@artinstack/migrator 0.1.7 → 0.1.9

package/README.md

@@ -16,7 +16,8 @@ src/
   sinks/            filesystem export, MigrationSink interface
   cli/              artinstack-migrate
   transformers/     HtmlToGrapes, css-to-styles, inline image rewrite, media ref expand
-  lib/              content-asset-urls, migration-media-ref, origin-url-rewrite
+  lib/              media-urls, utility (shared helpers)
+  test/             unit tests (mirrors src/ layout; vitest only)
 ```
 
 ## Install

package/dist/{bundle-uAAHehbv.d.ts → bundle-B3XS20r_.d.ts}

@@ -1,4 +1,4 @@
-import { N as NormalizedPost, a as NormalizedPage, b as NormalizedAsset, c as NormalizedPortfolio, d as NormalizedCategory, e as NormalizedTag, f as NormalizedEntity } from './types-DWOP8Dcy.js';
+import { N as NormalizedPost, a as NormalizedPage, b as NormalizedAsset, c as NormalizedPortfolio, d as NormalizedCategory, e as NormalizedTag, f as NormalizedEntity } from './types-Ce4r6zqt.js';
 
 interface EntityBundle {
     posts: NormalizedPost[];

package/dist/{chunk-BOYB6XRA.js → chunk-BONZ3U3I.js}

@@ -1,57 +1,9 @@
 import {
-  normalizeAssetUrl
-} from "./chunk-XYP3VYDH.js";
-
-// src/lib/migration-media-ref.ts
-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);
-  };
-}
-
-// src/lib/migration-media-url-index.ts
-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) {
-  const normalized = normalizeAssetUrl(src);
-  if (!normalized) return void 0;
-  return urlIndex.get(normalized) ?? urlIndex.get(src) ?? (urlPathname(normalized) ? urlIndex.get(urlPathname(normalized)) : void 0);
-}
+  createMigrationMediaRefReplaceWith,
+  isMigrationMediaRef,
+  normalizeAssetUrl,
+  resolveMigrationMediaSourceId
+} from "./chunk-S4GMDRGX.js";
 
 // src/transformers/rewrite-inline-images.ts
 import * as cheerio from "cheerio";
@@ -150,7 +102,11 @@ function stampMigrationMediaRefs(html, options) {
     html,
     {
       resolveAsset: (src) => {
-        const sourceAssetId = resolveMigrationMediaSourceId(src, options.urlToSourceId);
+        const sourceAssetId = resolveMigrationMediaSourceId(
+          src,
+          options.urlToSourceId,
+          options.originUrlRewrite
+        );
         if (!sourceAssetId) return void 0;
         return { originalSrc: src, sourceAssetId };
       },
@@ -162,14 +118,7 @@ function stampMigrationMediaRefs(html, options) {
 }
 
 export {
-  MIGRATION_MEDIA_REF_SCHEME,
-  formatMigrationMediaRef,
-  isMigrationMediaRef,
-  parseMigrationMediaRef,
-  createMigrationMediaRefReplaceWith,
-  buildMigrationMediaUrlIndex,
-  resolveMigrationMediaSourceId,
   rewriteInlineImages,
   stampMigrationMediaRefs
 };
-//# sourceMappingURL=chunk-BOYB6XRA.js.map
+//# sourceMappingURL=chunk-BONZ3U3I.js.map

package/dist/chunk-BONZ3U3I.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/transformers/rewrite-inline-images.ts"],"sourcesContent":["import * as cheerio from \"cheerio\";\n\nimport {\n  buildMigrationMediaUrlIndex,\n  createMigrationMediaRefReplaceWith,\n  isMigrationMediaRef,\n  normalizeAssetUrl,\n  resolveMigrationMediaSourceId,\n  type OriginUrlRewriteConfig,\n} from \"../lib/media-urls.js\";\n\nexport interface RewriteInlineImageRef {\n  originalSrc: string;\n  sourceAssetId?: string;\n}\n\nexport interface UploadedAssetRef {\n  targetId: string;\n  publicUrl?: string;\n}\n\nexport interface RewriteInlineImagesOptions {\n  resolveAsset: (src: string) => RewriteInlineImageRef | undefined;\n  /**\n   * Replace a resolved source id with a migration ref or CDN URL.