ci: support shard annotations in checks + endpoint (pageKey inference)

scripts/build-annotations-index.mjs

@@ -117,7 +117,7 @@ async function walk(dir) {
 function inferExpectedFromRel(relNoExt) {
   const parts = relNoExt.split("/").filter(Boolean);
   const last = parts.at(-1) || "";
-  const isShard = /^p-\d+-/i.test(last);
+  const isShard = parts.length > 1 && /^p-\d+-/i.test(last); // ✅ durcissement
   const pageKey = isShard ? parts.slice(0, -1).join("/") : relNoExt;
   const paraId = isShard ? last : null;
   return { isShard, pageKey, paraId };
@@ -141,7 +141,6 @@ function validateAndNormalizeDoc(doc, relFile, expectedPageKey, expectedParaId)
   }
 
   if (expectedParaId) {
-    // invariant shard : exactement 1 clé, celle du filename
     const keys = Object.keys(doc.paras || {}).map(String);
     assert(
       keys.includes(expectedParaId),
@@ -165,15 +164,14 @@ async function main() {
   const files = await walk(ANNO_ROOT);
 
   for (const fp of files) {
-    const rel = normPath(path.relative(ANNO_ROOT, fp));              // e.g. archicrat-ia/chapitre-4/p-11-...
-    const relNoExt = rel.replace(/\.ya?ml$/i, "");                    // no ext
+    const rel = normPath(path.relative(ANNO_ROOT, fp));
+    const relNoExt = rel.replace(/\.ya?ml$/i, "");
     const { isShard, pageKey, paraId } = inferExpectedFromRel(relNoExt);
 
     try {
       const raw = await fs.readFile(fp, "utf8");
       const doc = YAML.parse(raw) || {};
 
-      // ignore non schema:1
       if (!isObj(doc) || doc.schema !== 1) continue;
 
       validateAndNormalizeDoc(
@@ -209,7 +207,6 @@ async function main() {
     }
   }
 
-  // sort paras per page
   for (const [pageKey, pg] of Object.entries(pages)) {
     const keys = Object.keys(pg.paras || {});
     keys.sort((a, b) => {
@@ -235,7 +232,6 @@ async function main() {
     errors,
   };
 
-  // CI behaviour: if ANY error => fail build
   if (errors.length) {
     throw new Error(`${errors[0].file}: ${errors[0].error}`);
   }

scripts/check-annotations-media.mjs

@@ -48,6 +48,9 @@ async function main() {
   let missing = 0;
   const notes = [];
 
+  // Optim: éviter de vérifier 100 fois le même fichier media
+  const seenMedia = new Set(); // src string
+
   for (const f of files) {
     const rel = path.relative(CWD, f).replace(/\\/g, "/");
     const raw = await fs.readFile(f, "utf8");
@@ -70,6 +73,10 @@ async function main() {
         const src = String(m?.src || "");
         if (!src.startsWith("/media/")) continue; // externes ok, ou autres conventions futures
 
+        // dédupe
+        if (seenMedia.has(src)) continue;
+        seenMedia.add(src);
+
         checked++;
         const p = toPublicPathFromUrl(src);
         if (!p) continue;
@@ -94,4 +101,4 @@ async function main() {
 main().catch((e) => {
   console.error("FAIL: check-annotations-media crashed:", e);
   process.exit(1);
-});
+});

scripts/check-annotations.mjs

@@ -27,11 +27,6 @@ function escRe(s) {
   return String(s).replace(/[.*+?^${}()|[\]\\]/g, "\\$&");
 }
 
-function inferPageKeyFromFile(fileAbs) {
-  const rel = path.relative(ANNO_DIR, fileAbs).replace(/\\/g, "/");
-  return rel.replace(/\.(ya?ml|json)$/i, "");
-}
-
 function normalizePageKey(s) {
   return String(s || "").replace(/^\/+/, "").replace(/\/+$/, "");
 }
@@ -40,6 +35,31 @@ function isPlainObject(x) {
   return !!x && typeof x === "object" && !Array.isArray(x);
 }
 
