@artinstack/migrator 0.1.6 → 0.1.8

package/dist/{chunk-XYP3VYDH.js → chunk-WHGUE5FC.js}

@@ -1,5 +1,30 @@
-// src/lib/content-asset-urls.ts
+// src/lib/media-urls.ts
 import * as cheerio from "cheerio";
+function rewriteOriginUrlsInText(text, config) {
+  if (!text || config.rules.length === 0) return text;
+  let result = text;
+  for (const rule of config.rules) {
+    if (typeof rule.match === "string") {
+      if (!rule.match) continue;
+      result = result.split(rule.match).join(rule.replace);
+      continue;
+    }
+    result = result.replace(rule.match, rule.replace);
+  }
+  return result;
+}
+function createWpContentGatewayRewrite(gatewayBase, publicOrigin) {
+  const normalizedGateway = gatewayBase.replace(/\/$/, "");
+  const normalizedPublic = publicOrigin.replace(/\/$/, "");
+  return {
+    rules: [
+      {
+        match: `${normalizedGateway}/wp-content/`,
+        replace: `${normalizedPublic}/wp-content/`
+      }
+    ]
+  };
+}
 var IMAGE_EXTENSIONS = "jpe?g|png|gif|webp|avif|svg";
 var IMAGE_EXTENSION_PATTERN = new RegExp(String.raw`\.(?:${IMAGE_EXTENSIONS})\b`, "i");
 var QUOTED_IMAGE_PATH = String.raw`[^"']+\.(?:${IMAGE_EXTENSIONS})(?:\?[^"'#]*)?(?:#.*)?`;
@@ -146,14 +171,97 @@ function discoverContentAssetUrls(content) {
 function extractInlineImageSrcs(content) {
   return discoverContentAssetUrls(content);
 }
+var MIGRATION_MEDIA_REF_SCHEME = "artinstack-migration://asset/";
+function formatMigrationMediaRef(sourceAssetId) {
+  return `${MIGRATION_MEDIA_REF_SCHEME}${encodeURIComponent(sourceAssetId)}`;
+}
+function isMigrationMediaRef(value) {
+  return value.trim().startsWith(MIGRATION_MEDIA_REF_SCHEME);
+}
+function parseMigrationMediaRef(value) {
+  const trimmed = value.trim();
+  if (!trimmed.startsWith(MIGRATION_MEDIA_REF_SCHEME)) return void 0;
+  const encoded = trimmed.slice(MIGRATION_MEDIA_REF_SCHEME.length);
+  if (!encoded) return void 0;
+  try {
+    return decodeURIComponent(encoded);
+  } catch {
+    return void 0;
+  }
+}
+function createMigrationMediaRefReplaceWith() {
+  return (ref) => {
+    if (!ref.sourceAssetId) return "";
+    return formatMigrationMediaRef(ref.sourceAssetId);
+  };
+}
+function canonicalizeInlineAssetUrl(raw, originUrlRewrite) {
+  let value = raw.trim();
+  if (!value || value.startsWith("data:")) return void 0;
+  if (originUrlRewrite) {
+    value = rewriteOriginUrlsInText(value, originUrlRewrite);
+  }
+  const canonicalUrl = normalizeAssetUrl(value);
+  if (!canonicalUrl) return void 0;
+  return {
+    canonicalUrl,
+    sourceId: `url:${canonicalUrl}`
+  };
+}
+function urlPathname(url) {
+  try {
+    return new URL(url, "http://migration.local").pathname;
+  } catch {
+    return void 0;
+  }
+}
+function buildMigrationMediaUrlIndex(entries) {
+  const index = /* @__PURE__ */ new Map();
+  for (const entry of entries) {
+    index.set(entry.sourceUrl, entry.sourceId);
+    const normalized = normalizeAssetUrl(entry.sourceUrl);
+    if (normalized) index.set(normalized, entry.sourceId);
+    const pathname = urlPathname(entry.sourceUrl);
+    if (pathname) index.set(pathname, entry.sourceId);
+  }
+  return index;
+}
+function resolveMigrationMediaSourceId(src, urlIndex, originUrlRewrite) {
+  const canonical = canonicalizeInlineAssetUrl(src, originUrlRewrite);
+  const normalized = canonical?.canonicalUrl ?? normalizeAssetUrl(src);
+  if (!normalized) return void 0;
+  return urlIndex.get(normalized) ?? urlIndex.get(src) ?? (urlPathname(normalized) ? urlIndex.get(urlPathname(normalized)) : void 0);
+}
+function buildContentMediaUrlIndex(entries, originUrlRewrite) {
+  const canonicalEntries = [];
+  for (const entry of entries) {
+    const canonical = canonicalizeInlineAssetUrl(entry.sourceUrl, originUrlRewrite);
+    canonicalEntries.push({
+      sourceUrl: canonical?.canonicalUrl ?? entry.sourceUrl,
+      sourceId: entry.sourceId
+    });
+  }
