["", "## Duplicates (same issue referenced)", ""])
-        for a, b in all_duplicates:
-            lines.append(f"- PR #{a} and PR #{b}")
-
-    if all_stale:
-        lines.extend(["", "## Stale PRs (referenced issue is closed)", ""])
-        for s in all_stale:
-            lines.append(f"- {s['repo']} PR #{s['pr']} → issue #{s['issue']} (closed)")
-
-    # Per-repo detail
-    for r in results:
-        if r.get("error"):
-            lines.extend(["", f"## {r['repo']} — ERROR", "", f"```{r['error']}```"])
-            continue
-
-        lines.extend([f"", f"## {r['repo']} ({r.get('total_prs', 0)} open PRs)", ""])
-        for cat, prs in r.get("categorized", {}).items():
-            if not prs:
-                continue
-            lines.append(f"
+if __name__ == "__main__":
+    main()

tests/test_pr_triage.py

@@ -0,0 +1,45 @@
+"""Tests for PR triage automation (#659)."""
+import pytest
+
+class TestCategorize:
+    def _pr(self, title="", body=""):
+        return {"title": title, "body": body, "labels": []}
+
+    def test_training(self):
+        from scripts.pr_triage import categorize
+        assert categorize(self._pr("Add DPO pairs")) == "training-data"
+
+    def test_bug(self):
+        from scripts.pr_triage import categorize
+        assert categorize(self._pr("fix: crash")) == "bug-fix"
+
+    def test_feature(self):
+        from scripts.pr_triage import categorize
+        assert categorize(self._pr("feat: dark mode")) == "feature"
+
+    def test_other(self):
+        from scripts.pr_triage import categorize
+        assert categorize(self._pr("random")) == "other"
+
+class TestRefs:
+    def test_simple(self):
+        from scripts.pr_triage import refs
+        assert 123 in refs({"title": "Fix #123", "body": ""})
+
+    def test_multiple(self):
+        from scripts.pr_triage import refs
+        r = refs({"title": "", "body": "Closes #100, Refs #200"})
+        assert 100 in r and 200 in r
+
+class TestDuplicates:
+    def test_found(self):
+        from scripts.pr_triage import find_duplicates
+        prs = [{"title": "", "body": "Fix #1", "number": 1, "head": {"ref": "a"}, "created_at": "2026-01-01T00:00:00Z", "updated_at": "2026-01-01T00:00:00Z", "user": {}},
+               {"title": "", "body": "Refs #1", "number": 2, "head": {"ref": "b"}, "created_at": "2026-01-01T00:00:00Z", "updated_at": "2026-01-01T00:00:00Z", "user": {}}]
+        assert len(find_duplicates(prs)) == 1
+
+    def test_none(self):
+        from scripts.pr_triage import find_duplicates
+        prs = [{"title": "", "body": "Fix #1", "number": 1, "head": {"ref": "a"}, "created_at": "2026-01-01T00:00:00Z", "updated_at": "2026-01-01T00:00:00Z", "user": {}},
+               {"title": "", "body": "Fix #2", "number": 2, "head": {"ref": "b"}, "created_at": "2026-01-01T00:00:00Z", "updated_at": "2026-01-01T00:00:00Z", "user": {}}]
+        assert find_duplicates(prs) == []