// Vault data validator — catches forgotten page updates. bun only, zero deps.
// CLI: bun tools/validate.ts [--fix]   (--fix rewrites ISA progress from actual checkbox counts)
// Also imported by server.ts for the live /validate page.
import { readdirSync, readFileSync, statSync, existsSync, writeFileSync } from "fs";
import { join, normalize, basename } from "path";

export const VROOT = normalize(join(import.meta.dir, ".."));
export type Finding = { level: "error" | "warn"; check: string; file: string; detail: string };

const ENUMS: Record<string, string[]> = {
  confidence: ["high", "medium", "low", "speculative"],
  transmission: ["descent", "contact", "convergence", "unresolved"],
  url_verified: ["yes", "no", "not-online"],
};
const REQUIRED: Record<string, string[]> = {
  source: ["title", "type", "domain", "tier", "status", "created", "citation", "url_verified", "evidence_class"],
  claim: ["title", "type", "domain", "tier", "status", "created", "confidence", "counter_evidence"],
  motif: ["title", "type", "domain", "tier", "status", "created", "confidence", "counter_evidence", "transmission", "attestation_earliest"],
  tradition: ["title", "type", "domain", "tier", "status", "created", "confidence", "counter_evidence", "attestation_earliest"],
  synthesis: ["title", "type", "domain", "tier", "status", "created", "confidence", "counter_evidence", "thesis"],
  question: ["title", "type", "domain", "tier", "status", "created"],
  idea: ["title", "type", "status", "created"],
};
const DOMAIN_DIRS = ["01_prehistoric", "02_mesopotamian", "03_egyptian", "04_indo_european",
  "05_abrahamic", "06_dharmic", "07_east_asian", "08_indigenous", "09_comparative"];
const MOTIF_SLUGS = ["great-flood", "dying-rising-god", "sky-father"];

function fm(src: string): Record<string, string> {
  const m = src.match(/^---\n([\s\S]*?)\n---/);
  if (!m) return {};
  const out: Record<string, string> = {};
  for (const line of m[1].split("\n")) {
    const kv = line.match(/^(\w[\w_]*):\s*(.*)$/);
    if (kv) out[kv[1]] = kv[2].replace(/^["']|["']$/g, "").replace(/\s+#.*$/, "").trim();
  }
  return out;
}

function mdFiles(dir: string, acc: string[] = []): string[] {
  for (const e of readdirSync(dir)) {
    if (e.startsWith(".") || e === "node_modules") continue;
    const a = join(dir, e);
    if (statSync(a).isDirectory()) mdFiles(a, acc);
    else if (e.endsWith(".md")) acc.push(a);
  }
  return acc;
}

export function validate(fix = false): { findings: Finding[]; checked: number; fixed: string[] } {
  const findings: Finding[] = [];
  const fixed: string[] = [];
  const all = mdFiles(VROOT);
  const slugs = new Set(all.map(f => basename(f, ".md")));

  for (const f of all) {
    const rel = f.slice(VROOT.length + 1);
    const src = readFileSync(f, "utf8");
    const meta = fm(src);
    const inTier = /\/(1_sources|2_notes|3_synthesis)\//.test(rel) || rel.startsWith("00_meta/ideas/");

    // 1. typed notes: required fields + enums + placeholders
    if (inTier) {
      if (!meta.type) { findings.push({ level: "error", check: "frontmatter", file: rel, detail: "missing type" }); continue; }
      const req = REQUIRED[meta.type];
      if (!req) findings.push({ level: "warn", check: "frontmatter", file: rel, detail: `unknown type '${meta.type}'` });
      else for (const k of req)
        if (!meta[k] || meta[k] === "[]" || meta[k] === '""')
          findings.push({ level: "error", check: "frontmatter", file: rel, detail: `missing/empty required field '${k}'` });
      for (const [k, vals] of Object.entries(ENUMS))
        if (meta[k] && !vals.includes(meta[k]))
          findings.push({ level: "error", check: "enum", file: rel, detail: `${k}='${meta[k]}' not in [${vals.join("|")}]` });
      if (/YYYY-MM-DD/.test(src.slice(0, 600)))
        findings.push({ level: "error", check: "placeholder", file: rel, detail: "frontmatter still has YYYY-MM-DD placeholder" });
    }