+  return buildMigrationMediaUrlIndex(canonicalEntries);
+}
 
 export {
+  rewriteOriginUrlsInText,
+  createWpContentGatewayRewrite,
   discoverRawImgSrcs,
   normalizeAssetUrl,
   isLikelyImageUrl,
   discoverFeaturedAssetCandidateUrls,
   resolveFeaturedContentAssetUrl,
   discoverContentAssetUrls,
-  extractInlineImageSrcs
+  extractInlineImageSrcs,
+  MIGRATION_MEDIA_REF_SCHEME,
+  formatMigrationMediaRef,
+  isMigrationMediaRef,
+  parseMigrationMediaRef,
+  createMigrationMediaRefReplaceWith,
+  canonicalizeInlineAssetUrl,
+  buildMigrationMediaUrlIndex,
+  resolveMigrationMediaSourceId,
+  buildContentMediaUrlIndex
 };
-//# sourceMappingURL=chunk-XYP3VYDH.js.map
+//# sourceMappingURL=chunk-WHGUE5FC.js.map

package/dist/chunk-WHGUE5FC.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/lib/media-urls.ts"],"sourcesContent":["import * as cheerio from \"cheerio\";\n\n// --- Origin URL rewrite (gateway → public origin before parse/discovery) ---\n\nexport interface OriginUrlRewriteRule {\n  /** Literal substring or regex matched against the full text block. */\n  match: string | RegExp;\n  replace: string;\n}\n\nexport interface OriginUrlRewriteConfig {\n  rules: OriginUrlRewriteRule[];\n}\n\n/** Swap legacy gateway/staging host fragments before parse, fetch, or asset discovery. */\nexport function rewriteOriginUrlsInText(text: string, config: OriginUrlRewriteConfig): string {\n  if (!text || config.rules.length === 0) return text;\n\n  let result = text;\n  for (const rule of config.rules) {\n    if (typeof rule.match === \"string\") {\n      if (!rule.match) continue;\n      result = result.split(rule.match).join(rule.replace);\n      continue;\n    }\n    result = result.replace(rule.match, rule.replace);\n  }\n  return result;\n}\n\n/** Build a rule that rewrites API-gateway `/prod/wp-content/` paths to a public origin. */\nexport function createWpContentGatewayRewrite(gatewayBase: string, publicOrigin: string): OriginUrlRewriteConfig {\n  const normalizedGateway = gatewayBase.replace(/\\/$/, \"\");\n  const normalizedPublic = publicOrigin.replace(/\\/$/, \"\");\n  return {\n    rules: [\n      {\n        match: `${normalizedGateway}/wp-content/`,\n        replace: `${normalizedPublic}/wp-content/`,\n      },\n    ],\n  };\n}\n\n// --- Content asset URL discovery & normalization ---\n\nconst IMAGE_EXTENSIONS = \"jpe?g|png|gif|webp|avif|svg\";\n/** Image file extension in a path or URL (allows trailing `?query` / `#hash`). */\nconst IMAGE_EXTENSION_PATTERN = new RegExp(String.raw`\\.(?:${IMAGE_EXTENSIONS})\\b`, \"i\");\n\n/** Captured value must contain an image extension — skips `url=\"…/about\"`, `<iframe src=\"…youtube…\">`, etc. */\nconst QUOTED_IMAGE_PATH = String.raw`[^\"']+\\.(?:${IMAGE_EXTENSIONS})(?:\\?[^\"'#]*)?(?:#.*)?`;\n\nconst SHORTCODE_IMAGE_PARAM_PATTERN = new RegExp(\n  String.raw`\\b(?:image|bg_image|background_image|url)\\s*=\\s*[\"'](${QUOTED_IMAGE_PATH})[\"']`,\n  \"gi\",\n);\n\n/** Bare `src=\"…jpg\"` outside `<img>` (shortcode fragments); `<img src>` handled by cheerio. */\nconst BARE_SRC_PARAM_PATTERN = new RegExp(\n  String.raw`\\bsrc\\s*=\\s*[\"'](${QUOTED_IMAGE_PATH})[\"']`,\n  \"gi\",\n);\n\nconst DATA_BG_IMAGE_PATTERN = /\\bdata-bg-image\\s*=\\s*[\"']([^\"']+)[\"']/gi;\n\n/** Inline CSS `background` / `background-image: url(…)` (quoted or bare). */\nconst BACKGROUND_IMAGE_URL_PATTERN =\n  /background(?:-image)?\\s*:[^;]*?url\\s*\\(\\s*(['\"]?)