@aeptus/aexeo-emdash 0.8.19 → 0.9.0

package/dist/configured.js

@@ -5,13 +5,15 @@ var __export = (target, all) => {
 };
 
 // src/adapter.ts
-async function adaptContentItemWithContext(content, contentApi, defaultLocale) {
+async function adaptContentItemWithContext(content, contentApi, defaultLocale, blockMappings, contentFields) {
   const publicUrl = content.status === "published" ? await contentApi.getPublicUrl(content.type, content.id) : null;
   const route = publicUrl === null ? deriveRoute(content) : routeFromPublicUrl(publicUrl);
   const adapted = adaptContentItem(content, {
     route,
     publicUrl,
-    ...defaultLocale === void 0 ? {} : { defaultLocale }
+    ...defaultLocale === void 0 ? {} : { defaultLocale },
+    ...blockMappings === void 0 ? {} : { blockMappings },
+    ...contentFields === void 0 ? {} : { contentFields }
   });
   const translationSet = await contentApi.getTranslations(
     content.type,
@@ -53,8 +55,11 @@ function adaptContentItem(content, options = {}) {
   if (typeof locale === "string" && locale.length > 0) {
     document.lang = locale;
   }
-  const body = extractPortableText(content.data);
-  if (body !== null) document.body = body;
+  const body = options.contentFields === void 0 ? extractPortableText(content.data, options.blockMappings) : extractSelectedContent(content.data, options.contentFields, options.blockMappings);
+  if (body !== null) {
+    document.body = body.blocks;
+    if (body.capabilities !== void 0) document.body_capabilities = body.capabilities;
+  }
   return {
     document,
     meta: {
@@ -63,7 +68,8 @@ function adaptContentItem(content, options = {}) {
       slug: content.slug,
       status: content.status,
       title,
-      publicUrl: options.publicUrl ?? null
+      publicUrl: options.publicUrl ?? null,
+      ...content.updatedAt === void 0 ? {} : { updatedAt: content.updatedAt }
     }
   };
 }
@@ -82,24 +88,53 @@ function routeFromPublicUrl(publicUrl) {
 function stringOrEmpty(value) {
   return typeof value === "string" ? value : "";
 }
-function extractPortableText(data) {
+function extractSelectedContent(data, fields, mappings) {
+  if (fields.length === 0 || new Set(fields).size !== fields.length || fields.some((field) => typeof field !== "string" || field.length === 0)) return null;
+  const all = ["text", "headings", "paragraphs", "links"];
+  let capabilities = all;
+  const blocks = [];
+  let observed = false;
+  for (const field of fields) {
+    const value = data[field];
+    if (value === void 0 || value === null) continue;
+    observed = true;
+    const part = extractFirstPortableText({ content: value }, mappings);
+    if (part === null) return { blocks, capabilities: [] };
+    blocks.push(...part.blocks);
+    capabilities = capabilities.filter((capability) => part.capabilities === void 0 || part.capabilities.includes(capability));
+  }
+  if (!observed) return null;
+  return capabilities.length === all.length ? { blocks } : { blocks, capabilities: [...capabilities] };
+}
+function extractPortableText(data, mappings) {
+  const result = extractFirstPortableText(data, mappings);
+  if (result === null) return null;
+  const candidates = Object.entries(data).filter(
+    ([field, value]) => isPortableTextArray(value) || isEmdashBlocksArray(value, field) || (field === "body" || field === "content") && typeof value === "string" && value.trim().length > 0
+  );
+  return candidates.length > 1 ? { blocks: result.blocks, capabilities: [] } : result;
+}
+function extractFirstPortableText(data, mappings) {
+  let foundEmptyBody = false;
   for (const slug of ["body", "content"]) {
     const value = data[slug];
-    if (isPortableTextArray(value)) return value;
-    if (isEmdashBlocksArray(value, slug)) return blocksToPortableText(value);
+    if (Array.isArray(value) && value.length === 0 || typeof value === "string" && value.trim().length === 0) foundEmptyBody = true;
+    if (isPortableTextArray(value)) return { blocks: value };
+    if (isEmdashBlocksArray(value, slug)) return mapCustomBlocks(value, mappings);
     if (typeof value === "string" && value.trim().length > 0) {
-      return [portableTextBlock(value.trim(), `${slug}-text`)];
+      return { blocks: [portableTextBlock(value.trim(), `${slug}-text`)], capabilities: ["text"] };
     }
   }
   const blocks = data["blocks"];
-  if (isPortableTextArray(blocks)) return blocks;
-  if (isEmdashBlocksArray(blocks, "blocks")) return blocksToPortableText(blocks);
+  if (Array.isArray(blocks) && blocks.length === 0) foundEmptyBody = true;
