mcp-scraper 0.57.2 → 0.58.0

package/dist/{extract-bundle-H7JEELY7.js → extract-bundle-XRRN6NAZ.js}

@@ -10,8 +10,8 @@ import {
 } from "./chunk-2K74LVV7.js";
 import {
   extractJobLimitInfo
-} from "./chunk-5XYBKE3X.js";
-import "./chunk-R6CO77BN.js";
+} from "./chunk-KMGRM4FL.js";
+import "./chunk-2U2ZDDSI.js";
 import "./chunk-NDIAJIKQ.js";
 import {
   getDb
@@ -565,4 +565,4 @@ export {
   SITE_EXTRACT_PAGE_CHUNK,
   assembleExtractArtifacts
 };
-//# sourceMappingURL=extract-bundle-H7JEELY7.js.map
+//# sourceMappingURL=extract-bundle-XRRN6NAZ.js.map

package/dist/{server-OZJXK7AB.js → server-J4YGBOTS.js}

@@ -115,6 +115,7 @@ import {
   CommonsClaimPublicationInputSchema,
   CommonsPreparePublicationInputSchema,
   CommonsPublishEditorialInputSchema,
+  CommonsUpdateEditorialArticleInputSchema,
   CommonsValidatePublicationInputSchema,
   CreateArtifactTemplateInputSchema,
   CreateEditorialReadingRoomInputSchema,
@@ -169,7 +170,7 @@ import {
   resolveDeploymentProfile,
   resolveLocalSourcebookSchemaType,
   transcribeMediaUrl
-} from "./chunk-67JUTUQD.js";
+} from "./chunk-P32KEY3D.js";
 import {
   auditImageUrls,
   auditImages,
@@ -225,7 +226,7 @@ import {
 } from "./chunk-FG6LJGJ7.js";
 import {
   PACKAGE_VERSION
-} from "./chunk-QN4W5SX6.js";
+} from "./chunk-EJ6KU4PC.js";
 import {
   abandonExtractSettlement,
   countSuccessfulPages,
@@ -248,7 +249,7 @@ import {
   setExtractJobTotal,
   settleExtractJob,
   terminalExtractJobStatus
-} from "./chunk-5XYBKE3X.js";
+} from "./chunk-KMGRM4FL.js";
 import {
   BROWSER_OPEN_MIN_BALANCE_MC,
   CONCURRENCY_PACK_INVOICE_DESCRIPTION,
@@ -301,7 +302,7 @@ import {
   tierChangeLeavesPrivateOffer,
   tierCreditsPerInterval,
   vendorCostUsdToMc
-} from "./chunk-R6CO77BN.js";
+} from "./chunk-2U2ZDDSI.js";
 import {
   CaptchaError,
   MapsSearchIncompletePageError,
@@ -9393,7 +9394,7 @@ var siteExtractFn = inngest.createFunction(
       });
     }) : null;
     const finalized = await step.run("finalize", async () => {
-      const { assembleExtractArtifacts } = await import("./extract-bundle-H7JEELY7.js");
+      const { assembleExtractArtifacts } = await import("./extract-bundle-XRRN6NAZ.js");
       const current = await getExtractJob(jobId2);
       if (!current) throw new Error("extract job disappeared before finalization");
       const stored = await assembleExtractArtifacts(current, { branding, imageAudit });
@@ -12884,6 +12885,7 @@ function mapSubmission(row) {
     businessName: String(rowValue(row, "business_name")),
     websiteUrl: String(rowValue(row, "website_url")),
     status: String(rowValue(row, "status")),
+    acquisitionError: rowValue(row, "acquisition_error") == null ? null : String(rowValue(row, "acquisition_error")),
     draftRevision: Number(rowValue(row, "draft_revision")),
     publishedRevision: rowValue(row, "published_revision") == null ? null : Number(rowValue(row, "published_revision")),
     coverage: parseJson2(rowValue(row, "coverage_json")),
@@ -14576,6 +14578,15 @@ function nameOverlap(left, right) {
   const intersection = [...a].filter((token6) => b.has(token6)).length;
   return intersection / Math.max(a.size, b.size);
 }
+function nameContainment(left, right) {
+  const a = tokens(left);
+  const b = tokens(right);
+  if (!a.size || !b.size) return 0;
+  const smaller = a.size <= b.size ? a : b;
+  if (smaller.size < 2) return 0;
+  const intersection = [...a].filter((token6) => b.has(token6)).length;
+  return intersection / smaller.size;
+}
 function validateLocalSourcebookMapsIdentity(submission, maps) {
   const expectedHost = host(submission.websiteUrl);
   const mapsHost = host(maps.website);
@@ -14585,7 +14596,9 @@ function validateLocalSourcebookMapsIdentity(submission, maps) {
