@artinstack/migrator 0.1.21 → 0.1.23

package/dist/{chunk-H62Y7JAJ.js → chunk-E2EUJY3C.js}

@@ -53,6 +53,11 @@ var squarespaceClientOptionsSchema = z.object({
   retryBaseDelayMs: z.number().int().min(0).default(500),
   maxRetryDelayMs: z.number().int().min(0).default(8e3),
   requestIntervalMs: z.number().int().min(0).default(200),
+  /**
+   * When json-pretty yields an empty classic layout (common on 7.1 section pages),
+   * fetch the rendered HTML once and parse `.sqs-block` content from `#sections`.
+   */
+  htmlFallback: z.boolean().default(true),
   fetchImpl: z.custom().optional()
 });
 function isRecord(value) {
@@ -95,6 +100,37 @@ function inferBlockTypeFromClassName(className) {
   if (raw === "horizontalrule") return "line";
   return raw.replace(/-block$/g, "");
 }
+function isEmptyClassicMainContent(html) {
+  if (!html?.trim()) return true;
+  const $ = cheerio.load(html, { xml: false });
+  if ($(".sqs-block").length > 0) return false;
+  if ($(".sqs-layout.empty").length > 0) return true;
+  return html.replace(/\s+/g, " ").trim().length < 280;
+}
+function pageContentIsEmpty(page) {
+  if ((page.blocks?.length ?? 0) > 0) return false;
+  return isEmptyClassicMainContent(page.contentHtml);
+}
+function extractPageBodyHtml(html) {
+  if (!html.trim()) return "";
+  const $ = cheerio.load(html, { xml: false });
+  const sections = $("#sections").first();
+  if (sections.length) return sections.html() ?? "";
+  const article = $("article.sections").first();
+  if (article.length) return article.html() ?? "";
+  const main = $("main#page, main[role='main'], main").first();
+  if (main.length) return main.html() ?? "";
+  return "";
+}
+function extractPageContentFromHtml(html) {
+  const body = extractPageBodyHtml(html);
+  if (!body.trim()) return { contentHtml: "" };
+  const parsedBlocks = extractBlocksFromBodyHtml(body);
+  if (parsedBlocks.length > 0) {
+    return { blocks: parsedBlocks };
+  }
+  return { contentHtml: body };
+}
 function extractBlocksFromBodyHtml(html) {
   if (!html.trim()) return [];
   const $ = cheerio.load(html, { xml: false });
@@ -238,19 +274,60 @@ function isBlogCollection(collection) {
   const typeName = String(collection.typeName ?? collection.typeLabel ?? "").toLowerCase();
   return typeName.includes("blog");
 }
+function isGalleryCollection(collection) {
+  if (!collection) return false;
+  if (isBlogCollection(collection)) return false;
+  const typeName = String(collection.typeName ?? "").toLowerCase();
+  const typeLabel = String(collection.typeLabel ?? "").toLowerCase();
+  return typeName.includes("gallery") || typeLabel.includes("gallery") || typeName === "portfolio" || typeLabel === "portfolio";
+}
 function isStaticPageItem(item, collection) {
   const recordTypeLabel = String(item.recordTypeLabel ?? "").toLowerCase();
   if (recordTypeLabel.includes("page")) return true;
   const collectionType = String(collection?.typeName ?? collection?.typeLabel ?? "").toLowerCase();
   return collectionType === "page" || collectionType.includes("page-collection");
 }
-function mapWireItemToPost(item) {
+function stripSimpleHtml(value) {
+  return value.replace(/<[^>]+>/g, " ").replace(/\s+/g, " ").trim();
+}
+function mapWireGalleryItem(item) {
+  const resolved = typeof item.assetUrl === "string" && item.assetUrl.trim() ? item.assetUrl.trim() : typeof item.imageUrl === "string" && item.imageUrl.trim() ? item.imageUrl.trim() : "";
+  if (!resolved.startsWith("http")) return void 0;
+  const rawCaption = typeof item.body === "string" ? stripSimpleHtml(item.body) : typeof item.excerpt === "string" ? item.excerpt.trim() : typeof item.caption === "string" ? item.caption.trim() : void 0;
+  return {
+    id: String(item.id ?? item.systemDataId ?? item.urlId ?? ""),
+    imageUrl: resolved,
+    altText: typeof item.altText === "string" ? item.altText : typeof item.title === "string" ? item.title : void 0,
+    caption: rawCaption || void 0
+  };
+}
+function mapWireGalleryCollection(collection, items, context) {