+  if (isPortableTextArray(blocks)) return { blocks };
+  if (isEmdashBlocksArray(blocks, "blocks")) return mapCustomBlocks(blocks, mappings);
   for (const [field, value] of Object.entries(data)) {
     if (field === "body" || field === "content" || field === "blocks") continue;
-    if (isPortableTextArray(value)) return value;
-    if (isEmdashBlocksArray(value, field)) return blocksToPortableText(value);
+    if (isPortableTextArray(value)) return { blocks: value };
+    if (isEmdashBlocksArray(value, field)) return mapCustomBlocks(value, mappings);
   }
-  return null;
+  return foundEmptyBody ? { blocks: [] } : null;
 }
 function isPortableTextArray(value) {
   return Array.isArray(value) && value.length > 0 && value.every(
@@ -114,6 +149,52 @@ function isEmdashBlocksArray(value, field) {
     (item) => isRecord(item) && typeof item["_type"] === "string" && typeof item["_key"] === "string" && (typeof item["_version"] === "number" || field === "blocks")
   );
 }
+function mapCustomBlocks(blocks, mappings) {
+  const unavailable = () => ({ blocks: blocksToPortableText(blocks), capabilities: [] });
+  const result = [];
+  const mapFields = (record, fields, depth) => {
+    if (depth > 8 || fields.length === 0) return false;
+    for (const entry of fields) {
+      if (!entry || typeof entry.field !== "string" || !entry.field || !["heading", "paragraph", "richText", "link", "repeat"].includes(entry.kind)) return false;
+      if (entry.kind === "heading" && entry.level !== void 0 && ![1, 2, 3, 4, 5, 6].includes(entry.level)) return false;
+      if (entry.kind === "link" && (typeof entry.hrefField !== "string" || !entry.hrefField)) return false;
+      if (entry.kind === "repeat" && (!Array.isArray(entry.fields) || entry.fields.length === 0)) return false;
+      const value = record[entry.field];
+      if ((value === void 0 || value === null || value === "") && entry.optional) {
+        if (entry.kind === "link" && record[entry.hrefField] != null && record[entry.hrefField] !== "") return false;
+        continue;
+      }
+      if (entry.kind === "richText") {
+        if (!Array.isArray(value) || value.length > 0 && !isPortableTextArray(value)) return false;
+        result.push(...value);
+      } else if (entry.kind === "heading" || entry.kind === "paragraph") {
+        if (typeof value !== "string") return false;
+        result.push({ ...portableTextBlock(value, ""), style: entry.kind === "heading" ? `h${entry.level ?? 2}` : "normal" });
+      } else if (entry.kind === "link") {
+        const href = record[entry.hrefField];
+        if (typeof value !== "string" || typeof href !== "string" || !href) return false;
+        result.push({
+          _type: "block",
+          children: [{ _type: "span", text: value, marks: ["link"] }],
+          markDefs: [{ _key: "link", _type: "link", href }]
+        });
+      } else if (entry.kind === "repeat") {
+        if (!Array.isArray(value) || !Array.isArray(entry.fields)) return false;
+        for (const item of value) {
+          if (!isRecord(item) || !mapFields(item, entry.fields, depth + 1)) return false;
+        }
+      } else {
+        return false;
+      }
+    }
+    return true;
+  };
+  for (const block of blocks) {
+    const fields = mappings?.[String(block["_type"])];
+    if (!Array.isArray(fields) || !mapFields(block, fields, 0)) return unavailable();
+  }
+  return { blocks: result };
+}
 function blocksToPortableText(blocks) {
   const paragraphs = [];
   for (const block of blocks) {
@@ -197,35 +278,107 @@ function isRecord(value) {
   return typeof value === "object" && value !== null && !Array.isArray(value);
 }
 
+// src/version.ts
+var PACKAGE_VERSION = "0.9.0";
+
 // src/plugin.ts
 var SETTINGS_TOKEN_KEY = "eval_token";
+var SETTINGS_BLOCK_MAPPINGS_KEY = "block_mappings";
+var SETTINGS_CONTENT_FIELDS_KEY = "content_fields_by_collection";
 var PLUGIN_SETTINGS_SCHEMA = {
+  [SETTINGS_CONTENT_FIELDS_KEY]: {
+    type: "string",
+    label: "Content fields by collection (JSON)",
+    description: "Ordered rendered body fields per collection.",
+    multiline: true
+  },
+  [SETTINGS_BLOCK_MAPPINGS_KEY]: {
+    type: "string",
+    label: "Custom block mappings (JSON)",
+    description: "Ordered mappings for the rendered fields of each custom block type.",
+    multiline: true
+  },
   [SETTINGS_TOKEN_KEY]: {
     type: "secret",
     label: "EVAL_TOKEN",
     description: "Bearer token for your aexeo-crawl-worker Worker. Encrypted at rest by the emdash host."
