@artinstack/migrator 0.1.3 → 0.1.4

package/README.md

@@ -11,6 +11,7 @@ See [docs/architecture.md](./docs/architecture.md) for the high-level blueprint:
 ```
 src/
   parsers/          WordPress, SmugMug, Squarespace, Wix → normalizer DTOs
+    wordpress/      WXR parse, builder flattening (theme registry)
   normalizer/       Canonical DTOs + portable idempotency types
   sinks/            filesystem export, MigrationSink interface
   cli/              artinstack-migrate
@@ -56,6 +57,8 @@ artinstack-migrate validate <platform> <export-file>
 | `--dry-run` | Parse and analyze only; no export files |
 | `--report <dir>` | With `--dry-run`, write `conflicts.json` and `migration-report.json` |
 | `--offline` | Skip network HEAD requests for asset size estimates |
+| `--rewrite-gateway <url>` | WordPress: legacy API-gateway base (use with `--rewrite-public`) |
+| `--rewrite-public <url>` | WordPress: public origin for `/wp-content/` asset paths |
 | `--sink filesystem` | Run through `MigrationSink` before writing (requires `--out`) |
 | `--urls <file>` | Wix only: URL list or `sitemap.xml` for static page snapshots |
 
@@ -68,6 +71,12 @@ artinstack-migrate wordpress export.xml --out ./output
 # Preview conflicts without writing content
 artinstack-migrate wordpress export.xml --dry-run --report ./preview/
 
+# WordPress: rewrite legacy gateway URLs before dry-run / export (e.g. API Gateway → public CDN)
+artinstack-migrate wordpress export.xml \
+  --rewrite-gateway "https://gateway.example/prod" \
+  --rewrite-public "https://www.example.com" \
+  --dry-run --report ./preview/
+
 # Validate export structure (JSON result on stdout, exit 0/1)
 artinstack-migrate validate wordpress export.xml
 
@@ -119,8 +128,10 @@ pnpm dev          # watch build
 | Piece | `@artinstack/migrator` | Host application |
 |-------|------------------------|------------------|
 | Parsers + normalizer DTOs | Yes | No |
+| WordPress builder flattening + origin URL rewrite (pre-DTO) | Yes | Optional same config on adapter input |
 | CLI + filesystem JSON export | Yes | No |
 | `MigrationSink` interface | Yes | Implementation |
+| Dynamic shortcodes (`[portfolio]`, `[recent_posts]`), forms, sanitize | No | Yes |
 | Jobs, worker, credentials, UI | No | Yes |
 
 ## License

package/dist/{chunk-QEXTXHFG.js → chunk-Q6M5JEL3.js}

@@ -1,11 +1,12 @@
 import {
   SquarespaceCollectionClient,
+  WORDPRESS_BUILDER_REGISTRY,
   enumerateSquarespaceEntities,
   linkToPath,
   sanitizeSlug,
   summarizeSquarespaceExport,
   validateSquarespaceExportFile
-} from "./chunk-HH7666MQ.js";
+} from "./chunk-XKWWXKP3.js";
 import {
   discoverContentAssetUrls,
   normalizeAssetUrl
@@ -43,45 +44,52 @@ import { readFile } from "fs/promises";
 import { basename } from "path";
 import { XMLParser } from "fast-xml-parser";
 
-// src/parsers/wordpress/builders/registry.ts
-var WORDPRESS_BUILDER_REGISTRY = [
-  {
-    id: "tatsu",
-    detect: /\[(?:\/)?tatsu_/i,
-    contentRules: [
-      { shortcodePrefix: "tatsu_image", urlParams: ["image", "url", "src"], tag: "img" },
-      { shortcodePrefix: "tatsu_video", urlParams: ["video", "src", "url"], tag: "video" }
-    ],
-    scaffoldingPrefix: "tatsu_"
-  },
-  {
-    id: "divi",
-    detect: /\[(?:\/)?et_pb_/i,
-    contentRules: [{ shortcodePrefix: "et_pb_image", urlParams: ["src", "url"], tag: "img" }],
-    scaffoldingPrefix: "et_pb_"
-  },
-  {
-    id: "elementor",
-    detect: /\[(?:\/)?elementor[-_]/i,
-    contentRules: [
-      { shortcodePrefix: "elementor-widget", urlParams: ["url", "src", "image"], tag: "img" }
-    ],
-    scaffoldingPrefix: "elementor_"
-  }
-];
-
 // src/parsers/wordpress/builders/flatten.ts
 function escapeRegExp(value) {
   return value.replace(/[.*+?^${}()|[\]\\]/g, "\\$&");
 }
+function extractQuotedParam(params, name) {
+  const pattern = new RegExp(`\\b${escapeRegExp(name)}\\s*=\\s*`, "i");
+  const match = pattern.exec(params);
+  if (!match) return void 0;
+  let index = match.index + match[0].length;
+  while (index < params.length && /\s/.test(params[index])) index += 1;
