CMG-824 | Delta Migration Support for Wordpress CMS

api/package.json

@@ -51,7 +51,7 @@
     "html-to-json-parser": "^2.0.1",
     "jsdom": "^24.1.0",
     "jsonwebtoken": "^9.0.3",
-    "lodash": "^4.17.21",
+    "lodash": "^4.18.1",
     "lowdb": "^7.0.1",
     "mkdirp": "^3.0.1",
     "mysql2": "^3.16.2",

api/src/services/wordpress.service.ts

@@ -518,6 +518,11 @@ async function saveEntry(fields: any, entry: any,  file_path: string, assetData
   const $ = cheerio.load(xmlData, { xmlMode: true });
   const items = $('item');
   const entryData: Record<string, any> = {};
+  const fieldList = Array.isArray(fields) ? fields : [];
+  const hasField = (uid: string) =>
+    fieldList.some(
+      (field: any) => field?.contentstackFieldUid === uid || field?.uid === uid
+    );
 
   try {
     if(entry ){
@@ -584,16 +589,20 @@ async function saveEntry(fields: any, entry: any,  file_path: string, assetData
           // Pass individual content to createSchema
           entryData[uid] = await createSchema(fields, blocksJson, item?.title, uid, assetData, duplicateBlockMappings);
           const categoryReference = extractCategoryReference(item?.['category']);
-          if (categoryReference?.length > 0) {
-            entryData[uid]['taxonomies'] = taxonomies;
+          if (hasField("taxonomies") && categoryReference?.length > 0) {
+            entryData[uid]["taxonomies"] = taxonomies;
           }
           const termsReference = extractTermsReference(item?.['category']);
-          if(termsReference?.length > 0) {
-            entryData[uid]['terms'] = terms;
+          if (hasField("terms") && termsReference?.length > 0) {
+            entryData[uid]["terms"] = terms;
+          }
+          entryData[uid]["tags"] = tags?.map((tag: any) => tag?.text) || [];
+          if (hasField("author")) {
+            entryData[uid]["author"] =
+              authorData?.filter((author: any) => author?.uid) || [];
           }
-          entryData[uid]['tags'] = tags?.map((tag: any) => tag?.text);
-          entryData[uid]['author'] = authorData;
           entryData[uid]['locale'] = locale;
+          entryData[uid]["publish_details"] = [];
           entryData[uid]['publish_details'] = [];
 
 

upload-api/migration-wordpress/index.js

@@ -3,9 +3,11 @@ var __importDefault = (this && this.__importDefault) || function (mod) {
     return (mod && mod.__esModule) ? mod : { "default": mod };
 };
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.extractLocale = exports.extractContentTypes = void 0;
+exports.extractEntries = exports.extractLocale = exports.extractContentTypes = void 0;
 const contentTypes_1 = __importDefault(require("./libs/contentTypes"));
 exports.extractContentTypes = contentTypes_1.default;
 //import contentTypeMaker from './libs/contentTypeMapper';
 const extractLocale_1 = __importDefault(require("./libs/extractLocale"));
 exports.extractLocale = extractLocale_1.default;
+const extractEntries_1 = __importDefault(require('./libs/extractEntries'));
+exports.extractEntries = extractEntries_1.default;

upload-api/migration-wordpress/index.ts

@@ -1,9 +1,11 @@
 import extractContentTypes from './libs/contentTypes';
 //import contentTypeMaker from './libs/contentTypeMapper';
 import extractLocale from './libs/extractLocale';
+import extractEntries from './libs/extractEntries';
 
 export {
-    extractContentTypes,
-    //contentTypeMaker,
-    extractLocale
-}
+  extractContentTypes,
+  //contentTypeMaker,
+  extractLocale,
+  extractEntries
+};