   }
 };
-async function handleAfterSaveConfigured(event, ctx, evaluator, suppressionFilter) {
+function recordContentCoverage(ctx, documents) {
+  const counts = { documents: documents.length, complete: 0, partial: 0, unavailable: 0 };
+  for (const document of documents) {
+    if (document.body === void 0 || document.body_capabilities?.length === 0) counts.unavailable++;
+    else if (document.body_capabilities === void 0 || ["text", "headings", "paragraphs", "links"].every(
+      (capability) => document.body_capabilities?.includes(capability)
+    )) counts.complete++;
+    else counts.partial++;
+  }
+  try {
+    ctx.log?.info?.("aexeo.content_coverage", counts);
+  } catch {
+  }
+}
+async function handleAfterDelete(event, ctx) {
+  for (const entry of await ctx.kv.list("document:")) {
+    const stored = toStoredDocument(entry.value);
+    if (stored !== null && stored.meta.id === event.id && stored.meta.collection === event.collection) {
+      await ctx.kv.delete(entry.key);
+      await ctx.kv.delete(findingsKey(stored.document.route));
+    }
+  }
+}
+var SITE_GRAPH_RULE_IDS = /* @__PURE__ */ new Set([
+  "LNK001",
+  // internal link target missing from the site
+  "LNK002",
+  // no inbound links
+  "LNK003",
+  // weak anchor, checked only when the target route exists
+  "LNK004",
+  // too few inbound links
+  "SEO009",
+  // hreflang alternate target missing from the site
+  "SEO012",
+  // hreflang alternate does not link back
+  "SEO016",
+  // canonical points at another crawlable route
+  "SEO017",
+  // title and description shared with another route
+  "CNT005"
+  // visible text duplicated on another route
+]);
+async function handleAfterSaveConfigured(event, ctx, evaluator, suppressionFilter, blockMappings, contentFields) {
   const adapted = await adaptContentItemWithContext(
     event.content,
     ctx.content,
-    ctx.site?.locale
+    ctx.site?.locale,
+    blockMappings,
+    contentFields
   );
   const document = adapted.document;
+  recordContentCoverage(ctx, [document]);
   const { kv, log } = ctx;
   const stored = { document, meta: adapted.meta };
-  await persistDocument(ctx, stored);
-  const result = await evaluator([document]);
+  if (!await persistDocument(ctx, stored)) {
+    log?.info?.(`Aexeo afterSave skipped: ${document.route} has a newer save`);
+    return;
+  }
+  const result = await evaluator([document], { collections: [adapted.meta.collection] });
   if (!result.ok) {
     log?.error?.(`afterSave evaluator failure (${result.reason})`, {
       detail: result.detail
     });
     return;
   }
-  const pageFindings = result.findings.filter(
-    (finding) => finding.scope === "page"
+  const storedGraphFindings = (await readFindings(kv, document.route)).filter(
+    (finding) => SITE_GRAPH_RULE_IDS.has(finding.rule_id)
   );
+  const pageFindings = [
+    ...result.findings.filter(
+      (finding) => finding.scope === "page" && !SITE_GRAPH_RULE_IDS.has(finding.rule_id)
+    ),
+    ...storedGraphFindings
+  ];
   const filtered = suppressionFilter === void 0 ? pageFindings : suppressionFilter.apply(
     {
       route: document.route,
@@ -234,6 +387,10 @@ async function handleAfterSaveConfigured(event, ctx, evaluator, suppressionFilte
     },
     pageFindings
   );
+  if (await isSuperseded(kv, document.route, adapted.meta.updatedAt)) {
+    log?.info?.(`Aexeo afterSave discarded: ${document.route} changed while evaluating`);
+    return;
+  }
   await kv.set(findingsKey(document.route), {
     route: document.route,
     findings: filtered
@@ -250,7 +407,9 @@ async function evaluateAndPersistAll(ctx, options) {
     totalFindings: 0,
     errors: []
   };
+  const storedAtStart = await readAllStoredDocuments(kv);
   const documents = [];
+  const documentCollections = [];
   const adaptedByRoute = /* @__PURE__ */ new Map();
   const documentRoutes = /* @__PURE__ */ new Set();
   for (const collection of collections) {
@@ -265,9 +424,12 @@ async function evaluateAndPersistAll(ctx, options) {
           const adapted = await adaptContentItemWithContext(
             item,
             ctx.content,
-            ctx.site?.locale
+            ctx.site?.locale,
+            options.blockMappings,
+            options.contentFieldsByCollection?.[collection]
           );
           documents.push(adapted.document);
+          documentCollections.push(adapted.meta.collection);
           documentRoutes.add(adapted.document.route);
           adaptedByRoute.set(adapted.document.route, {
             document: adapted.document,
@@ -283,10 +445,14 @@ async function evaluateAndPersistAll(ctx, options) {
     } while (cursor !== void 0);
   }
   summary.documentsScanned = documents.length;
+  recordContentCoverage(ctx, documents);
   for (const stored of adaptedByRoute.values()) {
-    await kv.set(documentKey(stored.document.route), stored);
+    const current = await readStoredDocument(kv, stored.document.route);
+    if (!isOlderRevision(stored.meta.updatedAt, current?.meta.updatedAt)) {
+      await kv.set(documentKey(stored.document.route), stored);
+    }
   }
