anno: build-annotations-index supports shard annotations

scripts/build-annotations-index.mjs

@@ -1,28 +1,106 @@
+#!/usr/bin/env node
 // scripts/build-annotations-index.mjs
+// Construit dist/annotations-index.json à partir de src/annotations/**/*.yml
+// Supporte:
+//   - monolith : src/annotations/<pageKey>.yml
+//   - shard    : src/annotations/<pageKey>/<paraId>.yml   (paraId = p-<n>-...)
+// Invariants:
+//   - doc.schema === 1
+//   - doc.page (si présent) == pageKey déduit du chemin
+//   - shard: doc.paras doit contenir EXACTEMENT la clé paraId (sinon fail)
+//
+// Deep-merge non destructif (media/refs/comments dédupliqués), tri stable.
+
 import fs from "node:fs/promises";
 import path from "node:path";
 import YAML from "yaml";
 
-function parseArgs(argv) {
+const ROOT = process.cwd();
-  const out = {
+const ANNO_ROOT = path.join(ROOT, "src", "annotations");
-    inDir: "src/annotations",
+const DIST_DIR = path.join(ROOT, "dist");
-    outFile: "dist/annotations-index.json",
+const OUT = path.join(DIST_DIR, "annotations-index.json");
-  };
 
-  for (let i = 0; i < argv.length; i++) {
+function assert(cond, msg) {
-    const a = argv[i];
+  if (!cond) throw new Error(msg);
+}
 
-    if (a === "--in" && argv[i + 1]) out.inDir = argv[++i];
+function isObj(x) {
-    else if (a.startsWith("--in=")) out.inDir = a.slice("--in=".length);
+  return !!x && typeof x === "object" && !Array.isArray(x);
+}
+function isArr(x) {
+  return Array.isArray(x);
+}
 
-    if (a === "--out" && argv[i + 1]) out.outFile = argv[++i];
+function normPath(s) {
-    else if (a.startsWith("--out=")) out.outFile = a.slice("--out=".length);
+  return String(s || "")
+    .replace(/\\/g, "/")
+    .replace(/^\/+|\/+$/g, "");
+}
+
+function paraNum(pid) {
+  const m = String(pid).match(/^p-(\d+)-/i);
+  return m ? Number(m[1]) : Number.POSITIVE_INFINITY;
+}
+
+function stableSortByTs(arr) {
+  if (!Array.isArray(arr)) return;
+  arr.sort((a, b) => {
+    const ta = Date.parse(a?.ts || "") || 0;
+    const tb = Date.parse(b?.ts || "") || 0;
+    if (ta !== tb) return ta - tb;
+    return JSON.stringify(a).localeCompare(JSON.stringify(b));
+  });
+}
+
+function keyMedia(x) { return String(x?.src || ""); }
+function keyRef(x) {
+  return `${x?.url || ""}||${x?.label || ""}||${x?.kind || ""}||${x?.citation || ""}`;
+}
+function keyComment(x) { return String(x?.text || "").trim(); }
+
+function uniqUnion(dst, src, keyFn) {
+  const out = isArr(dst) ? [...dst] : [];
+  const seen = new Set(out.map((x) => keyFn(x)));
+  for (const it of (isArr(src) ? src : [])) {
+    const k = keyFn(it);
+    if (!k) continue;
+    if (!seen.has(k)) {
+      seen.add(k);
+      out.push(it);
+    }
   }
   return out;
 }
 
-async function exists(p) {
+function deepMergeEntry(dst, src) {
-  try { await fs.access(p); return true; } catch { return false; }
+  if (!isObj(dst) || !isObj(src)) return;
+
+  for (const [k, v] of Object.entries(src)) {
+    if (k === "media" && isArr(v)) { dst.media = uniqUnion(dst.media, v, keyMedia); continue; }
+    if (k === "refs" && isArr(v)) { dst.refs = uniqUnion(dst.refs, v, keyRef); continue; }
+    if (k === "comments_editorial" && isArr(v)) { dst.comments_editorial = uniqUnion(dst.comments_editorial, v, keyComment); continue; }
+
+    if (isObj(v)) {
+      if (!isObj(dst[k])) dst[k] = {};
+      deepMergeEntry(dst[k], v);
+      continue;
+    }
+
+    if (isArr(v)) {
+      const cur = isArr(dst[k]) ? dst[k] : [];
+      const seen = new Set(cur.map((x) => JSON.stringify(x)));
+      const out = [...cur];
+      for (const it of v) {
+        const s = JSON.stringify(it);
+        if (!seen.has(s)) { seen.add(s); out.push(it); }
+      }
+      dst[k] = out;
+      continue;
+    }
+
+    // scalar: set only if missing/empty
+    if (!(k in dst) || dst[k] == null || dst[k] === "") dst[k] = v;
+  }
 }
 
 async function walk(dir) {
@@ -30,111 +108,119 @@ async function walk(dir) {
   const ents = await fs.readdir(dir, { withFileTypes: true });
   for (const e of ents) {
     const p = path.join(dir, e.name);
-    if (e.isDirectory()) out.push(...(await walk(p)));
+    if (e.isDirectory()) out.push(...await walk(p));
-    else out.push(p);
+    else if (e.isFile() && /\.ya?ml$/i.test(e.name)) out.push(p);
   }
   return out;
 }
 
