Reduce Spotless fix to target update

Keep the upstream/main simplification of fix_ci.py and apply only the intended spotlessApply target change.

Author: trask

.github/scripts/pr-triage/fix_ci.py

@@ -18,22 +18,18 @@
     download_actions_job_log,
     extract_job_id,
     gh_json,
-    gradlew_cmd,
     invoke_copilot,
     make_temp_dir,
     progress,
     push,
-    run,
     run_pr_workflow,
     status_porcelain,
     untracked_files,
     write_json,
 )
 
 
-ERROR_PATTERN = re.compile(r"error:|Task .*FAILED|FAILURE: Build failed|\[ERROR\]|markdownlint", re.IGNORECASE)
 AGGREGATE_CHECK_SUFFIX = "required-status-check"
-MAX_LOGS_PER_JOB_FAMILY = 3
 
 
 def parse_args() -> argparse.Namespace:
@@ -59,206 +55,50 @@ def job_family(name: str) -> str:
     return re.sub(r"\s+\([^)]*\)$", "", name).strip()
 
 
-def matrix_tokens(name: str) -> set[str]:
-    match = re.search(r"\(([^)]*)\)\s*$", name)
-    if not match:
-        return set()
-    return {token.strip().lower() for token in match.group(1).split(",") if token.strip()}
-
-
-def selected_log_indexes(checks: list[dict[str, Any]], summary: Summary) -> set[int]:
-    progress(f"Selecting representative logs for {len(checks)} failing checks")
-    families: dict[str, list[int]] = {}
-    for index, check in enumerate(checks):
-        name = check.get("name") or "unknown"
-        families.setdefault(job_family(name), []).append(index)
-
-    selected: set[int] = set()
-    for family, indexes in families.items():
-        downloadable = [index for index in indexes if extract_job_id(checks[index]) is not None]
-        if len(downloadable) <= MAX_LOGS_PER_JOB_FAMILY:
-            selected.update(downloadable)
-            continue
-
-        picked = [downloadable[0]]
-        seen_tokens = set(matrix_tokens(checks[downloadable[0]].get("name") or ""))
-        remaining = downloadable[1:]
-        while remaining and len(picked) < MAX_LOGS_PER_JOB_FAMILY:
-            best_index = max(
-                remaining,
-                key=lambda index: len(matrix_tokens(checks[index].get("name") or "") - seen_tokens),
-            )
-            picked.append(best_index)
-            seen_tokens.update(matrix_tokens(checks[best_index].get("name") or ""))
-            remaining.remove(best_index)
-
-        selected.update(picked)
-        summary.notes.append(
-            f"Downloaded {len(picked)} representative logs for {family}; "
-            f"skipped {len(downloadable) - len(picked)} sibling logs"
-        )
-    return selected
-
-
-def extract_snippet(log_text: str, max_lines: int = 160) -> str:
-    lines = log_text.splitlines()
-    selected: list[str] = []
-    for index, line in enumerate(lines):
-        if ERROR_PATTERN.search(line):
-            start = max(0, index - 2)
-            end = min(len(lines), index + 21)
-            selected.extend(lines[start:end])
-            selected.append("---")
-            if len(selected) >= max_lines:
-                break
-    if selected:
-        return "\n".join(selected[:max_lines]) + "\n"
-    return "\n".join(lines[-max_lines:]) + "\n"
-
-
-def write_ci_bundle(
-    pr: int, checks: list[dict[str, Any]], directory: Path, summary: Summary
-) -> tuple[list[dict[str, Any]], Path]:
+def write_ci_bundle(pr: int, checks: list[dict[str, Any]], directory: Path, summary: Summary) -> list[dict[str, Any]]:
     progress(f"Preparing CI failure bundle in {directory}")
     repo = detect_repo(summary)
     owner, repo_name = repo.split("/", 1)
-    download_indexes = selected_log_indexes(checks, summary)
     logs_dir = directory / "logs"
-    snippets_dir = directory / "snippets"
     logs_dir.mkdir(parents=True, exist_ok=True)
-    snippets_dir.mkdir(parents=True, exist_ok=True)
 
     bundle_checks: list[dict[str, Any]] = []
-    for index, check in enumerate(checks):
+    seen_families: set[str] = set()
+    for check in checks:
         name = check.get("name") or "unknown"
         job_id = extract_job_id(check)
         summary.failures.append(f"{name} ({job_id or 'no job id'})")
-        entry: dict[str, Any] = {
-            "name": name,
-            "family": job_family(name),
-            "job_id": job_id,
-            "details_url": check.get("detailsUrl") or check.get("details_url"),
-            "database_id": check.get("databaseId") or check.get("database_id"),
-            "log_sampled": index in download_indexes,
-        }
-        if job_id is not None and index in download_indexes:
-            progress(f"Sampling log for failed job: {name}")
+        entry: dict[str, Any] = {"name": name, "job_id": job_id}
+        family = job_family(name)
+        if job_id is None:
+            pass
+        elif family in seen_families:
+            progress(f"Skipping sibling log for failed job: {name}")
+            entry["log_note"] = "log download skipped; covered by representative sibling job"
+        else:
+            seen_families.add(family)
+            progress(f"Downloading log for failed job: {name}")
             log_path = logs_dir / f"{job_id}.log"
-            snippet_path = snippets_dir / f"{job_id}-errors.txt"
             download_actions_job_log(owner, repo_name, job_id, log_path, summary)
-            snippet_path.write_text(
-                extract_snippet(log_path.read_text(encoding="utf-8", errors="replace")),
-                encoding="utf-8",
-            )
             entry["log"] = str(log_path)
-            entry["snippet"] = str(snippet_path)
-        elif job_id is not None:
-            progress(f"Skipping sibling log for failed job: {name}")
-            entry["log_note"] = "log download skipped; covered by representative sibling job"
         bundle_checks.append(entry)
 