+  const id = String(collection.id ?? collection.urlId ?? collection.fullUrl ?? "gallery");
+  const title = String(collection.title ?? collection.navigationTitle ?? "Gallery");
+  const slug = sanitizeSlug(String(collection.urlId ?? collection.title ?? id));
+  const galleryItems = items.map((item) => mapWireGalleryItem(item)).filter((item) => !!item);
+  return {
+    id,
+    title,
+    slug,
+    url: resolveAbsoluteUrl(
+      typeof collection.fullUrl === "string" ? collection.fullUrl : void 0,
+      context?.fetchedUrl
+    ),
+    description: typeof collection.description === "string" ? stripSimpleHtml(collection.description) : void 0,
+    items: galleryItems
+  };
+}
+function mapWireItemToPost(item, options) {
   const content = mapWireItemContent(item);
   return {
     id: String(item.id ?? item.systemDataId ?? item.urlId ?? ""),
     title: String(item.title ?? "Untitled"),
     slug: sanitizeSlug(String(item.urlId ?? item.title ?? item.id ?? "post")),
-    url: typeof item.fullUrl === "string" ? item.fullUrl : void 0,
+    url: resolveAbsoluteUrl(
+      typeof item.fullUrl === "string" ? item.fullUrl : void 0,
+      options?.fallbackUrl
+    ),
     excerpt: typeof item.excerpt === "string" ? item.excerpt : void 0,
     publishedAt: mapPublishOn(item.publishOn ?? item.addedOn),
     status: mapWorkflowState(item.workflowState),
@@ -262,13 +339,26 @@ function mapWireItemToPost(item) {
     ...content
   };
 }
+function resolveAbsoluteUrl(raw, fallback) {
+  const candidate = raw?.trim() || fallback?.trim();
+  if (!candidate) return void 0;
+  try {
+    if (fallback) return new URL(candidate, fallback).toString();
+    return new URL(candidate).toString();
+  } catch {
+    return candidate.startsWith("http") ? candidate : void 0;
+  }
+}
 function mapWireItemToPage(item, options) {
   const content = mapWireItemContent(item);
   return {
     id: String(item.id ?? item.urlId ?? ""),
     title: String(item.title ?? "Untitled"),
     slug: sanitizeSlug(String(item.urlId ?? item.title ?? item.id ?? "page")),
-    url: typeof item.fullUrl === "string" ? item.fullUrl : options?.fallbackUrl,
+    url: resolveAbsoluteUrl(
+      typeof item.fullUrl === "string" ? item.fullUrl : void 0,
+      options?.fallbackUrl
+    ),
     status: mapWorkflowState(item.workflowState),
     isHomePage: options?.isHomePage,
     seoTitle: readSeo(item.seoData, "title"),
@@ -313,11 +403,16 @@ function mapJsonPrettyWire(wire, context) {
     site: siteFromWire(wire),
     pages: [],
     posts: [],
+    galleries: [],
     categories: mapWireCategories(collection),
     tags: []
   };
   if (Array.isArray(wire.items)) {
     const itemRecords = wire.items.map((entry) => asRecord(entry)).filter((entry) => !!entry);
+    if (collection && isGalleryCollection(collection)) {
+      partial.galleries = [mapWireGalleryCollection(collection, itemRecords, context)];
+      return partial;
+    }
     partial.tags = mapWireTags(itemRecords);
     for (const item2 of itemRecords) {
       if (isStaticPageItem(item2, collection)) {
@@ -325,7 +420,7 @@ function mapJsonPrettyWire(wire, context) {
           mapWireItemToPage(item2, { fallbackUrl: context?.fetchedUrl, isHomePage: context?.isHomePage })
         );
       } else {
-        partial.posts.push(mapWireItemToPost(item2));
+        partial.posts.push(mapWireItemToPost(item2, { fallbackUrl: context?.fetchedUrl }));
       }
     }
     return partial;
@@ -337,20 +432,29 @@ function mapJsonPrettyWire(wire, context) {
         mapWireItemToPage(item, { fallbackUrl: context?.fetchedUrl, isHomePage: context?.isHomePage })
       );
     } else {
-      partial.posts.push(mapWireItemToPost(item));
+      partial.posts.push(mapWireItemToPost(item, { fallbackUrl: context?.fetchedUrl }));
       partial.tags = mapWireTags([item]);
     }
     return partial;
   }
+  if (collection && isGalleryCollection(collection)) {
+    partial.galleries = [mapWireGalleryCollection(collection, [], context)];
+    return partial;