\n   * When omitted, defaults to OSS-14 `artinstack-migration://asset/…` refs.\n   */\n  replaceWith?: (ref: RewriteInlineImageRef, uploaded?: UploadedAssetRef) => string;\n  /**\n   * When true, skip URLs that cannot be matched to an uploaded vault target.\n   * Default: false when using migration refs; true when a custom `replaceWith` is supplied.\n   */\n  requireUploaded?: boolean;\n}\n\nexport interface RewriteInlineImagesResult {\n  html: string;\n  referencedSources: string[];\n  unresolved: string[];\n}\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\nfunction resolveRewriteOptions(\n  options: RewriteInlineImagesOptions,\n): Required<Pick<RewriteInlineImagesOptions, \"replaceWith\" | \"requireUploaded\">> {\n  const replaceWith = options.replaceWith ?? createMigrationMediaRefReplaceWith();\n  const requireUploaded = options.requireUploaded ?? Boolean(options.replaceWith);\n  return { replaceWith, requireUploaded };\n}\n\nfunction tryRewriteUrl(\n  src: string,\n  options: RewriteInlineImagesOptions,\n  uploadedBySourceId: Map<string, UploadedAssetRef>,\n  referencedSources: Set<string>,\n  unresolved: Set<string>,\n): string | undefined {\n  const normalized = normalizeAssetUrl(src);\n  if (!normalized) return undefined;\n\n  if (isMigrationMediaRef(normalized)) {\n    referencedSources.add(normalized);\n    return normalized;\n  }\n\n  referencedSources.add(normalized);\n  const ref = options.resolveAsset(normalized);\n  if (!ref?.sourceAssetId) {\n    unresolved.add(normalized);\n    return undefined;\n  }\n\n  const { replaceWith, requireUploaded } = resolveRewriteOptions(options);\n  const uploaded = uploadedBySourceId.get(ref.sourceAssetId);\n  if (requireUploaded && !uploaded) {\n    unresolved.add(normalized);\n    return undefined;\n  }\n\n  return replaceWith(ref, uploaded);\n}\n\nfunction rewriteBackgroundUrlsInStyle(\n  style: string,\n  options: RewriteInlineImagesOptions,\n  uploadedBySourceId: Map<string, UploadedAssetRef>,\n  referencedSources: Set<string>,\n  unresolved: Set<string>,\n): string {\n  return style.replace(BACKGROUND_IMAGE_URL_PATTERN, (full, quote: string, rawUrl: string) => {\n    const replaced = tryRewriteUrl(rawUrl.trim(), options, uploadedBySourceId, referencedSources, unresolved);\n    if (!replaced) return full;\n\n    const urlCall = quote\n      ? `url(${quote}${replaced}${quote})`\n      : `url(${replaced})`;\n    return full.replace(/url\\s*\\(\\s*(['\"]?)([^'\")]+)\\1\\s*\\)/i, urlCall);\n  });\n}\n\nfunction rewriteSrcset(\n  srcset: string,\n  options: RewriteInlineImagesOptions,\n  uploadedBySourceId: Map<string, UploadedAssetRef>,\n  referencedSources: Set<string>,\n  unresolved: Set<string>,\n): string {\n  return srcset\n    .split(\",\")\n    .map((entry) => {\n      const trimmed = entry.trim();\n      if (!trimmed) return entry;\n      const [urlPart, descriptor] = trimmed.split(/\\s+/, 2);\n      const replaced = tryRewriteUrl(urlPart ?? \"\", options, uploadedBySourceId, referencedSources, unresolved);\n      