+  const quote = params[index];
+  if (quote !== '"' && quote !== "'") return void 0;
+  index += 1;
+  let value = "";
+  while (index < params.length) {
+    const char = params[index];
+    if (char === "\\" && index + 1 < params.length) {
+      value += params[index + 1];
+      index += 2;
+      continue;
+    }
+    if (char === quote) break;
+    value += char;
+    index += 1;
+  }
+  const trimmed = value.trim();
+  return trimmed || void 0;
+}
 function extractShortcodeParam(params, names) {
   for (const name of names) {
-    const pattern = new RegExp(`\\b${escapeRegExp(name)}\\s*=\\s*["']([^"']+)["']`, "i");
-    const match = params.match(pattern);
-    if (match?.[1]?.trim()) return match[1].trim();
+    const value = extractQuotedParam(params, name);
+    if (value) return value;
   }
   return void 0;
 }
+function escapeHtmlText(text) {
+  return text.replace(/&/g, "&amp;").replace(/</g, "&lt;").replace(/>/g, "&gt;").replace(/"/g, "&quot;");
+}
+function textToHtml(text, tag) {
+  const paragraphs = text.split(/\n{2,}/).map((part) => part.trim()).filter(Boolean);
+  if (paragraphs.length === 0) return "";
+  return paragraphs.map((paragraph) => {
+    const inner = escapeHtmlText(paragraph).replace(/\n/g, "<br />");
+    return `<${tag}>${inner}</${tag}>`;
+  }).join("\n");
+}
 function emitHtmlTag(tag, url) {
   const normalized = normalizeAssetUrl(url) ?? url;
   const escaped = normalized.replace(/&/g, "&amp;").replace(/"/g, "&quot;").replace(/</g, "&lt;");
@@ -94,10 +102,10 @@ function emitHtmlTag(tag, url) {
       return `<iframe src="${escaped}" loading="lazy"></iframe>`;
   }
 }
-function convertContentBlocker(content, rule) {
+function convertUrlRule(content, rule) {
   const prefix = escapeRegExp(rule.shortcodePrefix);
   const pattern = new RegExp(
-    `\\[${prefix}([^\\]]*)\\]\\s*(?:\\[\\/${prefix}[^\\]]*\\])?`,
+    `\\[${prefix}\\b([^\\]]*)\\]\\s*(?:\\[\\/${prefix}\\b[^\\]]*\\])?`,
     "gi"
   );
   return content.replace(pattern, (block, params) => {
@@ -106,12 +114,63 @@ function convertContentBlocker(content, rule) {
     return emitHtmlTag(rule.tag, url);
   });
 }
-function stripScaffolding(content, prefix) {
+function convertTextRule(content, rule) {
+  const prefix = escapeRegExp(rule.shortcodePrefix);
+  const pattern = new RegExp(
+    `\\[${prefix}\\b([^\\]]*)\\]\\s*(?:\\[\\/${prefix}\\b[^\\]]*\\])?`,
+    "gis"
+  );
+  return content.replace(pattern, (block, params) => {
+    const parts = [];
+    for (const field of rule.fields) {
+      const text = extractQuotedParam(params, field.param);
+      if (!text) continue;
+      const html = textToHtml(text, field.tag);
+      if (html) parts.push(html);
+    }
+    return parts.length > 0 ? parts.join("\n") : block;
+  });
+}
+function convertWrapperRule(content, rule) {
+  const prefix = escapeRegExp(rule.shortcodePrefix);
+  const pattern = new RegExp(
+    `\\[${prefix}\\b([^\\]]*)\\]([\\s\\S]*?)\\[\\/${prefix}\\b[^\\]]*\\]`,
+    "gi"
+  );
+  return content.replace(pattern, (_, params, inner) => {
+    const parts = [];
+    if (rule.urlParams?.length) {
+      const url = extractShortcodeParam(params, rule.urlParams);
+      if (url) parts.push(emitHtmlTag("img", url));
+    }
+    parts.push(inner.trim());
+    return parts.filter(Boolean).join("\n");
+  });
+}
+function convertPlaceholderRule(content, rule) {
+  const prefix = escapeRegExp(rule.shortcodePrefix);
+  const pattern = new RegExp(
+    `\\[${prefix}\\b([^\\]]*)\\]\\s*(?:\\[\\/${prefix}\\b[^\\]]*\\])?`,
+    "gi"
+  );
+  return content.replace(pattern, rule.html);
+}
+function stripScaffoldingPrefix(content, prefix) {
   const escaped = escapeRegExp(prefix);
   const opener = new RegExp(`\\[${escaped}[a-z0-9_-]*[^\\]]*\\]`, "gi");
   const closer = new RegExp(`\\[\\/${escaped}[a-z0-9_-]*[^\\]]*\\]`, "gi");
   return content.replace(opener, "").replace(closer, "");
 }
+function stripLegacyTokens(content, tokens) {
+  let result = content;
+  for (const token of tokens) {
+    const escaped = escapeRegExp(token);