+  }
   if (collection && isBlogCollection(collection) === false) {
+    const mainContent = typeof wire.mainContent === "string" ? wire.mainContent : void 0;
+    const body = mainContent && !isEmptyClassicMainContent(mainContent) ? mainContent : typeof collection.description === "string" ? collection.description : void 0;
     partial.pages.push(
       mapWireItemToPage(
         {
           id: collection.id ?? collection.urlId ?? collection.fullUrl,
           title: collection.title ?? collection.navigationTitle,
           urlId: collection.urlId,
-          fullUrl: collection.fullUrl ?? context?.fetchedUrl,
-          body: collection.description,
+          fullUrl: resolveAbsoluteUrl(
+            typeof collection.fullUrl === "string" ? collection.fullUrl : void 0,
+            context?.fetchedUrl
+          ),
+          body,
           workflowState: collection.draft ? 2 : 1,
           seoData: collection.seoData
         },
@@ -380,6 +484,7 @@ function mergeSquarespaceExportPartials(partials) {
     exportedAt: (/* @__PURE__ */ new Date()).toISOString(),
     pages: [],
     posts: [],
+    galleries: [],
     categories: [],
     tags: []
   };
@@ -387,11 +492,13 @@ function mergeSquarespaceExportPartials(partials) {
     if (partial.site) merged.site = { ...merged.site, ...partial.site };
     merged.pages.push(...partial.pages ?? []);
     merged.posts.push(...partial.posts ?? []);
+    merged.galleries.push(...partial.galleries ?? []);
     merged.categories.push(...partial.categories ?? []);
     merged.tags.push(...partial.tags ?? []);
   }
   merged.pages = dedupeById(merged.pages);
   merged.posts = dedupeById(merged.posts ?? []);
+  merged.galleries = dedupeById(merged.galleries ?? []);
   merged.categories = dedupeBySlug(merged.categories ?? []);
   merged.tags = dedupeBySlug(merged.tags ?? []);
   return merged;
@@ -415,6 +522,7 @@ var SquarespaceCollectionClient = class {
   retryBaseDelayMs;
   maxRetryDelayMs;
   requestIntervalMs;
+  htmlFallback;
   fetchImpl;
   lastRequestAt = 0;
   constructor(options = {}) {
@@ -424,15 +532,24 @@ var SquarespaceCollectionClient = class {
     this.retryBaseDelayMs = parsed.retryBaseDelayMs;
     this.maxRetryDelayMs = parsed.maxRetryDelayMs;
     this.requestIntervalMs = parsed.requestIntervalMs;
+    this.htmlFallback = parsed.htmlFallback;
     this.fetchImpl = parsed.fetchImpl ?? fetch;
   }
   buildJsonPrettyUrl(pageUrl) {
     return buildJsonPrettyUrl(pageUrl, this.format);
   }
   async fetchWire(url) {
-    const response = await this.requestWithRetry(this.buildJsonPrettyUrl(url));
+    const response = await this.requestWithRetry(this.buildJsonPrettyUrl(url), {
+      accept: "application/json, text/html, */*"
+    });
     return response.json();
   }
+  async fetchHtml(url) {
+    const response = await this.requestWithRetry(url, {
+      accept: "text/html,application/xhtml+xml"
+    });
+    return response.text();
+  }
   async collectExport(targets) {
     if (targets.length === 0) {
       throw new Error("Squarespace collector requires at least one target URL");
@@ -449,7 +566,30 @@ var SquarespaceCollectionClient = class {
         );
       }
     }
-    return mergeSquarespaceExportPartials(partials);
+    const merged = mergeSquarespaceExportPartials(partials);
+    if (this.htmlFallback) {
+      await this.enrichEmptyPagesFromHtml(merged);
+    }
+    return merged;
+  }
+  async enrichEmptyPagesFromHtml(doc) {
+    for (const page of doc.pages) {
+      if (!pageContentIsEmpty(page)) continue;
+      const pageUrl = page.url?.trim();
+      if (!pageUrl) continue;
+      try {
+        const html = await this.fetchHtml(pageUrl);
+        const content = extractPageContentFromHtml(html);
+        if (content.blocks?.length) {
+          page.blocks = content.blocks;
+          delete page.contentHtml;
+        } else if (content.contentHtml?.trim()) {
+          page.contentHtml = content.contentHtml;
+          delete page.blocks;
+        }
+      } catch {