-function inferPageKeyFromFile(inDirAbs, fileAbs) {
+function inferExpectedFromRel(relNoExt) {
-  // src/annotations/<page>.yml -> "<page>"
+  const parts = relNoExt.split("/").filter(Boolean);
-  const rel = path.relative(inDirAbs, fileAbs).replace(/\\/g, "/");
+  const last = parts.at(-1) || "";
-  return rel.replace(/\.(ya?ml|json)$/i, "");
+  const isShard = /^p-\d+-/i.test(last);
+  const pageKey = isShard ? parts.slice(0, -1).join("/") : relNoExt;
+  const paraId = isShard ? last : null;
+  return { isShard, pageKey, paraId };
 }
 
-function assert(cond, msg) {
+function validateAndNormalizeDoc(doc, relFile, expectedPageKey, expectedParaId) {
-  if (!cond) throw new Error(msg);
+  assert(isObj(doc), `${relFile}: doc must be an object`);
-}
+  assert(doc.schema === 1, `${relFile}: schema must be 1`);
+  assert(isObj(doc.paras), `${relFile}: missing object key "paras"`);
 
-function isPlainObject(x) {
+  const gotPage = doc.page != null ? normPath(doc.page) : "";
-  return !!x && typeof x === "object" && !Array.isArray(x);
+  const expPage = normPath(expectedPageKey);
-}
 
-function normalizePageKey(s) {
+  if (gotPage) {
-  // pas de / en tête/fin
-  return String(s || "").replace(/^\/+/, "").replace(/\/+$/, "");
-}
-
-function validateAndNormalizeDoc(doc, pageKey, fileRel) {
-  assert(isPlainObject(doc), `${fileRel}: document must be an object`);
-  assert(doc.schema === 1, `${fileRel}: schema must be 1`);
-  if (doc.page != null) {
     assert(
-      normalizePageKey(doc.page) === pageKey,
+      gotPage === expPage,
-      `${fileRel}: page mismatch (page="${doc.page}" vs path="${pageKey}")`
+      `${relFile}: page mismatch (page="${doc.page}" vs path="${expectedPageKey}")`
+    );
+  } else {
+    doc.page = expPage;
+  }
+
+  if (expectedParaId) {
+    // invariant shard : exactement 1 clé, celle du filename
+    const keys = Object.keys(doc.paras || {}).map(String);
+    assert(
+      keys.includes(expectedParaId),
+      `${relFile}: shard mismatch: must contain paras["${expectedParaId}"]`
+    );
+    assert(
+      keys.length === 1 && keys[0] === expectedParaId,
+      `${relFile}: shard invariant violated: shard file must contain ONLY paras["${expectedParaId}"] (got: ${keys.join(", ")})`
     );
   }
-  assert(isPlainObject(doc.paras), `${fileRel}: missing object key "paras"`);
 
-  const parasOut = Object.create(null);
+  return doc;
-
-  for (const [paraId, entry] of Object.entries(doc.paras)) {
-    assert(/^p-\d+-/i.test(paraId), `${fileRel}: invalid para id "${paraId}"`);
-
-    // entry peut être vide, mais doit être un objet si présent
-    assert(entry == null || isPlainObject(entry), `${fileRel}: paras.${paraId} must be an object`);
-
-    const e = entry ? { ...entry } : {};
-
-    // Sanity checks (non destructifs : on n’écrase pas, on vérifie juste les types)
-    if (e.refs != null) assert(Array.isArray(e.refs), `${fileRel}: paras.${paraId}.refs must be an array`);
-    if (e.authors != null) assert(Array.isArray(e.authors), `${fileRel}: paras.${paraId}.authors must be an array`);
-    if (e.quotes != null) assert(Array.isArray(e.quotes), `${fileRel}: paras.${paraId}.quotes must be an array`);
-    if (e.media != null) assert(Array.isArray(e.media), `${fileRel}: paras.${paraId}.media must be an array`);
-    if (e.comments_editorial != null) assert(Array.isArray(e.comments_editorial), `${fileRel}: paras.${paraId}.comments_editorial must be an array`);
-
-    parasOut[paraId] = e;
-  }
-
-  return parasOut;
-}
-
-async function readDoc(fileAbs) {
-  const raw = await fs.readFile(fileAbs, "utf8");
-  if (/\.json$/i.test(fileAbs)) return JSON.parse(raw);
-  return YAML.parse(raw);
 }
 
 async function main() {
-  const { inDir, outFile } = parseArgs(process.argv.slice(2));
+  const pages = {};
-  const CWD = process.cwd();
+  const errors = [];
 
-  const inDirAbs = path.isAbsolute(inDir) ? inDir : path.join(CWD, inDir);
+  await fs.mkdir(DIST_DIR, { recursive: true });
-  const outAbs = path.isAbsolute(outFile) ? outFile : path.join(CWD, outFile);
 