([^'\")]+)\\1\\s*\\)/gi;\n\nconst HERO_URL_PARAM_PATTERN = new RegExp(\n  String.raw`\\b(?:bg_image|background_image)\\s*=\\s*[\"'](${QUOTED_IMAGE_PATH})[\"']`,\n  \"gi\",\n);\n\nconst INLINE_IMAGE_PARAM_PATTERN = new RegExp(\n  String.raw`\\bimage\\s*=\\s*[\"'](${QUOTED_IMAGE_PATH})[\"']`,\n  \"gi\",\n);\n\nconst IMG_TAG_SRC_PATTERN = /<img\\b[^>]*\\bsrc\\s*=\\s*[\"']([^\"']+)[\"']/gi;\n\ninterface FeaturedAssetCandidate {\n  url: string;\n  index: number;\n  tier: 0 | 1;\n}\n\nfunction ingestLikelyImageUrl(urls: Set<string>, raw: string | undefined): void {\n  const normalized = normalizeAssetUrl(raw ?? \"\");\n  if (normalized && isLikelyImageUrl(normalized)) {\n    urls.add(normalized);\n  }\n}\n\nfunction extractImgTagSrcs(content: string): string[] {\n  if (!content.trim()) return [];\n  const $ = cheerio.load(content, { xml: false });\n  const srcs: string[] = [];\n  $(\"img[src]\").each((_, el) => {\n    const src = $(el).attr(\"src\")?.trim();\n    if (src) srcs.push(src);\n  });\n  return srcs;\n}\n\nfunction hasImageExtension(value: string): boolean {\n  const withoutHash = value.split(\"#\", 1)[0] ?? value;\n  const withoutQuery = withoutHash.split(\"?\", 1)[0] ?? withoutHash;\n  return IMAGE_EXTENSION_PATTERN.test(withoutQuery);\n}\n\nfunction extractDataBgImageUrls(content: string): string[] {\n  const urls: string[] = [];\n  for (const match of content.matchAll(DATA_BG_IMAGE_PATTERN)) {\n    const raw = match[1]?.trim();\n    if (raw) urls.push(raw);\n  }\n  return urls;\n}\n\nfunction extractCssBackgroundImageUrls(content: string): string[] {\n  const urls: string[] = [];\n  for (const match of content.matchAll(BACKGROUND_IMAGE_URL_PATTERN)) {\n    const raw = match[2]?.trim();\n    if (raw) urls.push(raw);\n  }\n  return urls;\n}\n\n/** All `<img src>` values (including those not ingested as vault assets). */\nexport function discoverRawImgSrcs(content: string): string[] {\n  return extractImgTagSrcs(content).filter((src) => !src.startsWith(\"data:\"));\n}\n\n/** Normalize protocol-relative and trim; skip data URIs. */\nexport function normalizeAssetUrl(raw: string): string | undefined {\n  const trimmed = raw.trim();\n  if (!trimmed || trimmed.startsWith(\"data:\")) return undefined;\n  if (trimmed.startsWith(\"//\")) return `https:${trimmed}`;\n  return trimmed;\n}\n\n/** Heuristic: URL likely points at a raster/vector image asset, not a page link. */\nexport function isLikelyImageUrl(url: string): boolean {\n  if (!url || url.startsWith(\"data:\")) return false;\n\n  if (url.startsWith(\"/\")) {\n    return hasImageExtension(url);\n  }\n\n  if (!/^https?:\\/\\//i.test(url)) return false;\n\n  try {\n    const { pathname } = new URL(url);\n    if (hasImageExtension(pathname)) return true;\n  } catch {\n    // fall through — malformed absolute URL\n  }\n\n  return hasImageExtension(url);\n}\n\nfunction pushFeaturedCandidate(\n  candidates: FeaturedAssetCandidate[],\n  raw: string | undefined,\n  index: number,\n  tier: 0 | 1,\n): void {\n  const normalized = normalizeAssetUrl(raw ?? \"\");\n  if (!normalized || !isLikelyImageUrl(normalized)) return;\n  candidates.push({ url: normalized, index, tier });\n}\n\nfunction collectFeaturedAssetCandidates(content: string): FeaturedAssetCandidate[] {\n  const candidates: FeaturedAssetCandidate[] = [];\n\n  for (const match of content.matchAll(DATA_BG_IMAGE_PATTERN)) {\n    pushFeaturedCandidate(candidates, match[1], match.index ?? 0, 0);\n  }\n  for (const match of content.matchAll(BACKGROUND_IMAGE_URL_PATTERN)) {\n    pushFeaturedCandidate(candidates, match[2], match.index ?? 0, 0);\n  }\n  for (const match of content.matchAll(HERO_URL_PARAM_PATTERN)) {\n    pushFeaturedCandidate(candidates, match[1], match.index ?? 0, 0);\n  }\n  for (const match of content.matchAll(IMG_TAG_SRC_PATTERN)) {\n    pushFeaturedCandidate(candidates, match[1], match.index ?? 0, 1);\n  }\n  for (const match of content.matchAll(INLINE_IMAGE_PARAM_PATTERN)) {\n    pushFeaturedCandidate(candidates, match[1], match.index ?? 0, 1);\n  }\n\n  return candidates;\n}\n\n/**\n * Ordered featured-image candidates when `_thumbnail_id` is missing — heroes\n * (`data-bg-image`, CSS backgrounds, `bg_image=`) before inline assets; within\n * each tier, first in document order wins. Filename tokens (`_w`, `_2048`, …)\n * are not interpreted as quality signals.