+      }
+    }
   }
   async collectCollectionPages(target) {
     const partials = [];
@@ -466,13 +606,14 @@ var SquarespaceCollectionClient = class {
     }
     return partials;
   }
-  async requestWithRetry(url) {
+  async requestWithRetry(url, options) {
     let attempt = 0;
+    const accept = options?.accept ?? "application/json, text/html, */*";
     while (true) {
       await this.throttle();
       const response = await this.fetchImpl(url, {
         method: "GET",
-        headers: { Accept: "application/json" }
+        headers: { Accept: accept }
       });
       if (response.ok) {
         return response;
@@ -722,6 +863,98 @@ function assetFromUrl(url, exportedAt, index) {
     mimeType: guessMime(filename)
   };
 }
+function galleryPortfolioSourceId(parentSourceId, block, galleryIndex) {
+  const blockId = block.id?.trim();
+  if (blockId) return `gallery:${blockId}`;
+  return `gallery:${parentSourceId}:${galleryIndex}`;
+}
+function countGalleryPortfolios(blocks) {
+  if (!blocks?.length) return 0;
+  return blocks.filter(
+    (block) => block.type.toLowerCase() === "gallery" && (block.items?.length ?? 0) > 0
+  ).length;
+}
+function collectGalleryImageUrls(blocks) {
+  const urls = /* @__PURE__ */ new Set();
+  if (!blocks?.length) return urls;
+  for (const block of blocks) {
+    if (block.type.toLowerCase() !== "gallery") continue;
+    for (const item of block.items ?? []) {
+      if (item.imageUrl) urls.add(item.imageUrl);
+    }
+  }
+  return urls;
+}
+function galleryAssetSourceId(portfolioSourceId, item, sort) {
+  const itemId = item.id?.trim();
+  if (itemId) return itemId;
+  return `${portfolioSourceId}:${sort}`;
+}
+function* emitGalleriesFromBlocks(blocks, parent, exportedAt) {
+  if (!blocks?.length) return;
+  const nonEmptyGalleries = blocks.filter(
+    (block) => block.type.toLowerCase() === "gallery" && (block.items?.length ?? 0) > 0
+  );
+  if (nonEmptyGalleries.length === 0) return;
+  let galleryIndex = 0;
+  for (const block of blocks) {
+    if (block.type.toLowerCase() !== "gallery") continue;
+    const items = (block.items ?? []).filter((item) => Boolean(item.imageUrl?.trim()));
+    if (items.length === 0) continue;
+    const sourceId = galleryPortfolioSourceId(parent.sourceId, block, galleryIndex);
+    const slugBase = block.id?.trim() ? sanitizeSlug(block.id) : sanitizeSlug(`gallery-${galleryIndex + 1}`);
+    const title = nonEmptyGalleries.length === 1 ? parent.title : `${parent.title} \u2014 Gallery ${galleryIndex + 1}`;
+    yield {
+      type: "portfolio",
+      source: sourceMeta(sourceId, parent.url, exportedAt),
+      sourceId,
+      title,
+      slug: slugBase.startsWith("gallery") ? slugBase : `gallery-${slugBase}`
+    };
+    yield* emitGalleryAssets(sourceId, items, exportedAt);
+    galleryIndex += 1;
+  }
+}
+function galleryCollectionPortfolioSourceId(collectionId) {
+  return `gallery-collection:${collectionId}`;
+}
+function* emitGalleryAssets(portfolioSourceId, items, exportedAt) {
+  for (let sort = 0; sort < items.length; sort++) {
+    const item = items[sort];
+    const assetSourceId = galleryAssetSourceId(portfolioSourceId, item, sort);
+    const filename = filenameFromUrl(item.imageUrl, `${assetSourceId}.jpg`);
+    yield {
+      type: "asset",
+      source: sourceMeta(assetSourceId, item.imageUrl, exportedAt),
+      sourceId: assetSourceId,
+      sourceUrl: item.imageUrl,
+      filename,
+      mimeType: guessMime(filename),
+      caption: item.caption,
+      altText: item.altText,
+      portfolioSourceId,
+      sort
+    };
+  }
+}
+function* emitGalleryCollections(galleries, exportedAt) {
+  if (!galleries?.length) return;
+  for (const gallery of galleries) {
+    const items = gallery.items.filter((item) => Boolean(item.imageUrl?.trim()));
+    if (items.length === 0) continue;
+    const sourceId = galleryCollectionPortfolioSourceId(gallery.id);
+    const slugBase = sanitizeSlug(gallery.slug || gallery.id);
+    yield {
+      type: "portfolio",