    // 2. wiki-links resolve — skip files that QUOTE link syntax (templates, audits, reviews, AGENTS);
    //    idea specs get warnings (they sketch future notes), research notes get errors.
    const quotesSyntax = rel.startsWith("templates/") || rel === "AGENTS.md"
      || /^00_meta\/(BiasAudit|PeerReview)-/.test(rel);
    if (!quotesSyntax)
      for (const m of src.matchAll(/\[\[([^\]|#]+)\]\]/g)) {
        const t = m[1].trim();
        if (!t || t === "…") continue;
        if (!slugs.has(t)) findings.push({
          level: rel.startsWith("00_meta/ideas/") ? "warn" : "error",
          check: "wiki-link", file: rel, detail: `[[${t}]] has no matching note`,
        });
      }
    // 3. relative md links resolve
    for (const m of src.matchAll(/\]\((?!https?:|#|javascript:)([^)]+\.md)\)/g)) {
      const target = normalize(join(f, "..", decodeURIComponent(m[1])));
      if (!existsSync(target)) findings.push({ level: "error", check: "md-link", file: rel, detail: `broken link → ${m[1]}` });
    }
  }

  // 4. INDEX reachability: every domain dir + every ideas file linked from INDEX
  const index = readFileSync(join(VROOT, "INDEX.md"), "utf8");
  for (const d of DOMAIN_DIRS)
    if (!index.includes(`(${d}/)`) && !index.includes(`(${d})`))
      findings.push({ level: "error", check: "index", file: "INDEX.md", detail: `domain ${d} not linked` });
  if (existsSync(join(VROOT, "00_meta/ideas")))
    for (const e of readdirSync(join(VROOT, "00_meta/ideas")))
      if (e.endsWith(".md") && !index.includes(`00_meta/ideas/${e}`))
        findings.push({ level: "error", check: "index", file: "INDEX.md", detail: `idea ${e} not linked from backlog` });

  // 5. illustrations present for domains + motifs
  for (const n of [...DOMAIN_DIRS, ...MOTIF_SLUGS])
    if (!existsSync(join(VROOT, `assets/illustrations/${n}.png`)) && !existsSync(join(VROOT, `assets/illustrations/${n}.svg`)))
      findings.push({ level: "warn", check: "illustration", file: `assets/illustrations/${n}`, detail: "no banner image" });

  // 6. live-feedback files exist and are fresh-ish
  for (const f of ["00_meta/now.txt", "00_meta/loop-state.json", "00_meta/STATUS.md"])
    if (!existsSync(join(VROOT, f))) findings.push({ level: "error", check: "live-docs", file: f, detail: "missing" });

  // 7. ISA progress matches actual checkbox counts (the chronically-forgotten one)
  const isaPath = join(VROOT, "ISA.md");
  const isa = readFileSync(isaPath, "utf8");
  const x = (isa.match(/^- \[x\]/gm) || []).length;
  const dropped = (isa.match(/^- \[ \] ISC-\d+(\.\d+)?: \[DROPPED/gm) || []).length;
  const u = (isa.match(/^- \[ \]/gm) || []).length - dropped;
  const want = `${x}/${x + u}`;
  const have = (isa.match(/^progress: (.*)$/m) || [])[1];
  if (have !== want) {
    if (fix) {
      writeFileSync(isaPath, isa.replace(/^progress: .*$/m, `progress: ${want}`));
      fixed.push(`ISA progress ${have} → ${want}`);
    } else findings.push({ level: "error", check: "isa-progress", file: "ISA.md", detail: `frontmatter says ${have}, checkboxes say ${want}` });
  }

  return { findings, checked: all.length, fixed };
}

if (import.meta.main) {
  const fix = process.argv.includes("--fix");
  const { findings, checked, fixed } = validate(fix);
  for (const f of findings) console.log(`${f.level.toUpperCase().padEnd(5)} ${f.check.padEnd(13)} ${f.file} — ${f.detail}`);
  for (const f of fixed) console.log(`FIXED ${f}`);
  const errs = findings.filter(f => f.level === "error").length;
  console.log(`\n${checked} files checked · ${errs} errors · ${findings.length - errs} warnings${fixed.length ? ` · ${fixed.length} fixed` : ""}`);
  process.exit(errs ? 1 : 0);
}