-  const result = await options.evaluator(documents);
+  const result = await options.evaluator(documents, { collections: documentCollections });
   if (!result.ok) {
     log?.error?.(`aexeo evaluator failure (${result.reason})`, {
       detail: result.detail
@@ -307,6 +473,9 @@ async function evaluateAndPersistAll(ctx, options) {
   let totalAfterSuppression = 0;
   for (const [route, findings] of findingsByRoute) {
     const stored = adaptedByRoute.get(route);
+    if (stored !== void 0 && await isSuperseded(kv, route, stored.meta.updatedAt)) {
+      continue;
+    }
     const context = {
       route,
       ...stored?.meta.collection === void 0 ? {} : { collection: stored.meta.collection },
@@ -319,29 +488,52 @@ async function evaluateAndPersistAll(ctx, options) {
   }
   summary.totalFindings = totalAfterSuppression;
   if (summary.errors.length === 0) {
-    await removeStaleDocuments(ctx, collections, documentRoutes);
+    await removeStaleDocuments(ctx, collections, documentRoutes, storedAtStart);
   }
   return summary;
 }
+function isOlderRevision(revision, other) {
+  if (revision === void 0 || other === void 0) return false;
+  return Date.parse(revision) < Date.parse(other);
+}
+async function readStoredDocument(kv, route) {
+  return toStoredDocument(await kv.get(documentKey(route)));
+}
+async function isSuperseded(kv, route, revision) {
+  const current = await readStoredDocument(kv, route);
+  return current === null || isOlderRevision(revision, current.meta.updatedAt);
+}
 async function persistDocument(ctx, stored) {
   const route = stored.document.route;
-  const entries = await ctx.kv.list("document:");
-  for (const entry of entries) {
+  const sameEntry = (await ctx.kv.list("document:")).flatMap((entry) => {
     const previous = toStoredDocument(entry.value);
-    if (previous === null || previous.document.route === route || previous.meta.id !== stored.meta.id || previous.meta.collection !== stored.meta.collection) {
-      continue;
-    }
-    await ctx.kv.delete(entry.key);
+    return previous !== null && previous.meta.id === stored.meta.id && previous.meta.collection === stored.meta.collection ? [{ key: entry.key, previous }] : [];
+  });
+  if (sameEntry.some(
+    ({ previous }) => isOlderRevision(stored.meta.updatedAt, previous.meta.updatedAt)
+  )) {
+    return false;
+  }
+  for (const { key, previous } of sameEntry) {
+    if (previous.document.route === route) continue;
+    await ctx.kv.delete(key);
     await ctx.kv.delete(findingsKey(previous.document.route));
   }
   await ctx.kv.set(documentKey(route), stored);
+  return true;
 }
-async function removeStaleDocuments(ctx, collections, currentRoutes) {
+async function removeStaleDocuments(ctx, collections, currentRoutes, storedAtStart) {
   const scannedCollections = new Set(collections);
+  const revisionAtStart = new Map(
+    storedAtStart.map((stored) => [
+      stored.document.route,
+      stored.meta.updatedAt
+    ])
+  );
   const entries = await ctx.kv.list("document:");
   for (const entry of entries) {
     const stored = toStoredDocument(entry.value);
-    if (stored === null || !scannedCollections.has(stored.meta.collection) || currentRoutes.has(stored.document.route)) {
+    if (stored === null || stored.meta.collection !== "" && !scannedCollections.has(stored.meta.collection) || currentRoutes.has(stored.document.route) || !revisionAtStart.has(stored.document.route) || revisionAtStart.get(stored.document.route) !== stored.meta.updatedAt) {
       continue;
     }
     await ctx.kv.delete(entry.key);
@@ -438,6 +630,8 @@ async function handleRefresh(options) {
     summary = await evaluateAndPersistAll(options.ctx, {
       collections: options.collections,
       evaluator: options.evaluator,
+      blockMappings: options.blockMappings,
+      contentFieldsByCollection: options.contentFieldsByCollection,
       // Always forwarded. See RefreshOptions.suppressionFilter.
       suppressionFilter: options.suppressionFilter
     });
@@ -578,7 +772,7 @@ function findingsTable(rows) {
     }))
   };
 }
-async function renderScoreWidget(kv, tuning) {
+async function renderScoreWidget(kv, tuning, log) {
   const documents = await readAllDocuments(kv);
   if (documents.length === 0) {
     return {
@@ -591,28 +785,48 @@ async function renderScoreWidget(kv, tuning) {
       ]
     };
   }
-  const score = await tuning.score(documents, kv);
+  let score;
+  try {
+    score = await tuning.score(documents, kv);
+  } catch (error) {
+    log?.error("Aexeo score widget failed to initialize", { error });
+    return {
+      blocks: [
+        { type: "banner", title: "SEO score is temporarily unavailable", variant: "alert" },
+        { type: "context", text: "The authoring interface remains available. Retry after the evaluator runtime has been checked." }
+      ]
+    };
+  }
+  if (score === null) {
+    return {
+      blocks: [
+        { type: "header", text: "SEO score" },
+        {
+          type: "context",
+          text: "The site score is unavailable in sandboxed mode. Findings are not affected."