\n */\nexport function discoverFeaturedAssetCandidateUrls(content: string): string[] {\n  if (!content.trim()) return [];\n\n  const ranked = [...collectFeaturedAssetCandidates(content)].sort((left, right) => {\n    if (left.tier !== right.tier) return left.tier - right.tier;\n    return left.index - right.index;\n  });\n\n  const urls: string[] = [];\n  const seen = new Set<string>();\n  for (const candidate of ranked) {\n    if (seen.has(candidate.url)) continue;\n    seen.add(candidate.url);\n    urls.push(candidate.url);\n  }\n  return urls;\n}\n\n/** Best featured-image URL from post/page HTML when attachment id is unavailable. */\nexport function resolveFeaturedContentAssetUrl(content: string): string | undefined {\n  return discoverFeaturedAssetCandidateUrls(content)[0];\n}\n\n/**\n * Generic content-discovery pass: collect image URLs from HTML `<img>` tags,\n * section hero markers (`data-bg-image`), inline CSS backgrounds, and common\n * shortcode/builder attributes (`src=`, `image=`, `bg_image=`, …) without\n * parsing builder-specific structure (Tatsu, Elementor, etc.).\n */\nexport function discoverContentAssetUrls(content: string): string[] {\n  if (!content.trim()) return [];\n\n  const urls = new Set<string>();\n\n  for (const raw of extractImgTagSrcs(content)) {\n    ingestLikelyImageUrl(urls, raw);\n  }\n\n  for (const match of content.matchAll(SHORTCODE_IMAGE_PARAM_PATTERN)) {\n    ingestLikelyImageUrl(urls, match[1]);\n  }\n\n  for (const match of content.matchAll(BARE_SRC_PARAM_PATTERN)) {\n    ingestLikelyImageUrl(urls, match[1]);\n  }\n\n  for (const raw of extractDataBgImageUrls(content)) {\n    ingestLikelyImageUrl(urls, raw);\n  }\n\n  for (const raw of extractCssBackgroundImageUrls(content)) {\n    ingestLikelyImageUrl(urls, raw);\n  }\n\n  return [...urls];\n}\n\n/** @deprecated Use discoverContentAssetUrls — kept for call-site clarity during transition. */\nexport function extractInlineImageSrcs(content: string): string[] {\n  return discoverContentAssetUrls(content);\n}\n\n// --- Migration media refs (`artinstack-migration://asset/{sourceId}`) ---\n\n/** Pseudo-URL scheme for portable migration asset pointers (not WordPress shortcodes). */\nexport const MIGRATION_MEDIA_REF_SCHEME = \"artinstack-migration://asset/\";\n\n/** Build `artinstack-migration://asset/{sourceId}` (percent-encodes the normalizer source id). */\nexport function formatMigrationMediaRef(sourceAssetId: string): string {\n  return `${MIGRATION_MEDIA_REF_SCHEME}${encodeURIComponent(sourceAssetId)}`;\n}\n\nexport function isMigrationMediaRef(value: string): boolean {\n  return value.trim().startsWith(MIGRATION_MEDIA_REF_SCHEME);\n}\n\n/** Parse a migration media ref back to the normalizer `sourceId`, or `undefined` if not a ref. */\nexport function parseMigrationMediaRef(value: string): string | undefined {\n  const trimmed = value.trim();\n  if (!trimmed.startsWith(MIGRATION_MEDIA_REF_SCHEME)) return undefined;\n  const encoded = trimmed.slice(MIGRATION_MEDIA_REF_SCHEME.length);\n  if (!encoded) return undefined;\n  try {\n    return decodeURIComponent(encoded);\n  } catch {\n    return undefined;\n  }\n}\n\n/** Default `replaceWith` for `rewriteInlineImages` / `stampMigrationMediaRefs` (OSS-14). */\nexport function createMigrationMediaRefReplaceWith(): (\n  ref: { sourceAssetId?: string },\n) => string {\n  return (ref) => {\n    if (!ref.sourceAssetId) return \"\";\n    return formatMigrationMediaRef(ref.sourceAssetId);\n  };\n}\n\n// --- Canonical inline keys & lookup index (OSS-15) ---\n\nexport interface CanonicalInlineAssetUrl {\n  /** Canonical absolute URL stored on `NormalizedAsset.sourceUrl`. */\n  canonicalUrl: string;\n  /** Normalizer id: `url:{canonicalUrl}`. */\n  sourceId: string;\n}\n\n/**\n * OSS-15: one canonical key for inline `url:` assets — apply origin rewrite then\n * `normalizeAssetUrl` so discovery, refs, and vault entities share the same id.