if (!replaced) return entry;\n      return descriptor ? `${replaced} ${descriptor}` : replaced;\n    })\n    .join(\", \");\n}\n\n/** Rewrite `<img src>` / `srcset`, `data-bg-image`, and inline CSS backgrounds using uploaded asset targets. */\nexport function rewriteInlineImages(\n  html: string,\n  options: RewriteInlineImagesOptions,\n  uploadedBySourceId: Map<string, UploadedAssetRef>,\n): RewriteInlineImagesResult {\n  if (!html.trim()) {\n    return { html, referencedSources: [], unresolved: [] };\n  }\n\n  const $ = cheerio.load(html, { xml: false });\n  const referencedSources = new Set<string>();\n  const unresolved = new Set<string>();\n\n  $(\"img\").each((_, element) => {\n    const img = $(element);\n    const src = img.attr(\"src\")?.trim();\n    if (src && !src.startsWith(\"data:\")) {\n      const replaced = tryRewriteUrl(src, options, uploadedBySourceId, referencedSources, unresolved);\n      if (replaced) img.attr(\"src\", replaced);\n    }\n\n    const srcset = img.attr(\"srcset\")?.trim();\n    if (srcset) {\n      img.attr(\"srcset\", rewriteSrcset(srcset, options, uploadedBySourceId, referencedSources, unresolved));\n    }\n  });\n\n  $(\"[data-bg-image]\").each((_, element) => {\n    const node = $(element);\n    const bgImage = node.attr(\"data-bg-image\")?.trim();\n    if (!bgImage || bgImage.startsWith(\"data:\")) return;\n    const replaced = tryRewriteUrl(bgImage, options, uploadedBySourceId, referencedSources, unresolved);\n    if (replaced) node.attr(\"data-bg-image\", replaced);\n  });\n\n  $(\"[style]\").each((_, element) => {\n    const node = $(element);\n    const style = node.attr(\"style\");\n    if (!style?.includes(\"background\")) return;\n    const rewritten = rewriteBackgroundUrlsInStyle(\n      style,\n      options,\n      uploadedBySourceId,\n      referencedSources,\n      unresolved,\n    );\n    if (rewritten !== style) node.attr(\"style\", rewritten);\n  });\n\n  return {\n    html: $.root().html() ?? html,\n    referencedSources: [...referencedSources],\n    unresolved: [...unresolved],\n  };\n}\n\nexport interface StampMigrationMediaRefsOptions {\n  /** Pre-built url/pathname → sourceId map (from attachments + inline assets). */\n  urlToSourceId: Map<string, string>;\n  /** Canonicalize lookup keys during stamp (OSS-15). */\n  originUrlRewrite?: OriginUrlRewriteConfig;\n  replaceWith?: RewriteInlineImagesOptions[\"replaceWith\"];\n  requireUploaded?: boolean;\n}\n\n/**\n * OSS-14 — replace resolved `wp-content/uploads` URLs with `artinstack-migration://asset/…`\n * refs. Does not invent refs for unknown URLs (left unchanged + listed in `unresolved`).\n */\nexport function stampMigrationMediaRefs(\n  html: string,\n  options: StampMigrationMediaRefsOptions,\n): RewriteInlineImagesResult {\n  return rewriteInlineImages(\n    html,\n    {\n      resolveAsset: (src) => {\n        const sourceAssetId = resolveMigrationMediaSourceId(\n          src,\n          options.urlToSourceId,\n          options.originUrlRewrite,\n        );\n        if (!sourceAssetId) return undefined;\n        return { originalSrc: src, sourceAssetId };\n      },\n      replaceWith: options.replaceWith,\n      requireUploaded: options.requireUploaded ?? false,\n    },\n    