   if (mapsHost && expectedHost && mapsHost !== expectedHost && !mapsHost.endsWith(`.${expectedHost}`) && !expectedHost.endsWith(`.${mapsHost}`)) {
     throw new Error(`Exact-place match rejected: Google Maps linked ${mapsHost}, not the submitted domain ${expectedHost}.`);
   }
-  if (overlap < 0.6) {
+  const containment = nameContainment(submission.businessName, actualName);
+  const domainConfirmsIdentity = Boolean(mapsHost && expectedHost);
+  if (overlap < 0.6 && containment < 0.9 && !domainConfirmsIdentity) {
     throw new Error(`Exact-place match rejected: Google Maps returned \u201C${actualName},\u201D which does not sufficiently match \u201C${submission.businessName}.\u201D`);
   }
   if (addressState && addressState !== submission.state.toLowerCase()) {
@@ -35905,6 +35918,92 @@ async function publishCommonsEditorial(input, user) {
   if (!edition) throw new CommonsPublicationError("publication_publish_failed", "The published edition was not persisted.");
   return publicationResult({ ...publication, latestEditionId: editionId, updatedAt: now }, edition, false);
 }
+async function updateCommonsEditorialArticle(input, user) {
+  await ensureCommonsSchema();
+  const subdomain = normalizePublicationSubdomain(input.publicationSubdomain);
+  const publication = await getCommonsPublicationBySubdomain(subdomain);
+  if (!publication) throw new CommonsPublicationError("publication_not_found", "Claim this publication name before editing an edition.", 404);
+  if (publication.ownerUserId !== Number(user.id)) {
+    throw new CommonsPublicationError("publication_not_owned", "Only the account that claimed this publication can edit it.", 404);
+  }
+  const idempotencyKey3 = requiredIdempotencyKey(input.idempotencyKey);
+  const replay = await getEditionByIdempotency(Number(user.id), idempotencyKey3);
+  if (replay) return publicationResult(publication, replay, true);
+  const latestOverall = publication.latestEditionId ? await getPublicationEditionById(publication.latestEditionId) : null;
+  const editionSlug = normalizeEditionSlug(input.editionSlug || latestOverall?.editionSlug || "");
+  if (!editionSlug) throw new CommonsPublicationError("publication_edition_required", "This publication has no edition yet; publish one before editing an article.", 409);
+  const latest = await getLatestEdition(publication.id, editionSlug);
+  if (!latest) throw new CommonsPublicationError("publication_edition_not_found", `Edition ${editionSlug} does not exist.`, 404);
+  if (input.baseRevision !== latest.revision) {
+    throw new CommonsPublicationError("publication_revision_conflict", `Edition ${editionSlug} is revision ${latest.revision}; read it again before editing.`, 409);
+  }
+  const articles = latest.articles;
+  const index = articles.findIndex((article) => article.slug === input.articleSlug);
+  if (index === -1) {
+    throw new CommonsPublicationError("publication_article_not_found", `Edition ${editionSlug} has no article with slug ${input.articleSlug}. Known slugs: ${articles.map((article) => article.slug).join(", ")}.`, 404);
+  }
+  const current = articles[index];
+  const nextArticles = [...articles];
+  nextArticles[index] = {
+    ...current,
+    ...input.markdown === void 0 ? {} : { markdown: input.markdown },
+    ...input.title === void 0 ? {} : { title: input.title },
+    ...input.summary === void 0 ? {} : { summary: input.summary },
+    ...input.category === void 0 ? {} : { category: input.category },
+    ...input.kicker === void 0 ? {} : { kicker: input.kicker },
+    ...input.order === void 0 ? {} : { order: input.order }
+  };
+  const editionInput = { site: latest.site, deck: latest.deck, articles: nextArticles };
+  const canonicalUrl = editionPublicUrl(subdomain, editionSlug);
+  const rendered = renderEditorialReadingRoom(editionInput);