+        }
+      ]
+    };
+  }
   const blocks = [
     {
       type: "stats",
       items: [
-        { label: "Overall", value: `${score.overall_score}` },
-        { label: "Citation", value: `${score.citation_readiness_score}` },
+        {
+          label: "Citation",
+          value: score.citation_readiness_score?.toString() ?? "unknown"
+        },
         {
           label: tuning.truthLabel(score),
-          value: `${score.truth_consistency_score}`
+          value: score.truth_consistency_score?.toString() ?? "unknown"
         },
-        { label: "Answers", value: `${score.answer_pack_score}` }
+        {
+          label: "Answers",
+          value: `${score.answer_pack.routes_with_answer_blocks}/${score.answer_pack.routes_analyzed} routes`
+        }
       ]
     }
   ];
-  if (score.overall_score < 60) {
-    blocks.unshift({
-      type: "banner",
-      title: `Site score is ${score.overall_score} \u2014 below the 60 quality threshold`,
-      variant: "alert"
-    });
-  }
   if (score.blockers.length > 0) {
     blocks.push({
       type: "context",
@@ -695,11 +909,65 @@ __export(aexeo_emdash_bridge_bg_exports, {
   __wbg_set_wasm: () => __wbg_set_wasm,
   __wbg_stack_3b0d974bbf31e44f: () => __wbg_stack_3b0d974bbf31e44f,
   __wbindgen_init_externref_table: () => __wbindgen_init_externref_table,
+  evaluateContent: () => evaluateContent,
+  evaluateContentCollection: () => evaluateContentCollection,
   evaluateDocuments: () => evaluateDocuments,
   generateFactsPrompt: () => generateFactsPrompt,
   scoreIntelligence: () => scoreIntelligence,
   validateFactsManifest: () => validateFactsManifest
 });
+function evaluateContent(document_json, preset_name, sources_json, evidence_json) {
+  let deferred6_0;
+  let deferred6_1;
+  try {
+    const ptr0 = passStringToWasm0(document_json, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const len0 = WASM_VECTOR_LEN;
+    const ptr1 = passStringToWasm0(preset_name, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const len1 = WASM_VECTOR_LEN;
+    var ptr2 = isLikeNone(sources_json) ? 0 : passStringToWasm0(sources_json, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    var len2 = WASM_VECTOR_LEN;
+    var ptr3 = isLikeNone(evidence_json) ? 0 : passStringToWasm0(evidence_json, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    var len3 = WASM_VECTOR_LEN;
+    const ret = wasm.evaluateContent(ptr0, len0, ptr1, len1, ptr2, len2, ptr3, len3);
+    var ptr5 = ret[0];
+    var len5 = ret[1];
+    if (ret[3]) {
+      ptr5 = 0;
+      len5 = 0;
+      throw takeFromExternrefTable0(ret[2]);
+    }
+    deferred6_0 = ptr5;
+    deferred6_1 = len5;
+    return getStringFromWasm0(ptr5, len5);
+  } finally {
+    wasm.__wbindgen_free(deferred6_0, deferred6_1, 1);
+  }
+}
+function evaluateContentCollection(documents_json, preset_name, sources_json) {
+  let deferred5_0;
+  let deferred5_1;
+  try {
+    const ptr0 = passStringToWasm0(documents_json, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const len0 = WASM_VECTOR_LEN;
+    const ptr1 = passStringToWasm0(preset_name, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const len1 = WASM_VECTOR_LEN;
+    var ptr2 = isLikeNone(sources_json) ? 0 : passStringToWasm0(sources_json, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    var len2 = WASM_VECTOR_LEN;
+    const ret = wasm.evaluateContentCollection(ptr0, len0, ptr1, len1, ptr2, len2);
+    var ptr4 = ret[0];
+    var len4 = ret[1];
+    if (ret[3]) {
+      ptr4 = 0;
+      len4 = 0;
+      throw takeFromExternrefTable0(ret[2]);
+    }
+    deferred5_0 = ptr4;
+    deferred5_1 = len4;
+    return getStringFromWasm0(ptr4, len4);
+  } finally {
+    wasm.__wbindgen_free(deferred5_0, deferred5_1, 1);
+  }
+}
 function evaluateDocuments(documents_json, config_json) {
   let deferred4_0;
   let deferred4_1;
@@ -969,6 +1237,10 @@ async function validateFactsManifest2(manifestJson, documentsJson) {
   await ensureInitialized();
   return validateFactsManifest(manifestJson, documentsJson);
 }
+async function evaluateContent2(documentJson, presetName, sourcesJson, evidenceJson) {
+  await ensureInitialized();
+  return evaluateContent(documentJson, presetName, sourcesJson, evidenceJson);
+}
 
 // src/data-route.ts
 async function handleDataRoute(ctx, options) {
@@ -977,6 +1249,8 @@ async function handleDataRoute(ctx, options) {
     summary = await evaluateAndPersistAll(ctx, {
       collections: options.collections,
       evaluator: options.evaluator,
+      blockMappings: options.blockMappings,
+      contentFieldsByCollection: options.contentFieldsByCollection,