+    const opener = new RegExp(`\\[${escaped}\\b[^\\]]*\\]`, "gi");
+    const closer = new RegExp(`\\[\\/${escaped}\\b[^\\]]*\\]`, "gi");
+    result = result.replace(opener, "").replace(closer, "");
+  }
+  return result;
+}
 function detectThemes(content, registry) {
   return registry.filter((theme) => theme.detect.test(content));
 }
@@ -126,10 +185,24 @@ function flattenWordPressBuilders(content, options = {}) {
   }
   let html = content;
   for (const theme of themes) {
-    for (const rule of theme.contentRules) {
-      html = convertContentBlocker(html, rule);
+    for (const rule of theme.wrapperRules ?? []) {
+      html = convertWrapperRule(html, rule);
+    }
+    for (const rule of theme.textRules ?? []) {
+      html = convertTextRule(html, rule);
+    }
+    for (const rule of theme.urlRules ?? []) {
+      html = convertUrlRule(html, rule);
+    }
+    for (const rule of theme.placeholderRules ?? []) {
+      html = convertPlaceholderRule(html, rule);
+    }
+    for (const prefix of theme.scaffoldingPrefixes ?? []) {
+      html = stripScaffoldingPrefix(html, prefix);
+    }
+    if (theme.legacyScaffoldingTokens?.length) {
+      html = stripLegacyTokens(html, theme.legacyScaffoldingTokens);
     }
-    html = stripScaffolding(html, theme.scaffoldingPrefix);
   }
   html = html.replace(/\n{3,}/g, "\n\n").trim();
   return {
@@ -140,6 +213,14 @@ function flattenWordPressBuilders(content, options = {}) {
 
 // src/parsers/wordpress/parse-wxr.ts
 var PLATFORM = "wordpress";
+var WOOCOMMERCE_STUB_PAGE_SLUGS = /* @__PURE__ */ new Set(["cart", "checkout", "my-account"]);
+var WOOCOMMERCE_STUB_SHORTCODE = /^\[woocommerce_(?:cart|checkout|my_account)\]\s*$/i;
+function isWooCommerceStubPage(slug, contentHtml) {
+  if (WOOCOMMERCE_STUB_PAGE_SLUGS.has(slug)) return true;
+  const trimmed = contentHtml.trim();
+  if (!trimmed) return false;
+  return WOOCOMMERCE_STUB_SHORTCODE.test(trimmed);
+}
 function asArray(value) {
   if (value === void 0) return [];
   return Array.isArray(value) ? value : [value];
@@ -316,6 +397,11 @@ function resolveFeaturedAssetSourceId(thumbnailId, attachmentIndex, contentHtml)
   const firstInline = discoverContentAssetUrls(contentHtml)[0];
   return firstInline ? `url:${firstInline}` : void 0;
 }
+function maybeRewriteUrl(url, config) {
+  if (!url) return void 0;
+  if (!config) return url;
+  return rewriteOriginUrlsInText(url, config);
+}
 async function* enumerateWxrEntities(options) {
   const xml = await readFile(options.filePath, "utf8");
   const items = parseItems(xml);
@@ -346,9 +432,12 @@ async function* enumerateWxrEntities(options) {
     const postType = textValue(item.post_type);
     if (postType !== "post" && postType !== "page") continue;
     const id = textValue(item.post_id);
-    const link = textValue(item.link);
+    const link = maybeRewriteUrl(textValue(item.link), options.originUrlRewrite);
     const slug = sanitizeSlug(textValue(item.post_name) || textValue(item.title) || id);
     const contentHtml = preprocessContent(getContentEncoded(item), options);
+    if (postType === "page" && options.skipWooCommerceStubPages !== false && isWooCommerceStubPage(slug, contentHtml)) {
+      continue;
+    }
     for (const asset of collectInlineAssets(
       contentHtml,
       attachmentIndex,
@@ -446,10 +535,13 @@ function resolveWxrOptions(input) {
     return {
       filePath: String(obj.path),
       originUrlRewrite: obj.originUrlRewrite,
-      flattenBuilders: obj.flattenBuilders
+      flattenBuilders: obj.flattenBuilders,
+      skipWooCommerceStubPages: obj.skipWooCommerceStubPages
     };
   }
-  throw new Error("WordPress adapter requires input path (string or { path, originUrlRewrite?, flattenBuilders? })");
+  throw new Error(
+    "WordPress adapter requires input path (string or { path, originUrlRewrite?, flattenBuilders?, skipWooCommerceStubPages? })"
+  );
 }
 var wordpressAdapter = {
   platform: "wordpress",
@@ -2441,4 +2533,4 @@ export {
   wixAdapter,
   getAdapter
 };
-//# sourceMappingURL=chunk-QEXTXHFG.js.map
+//# sourceMappingURL=chunk-Q6M5JEL3.js.map