+  const html = addPublicMetadata(rendered.html, canonicalUrl, publication.title);
+  const editionId = `tced_${randomUUID18()}`;
+  const revision = latest.revision + 1;
+  const now = (/* @__PURE__ */ new Date()).toISOString();
+  await getDb().batch([
+    {
+      sql: `
+        INSERT INTO commons_publication_editions (
+          id, publication_id, owner_user_id, edition_slug, revision, title, site_json, deck,
+          articles_json, html, filename, sha256, article_count, word_count, bytes, warnings_json,
+          idempotency_key, created_at, published_at
+        ) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
+      `,
+      args: [
+        editionId,
+        publication.id,
+        Number(user.id),
+        editionSlug,
+        revision,
+        latest.site.title,
+        JSON.stringify(latest.site),
+        latest.deck,
+        JSON.stringify(nextArticles),
+        html,
+        rendered.filename,
+        createHash16("sha256").update(html).digest("hex"),
+        rendered.articleCount,
+        rendered.wordCount,
+        Buffer.byteLength(html),
+        JSON.stringify(rendered.warnings),
+        idempotencyKey3,
+        now,
+        now
+      ]
+    },
+    {
+      sql: "UPDATE commons_publications SET latest_edition_id = ?, updated_at = ? WHERE id = ? AND owner_user_id = ?",
+      args: [editionId, now, publication.id, Number(user.id)]
+    }
+  ], "write");
+  const edition = await getPublicationEditionById(editionId);
+  if (!edition) throw new CommonsPublicationError("publication_publish_failed", "The edited edition was not persisted.");
+  return {
+    ...publicationResult({ ...publication, latestEditionId: editionId, updatedAt: now }, edition, false),
+    updatedArticle: { slug: input.articleSlug, previousRevision: latest.revision, revision },
+    unchangedArticles: articles.length - 1
+  };
+}
 async function getCommonsPublicationBySubdomain(subdomainInput) {
   await ensureCommonsSchema();
   const subdomain = normalizePublicationSubdomain(subdomainInput);
@@ -36002,6 +36101,7 @@ function rowToEdition(row, subdomain) {
     revision: Number(row.revision),
     title: String(row.title),
     site: parseJson4(row.site_json, {}),
+    articles: parseJson4(row.articles_json, []),
     deck: String(row.deck),
     filename: String(row.filename),
     sha256: String(row.sha256),
@@ -36239,6 +36339,7 @@ var PreparePublicationSchema = z32.object(CommonsPreparePublicationInputSchema).
 var ValidatePublicationSchema = z32.object(CommonsValidatePublicationInputSchema).strict();
 var ClaimPublicationSchema = z32.object(CommonsClaimPublicationInputSchema).strict();
 var PublishEditorialSchema = z32.object(CommonsPublishEditorialInputSchema).strict();
+var UpdateEditorialArticleSchema = z32.object(CommonsUpdateEditorialArticleInputSchema).strict();
 commonsApp.get("/health", async (c) => {
   await ensureCommonsSchema();
   return c.json({ ok: true, data: commonsDatabaseReport() });
@@ -36274,11 +36375,16 @@ commonsApp.post("/publications/publish", auth2, async (c) => {
   if (!parsed.success) return validationError(c, parsed.error);
   return publicationOperation(c, () => publishCommonsEditorial(parsed.data, c.get("user")), 201);
 });
+commonsApp.post("/publications/articles", auth2, async (c) => {
+  const parsed = UpdateEditorialArticleSchema.safeParse(await c.req.json().catch(() => ({})));
+  if (!parsed.success) return validationError(c, parsed.error);
+  return publicationOperation(c, () => updateCommonsEditorialArticle(parsed.data, c.get("user")), 201);
+});
 commonsApp.get("/publications/me", auth2, async (c) => {
   const publication = await getCommonsPublicationForOwner(Number(c.get("user").id));