       ...options.suppressionFilter === void 0 ? {} : { suppressionFilter: options.suppressionFilter }
     });
     await persistManifestStateFindings(ctx);
@@ -1245,12 +1519,16 @@ async function handleSave(ctx, body) {
   return { saved: true };
 }
 
-// src/presence-route.ts
-import { PluginRouteError as PluginRouteError2 } from "emdash";
-
 // src/presence-check.ts
 var FETCH_TIMEOUT_MS = 5e3;
 var USER_AGENT = "aexeo-emdash-plugin/0.8 (+https://github.com/schiste/Aexeo; entity-presence diagnostic)";
+var PRESENCE_HOSTS = [
+  "en.wikipedia.org",
+  "www.wikidata.org",
+  "api.github.com",
+  "rdap.org",
+  "index.commoncrawl.org"
+];
 var SOURCE_NAMES = [
   "wikipedia",
   "wikidata",
@@ -1307,7 +1585,7 @@ async function checkWikipedia(input, fetcher) {
       return notFound2("wikipedia");
     }
     const title = String(titles[0]);
-    if (!fuzzyMatch(title, input.name)) {
+    if (!namesEntity(title, input)) {
       return notFound2("wikipedia");
     }
     const articleUrl = String(urls[0] ?? "");
@@ -1337,7 +1615,7 @@ async function checkWikidata(input, fetcher) {
       return notFound2("wikidata");
     }
     const label = top.label ?? input.name;
-    if (!fuzzyMatch(label, input.name)) {
+    if (!namesEntity(label, input)) {
       return notFound2("wikidata");
     }
     return found("wikidata", {
@@ -1426,9 +1704,27 @@ async function checkCommonCrawl(input, fetcher) {
       "no website URL in manifest, can't query crawl index"
     );
   }
-  const index = "CC-MAIN-2026-15";
+  let index;
+  try {
+    const res = await timedFetch(
+      fetcher,
+      "https://index.commoncrawl.org/collinfo.json"
+    );
+    if (!res.ok) return unreachable("common_crawl", `HTTP ${res.status}`);
+    const newest = newestCrawlId(await res.json());
+    if (newest === null) {
+      return unreachable(
+        "common_crawl",
+        "no crawl index listed in collinfo.json"
+      );
+    }
+    index = newest;
+  } catch (error) {
+    return unreachable("common_crawl", errMsg(error));
+  }
   const url = new URL(`https://index.commoncrawl.org/${index}-index`);
   url.searchParams.set("url", host);
+  url.searchParams.set("matchType", "domain");
   url.searchParams.set("output", "json");
   url.searchParams.set("limit", "1");
   try {
@@ -1455,6 +1751,16 @@ async function checkCommonCrawl(input, fetcher) {
     return unreachable("common_crawl", errMsg(error));
   }
 }
+function newestCrawlId(collinfo) {
+  if (!Array.isArray(collinfo)) return null;
+  let newest = null;
+  for (const crawl of collinfo) {
+    const id = crawl?.id;
+    if (typeof id !== "string" || !/^CC-MAIN-[A-Za-z0-9-]+$/.test(id)) continue;
+    if (newest === null || id > newest) newest = id;
+  }
+  return newest;
+}
 async function timedFetch(fetcher, url, init) {
   const controller = new AbortController();
   const timer = setTimeout(() => controller.abort(), FETCH_TIMEOUT_MS);
@@ -1500,10 +1806,45 @@ function errMsg(error) {
   }
   return String(error);
 }
-function fuzzyMatch(a, b) {
-  const aL = a.toLowerCase().trim();
-  const bL = b.toLowerCase().trim();
-  return aL.includes(bL) || bL.includes(aL);
+function namesEntity(candidate, input) {
+  const normalized = normalizeEntityName(candidate);
+  return [input.name, ...input.aliases ?? []].some(
+    (name) => normalizeEntityName(name) === normalized
+  );
+}
+var LEGAL_FORMS = /* @__PURE__ */ new Set([
+  "ab",
+  "ag",
+  "bv",
+  "co",
+  "company",
+  "corp",
+  "corporation",
+  "gmbh",
+  "inc",
+  "incorporated",
+  "kk",
+  "limited",
+  "llc",
+  "llp",
+  "lp",
+  "ltd",
+  "nv",
+  "oy",
+  "plc",
+  "pty",
+  "sa",
+  "sarl",
+  "sas",
+  "srl"
+]);
+function normalizeEntityName(name) {
+  const words = name.replace(/\([^)]*\)/g, " ").toLowerCase().split(/[^\p{L}\p{N}]+/u).filter((word) => word.length > 0);
+  let end = words.length;
+  while (end > 0 && LEGAL_FORMS.has(words[end - 1])) end -= 1;
+  const start = words[0] === "the" ? 1 : 0;
+  const core = words.slice(start, end);
+  return (core.length > 0 ? core : words).join(" ");
 }
 function sanitizeGitHubHandle(name) {
   const handle = name.replace(/[\s_]+/g, "-");
@@ -1542,6 +1883,7 @@ function formatCdxTimestamp(ts) {
 }
 
 // src/presence-route.ts
+import { PluginRouteError as PluginRouteError2 } from "emdash";
 var CACHE_TTL_MS = 24 * 60 * 60 * 1e3;
 var PRESENCE_KEY = "presence:current";
 async function handlePresenceRoute(ctx) {
@@ -1664,8 +2006,61 @@ function globToMatcher(pattern) {
   return (route) => re.test(route);
 }
 
-// src/version.ts
-var PACKAGE_VERSION = "0.8.19";
+// src/content.ts
+async function evaluateContent3(document, preset, sources = {}, evidenceSources, onTelemetry) {