\n */\nexport function canonicalizeInlineAssetUrl(\n  raw: string,\n  originUrlRewrite?: OriginUrlRewriteConfig,\n): CanonicalInlineAssetUrl | undefined {\n  let value = raw.trim();\n  if (!value || value.startsWith(\"data:\")) return undefined;\n\n  if (originUrlRewrite) {\n    value = rewriteOriginUrlsInText(value, originUrlRewrite);\n  }\n\n  const canonicalUrl = normalizeAssetUrl(value);\n  if (!canonicalUrl) return undefined;\n\n  return {\n    canonicalUrl,\n    sourceId: `url:${canonicalUrl}`,\n  };\n}\n\nfunction urlPathname(url: string): string | undefined {\n  try {\n    return new URL(url, \"http://migration.local\").pathname;\n  } catch {\n    return undefined;\n  }\n}\n\n/**\n * Map normalized upload URLs (and pathnames) → normalizer `sourceId`.\n * Attachment ids are WXR `post_id` strings; inline discoveries use `url:{src}`.\n */\nexport function buildMigrationMediaUrlIndex(\n  entries: Iterable<{ sourceUrl: string; sourceId: string }>,\n): Map<string, string> {\n  const index = new Map<string, string>();\n\n  for (const entry of entries) {\n    index.set(entry.sourceUrl, entry.sourceId);\n    const normalized = normalizeAssetUrl(entry.sourceUrl);\n    if (normalized) index.set(normalized, entry.sourceId);\n    const pathname = urlPathname(entry.sourceUrl);\n    if (pathname) index.set(pathname, entry.sourceId);\n  }\n\n  return index;\n}\n\nexport function resolveMigrationMediaSourceId(\n  src: string,\n  urlIndex: Map<string, string>,\n  originUrlRewrite?: OriginUrlRewriteConfig,\n): string | undefined {\n  const canonical = canonicalizeInlineAssetUrl(src, originUrlRewrite);\n  const normalized = canonical?.canonicalUrl ?? normalizeAssetUrl(src);\n  if (!normalized) return undefined;\n\n  return (\n    urlIndex.get(normalized) ??\n    urlIndex.get(src) ??\n    (urlPathname(normalized) ? urlIndex.get(urlPathname(normalized)!) : undefined)\n  );\n}\n\n/** Merge attachment + inline asset rows into one stamp/lookup index (OSS-15). */\nexport function buildContentMediaUrlIndex(\n  entries: Iterable<{ sourceUrl: string; sourceId: string }>,\n  originUrlRewrite?: OriginUrlRewriteConfig,\n): Map<string, string> {\n  const canonicalEntries: { sourceUrl: string; sourceId: string }[] = [];\n  for (const entry of entries) {\n    const canonical = canonicalizeInlineAssetUrl(entry.sourceUrl, originUrlRewrite);\n    canonicalEntries.push({\n      sourceUrl: canonical?.canonicalUrl ?? entry.sourceUrl,\n      sourceId: entry.sourceId,\n    });\n  }\n  return buildMigrationMediaUrlIndex(canonicalEntries);\n}\n"],"mappings":";AAAA,YAAY,aAAa;AAelB,SAAS,wBAAwB,MAAc,QAAwC;AAC5F,MAAI,CAAC,QAAQ,OAAO,MAAM,WAAW,EAAG,QAAO;AAE/C,MAAI,SAAS;AACb,aAAW,QAAQ,OAAO,OAAO;AAC/B,QAAI,OAAO,KAAK,UAAU,UAAU;AAClC,UAAI,CAAC,KAAK,MAAO;AACjB,eAAS,OAAO,MAAM,KAAK,KAAK,EAAE,KAAK,KAAK,OAAO;AACnD;AAAA,IACF;AACA,aAAS,OAAO,QAAQ,KAAK,OAAO,KAAK,OAAO;AAAA,EAClD;AACA,SAAO;AACT;AAGO,SAAS,8BAA8B,aAAqB,cAA8C;AAC/G,QAAM,oBAAoB,YAAY,QAAQ,OAAO,EAAE;AACvD,QAAM,mBAAmB,aAAa,QAAQ,OAAO,EAAE;AACvD,SAAO;AAAA,IACL,OAAO;AAAA,MACL;AAAA,QACE,OAAO,GAAG,iBAAiB;AAAA,QAC3B,SAAS,GAAG,gBAAgB;AAAA,MAC9B;AAAA,IACF;AAAA,EACF;AACF;AAIA,IAAM,mBAAmB;AAEzB,IAAM,0BAA0B,IAAI,OAAO,OAAO,WAAW,gBAAgB,OAAO,GAAG;AAGvF,IAAM,oBAAoB,OAAO,iBAAiB,gBAAgB;AAElE,IAAM,gCAAgC,IAAI;AAAA,EACxC,OAAO,2DAA2D,iBAAiB;AAAA,EACnF;AACF;AAGA,IAAM,yBAAyB,IAAI;AAAA,EACjC,OAAO,uBAAuB,iBAAiB;AAAA,EAC/C;AACF;AAEA,IAAM,wBAAwB;AAG9B,IAAM,+BACJ;AAEF,IAAM,yBAAyB,IAAI;AAAA,EACjC,OAAO,iDAAiD,iBAAiB;AAAA,EACzE;AACF;AAEA,IAAM,6BAA6B,IAAI;AAAA,EACrC,OAAO,yBAAyB,iBAAiB;AAAA,EACjD;AACF;AAEA,IAAM,sBAAsB;AAQ5B,SAAS,qBAAqB,MAAmB,KAA+B;AAC9E,QAAM,aAAa,kBAAkB,OAAO,EAAE;AAC9C,MAAI,cAAc,iBAAiB,UAAU,GAAG;AAC9C,SAAK,IAAI,UAAU;AAAA,EACrB;AACF;AAEA,SAAS,kBAAkB,SAA2B;AACpD,MAAI,CAAC,QAAQ,KAAK,EAAG,QAAO,CAAC;AAC7B,QAAM,IAAY,aAAK,SAAS,EAAE,KAAK,MAAM,CAAC;AAC9C,QAAM,OAAiB,CAAC;AACxB,IAAE,UAAU,EAAE,KAAK,CAAC,GAAG,OAAO;AAC5B,UAAM,MAAM,EAAE,EAAE,EAAE,KAAK,KAAK,GAAG,KAAK;AACpC,QAAI,IAAK,MAAK,KAAK,GAAG;AAAA,EACxB,CAAC;AACD,SAAO;AACT;AAEA,SAAS,kBAAkB,OAAwB;AACjD,QAAM,cAAc,MAAM,MAAM,KAAK,CAAC,EAAE,CAAC,KAAK;AAC9C,QAAM,eAAe,YAAY,MAAM,KAAK,CAAC,EAAE,CAAC,KAAK;AACrD,SAAO,wBAAwB,KAAK,YAAY;AAClD;AAEA,SAAS,uBAAuB,SAA2B;AACzD,QAAM,OAAiB,CAAC;AACxB,aAAW,SAAS,QAAQ,SAAS,qBAAqB,GAAG;AAC3D,UAAM,MAAM,MAAM,CAAC,GAAG,KAAK;AAC3B,QAAI,IAAK,MAAK,KAAK,GAAG;AAAA,EACxB;AACA,SAAO;AACT;AAEA,SAAS,8BAA8B,SAA2B;AAChE,QAAM,OAAiB,CAAC;AACxB,aAAW,SAAS,QAAQ,SAAS,4BAA4B,GAAG;AAClE,UAAM,MAAM,MAAM,CAAC,GAAG,KAAK;AAC3B,QAAI,IAAK,MAAK,KAAK,GAAG;AAAA,EACxB;AACA,SAAO;AACT;AAGO,SAAS,mBAAmB,SAA2B;AAC5D,SAAO,kBAAkB,OAAO,EAAE,OAAO,CAAC,QAAQ,CAAC,IAAI,WAAW,OAAO,CAAC;AAC5E;AAGO,SAAS,kBAAkB,KAAiC;AACjE,QAAM,UAAU,IAAI,KAAK;AACzB,MAAI,CAAC,WAAW,QAAQ,WAAW,OAAO,EAAG,QAAO;AACpD,MAAI,QAAQ,WAAW,IAAI,EAAG,QAAO,SAAS,OAAO;AACrD,SAAO;AACT;AAGO,SAAS,iBAAiB,KAAsB;AACrD,MAAI,CAAC,OAAO,IAAI,WAAW,OAAO,EAAG,QAAO;AAE5C,MAAI,IAAI,WAAW,GAAG,GAAG;AACvB,WAAO,kBAAkB,GAAG;AAAA,EAC9B;AAEA,MAAI,CAAC,gBAAgB,KAAK,GAAG,EAAG,QAAO;AAEvC,MAAI;AACF,UAAM,EAAE,SAAS,IAAI,IAAI,IAAI,GAAG;AAChC,QAAI,kBAAkB,QAAQ,EAAG,QAAO;AAAA,EAC1C,QAAQ;AAAA,EAER;AAEA,SAAO,kBAAkB,GAAG;AAC9B;AAEA,SAAS,sBACP,YACA,KACA,OACA,MACM;AACN,QAAM,aAAa,kBAAkB,OAAO,EAAE;AAC9C,MAAI,CAAC,cAAc,CAAC,iBAAiB,UAAU,EAAG;AAClD,aAAW,KAAK,EAAE,KAAK,YAAY,OAAO,KAAK,CAAC;AAClD;AAEA,SAAS,+BAA+B,SAA2C;AACjF,QAAM,aAAuC,CAAC;AAE9C,aAAW,SAAS,QAAQ,SAAS,qBAAqB,GAAG;AAC3D,0BAAsB,YAAY,MAAM,CAAC,GAAG,MAAM,SAAS,GAAG,CAAC;AAAA,EACjE;AACA,aAAW,SAAS,QAAQ,SAAS,4BAA4B,GAAG;AAClE,0BAAsB,YAAY,MAAM,CAAC,GAAG,MAAM,SAAS,GAAG,CAAC;AAAA,EACjE;AACA,aAAW,SAAS,QAAQ,SAAS,sBAAsB,GAAG;AAC5D,0BAAsB,YAAY,MAAM,CAAC,GAAG,MAAM,SAAS,GAAG,CAAC;AAAA,EACjE;AACA,aAAW,SAAS,QAAQ,SAAS,mBAAmB,GAAG;AACzD,0BAAsB,YAAY,MAAM,CAAC,GAAG,MAAM,SAAS,GAAG,CAAC;AAAA,EACjE;AACA,aAAW,SAAS,QAAQ,SAAS,0BAA0B,GAAG;AAChE,0BAAsB,YAAY,MAAM,CAAC,GAAG,MAAM,SAAS,GAAG,CAAC;AAAA,EACjE;AAEA,SAAO;AACT;AAQO,SAAS,mCAAmC,SAA2B;AAC5E,MAAI,CAAC,QAAQ,KAAK,EAAG,QAAO,CAAC;AAE7B,QAAM,SAAS,CAAC,GAAG,+BAA+B,OAAO,CAAC,EAAE,KAAK,CAAC,MAAM,UAAU;AAChF,QAAI,KAAK,SAAS,MAAM,KAAM,QAAO,KAAK,OAAO,MAAM;AACvD,WAAO,KAAK,QAAQ,MAAM;AAAA,EAC5B,CAAC;AAED,QAAM,OAAiB,CAAC;AACxB,QAAM,OAAO,oBAAI,IAAY;AAC7B,aAAW,aAAa,QAAQ;AAC9B,QAAI,KAAK,IAAI,UAAU,GAAG,EAAG;AAC7B,SAAK,IAAI,UAAU,GAAG;AACtB,SAAK,KAAK,UAAU,GAAG;AAAA,EACzB;AACA,SAAO;AACT;AAGO,SAAS,+BAA+B,SAAqC;AAClF,SAAO,mCAAmC,OAAO,EAAE,CAAC;AACtD;AAQO,SAAS,yBAAyB,SAA2B;AAClE,MAAI,CAAC,QAAQ,KAAK,EAAG,QAAO,CAAC;AAE7B,QAAM,OAAO,oBAAI,IAAY;AAE7B,aAAW,OAAO,kBAAkB,OAAO,GAAG;AAC5C,yBAAqB,MAAM,GAAG;AAAA,EAChC;AAEA,aAAW,SAAS,QAAQ,SAAS,6BAA6B,GAAG;AACnE,yBAAqB,MAAM,MAAM,CAAC,CAAC;AAAA,EACrC;AAEA,aAAW,SAAS,QAAQ,SAAS,sBAAsB,GAAG;AAC5D,yBAAqB,MAAM,MAAM,CAAC,CAAC;AAAA,EACrC;AAEA,aAAW,OAAO,uBAAuB,OAAO,GAAG;AACjD,yBAAqB,MAAM,GAAG;AAAA,EAChC;AAEA,aAAW,OAAO,8BAA8B,OAAO,GAAG;AACxD,yBAAqB,MAAM,GAAG;AAAA,EAChC;AAEA,SAAO,CAAC,GAAG,IAAI;AACjB;AAGO,SAAS,uBAAuB,SAA2B;AAChE,SAAO,yBAAyB,OAAO;AACzC;AAKO,IAAM,6BAA6B;AAGnC,SAAS,wBAAwB,eAA+B;AACrE,SAAO,GAAG,0BAA0B,GAAG,mBAAmB,aAAa,CAAC;AAC1E;AAEO,SAAS,oBAAoB,OAAwB;AAC1D,SAAO,MAAM,KAAK,EAAE,WAAW,0BAA0B;AAC3D;AAGO,SAAS,uBAAuB,OAAmC;AACxE,QAAM,UAAU,MAAM,KAAK;AAC3B,MAAI,CAAC,QAAQ,WAAW,0BAA0B,EAAG,QAAO;AAC5D,QAAM,UAAU,QAAQ,MAAM,2BAA2B,MAAM;AAC/D,MAAI,CAAC,QAAS,QAAO;AACrB,MAAI;AACF,WAAO,mBAAmB,OAAO;AAAA,EACnC,QAAQ;AACN,WAAO;AAAA,EACT;AACF;AAGO,SAAS,qCAEJ;AACV,SAAO,CAAC,QAAQ;AACd,QAAI,CAAC,IAAI,cAAe,QAAO;AAC/B,WAAO,wBAAwB,IAAI,aAAa;AAAA,EAClD;AACF;AAeO,SAAS,2BACd,KACA,kBACqC;AACrC,MAAI,QAAQ,IAAI,KAAK;AACrB,MAAI,CAAC,SAAS,MAAM,WAAW,OAAO,EAAG,QAAO;AAEhD,MAAI,kBAAkB;AACpB,YAAQ,wBAAwB,OAAO,gBAAgB;AAAA,EACzD;AAEA,QAAM,eAAe,kBAAkB,KAAK;AAC5C,MAAI,CAAC,aAAc,QAAO;AAE1B,SAAO;AAAA,IACL;AAAA,IACA,UAAU,OAAO,YAAY;AAAA,EAC/B;AACF;AAEA,SAAS,YAAY,KAAiC;AACpD,MAAI;AACF,WAAO,IAAI,IAAI,KAAK,wBAAwB,EAAE;AAAA,EAChD,QAAQ;AACN,WAAO;AAAA,EACT;AACF;AAMO,SAAS,4BACd,SACqB;AACrB,QAAM,QAAQ,oBAAI,IAAoB;AAEtC,aAAW,SAAS,SAAS;AAC3B,UAAM,IAAI,MAAM,WAAW,MAAM,QAAQ;AACzC,UAAM,aAAa,kBAAkB,MAAM,SAAS;AACpD,QAAI,WAAY,OAAM,IAAI,YAAY,MAAM,QAAQ;AACpD,UAAM,WAAW,YAAY,MAAM,SAAS;AAC5C,QAAI,SAAU,OAAM,IAAI,UAAU,MAAM,QAAQ;AAAA,EAClD;AAEA,SAAO;AACT;AAEO,SAAS,8BACd,KACA,UACA,kBACoB;AACpB,QAAM,YAAY,2BAA2B,KAAK,gBAAgB;AAClE,QAAM,aAAa,WAAW,gBAAgB,kBAAkB,GAAG;AACnE,MAAI,CAAC,WAAY,QAAO;AAExB,SACE,SAAS,IAAI,UAAU,KACvB,SAAS,IAAI,GAAG,MACf,YAAY,UAAU,IAAI,SAAS,IAAI,YAAY,UAAU,CAAE,IAAI;AAExE;AAGO,SAAS,0BACd,SACA,kBACqB;AACrB,QAAM,mBAA8D,CAAC;AACrE,aAAW,SAAS,SAAS;AAC3B,UAAM,YAAY,2BAA2B,MAAM,WAAW,gBAAgB;AAC9E,qBAAiB,KAAK;AAAA,MACpB,WAAW,WAAW,gBAAgB,MAAM;AAAA,MAC5C,UAAU,MAAM;AAAA,IAClB,CAAC;AAAA,EACH;AACA,SAAO,4BAA4B,gBAAgB;AACrD;","names":[]}