-  // antifragile
+  const files = await walk(ANNO_ROOT);
-  if (!(await exists(inDirAbs))) {
-    console.log(`ℹ️  annotations-index: skip (input missing): ${inDir}`);
-    process.exit(0);
-  }
 
-  const files = (await walk(inDirAbs)).filter((p) => /\.(ya?ml|json)$/i.test(p));
+  for (const fp of files) {
-  if (!files.length) {
+    const rel = normPath(path.relative(ANNO_ROOT, fp));              // e.g. archicrat-ia/chapitre-4/p-11-...
-    console.log(`ℹ️  annotations-index: skip (no .yml/.yaml/.json found in): ${inDir}`);
+    const relNoExt = rel.replace(/\.ya?ml$/i, "");                    // no ext
-    process.exit(0);
+    const { isShard, pageKey, paraId } = inferExpectedFromRel(relNoExt);
-  }
 
-  const pages = Object.create(null);
-  let paraCount = 0;
-
-  for (const f of files) {
-    const fileRel = path.relative(CWD, f).replace(/\\/g, "/");
-    const pageKey = normalizePageKey(inferPageKeyFromFile(inDirAbs, f));
-    assert(pageKey, `${fileRel}: cannot infer page key`);
-
-    let doc;
     try {
-      doc = await readDoc(f);
+      const raw = await fs.readFile(fp, "utf8");
+      const doc = YAML.parse(raw) || {};
+
+      // ignore non schema:1
+      if (!isObj(doc) || doc.schema !== 1) continue;
+
+      validateAndNormalizeDoc(
+        doc,
+        `src/annotations/${rel}`,
+        pageKey,
+        isShard ? paraId : null
+      );
+
+      const pg = (pages[pageKey] ??= { paras: {} });
+
+      if (isShard) {
+        const entry = doc.paras[paraId];
+        if (!isObj(pg.paras[paraId])) pg.paras[paraId] = {};
+        if (isObj(entry)) deepMergeEntry(pg.paras[paraId], entry);
+
+        stableSortByTs(pg.paras[paraId].media);
+        stableSortByTs(pg.paras[paraId].refs);
+        stableSortByTs(pg.paras[paraId].comments_editorial);
+      } else {
+        for (const [pid, entry] of Object.entries(doc.paras || {})) {
+          const p = String(pid);
+          if (!isObj(pg.paras[p])) pg.paras[p] = {};
+          if (isObj(entry)) deepMergeEntry(pg.paras[p], entry);
+
+          stableSortByTs(pg.paras[p].media);
+          stableSortByTs(pg.paras[p].refs);
+          stableSortByTs(pg.paras[p].comments_editorial);
+        }
+      }
     } catch (e) {
-      throw new Error(`${fileRel}: parse failed: ${String(e?.message ?? e)}`);
+      errors.push({ file: `src/annotations/${rel}`, error: String(e?.message || e) });
     }
+  }
 
-    const paras = validateAndNormalizeDoc(doc, pageKey, fileRel);
+  // sort paras per page
-
+  for (const [pageKey, pg] of Object.entries(pages)) {
-    // 1 fichier = 1 page (canon)
+    const keys = Object.keys(pg.paras || {});
-    assert(!pages[pageKey], `${fileRel}: duplicate page "${pageKey}" (only one file per page)`);
+    keys.sort((a, b) => {
-    pages[pageKey] = { paras };
+      const ia = paraNum(a);
-    paraCount += Object.keys(paras).length;
+      const ib = paraNum(b);
+      if (Number.isFinite(ia) && Number.isFinite(ib) && ia !== ib) return ia - ib;
+      return String(a).localeCompare(String(b));
+    });
+    const next = {};
+    for (const k of keys) next[k] = pg.paras[k];
+    pg.paras = next;
   }
 
   const out = {
@@ -143,17 +229,22 @@ async function main() {
     pages,
     stats: {
       pages: Object.keys(pages).length,
-      paras: paraCount,
+      paras: Object.values(pages).reduce((n, p) => n + Object.keys(p.paras || {}).length, 0),
+      errors: errors.length,
     },
+    errors,
   };
 
-  await fs.mkdir(path.dirname(outAbs), { recursive: true });
+  // CI behaviour: if ANY error => fail build
-  await fs.writeFile(outAbs, JSON.stringify(out), "utf8");
+  if (errors.length) {
+    throw new Error(`${errors[0].file}: ${errors[0].error}`);
+  }
 
-  console.log(`✅ annotations-index: pages=${out.stats.pages} paras=${out.stats.paras} -> ${path.relative(CWD, outAbs)}`);
+  await fs.writeFile(OUT, JSON.stringify(out), "utf8");
+  console.log(`✅ annotations-index: pages=${out.stats.pages} paras=${out.stats.paras} -> dist/annotations-index.json`);
 }
 
 main().catch((e) => {
-  console.error("FAIL: build-annotations-index crashed:", e);
+  console.error(`FAIL: build-annotations-index crashed: ${e?.stack || e?.message || e}`);
   process.exit(1);
-});
+});