upload-api/migration-wordpress/libs/extractEntries.js

@@ -0,0 +1,117 @@
+"use strict";
+const fs = require("fs");
+const path = require("path");
+const config = require("../config/index.json");
+
+const { contentTypes: contentTypesConfig } = config?.modules ?? {};
+const contentTypeFolderPath = path.resolve(config?.data, contentTypesConfig?.dirName);
+const EXCLUDED_POST_TYPES = new Set(['attachment', 'wp_global_styles', 'wp_navigation']);
+
+const normalizeArray = (value) => {
+  if (!value) return [];
+  return Array.isArray(value) ? value : [value];
+};
+
+const idCorrector = (id) => {
+  const normalized = id?.replace(/[-{}]/g, '');
+  return normalized ? normalized.toLowerCase() : id;
+};
+
+const getEntryName = (item) => {
+  if (typeof item?.title === 'string' && item.title.trim()) return item.title.trim();
+  if (item?.title?.text) return String(item.title.text).trim();
+  if (typeof item?.['wp:post_name'] === 'string' && item['wp:post_name'].trim()) {
+    return item['wp:post_name'].trim();
+  }
+  return 'Untitled Entry';
+};
+
+/** Align with api wordpress.service entry uid: idCorrector(`posts_${wp:post_id}`). */
+const getSourceEntryUid = (item) => {
+  const postId = item?.['wp:post_id'];
+  if (postId != null && String(postId).trim() !== '') {
+    return idCorrector(`posts_${postId}`);
+  }
+  const candidate =
+    item?.guid?.text ?? item?.guid ?? item?.link ?? getEntryName(item);
+  return idCorrector(String(candidate || ''));
+};
+
+const getEntryLanguage = (item, channelLanguage) => {
+  const postMeta = normalizeArray(item?.['wp:postmeta']);
+  const languageMeta = postMeta.find((meta) => {
+    const key = String(meta?.['wp:meta_key'] || '').toLowerCase();
+    return key === 'language' || key === '_language' || key === 'locale' || key === '_locale';
+  });
+
+  const metaLanguage = languageMeta?.['wp:meta_value'];
+  if (typeof metaLanguage === 'string' && metaLanguage.trim()) {
+    return metaLanguage.trim();
+  }
+
+  if (typeof channelLanguage === 'string' && channelLanguage.trim()) {
+    return channelLanguage.trim();
+  }
+
+  return 'en-us';
+};
+
+const extractEntries = async (filePath, contentTypeData = []) => {
+  try {
+    const rawData = await fs.promises.readFile(filePath, 'utf8');
+    const jsonData = JSON.parse(rawData);
+    const items = normalizeArray(jsonData?.rss?.channel?.item);
+    const channelLanguage = jsonData?.rss?.channel?.language;
+
+    const groupedByType = items?.reduce((acc, item) => {
+      const postType = item?.['wp:post_type'] || 'unknown';
+      if (EXCLUDED_POST_TYPES.has(postType)) return acc;
+      if (!acc[postType]) acc[postType] = [];
+      acc[postType].push(item);
+      return acc;
+    }, {});
+
+    const updatedTypes = contentTypeData.map((ct) => ({ ...ct }));
+
+    for (const [type, entries] of Object.entries(groupedByType)) {
+      const entryMapping = normalizeArray(entries)
+        .map((item) => {
+          const otherCmsEntryUid = getSourceEntryUid(item);
+          if (!otherCmsEntryUid) return null;
+          return {
+            contentTypeUid: type,
+            entryName: getEntryName(item),
+            otherCmsEntryUid: `posts_${otherCmsEntryUid}`,
+            otherCmsCTName: type,
+            language: getEntryLanguage(item, channelLanguage),
+            isUpdate: false
+          };
+        })
+        .filter(Boolean);
+
+      const contentTypeFilePath = path.join(contentTypeFolderPath, `${type.toLowerCase()}.json`);
+      if (fs.existsSync(contentTypeFilePath)) {
+        const ctFile = JSON.parse(await fs.promises.readFile(contentTypeFilePath, 'utf8'));
+        ctFile.entryMapping = entryMapping;
+        await fs.promises.writeFile(contentTypeFilePath, JSON.stringify(ctFile, null, 4), 'utf8');
+      }
+
+      const index = updatedTypes.findIndex(
+        (ct) =>
+          ct?.otherCmsUid?.toLowerCase?.() === type.toLowerCase() ||
+          ct?.contentstackUid?.toLowerCase?.() === type.toLowerCase()
+      );
+      if (index >= 0) {
+        updatedTypes[index] = { ...updatedTypes[index], entryMapping };
+      }
+    }
+
+    return updatedTypes;
+  } catch (error) {
+    console.error('Error while extracting WordPress entries:', error?.message || error);
+    return contentTypeData;
+  }
+};
+
+module.exports = extractEntries;
+module.exports.default = extractEntries;