     write_json(directory / "summary.json", {"repo": repo, "pr": pr, "failed_checks": bundle_checks})
-    plan = directory / "ci-plan.md"
-    plan.write_text(render_ci_plan(pr, bundle_checks), encoding="utf-8")
     summary.temp_dir = str(directory)
-    return bundle_checks, plan
+    return bundle_checks
 
 
-def render_ci_plan(pr: int, checks: list[dict[str, Any]]) -> str:
-    lines = [f"# CI Failure Analysis Plan for PR #{pr}", "", "## Failed Jobs", ""]
+def render_failed_jobs(checks: list[dict[str, Any]]) -> str:
+    lines: list[str] = []
     for check in checks:
         lines.append(f"- {check['name']} (job ID: {check.get('job_id')})")
-        if check.get("snippet"):
-            lines.append(f"  - Snippet: {check['snippet']}")
         if check.get("log"):
-            lines.append(f"  - Full log: {check['log']}")
+            lines.append(f"  - Log: {check['log']}")
         if check.get("log_note"):
             lines.append(f"  - {check['log_note']}")
-    lines.extend(["", "## Notes", "", "- Python downloaded logs before Copilot handoff.", ""])
     return "\n".join(lines)
 
 
-def maybe_apply_deterministic_fixes(bundle_dir: Path, plan_path: Path, summary: Summary) -> list[str]:
-    text = plan_path.read_text(encoding="utf-8")
-    for snippet in (bundle_dir / "snippets").glob("*.txt"):
-        text += "\n" + snippet.read_text(encoding="utf-8", errors="replace")
-
-    text = text.lower()
-    fix_kinds = []
-    if "spotless" in text:
-        run(gradlew_cmd("spotlessApply"), summary)
-        summary.notes.append("Applied deterministic spotless fix based on CI logs")
-        fix_kinds.append("spotless")
-    if "fossa" in text or "generatefossaconfiguration" in text or ".fossa.yml" in text:
-        run(gradlew_cmd("generateFossaConfiguration"), summary)
-        summary.notes.append("Applied deterministic FOSSA configuration fix based on CI logs")
-        fix_kinds.append("fossa")
-    return fix_kinds
-
-
-def ci_fix_commit_message(checks: list[dict[str, Any]], changed_paths: list[str]) -> list[str]:
-    families = sorted({job_family(check.get("name") or "unknown") for check in checks})
-    if len(families) == 1:
-        subject = f"Fix CI failure in {families[0]}"
-    else:
-        subject = f"Fix CI failures in {len(families)} job families"
-    body_lines = ["Failed jobs:"]
-    body_lines.extend(f"- {family}" for family in families[:8])
-    if len(families) > 8:
-        body_lines.append(f"- ... and {len(families) - 8} more")
-    body_lines.append("")
-    body_lines.append("Changed files:")
-    body_lines.extend(f"- {path}" for path in changed_paths[:12])
-    if len(changed_paths) > 12:
-        body_lines.append(f"- ... and {len(changed_paths) - 12} more")
-    return [subject, "\n".join(body_lines)]
-
-
-def append_deterministic_commit_details(body_lines: list[str], checks: list[dict[str, Any]], changed_paths: list[str], file_heading: str) -> None:
-    body_lines.extend(["", "Failed jobs:"])
-    body_lines.extend(f"- {family}" for family in sorted({job_family(check.get("name") or "unknown") for check in checks})[:8])
-    body_lines.append("")
-    body_lines.append(file_heading)
-    body_lines.extend(f"- {path}" for path in changed_paths[:12])
-    if len(changed_paths) > 12:
-        body_lines.append(f"- ... and {len(changed_paths) - 12} more")
-
-
-def deterministic_ci_fix_commit_message(fix_kinds: list[str], checks: list[dict[str, Any]], changed_paths: list[str]) -> list[str]:
-    if fix_kinds == ["spotless"]:
-        subject = "Apply spotless formatting"
-        body_lines = [
-            "CI reported Spotless formatting violations.",
-            "",
-            "Ran:",
-            "- ./gradlew spotlessApply",
-        ]
-        append_deterministic_commit_details(body_lines, checks, changed_paths, "Formatted files:")
-        return [subject, "\n".join(body_lines)]
-    if fix_kinds == ["fossa"]:
-        subject = "Regenerate FOSSA configuration"
-        body_lines = [
-            "CI reported that the FOSSA configuration was out of date.",
-            "",
-            "Ran:",
-            "- ./gradlew generateFossaConfiguration",
-        ]
-        append_deterministic_commit_details(body_lines, checks, changed_paths, "Updated files:")
-        return [subject, "\n".join(body_lines)]
-    if set(fix_kinds) == {"spotless", "fossa"}:
-        subject = "Apply deterministic CI fixes"
-        body_lines = [
-            "CI reported deterministic Spotless and FOSSA configuration failures.",
-            "",
-            "Ran:",
-            "- ./gradlew spotlessApply",
-            "- ./gradlew generateFossaConfiguration",
-        ]
-        append_deterministic_commit_details(body_lines, checks, changed_paths, "Updated files:")
-        return [subject, "\n".join(body_lines)]
-    return ci_fix_commit_message(checks, changed_paths)
-
-
 def read_commit_message(path: Path) -> list[str]:
     if not path.exists():
         raise RuntimeError(f"Copilot did not write a commit message to {path}")
@@ -286,12 +126,13 @@ def read_prompt_improvement(path: Path, summary: Summary) -> None:
             summary.notes.append(line)
 