new Map(),\n  );\n}\n\n/** Build a url index from attachment rows and/or normalized assets. */\nexport { buildMigrationMediaUrlIndex };\n"],"mappings":";;;;;;;;AAAA,YAAY,aAAa;AA0CzB,IAAM,+BACJ;AAEF,SAAS,sBACP,SAC+E;AAC/E,QAAM,cAAc,QAAQ,eAAe,mCAAmC;AAC9E,QAAM,kBAAkB,QAAQ,mBAAmB,QAAQ,QAAQ,WAAW;AAC9E,SAAO,EAAE,aAAa,gBAAgB;AACxC;AAEA,SAAS,cACP,KACA,SACA,oBACA,mBACA,YACoB;AACpB,QAAM,aAAa,kBAAkB,GAAG;AACxC,MAAI,CAAC,WAAY,QAAO;AAExB,MAAI,oBAAoB,UAAU,GAAG;AACnC,sBAAkB,IAAI,UAAU;AAChC,WAAO;AAAA,EACT;AAEA,oBAAkB,IAAI,UAAU;AAChC,QAAM,MAAM,QAAQ,aAAa,UAAU;AAC3C,MAAI,CAAC,KAAK,eAAe;AACvB,eAAW,IAAI,UAAU;AACzB,WAAO;AAAA,EACT;AAEA,QAAM,EAAE,aAAa,gBAAgB,IAAI,sBAAsB,OAAO;AACtE,QAAM,WAAW,mBAAmB,IAAI,IAAI,aAAa;AACzD,MAAI,mBAAmB,CAAC,UAAU;AAChC,eAAW,IAAI,UAAU;AACzB,WAAO;AAAA,EACT;AAEA,SAAO,YAAY,KAAK,QAAQ;AAClC;AAEA,SAAS,6BACP,OACA,SACA,oBACA,mBACA,YACQ;AACR,SAAO,MAAM,QAAQ,8BAA8B,CAAC,MAAM,OAAe,WAAmB;AAC1F,UAAM,WAAW,cAAc,OAAO,KAAK,GAAG,SAAS,oBAAoB,mBAAmB,UAAU;AACxG,QAAI,CAAC,SAAU,QAAO;AAEtB,UAAM,UAAU,QACZ,OAAO,KAAK,GAAG,QAAQ,GAAG,KAAK,MAC/B,OAAO,QAAQ;AACnB,WAAO,KAAK,QAAQ,uCAAuC,OAAO;AAAA,EACpE,CAAC;AACH;AAEA,SAAS,cACP,QACA,SACA,oBACA,mBACA,YACQ;AACR,SAAO,OACJ,MAAM,GAAG,EACT,IAAI,CAAC,UAAU;AACd,UAAM,UAAU,MAAM,KAAK;AAC3B,QAAI,CAAC,QAAS,QAAO;AACrB,UAAM,CAAC,SAAS,UAAU,IAAI,QAAQ,MAAM,OAAO,CAAC;AACpD,UAAM,WAAW,cAAc,WAAW,IAAI,SAAS,oBAAoB,mBAAmB,UAAU;AACxG,QAAI,CAAC,SAAU,QAAO;AACtB,WAAO,aAAa,GAAG,QAAQ,IAAI,UAAU,KAAK;AAAA,EACpD,CAAC,EACA,KAAK,IAAI;AACd;AAGO,SAAS,oBACd,MACA,SACA,oBAC2B;AAC3B,MAAI,CAAC,KAAK,KAAK,GAAG;AAChB,WAAO,EAAE,MAAM,mBAAmB,CAAC,GAAG,YAAY,CAAC,EAAE;AAAA,EACvD;AAEA,QAAM,IAAY,aAAK,MAAM,EAAE,KAAK,MAAM,CAAC;AAC3C,QAAM,oBAAoB,oBAAI,IAAY;AAC1C,QAAM,aAAa,oBAAI,IAAY;AAEnC,IAAE,KAAK,EAAE,KAAK,CAAC,GAAG,YAAY;AAC5B,UAAM,MAAM,EAAE,OAAO;AACrB,UAAM,MAAM,IAAI,KAAK,KAAK,GAAG,KAAK;AAClC,QAAI,OAAO,CAAC,IAAI,WAAW,OAAO,GAAG;AACnC,YAAM,WAAW,cAAc,KAAK,SAAS,oBAAoB,mBAAmB,UAAU;AAC9F,UAAI,SAAU,KAAI,KAAK,OAAO,QAAQ;AAAA,IACxC;AAEA,UAAM,SAAS,IAAI,KAAK,QAAQ,GAAG,KAAK;AACxC,QAAI,QAAQ;AACV,UAAI,KAAK,UAAU,cAAc,QAAQ,SAAS,oBAAoB,mBAAmB,UAAU,CAAC;AAAA,IACtG;AAAA,EACF,CAAC;AAED,IAAE,iBAAiB,EAAE,KAAK,CAAC,GAAG,YAAY;AACxC,UAAM,OAAO,EAAE,OAAO;AACtB,UAAM,UAAU,KAAK,KAAK,eAAe,GAAG,KAAK;AACjD,QAAI,CAAC,WAAW,QAAQ,WAAW,OAAO,EAAG;AAC7C,UAAM,WAAW,cAAc,SAAS,SAAS,oBAAoB,mBAAmB,UAAU;AAClG,QAAI,SAAU,MAAK,KAAK,iBAAiB,QAAQ;AAAA,EACnD,CAAC;AAED,IAAE,SAAS,EAAE,KAAK,CAAC,GAAG,YAAY;AAChC,UAAM,OAAO,EAAE,OAAO;AACtB,UAAM,QAAQ,KAAK,KAAK,OAAO;AAC/B,QAAI,CAAC,OAAO,SAAS,YAAY,EAAG;AACpC,UAAM,YAAY;AAAA,MAChB;AAAA,MACA;AAAA,MACA;AAAA,MACA;AAAA,MACA;AAAA,IACF;AACA,QAAI,cAAc,MAAO,MAAK,KAAK,SAAS,SAAS;AAAA,EACvD,CAAC;AAED,SAAO;AAAA,IACL,MAAM,EAAE,KAAK,EAAE,KAAK,KAAK;AAAA,IACzB,mBAAmB,CAAC,GAAG,iBAAiB;AAAA,IACxC,YAAY,CAAC,GAAG,UAAU;AAAA,EAC5B;AACF;AAeO,SAAS,wBACd,MACA,SAC2B;AAC3B,SAAO;AAAA,IACL;AAAA,IACA;AAAA,MACE,cAAc,CAAC,QAAQ;AACrB,cAAM,gBAAgB;AAAA,UACpB;AAAA,UACA,QAAQ;AAAA,UACR,QAAQ;AAAA,QACV;AACA,YAAI,CAAC,cAAe,QAAO;AAC3B,eAAO,EAAE,aAAa,KAAK,cAAc;AAAA,MAC3C;AAAA,MACA,aAAa,QAAQ;AAAA,MACrB,iBAAiB,QAAQ,mBAAmB;AAAA,IAC9C;AAAA,IACA,oBAAI,IAAI;AAAA,EACV;AACF;","names":[]}

package/dist/chunk-EJTWYEAX.js

@@ -0,0 +1 @@
+//# sourceMappingURL=chunk-EJTWYEAX.js.map

package/dist/chunk-EJTWYEAX.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":[],"sourcesContent":[],"mappings":"","names":[]}