package/dist/chunk-XRCF73DA.js

@@ -0,0 +1,24 @@
+// src/lib/utility.ts
+function sanitizeSlug(raw) {
+  return raw.trim().toLowerCase().replace(/&[^;]+;/g, "").replace(/[^a-z0-9]+/g, "-").replace(/^-+|-+$/g, "").slice(0, 200);
+}
+function linkToPath(link) {
+  if (!link) return void 0;
+  try {
+    const url = new URL(link);
+    const path = url.pathname;
+    if (!path || path === "/") return "/";
+    return path.endsWith("/") ? path : `${path}/`;
+  } catch {
+    if (link.startsWith("/")) {
+      return link.endsWith("/") ? link : `${link}/`;
+    }
+    return void 0;
+  }
+}
+
+export {
+  sanitizeSlug,
+  linkToPath
+};
+//# sourceMappingURL=chunk-XRCF73DA.js.map

package/dist/chunk-XRCF73DA.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/lib/utility.ts"],"sourcesContent":["/** Lowercase URL-safe slug from WordPress post_name or title. */\nexport function sanitizeSlug(raw: string): string {\n  return raw\n    .trim()\n    .toLowerCase()\n    .replace(/&[^;]+;/g, \"\")\n    .replace(/[^a-z0-9]+/g, \"-\")\n    .replace(/^-+|-+$/g, \"\")\n    .slice(0, 200);\n}\n\n/** Normalize absolute permalink to root-relative path. */\nexport function linkToPath(link: string | undefined): string | undefined {\n  if (!link) return undefined;\n  try {\n    const url = new URL(link);\n    const path = url.pathname;\n    if (!path || path === \"/\") return \"/\";\n    return path.endsWith(\"/\") ? path : `${path}/`;\n  } catch {\n    if (link.startsWith(\"/\")) {\n      return link.endsWith(\"/\") ? link : `${link}/`;\n    }\n    return undefined;\n  }\n}\n"],"mappings":";AACO,SAAS,aAAa,KAAqB;AAChD,SAAO,IACJ,KAAK,EACL,YAAY,EACZ,QAAQ,YAAY,EAAE,EACtB,QAAQ,eAAe,GAAG,EAC1B,QAAQ,YAAY,EAAE,EACtB,MAAM,GAAG,GAAG;AACjB;AAGO,SAAS,WAAW,MAA8C;AACvE,MAAI,CAAC,KAAM,QAAO;AAClB,MAAI;AACF,UAAM,MAAM,IAAI,IAAI,IAAI;AACxB,UAAM,OAAO,IAAI;AACjB,QAAI,CAAC,QAAQ,SAAS,IAAK,QAAO;AAClC,WAAO,KAAK,SAAS,GAAG,IAAI,OAAO,GAAG,IAAI;AAAA,EAC5C,QAAQ;AACN,QAAI,KAAK,WAAW,GAAG,GAAG;AACxB,aAAO,KAAK,SAAS,GAAG,IAAI,OAAO,GAAG,IAAI;AAAA,IAC5C;AACA,WAAO;AAAA,EACT;AACF;","names":[]}

package/dist/{chunk-IPYHS5R2.js → chunk-Z3L6N63Y.js}

@@ -8,12 +8,18 @@ import {
 } from "./chunk-HI7JHWZU.js";
 import {
   rewriteInlineImages
-} from "./chunk-EXYXTAZM.js";
+} from "./chunk-KYNKJ4XV.js";
 import {
+  linkToPath,
+  sanitizeSlug
+} from "./chunk-XRCF73DA.js";
+import {
+  createMigrationMediaRefReplaceWith,
   discoverContentAssetUrls,
   discoverRawImgSrcs,
+  isMigrationMediaRef,
   normalizeAssetUrl
-} from "./chunk-XYP3VYDH.js";
+} from "./chunk-WHGUE5FC.js";
 
 // src/sinks/types.ts
 var MIGRATION_WRITE_STAGES = [
@@ -35,25 +41,6 @@ import * as cheerio2 from "cheerio";
 import { readFile } from "fs/promises";
 import { basename } from "path";
 
-// src/lib/utility.ts
-function sanitizeSlug(raw) {
-  return raw.trim().toLowerCase().replace(/&[^;]+;/g, "").replace(/[^a-z0-9]+/g, "-").replace(/^-+|-+$/g, "").slice(0, 200);
-}
-function linkToPath(link) {
-  if (!link) return void 0;
-  try {
-    const url = new URL(link);
-    const path = url.pathname;
-    if (!path || path === "/") return "/";
-    return path.endsWith("/") ? path : `${path}/`;
-  } catch {
-    if (link.startsWith("/")) {
-      return link.endsWith("/") ? link : `${link}/`;
-    }
-    return void 0;
-  }
-}
-
 // src/parsers/squarespace/collect.ts
