pr-monitor/server/failures.ts

import { unzipSync, strFromU8 } from 'fflate';
import type { FailureReport, PrSnapshot, SpecFailure } from '../shared/types.js';
import type { CompiledFailureRule, RepoRules } from './config.js';
import type { Db } from './db.js';
import { restJson, restZip } from './github/rest.js';

const MAX_ARTIFACT_BYTES = 80 * 1024 * 1024;
const MAX_STORED_FAILURES = 1000;
const MSG_LEN = 240;

interface ArtifactInfo {
  id: number;
  name: string;
  expired: boolean;
  size_in_bytes: number;
}

function firstLine(text: string): string {
  const line = text.split('\n').find((l) => l.trim() !== '') ?? '';
  return line.trim().slice(0, MSG_LEN);
}

/* ---------- parsers ---------- */

interface RspecJson {
  examples?: {
    status: string;
    file_path: string;
    line_number: number;
    full_description: string;
    exception?: { message?: string };
  }[];
}

function parseRspecZip(zip: Uint8Array): SpecFailure[] {
  const files = unzipSync(zip, { filter: (f) => f.name.endsWith('.json') });
  const failures: SpecFailure[] = [];
  for (const [, data] of Object.entries(files)) {
    let parsed: RspecJson;
    try {
      parsed = JSON.parse(strFromU8(data)) as RspecJson;
    } catch {
      continue;
    }
    for (const ex of parsed.examples ?? []) {
      if (ex.status !== 'failed') continue;
      failures.push({
        path: ex.file_path.replace(/^\.\//, ''),
        line: ex.line_number ?? null,
        name: ex.full_description,
        message: firstLine(ex.exception?.message ?? ''),
      });
    }
  }
  return failures;
}

/** Playwright HTML-report artifacts carry one `data/<hash>.md` per failed test
 *  (the "copy prompt" attachment) with Name / Location / Error details blocks. */
function parsePlaywrightZip(zip: Uint8Array): SpecFailure[] {
  const files = unzipSync(zip, {
    filter: (f) => f.name.startsWith('data/') && f.name.endsWith('.md'),
  });
  const failures: SpecFailure[] = [];
  const seen = new Set<string>();
  for (const [, data] of Object.entries(files)) {
    const text = strFromU8(data);
    const name = /^- Name: (.+)$/m.exec(text)?.[1]?.trim();
    const location = /^- Location: (.+)$/m.exec(text)?.[1]?.trim();
    if (!name || seen.has(name + location)) continue;
    seen.add(name + location);
    const errorBlock = /# Error details\s*```([\s\S]*?)```/.exec(text)?.[1] ?? '';
    const locMatch = location ? /^(.*?):(\d+)(?::\d+)?$/.exec(location) : null;
    failures.push({
      path: locMatch?.[1] ?? location ?? '',
      line: locMatch?.[2] ? Number(locMatch[2]) : null,
      name: name.replace(/ >> /g, ' › '),
      message: firstLine(errorBlock),
    });
  }
  return failures;
}

/* ---------- orchestration ---------- */

function runIdFromUrl(url: string | null): string | null {
  if (!url) return null;
  return /\/actions\/runs\/(\d+)\//.exec(url)?.[1] ?? null;
}

function artifactMatcher(pattern: string, prNumber: number): (name: string) => boolean {
  const concrete = pattern.replaceAll('{number}', String(prNumber));
  if (!concrete.includes('*')) return (n) => n === concrete;
  const regex = new RegExp(
    `^${concrete.replace(/[.+^${}()|[\]\\]/g, '\\$&').replace(/\*/g, '.*')}$`,
  );
  return (n) => regex.test(n);
}

/** Sync failure reports for one PR: for each configured rule with at least one
 *  failing check, download + parse that run's artifacts unless the stored report
 *  already covers the same run. Reports for now-passing rules are deleted. */
export async function syncFailureReports(
  db: Db,
  rules: RepoRules | undefined,
  repo: string,
  snapshot: PrSnapshot,
): Promise<boolean> {
  if (!rules || rules.failureRules.length === 0) return false;

  const failingChecks = snapshot.categories
    .flatMap((c) => c.checks)
    .filter((c) => c.state === 'fail');

  const activeKeys: string[] = [];
  let changed = false;

  for (const rule of rules.failureRules) {
    const matches = failingChecks.filter((c) => rule.matchesCheck(c.name));
    if (matches.length === 0) continue;
    activeKeys.push(rule.key);

    const runId = matches.map((m) => runIdFromUrl(m.url)).find((r) => r !== null);
    if (!runId) continue;
    const existing = db.getFailureReportRunId(snapshot.id, rule.key);
    if (existing === runId) continue;

    try {
      const failures = await fetchAndParse(repo, runId, rule, snapshot.number);
      db.upsertFailureReport(snapshot.id, rule.key, rule.label, runId, {
        failed: failures.length,
        truncated: failures.length > MAX_STORED_FAILURES,
        failures: failures.slice(0, MAX_STORED_FAILURES),
      });
      changed = true;
      console.log(
        `[failures] ${repo}#${snapshot.number} ${rule.label}: ${failures.length} failing`,
      );
    } catch (err) {
      console.error(
        `[failures] ${repo}#${snapshot.number} ${rule.key}: ${err instanceof Error ? err.message : err}`,
      );
    }
  }

  changed = db.pruneFailureReports(snapshot.id, activeKeys) || changed;
  return changed;
}

async function fetchAndParse(
  repo: string,
  runId: string,
  rule: CompiledFailureRule,
  prNumber: number,
): Promise<SpecFailure[]> {
  const { artifacts } = await restJson<{ artifacts: ArtifactInfo[] }>(
    `/repos/${repo}/actions/runs/${runId}/artifacts?per_page=100`,
  );
  const match = artifactMatcher(rule.artifact, prNumber);
  const wanted = artifacts.filter((a) => match(a.name) && !a.expired);

  const failures: SpecFailure[] = [];
  for (const artifact of wanted) {
    if (artifact.size_in_bytes > MAX_ARTIFACT_BYTES) {
      console.warn(`[failures] skipping oversized artifact ${artifact.name}`);
      continue;
    }
    const zip = await restZip(`/repos/${repo}/actions/artifacts/${artifact.id}/zip`, MAX_ARTIFACT_BYTES);
    if (!zip) continue;
    failures.push(...(rule.parser === 'rspec' ? parseRspecZip(zip) : parsePlaywrightZip(zip)));
  }
  failures.sort((a, b) => a.path.localeCompare(b.path) || (a.line ?? 0) - (b.line ?? 0));
  return failures;
}

export function buildFailureReport(row: {
  key: string;
  label: string;
  fetched_at: string;
  failed: number;
  truncated: number;
  failures: string;
}): FailureReport {
  return {
    key: row.key,
    label: row.label,
    fetchedAt: row.fetched_at,
    failed: row.failed,
    truncated: row.truncated === 1,
    failures: JSON.parse(row.failures) as SpecFailure[],
  };
}