+  const started = onTelemetry ? performance.now() : 0;
+  const documentJson = JSON.stringify(document);
+  const sourcesJson = JSON.stringify(sources);
+  const result = evidenceSources === void 0 ? await evaluateContent2(documentJson, preset, sourcesJson) : await evaluateContent2(documentJson, preset, sourcesJson, JSON.stringify(evidenceSources));
+  const report = JSON.parse(result);
+  if (onTelemetry) {
+    const event = { evaluation_ms: performance.now() - started, source_locations_mapped: report.source_locations?.length ?? 0 };
+    for (const gate of report.gates) {
+      const checked = gate.measurements?.checked_destinations;
+      const invalid = gate.measurements?.invalid_destinations;
+      if (checked !== void 0 && invalid !== void 0) {
+        event.link_destinations ??= { checked: 0, invalid: 0 };
+        event.link_destinations.checked += checked;
+        event.link_destinations.invalid += invalid;
+      }
+    }
+    onTelemetry(event);
+  }
+  return report;
+}
+
+// src/content-evaluator.ts
+function withContentPreset(base, preset, evaluate) {
+  if (!preset) return base;
+  return async (documents, context) => {
+    const result = await base(documents, context);
+    if (!result.ok) return result;
+    const findings = [...result.findings];
+    try {
+      for (const [index, document] of documents.entries()) {
+        const collection = context?.collections[index];
+        if (preset.collections !== void 0) {
+          if (collection === void 0) throw new Error("collection is required for a scoped content preset");
+          if (!preset.collections.includes(collection)) continue;
+        }
+        const selected = collection !== void 0 && preset.byCollection !== void 0 && Object.hasOwn(preset.byCollection, collection) ? preset.byCollection[collection] : void 0;
+        const name = selected ?? preset.name;
+        if (!name) throw new Error(`empty preset name for collection ${collection}`);
+        const report = preset.evidenceSources === void 0 ? await evaluate(document, name, preset.sources) : await evaluate(document, name, preset.sources, preset.evidenceSources);
+        const incomplete = report.gates.filter((gate) => gate.status === "not_checked");
+        if (incomplete.length) return {
+          ok: false,
+          reason: "content_preset_incomplete",
+          detail: `${document.route}: ${incomplete.map((gate) => `${gate.gate_id}: ${gate.reason}`).join(", ")}`
+        };
+        for (const gate of report.gates) findings.push(...gate.findings);
+      }
+      return { ok: true, findings };
+    } catch (error) {
+      return { ok: false, reason: "content_preset_error", detail: error instanceof Error ? error.message : String(error) };
+    }
+  };
+}
 
 // src/configured.ts
 import { definePlugin } from "emdash";
@@ -1731,10 +2126,15 @@ function createPlugin(options = {}) {
     options.excludeCollections
   );
   const suppressionFilter = compileSuppressions(options.suppressions);
+  const evaluator = withContentPreset(inProcessEvaluator, options.contentPreset, evaluateContent3);
   return definePlugin({
     id: "aexeo-emdash",
     version: PACKAGE_VERSION,
-    capabilities: ["content:read"],
+    // The entity-presence route fetches its public sources directly (see
+    // presence-check.ts). rdap.org then redirects to the registry for the
+    // domain's TLD; those registries cannot be listed here.
+    capabilities: ["content:read", "network:request"],
+    allowedHosts: [...PRESENCE_HOSTS],
     hooks: {
       // afterSave processes one saved document at a time — collections
       // list isn't needed here. The hook always runs regardless of
@@ -1742,16 +2142,26 @@ function createPlugin(options = {}) {
       "content:afterSave": (event, ctx) => handleAfterSaveConfigured(
         event,
         ctx,
-        inProcessEvaluator,
-        suppressionFilter
-      )
+        evaluator,
+        suppressionFilter,
+        options.blockMappings,
+        options.contentFieldsByCollection?.[event.content.type]
+      ),
+      "content:afterDelete": (event, ctx) => handleAfterDelete(event, ctx)
     },
+    // emdash requires plugins:manage (admin) on a route that declares no
+    // permission. Reading findings only needs plugins:read (editor); anything
+    // that writes stays admin, stated explicitly so the split is visible.
     routes: {
       // The Refresh sweep (handleAdminRoute → handleRefresh) is the
       // one consumer of `collections`. Bind the resolved list here so
       // every request handles the same set without re-reading options.