+function isParaId(s) {
+  return /^p-\d+-/i.test(String(s || ""));
+}
+
+/**
+ * Supporte:
+ * - monolith: src/annotations/<pageKey>.yml -> pageKey = rel sans ext
+ * - shard   : src/annotations/<pageKey>/<paraId>.yml -> pageKey = dirname(rel), paraId = basename
+ *
+ * shard seulement si le fichier est dans un sous-dossier (anti cas pathologique).
+ */
+function inferFromFile(fileAbs) {
+  const rel = path.relative(ANNO_DIR, fileAbs).replace(/\\/g, "/");
+  const relNoExt = rel.replace(/\.(ya?ml|json)$/i, "");
+  const parts = relNoExt.split("/").filter(Boolean);
+  const base = parts[parts.length - 1] || "";
+  const dirParts = parts.slice(0, -1);
+
+  const isShard = dirParts.length > 0 && isParaId(base);
+  const pageKey = isShard ? dirParts.join("/") : relNoExt;
+  const paraId = isShard ? base : "";
+
+  return { pageKey: normalizePageKey(pageKey), paraId };
+}
+
 async function loadAliases() {
   if (!(await exists(ALIASES_PATH))) return {};
   try {
@@ -83,7 +103,11 @@ async function main() {
   const aliases = await loadAliases();
   const files = (await walk(ANNO_DIR)).filter((p) => /\.(ya?ml|json)$/i.test(p));
 
-  let pages = 0;
+  // perf: cache HTML par page (shards = beaucoup de fichiers pour 1 page)
+  const htmlCache = new Map();        // pageKey -> html
+  const missingDistPage = new Set();  // pageKey
+
+  let pagesSeen = new Set();
   let checked = 0;
   let failures = 0;
   const notes = [];
@@ -107,7 +131,7 @@ async function main() {
       continue;
     }
 
-    const pageKey = normalizePageKey(inferPageKeyFromFile(f));
+    const { pageKey, paraId: shardParaId } = inferFromFile(f);
 
     if (doc.page != null && normalizePageKey(doc.page) !== pageKey) {
       failures++;
@@ -121,20 +145,44 @@ async function main() {
       continue;
     }
 
+    // shard invariant (fort) : doit contenir paras[paraId]
+    if (shardParaId) {
+      if (!Object.prototype.hasOwnProperty.call(doc.paras, shardParaId)) {
+        failures++;
+        notes.push(`- SHARD MISMATCH: ${rel} (expected paras["${shardParaId}"] present)`);
+        continue;
+      }
+      // si extras -> warning (non destructif)
+      const keys = Object.keys(doc.paras);
+      if (!(keys.length === 1 && keys[0] === shardParaId)) {
+        notes.push(`- WARN shard has extra paras: ${rel} (expected only "${shardParaId}", got ${keys.join(", ")})`);
+      }
+    }
+
+    pagesSeen.add(pageKey);
+
     const distFile = path.join(DIST_DIR, pageKey, "index.html");
     if (!(await exists(distFile))) {
-      failures++;
-      notes.push(`- MISSING PAGE: dist/${pageKey}/index.html (from ${rel})`);
+      if (!missingDistPage.has(pageKey)) {
+        missingDistPage.add(pageKey);
+        failures++;
+        notes.push(`- MISSING PAGE: dist/${pageKey}/index.html (from ${rel})`);
+      } else {
+        notes.push(`- WARN missing page already reported: dist/${pageKey}/index.html (from ${rel})`);
+      }
       continue;
     }
 
-    pages++;
-    const html = await fs.readFile(distFile, "utf8");
+    let html = htmlCache.get(pageKey);
+    if (!html) {
+      html = await fs.readFile(distFile, "utf8");
+      htmlCache.set(pageKey, html);
+    }
 
     for (const paraId of Object.keys(doc.paras)) {
       checked++;
 
-      if (!/^p-\d+-/i.test(paraId)) {
+      if (!isParaId(paraId)) {
         failures++;
         notes.push(`- INVALID ID: ${rel} (${paraId})`);
         continue;
@@ -158,6 +206,7 @@ async function main() {
   }
 
   const warns = notes.filter((x) => x.startsWith("- WARN"));
+  const pages = pagesSeen.size;
 
   if (failures > 0) {
     console.error(`FAIL: annotations invalid (pages=${pages} checked=${checked} failures=${failures})`);
@@ -172,4 +221,4 @@ async function main() {
 main().catch((e) => {
   console.error("FAIL: annotations check crashed:", e);
   process.exit(1);
-});
+});