   if (!publication) return c.json({ ok: false, error: "publication_not_found", message: "This account has not claimed a Commons publication yet." }, 404);
   const editions = c.req.query("includeEditions") === "false" ? [] : await listCommonsPublicationEditions(publication.id);
-  return c.json({ ok: true, data: { publication, editions } });
+  return c.json({ ok: true, data: { publication, editions: withArticleSources(editions, c.req.query("includeArticles") === "true") } });
 });
 commonsApp.get("/publications/:subdomain/site", async (c) => {
   try {
@@ -36318,7 +36424,7 @@ commonsApp.get("/publications/:subdomain", async (c) => {
     const publication = await getCommonsPublicationBySubdomain(c.req.param("subdomain"));
     if (!publication) return c.json({ ok: false, error: "publication_not_found", message: "No Commons publication matched that name." }, 404);
     const editions = c.req.query("includeEditions") === "false" ? [] : await listCommonsPublicationEditions(publication.id);
-    return c.json({ ok: true, data: { publication, editions } });
+    return c.json({ ok: true, data: { publication, editions: withArticleSources(editions, false) } });
   } catch (error) {
     return publicationError(c, error);
   }
@@ -36549,6 +36655,13 @@ function publicationError(c, error) {
   }
   throw error;
 }
+function withArticleSources(editions, include) {
+  if (include) return editions;
+  return editions.map((edition) => {
+    const { articles: _articles, ...rest } = edition;
+    return rest;
+  });
+}
 function xmlEscape(value) {
   return value.replaceAll("&", "&amp;").replaceAll("<", "&lt;").replaceAll(">", "&gt;").replaceAll('"', "&quot;").replaceAll("'", "&apos;");
 }
@@ -47179,7 +47292,7 @@ app.get("/cron/tick", async (c) => {
     return c.json({ error: "Unauthorized" }, 401);
   }
   const startedAt = Date.now();
-  const { drainQueue } = await import("./worker-S2WXELUC.js");
+  const { drainQueue } = await import("./worker-QFSOMF2N.js");
   const budget = { maxJobs: 10, deadlineMs: startedAt + CRON_TICK_DRAIN_BUDGET_MS };
   const origin = `${new URL(c.req.url).protocol}//${new URL(c.req.url).host}`;
   const unfinished = [];
@@ -47249,8 +47362,8 @@ app.post("/api/internal/extract-refinalize/:id", async (c) => {
     finishExtractJob: finishExtractJob2,
     terminalExtractJobStatus: terminalExtractJobStatus2,
     extractJobLimitInfo: extractJobLimitInfo2
-  } = await import("./site-extract-repository-CFMSWAVL.js");
-  const { assembleExtractArtifacts } = await import("./extract-bundle-H7JEELY7.js");
+  } = await import("./site-extract-repository-XAA2FVDN.js");
+  const { assembleExtractArtifacts } = await import("./extract-bundle-XRRN6NAZ.js");
   const existing = await getExtractJob2(jobId2);
   if (!existing) return c.json({ error: "job not found" }, 404);
   if (existing.status !== "failed") {
@@ -47277,7 +47390,7 @@ app.post("/api/internal/extract-refinalize/:id", async (c) => {
   if (!updated) return c.json({ error: "job state changed during re-finalization" }, 409);
   let settlement = "already_settled_or_refunded";
   if (job.billedMc == null && job.userId != null) {
-    const { settleExtractJob: settleExtractJob3, countSuccessfulPages: countSuccessfulPages2 } = await import("./site-extract-repository-CFMSWAVL.js");
+    const { settleExtractJob: settleExtractJob3, countSuccessfulPages: countSuccessfulPages2 } = await import("./site-extract-repository-XAA2FVDN.js");
     const heldMc = Number(job.options.heldMc ?? 0);
     const successful = await countSuccessfulPages2(jobId2);
     const usedMc = Math.min(successful * MC_COSTS.page_scrape, heldMc);
@@ -47462,4 +47575,4 @@ app.get("/blog/:slug/", (c) => {
 export {
   app
 };
-//# sourceMappingURL=server-OZJXK7AB.js.map
+//# sourceMappingURL=server-J4YGBOTS.js.map