upload-api/migration-wordpress/libs/extractEntries.ts

@@ -0,0 +1,126 @@
+import fs from 'fs';
+import path from 'path';
+import config from '../config/index.json';
+
+const { contentTypes: contentTypesConfig } = config?.modules;
+const contentTypeFolderPath = path.resolve(config?.data, contentTypesConfig?.dirName);
+
+const EXCLUDED_POST_TYPES = new Set(['attachment', 'wp_global_styles', 'wp_navigation']);
+
+const normalizeArray = <T>(value: T | T[] | undefined): T[] => {
+  if (!value) return [];
+  return Array.isArray(value) ? value : [value];
+};
+
+const idCorrector = (id: string) => {
+  const normalized = id?.replace(/[-{}]/g, '');
+  return normalized ? normalized.toLowerCase() : id;
+};
+
+const getEntryName = (item: any): string => {
+  if (typeof item?.title === 'string' && item.title.trim()) {
+    return item.title.trim();
+  }
+  if (item?.title?.text) {
+    return String(item.title.text).trim();
+  }
+  if (typeof item?.['wp:post_name'] === 'string' && item['wp:post_name'].trim()) {
+    return item['wp:post_name'].trim();
+  }
+  return 'Untitled Entry';
+};
+
+/**
+ * Must match api WordPress entry keys: idCorrector(`posts_${wp:post_id}`) in wordpress.service.ts.
+ * Raw post id only caused otherCmsEntryUid to diverge from uid-map / entry JSON keys → entry mapper showed "-".
+ */
+const getSourceEntryUid = (item: any): string => {
+  const postId = item?.['wp:post_id'];
+  if (postId != null && String(postId).trim() !== '') {
+    return idCorrector(`posts_${postId}`);
+  }
+  const candidate =
+    item?.guid?.text ?? item?.guid ?? item?.link ?? getEntryName(item);
+  return idCorrector(String(candidate || ''));
+};
+
+const getEntryLanguage = (item: any, channelLanguage?: string): string => {
+  const postMeta = normalizeArray(item?.['wp:postmeta']);
+  const languageMeta = postMeta.find((meta: any) => {
+    const key = String(meta?.['wp:meta_key'] || '').toLowerCase();
+    return key === 'language' || key === '_language' || key === 'locale' || key === '_locale';
+  });
+
+  const metaLanguage = languageMeta?.['wp:meta_value'];
+  if (typeof metaLanguage === 'string' && metaLanguage.trim()) {
+    return metaLanguage.trim();
+  }
+
+  if (typeof channelLanguage === 'string' && channelLanguage.trim()) {
+    return channelLanguage.trim();
+  }
+
+  return 'en-us';
+};
+
+const extractEntries = async (filePath: string, contentTypeData: any[] = []) => {
+  try {
+    const rawData = await fs.promises.readFile(filePath, 'utf8');
+    const jsonData = JSON.parse(rawData);
+    const items = normalizeArray(jsonData?.rss?.channel?.item);
+    const channelLanguage = jsonData?.rss?.channel?.language;
+
+    const groupedByType = items?.reduce((acc: Record<string, any[]>, item: any) => {
+      const postType = item?.['wp:post_type'] || 'unknown';
+      if (EXCLUDED_POST_TYPES.has(postType)) return acc;
+      if (!acc[postType]) acc[postType] = [];
+      acc[postType].push(item);
+      return acc;
+    }, {});
+
+    const updatedTypes = contentTypeData?.map((ct) => ({ ...ct }));
+
+    for (const [type, entries] of Object.entries(groupedByType)) {
+      const entryMapping = normalizeArray(entries)
+        .map((item: any) => {
+          const otherCmsEntryUid = getSourceEntryUid(item);
+          if (!otherCmsEntryUid) return null;
+          return {
+            contentTypeUid: type,
+            entryName: getEntryName(item),
+            otherCmsEntryUid: `posts_${otherCmsEntryUid}`,
+            otherCmsCTName: type,
+            language: getEntryLanguage(item, channelLanguage),
+            isUpdate: false
+          };
+        })
+        .filter(Boolean);
+
+      const contentTypeFilePath = path.join(contentTypeFolderPath, `${type.toLowerCase()}.json`);
+      if (fs.existsSync(contentTypeFilePath)) {
+        const ctFile = JSON.parse(await fs.promises.readFile(contentTypeFilePath, 'utf8'));
+        ctFile.entryMapping = entryMapping;
+        await fs.promises.writeFile(contentTypeFilePath, JSON.stringify(ctFile, null, 4), 'utf8');
+      }
+
+      const index = updatedTypes.findIndex(
+        (ct: any) =>
+          ct?.otherCmsUid?.toLowerCase?.() === type.toLowerCase() ||
+          ct?.contentstackUid?.toLowerCase?.() === type.toLowerCase()
+      );
+      if (index >= 0) {
+        updatedTypes[index] = {
+          ...updatedTypes[index],
+          entryMapping
+        };
+      }
+    }
+
+    return updatedTypes;
+  } catch (error: any) {
+    console.error('Error while extracting WordPress entries:', error?.message || error);
+    return contentTypeData;
+  }
+};
+
+export default extractEntries;