 import * as cheerio from "cheerio";
 import { z } from "zod";
@@ -961,8 +948,35 @@ function fractionalLayoutMap(config) {
     bgParamName: config.bgParamName
   };
 }
+var WP_WIDGET_PLACEHOLDER = "\u200B";
+var WORDPRESS_WIDGET_REGISTRY = {
+  mapShortcodePrefixes: [
+    "blox_gmap",
+    "tatsu_gmap",
+    "tatsu_map",
+    "et_pb_map",
+    "vc_gmaps",
+    "vc_map"
+  ],
+  contactFormRules: [
+    { tag: "contact-form-7", source: "contact-form-7", idParam: "id" },
+    { tag: "contact_form", source: "contact-form-7", idParam: "id" },
+    { tag: "gravityform", source: "gravityforms", idParam: "id" },
+    { tag: "ninja_form", source: "ninja-forms", idParam: "id" },
+    { tag: "wpforms", source: "wpforms", idParam: "id" }
+  ],
+  videoShortcodePrefixes: [
+    "youtube",
+    "vimeo",
+    "embed",
+    "tatsu_video",
+    "et_pb_video",
+    "vc_video"
+  ],
+  portfolioShortcode: "portfolio",
+  galleryShortcode: "gallery"
+};
 var UNRESOLVABLE_SHORTCODE_PREFIXES = [
-  "portfolio",
   "recent_posts",
   "woocommerce_cart",
   "woocommerce_checkout",
@@ -986,8 +1000,7 @@ var WORDPRESS_BUILDER_REGISTRY = [
     ],
     urlRules: [
       { shortcodePrefix: "tatsu_image", urlParams: ["image", "url", "src"], tag: "img" },
-      { shortcodePrefix: "tatsu_single_image", urlParams: ["image", "url", "src"], tag: "img" },
-      { shortcodePrefix: "tatsu_video", urlParams: ["video", "src", "url"], tag: "video" }
+      { shortcodePrefix: "tatsu_single_image", urlParams: ["image", "url", "src"], tag: "img" }
     ],
     iconImageRules: [
       { shortcodePrefix: "tatsu_icon", imageParam: "icon_image", hrefParam: "href" }
@@ -1097,12 +1110,6 @@ var WORDPRESS_BUILDER_REGISTRY = [
     urlRules: [
       { shortcodePrefix: "blox_image", urlParams: ["image", "img", "src", "url"], tag: "img" }
     ],
-    placeholderRules: [
-      {
-        shortcodePrefix: "blox_gmap",
-        html: '<p data-unresolved-shortcode="blox_gmap"><!-- Map embed removed during migration --></p>'
-      }
-    ],
     scaffoldingPrefixes: ["blox_", "animate_icon"],
     legacyScaffoldingTokens: ["text", "icon", "linebreak", "grids", "testimonials"]
   }
@@ -1194,6 +1201,7 @@ function mediaUrlSet(bundle) {
 function findUnresolvedInlineImages(sourceId, contentHtml, mediaUrls) {
   const conflicts = [];
   for (const raw of discoverRawImgSrcs(contentHtml)) {
+    if (isMigrationMediaRef(raw)) continue;
     const normalized = normalizeAssetUrl(raw);
     if (!normalized) continue;
     if (!mediaUrls.has(normalized) && !mediaUrls.has(raw)) {
@@ -1442,6 +1450,8 @@ function mergeRewriteOptions(bundle, options) {
     const normalized = normalizeAssetUrl(asset.sourceUrl);
     if (normalized) urlToSourceId.set(normalized, asset.sourceId);
   }
+  const replaceWith = options.replaceWith ?? createMigrationMediaRefReplaceWith();
+  const requireUploaded = options.requireUploaded ?? Boolean(options.replaceWith);
   return {
     resolveAsset: (src) => {
       const resolved = options.resolveAsset(src);
@@ -1451,7 +1461,8 @@ function mergeRewriteOptions(bundle, options) {
       if (!sourceAssetId) return void 0;
       return { originalSrc: src, sourceAssetId };
     },
-    replaceWith: options.replaceWith
+    replaceWith,
+    requireUploaded
   };
 }
 function emptyBody() {
@@ -1682,8 +1693,6 @@ async function runDryRun(options) {
 
 export {
   MIGRATION_WRITE_STAGES,
-  sanitizeSlug,
-  linkToPath,
   SQUARESPACE_JSON_FORMAT,
   buildJsonPrettyUrl,
   mapJsonPrettyWire,
@@ -1691,6 +1700,8 @@ export {
   enumerateSquarespaceEntities,
   summarizeSquarespaceExport,
   validateSquarespaceExportFile,
+  WP_WIDGET_PLACEHOLDER,
+  WORDPRESS_WIDGET_REGISTRY,
   WORDPRESS_BUILDER_REGISTRY,
   emptyConflictReport,
   analyzeConflicts,
@@ -1711,4 +1722,4 @@ export {
   staleUrlsFromEstimate,
   runDryRun
 };
-//# sourceMappingURL=chunk-IPYHS5R2.js.map
+//# sourceMappingURL=chunk-Z3L6N63Y.js.map