+      //
+      // This route serves the pages and the dashboard widget, and also the
+      // refresh_findings action, which handleAdminRoute gates itself.
       admin: {
-        handler: (ctx) => handleAdminRoute(ctx, collections, suppressionFilter)
+        permission: "plugins:read",
+        handler: (ctx) => handleAdminRoute(ctx, collections, suppressionFilter, evaluator, options.blockMappings, options.contentFieldsByCollection)
       },
       // JSON data endpoint for the React adminEntry. Two routes:
       //   POST /_emdash/api/plugins/aexeo-emdash/data    — read current findings
@@ -1761,19 +2171,25 @@ function createPlugin(options = {}) {
       // emdash edit page and the public site, which Block Kit can't
       // express in any element type as of emdash 0.8.0.
       data: {
+        permission: "plugins:read",
         handler: async (ctx) => (await handleDataRoute(ctx, {
           collections,
-          evaluator: inProcessEvaluator,
+          evaluator,
           refresh: false,
-          suppressionFilter
+          suppressionFilter,
+          blockMappings: options.blockMappings,
+          contentFieldsByCollection: options.contentFieldsByCollection
         })).payload
       },
       refresh: {
+        permission: "plugins:manage",
         handler: (ctx) => handleDataRoute(ctx, {
           collections,
-          evaluator: inProcessEvaluator,
+          evaluator,
           refresh: true,
-          suppressionFilter
+          suppressionFilter,
+          blockMappings: options.blockMappings,
+          contentFieldsByCollection: options.contentFieldsByCollection
         })
       },
       // /facts route serves the truth-manifest authoring UI. It
@@ -1781,6 +2197,7 @@ function createPlugin(options = {}) {
       // onto one POST endpoint via a "kind" body field — see
       // facts-route.ts for the full contract.
       facts: {
+        permission: "plugins:manage",
         handler: (ctx) => handleFactsRoute(ctx)
       },
       // /presence route serves the layer-4 entity-presence
@@ -1790,6 +2207,7 @@ function createPlugin(options = {}) {
       // results in KV for 24h. Multiplexes data / refresh kinds —
       // see presence-route.ts.
       presence: {
+        permission: "plugins:manage",
         handler: (ctx) => handlePresenceRoute(ctx)
       }
     },
@@ -1821,6 +2239,7 @@ function createPlugin(options = {}) {
   });
 }
 var configured_default = createPlugin;
+var ADMIN_ROLE = 50;
 var adminViewTuning = {
   emptyFindingsText: "Once a Refresh runs, findings will list here.",
   emptyScoreText: "No documents indexed \u2014 click Refresh on the Aexeo findings page.",
@@ -1828,7 +2247,7 @@ var adminViewTuning = {
   score: async (documents, kv) => scoreLocally(documents, await readStoredFacts(kv)),
   truthLabel: formatTruthLabel
 };
-async function handleAdminRoute(ctx, collections, suppressionFilter) {
+async function handleAdminRoute(ctx, collections, suppressionFilter, evaluator, blockMappings, contentFieldsByCollection) {
   const body = normalizeInteraction(ctx.input);
   if (body.type === "page_load") {
     return handlePageLoad(ctx, body.page);
@@ -1838,12 +2257,28 @@ async function handleAdminRoute(ctx, collections, suppressionFilter) {
       return renderDocumentPanel(ctx.kv, body.value);
     }
     if (body.action_id === "refresh_findings") {
+      if ((ctx.user?.role ?? 0) < ADMIN_ROLE) {
+        const page = await renderFindingsPage(ctx.kv, adminViewTuning);
+        return {
+          ...page,
+          blocks: [
+            {
+              type: "banner",
+              title: "Only administrators can run Refresh.",
+              variant: "error"
+            },
+            ...page.blocks
+          ]
+        };
+      }
       return handleRefresh({
         ctx,
         collections,
-        evaluator: inProcessEvaluator,
+        evaluator,
         suppressionFilter,
-        tuning: adminViewTuning
+        tuning: adminViewTuning,
+        blockMappings,
+        contentFieldsByCollection
       });
     }
     if (body.action_id.startsWith("filter:")) {
@@ -1868,7 +2303,7 @@ async function handlePageLoad(ctx, page) {
     return renderFindingsPage(ctx.kv, adminViewTuning);
   }
   if (normalized === "widget:aexeo-score") {
-    return renderScoreWidget(ctx.kv, adminViewTuning);
+    return renderScoreWidget(ctx.kv, adminViewTuning, ctx.log);
   }
   if (normalized === "document") {
     return renderDocumentPanel(ctx.kv);