+      source: sourceMeta(sourceId, gallery.url, exportedAt),
+      sourceId,
+      title: gallery.title,
+      slug: slugBase.startsWith("gallery") ? slugBase : `gallery-${slugBase}`,
+      description: gallery.description
+    };
+    yield* emitGalleryAssets(sourceId, items, exportedAt);
+  }
+}
 function isSquarespaceExport(value) {
   if (!value || typeof value !== "object") return false;
   const record = value;
@@ -747,8 +980,8 @@ async function loadSquarespaceExport(options) {
   if (isSquarespaceExport(raw)) return raw;
   throw new Error("Invalid Squarespace export: expected exportVersion 1 with pages[]");
 }
-function* emitAssetsFromContent(contentHtml, explicitUrls, exportedAt) {
-  const seen = /* @__PURE__ */ new Set();
+function* emitAssetsFromContent(contentHtml, explicitUrls, exportedAt, excludeUrls) {
+  const seen = new Set(excludeUrls ?? []);
   const urls = [...explicitUrls, ...discoverContentAssetUrls(contentHtml)];
   let index = 0;
   for (const url of urls) {
@@ -773,7 +1006,13 @@ async function* emitPage(page, exportedAt) {
     seoTitle: page.seoTitle,
     seoDescription: page.seoDescription
   };
-  yield* emitAssetsFromContent(contentHtml, assetUrls, exportedAt);
+  yield* emitGalleriesFromBlocks(
+    page.blocks,
+    { sourceId: page.id, title: page.title, url: page.url },
+    exportedAt
+  );
+  const galleryUrls = collectGalleryImageUrls(page.blocks);
+  yield* emitAssetsFromContent(contentHtml, assetUrls, exportedAt, galleryUrls);
 }
 async function* emitPost(post, exportedAt) {
   const { contentHtml, assetUrls } = resolveContentHtml(post);
@@ -808,7 +1047,13 @@ async function* emitPost(post, exportedAt) {
       mimeType: guessMime(filename)
     };
   }
-  yield* emitAssetsFromContent(contentHtml, assetUrls, exportedAt);
+  yield* emitGalleriesFromBlocks(
+    post.blocks,
+    { sourceId: post.id, title: post.title, url: post.url },
+    exportedAt
+  );
+  const galleryUrls = collectGalleryImageUrls(post.blocks);
+  yield* emitAssetsFromContent(contentHtml, assetUrls, exportedAt, galleryUrls);
 }
 async function* enumerateSquarespaceEntities(options) {
   const doc = await loadSquarespaceExport(options);
@@ -837,13 +1082,29 @@ async function* enumerateSquarespaceEntities(options) {
   for (const post of doc.posts ?? []) {
     yield* emitPost(post, exportedAt);
   }
+  yield* emitGalleryCollections(doc.galleries, exportedAt);
+}
+function countCollectionPortfolios(galleries) {
+  if (!galleries?.length) return 0;
+  return galleries.filter(
+    (gallery) => gallery.items.some((item) => Boolean(item.imageUrl?.trim()))
+  ).length;
 }
 function summarizeSquarespaceExport(doc) {
+  let portfolios = 0;
+  for (const page of doc.pages) {
+    portfolios += countGalleryPortfolios(page.blocks);
+  }
+  for (const post of doc.posts ?? []) {
+    portfolios += countGalleryPortfolios(post.blocks);
+  }
+  portfolios += countCollectionPortfolios(doc.galleries);
   return {
     pages: doc.pages.length,
     posts: doc.posts?.length ?? 0,
     categories: doc.categories?.length ?? 0,
-    tags: doc.tags?.length ?? 0
+    tags: doc.tags?.length ?? 0,
+    portfolios
   };
 }
 async function validateSquarespaceExportFile(filePath) {
@@ -875,7 +1136,7 @@ async function validateSquarespaceExportFile(filePath) {
       posts: summary.posts,
       categories: summary.categories,
       tags: summary.tags,
-      portfolios: 0,
+      portfolios: summary.portfolios,
       assets: 0
     }
   };
@@ -1539,6 +1800,12 @@ export {
   MIGRATION_WRITE_STAGES,
   SQUARESPACE_JSON_FORMAT,
   buildJsonPrettyUrl,
+  isEmptyClassicMainContent,
+  pageContentIsEmpty,
+  extractPageBodyHtml,
+  extractPageContentFromHtml,
+  extractBlocksFromBodyHtml,
+  isGalleryCollection,
   mapJsonPrettyWire,
   SquarespaceCollectionClient,
   enumerateSquarespaceEntities,
@@ -1565,4 +1832,4 @@ export {
   resolveAdapterImportSummary,
   runDryRun
 };
-//# sourceMappingURL=chunk-H62Y7JAJ.js.map
+//# sourceMappingURL=chunk-E2EUJY3C.js.map