 
-def copilot_prompt(pr: int, plan_path: Path, commit_message_path: Path, prompt_improvement_path: Path) -> str:
+def copilot_prompt(pr: int, checks: list[dict[str, Any]], commit_message_path: Path, prompt_improvement_path: Path) -> str:
     return f"""You are fixing failing CI in opentelemetry-java-instrumentation.
 
-The PR branch (#{pr}) is already checked out. Read this CI bundle first:
+The PR branch (#{pr}) is already checked out. The following CI jobs are failing.
+Use the referenced log files as the source of truth:
 
-{plan_path}
+{render_failed_jobs(checks)}
 
 After fixing the issue, write a commit message to this exact file path:
 
@@ -319,9 +160,9 @@ def copilot_prompt(pr: int, plan_path: Path, commit_message_path: Path, prompt_i
 - Do not push.
 - Do not rebase, merge, amend, or use force operations.
 - Edit only files needed to fix the failures listed in the CI bundle.
-- Use the downloaded log snippets and full logs as the source of truth.
+- Use the downloaded log files as the source of truth.
 - For deterministic formatting or generated-file failures (for example Spotless
-  or FOSSA), run the corresponding Gradle task (for example `./gradlew spotless`
+    or FOSSA), run the corresponding Gradle task (for example `./gradlew spotlessApply`
   or `./gradlew generateFossaConfiguration`) instead of editing files by hand.
 - If the failures are flaky or infrastructure-only, do not invent a code fix; leave the tree clean and explain why.
 - When done, print a concise summary of the files changed and validation commands run.
@@ -338,19 +179,17 @@ def body(summary: Summary) -> int:
             return 0
 
         bundle_dir = make_temp_dir("otel-ci-fix", args.pr, args.keep_temp)
-        bundle_checks, plan_path = write_ci_bundle(args.pr, checks, bundle_dir, summary)
-        deterministic_fixes = maybe_apply_deterministic_fixes(bundle_dir, plan_path, summary)
+        bundle_checks = write_ci_bundle(args.pr, checks, bundle_dir, summary)
 
-        if not deterministic_fixes and args.skip_copilot:
+        if args.skip_copilot:
             summary.outcome = "downloaded CI logs; skipped Copilot handoff"
             return 0
 
         commit_message_path = bundle_dir / "commit-message.txt"
-        if not deterministic_fixes:
-            prompt_improvement_path = bundle_dir / "prompt-improvement.md"
-            response = invoke_copilot(copilot_prompt(args.pr, plan_path, commit_message_path, prompt_improvement_path), summary)
-            (bundle_dir / "copilot-response.txt").write_text(response + "\n", encoding="utf-8")
-            read_prompt_improvement(prompt_improvement_path, summary)
+        prompt_improvement_path = bundle_dir / "prompt-improvement.md"
+        response = invoke_copilot(copilot_prompt(args.pr, bundle_checks, commit_message_path, prompt_improvement_path), summary)
+        (bundle_dir / "copilot-response.txt").write_text(response + "\n", encoding="utf-8")
+        read_prompt_improvement(prompt_improvement_path, summary)
 
         diff_check(summary)
         if untracked_files(summary):
@@ -360,11 +199,7 @@ def body(summary: Summary) -> int:
             summary.outcome = "no code changes needed"
             return 0
 
-        commit_message = (
-            deterministic_ci_fix_commit_message(deterministic_fixes, bundle_checks, summary.changed_files)
-            if deterministic_fixes
-            else read_commit_message(commit_message_path)
-        )
+        commit_message = read_commit_message(commit_message_path)
         commit_all_tracked(commit_message, summary)
         if args.no_push:
             summary.push_result = "not pushed (--no-push)"
@@ -377,4 +212,4 @@ def body(summary: Summary) -> int:
 
 
 if __name__ == "__main__":
-    sys.exit(main())
+    sys.exit(main())