upload-api/migration-wordpress/package.json

@@ -1,5 +1,5 @@
 {
-    "name": "migration-v2-wordpress",
+    "name": "migration-wordpress",
     "version": "1.0.0",
     "description": "",
     "main": "build/index.js",

upload-api/migration-wordpress/utils/parseUtil.js

@@ -88,10 +88,23 @@ const setupWordPressBlocks = (rawContent) => __awaiter(void 0, void 0, void 0, f
     // Now import WordPress packages after setting up globals
     const wpBlocks = yield Promise.resolve().then(() => __importStar(require('@wordpress/blocks')));
     const { parse } = yield Promise.resolve().then(() => __importStar(require('@wordpress/blocks')));
-    const { registerCoreBlocks } = yield Promise.resolve().then(() => __importStar(require('@wordpress/block-library')));
     (_a = wpBlocks.__unstableSetDebugLevel) === null || _a === void 0 ? void 0 : _a.call(wpBlocks, 'none');
-    registerCoreBlocks();
-    const blocks = parse(rawContent);
-    return blocks;
+    try {
+        const { registerCoreBlocks } = yield Promise.resolve().then(() => __importStar(require('@wordpress/block-library')));
+        registerCoreBlocks();
+    }
+    catch (error) {
+        // Some environments cannot fully initialize Gutenberg internals (e.g. getSettings).
+        // Continue with parser-only mode so mapper generation does not fail.
+        console.warn('WordPress core blocks registration failed, using parser-only mode:', (error === null || error === void 0 ? void 0 : error.message) || error);
+    }
+    try {
+        const blocks = parse(rawContent);
+        return Array.isArray(blocks) ? blocks : [];
+    }
+    catch (error) {
+        console.warn('WordPress block parsing failed, returning empty blocks:', (error === null || error === void 0 ? void 0 : error.message) || error);
+        return [];
+    }
 });
 exports.setupWordPressBlocks = setupWordPressBlocks;