Introduce md2jcr to the helix importer

package.json

@@ -27,6 +27,7 @@
   },
   "devDependencies": {
     "@adobe/eslint-config-helix": "2.0.8",
+    "@adobe/helix-md2jcr": "0.0.1",
     "@adobe/helix-docx2md": "1.6.11",
     "@adobe/helix-mediahandler": "2.5.40",
     "@esm-bundle/chai": "4.3.4-fix.0",

src/importer/HTML2x.js

@@ -41,7 +41,7 @@ async function html2x(
   url,
   doc,
   transformCfg,
-  config = { toMd: true, toDocx: false },
+  config = { toMd: true, toDocx: false, toJcr: false },
   params = {},
 ) {
   const transformer = transformCfg || {};
@@ -156,12 +156,14 @@ async function html2x(
     storageHandler,
     skipDocxConversion: !config.toDocx,
     skipMDFileCreation: !config.toMd,
+    skipJcrFileCreation: !config.toJcr,
     logger,
     mdast2docxOptions: {
       stylesXML: config.docxStylesXML,
       image2png: config.image2png,
     },
     createDocumentFromString: config.createDocumentFromString,
+    components: params.components,
   });
 
   const pirs = await importer.import(url);
@@ -191,6 +193,10 @@ async function html2x(
       const docx = await storageHandler.get(pir.docx);
       res.docx = docx;
     }
+    if (config.toJcr && pir.jcr) {
+      const jcr = await storageHandler.get(pir.jcr);
+      res.jcr = jcr;
+    }
     return res;
   };
 
@@ -247,7 +253,18 @@ async function html2docx(url, document, transformCfg, config, params = {}) {
   return html2x(url, doc, transformCfg, { ...config, toMd: true, toDocx: true }, params);
 }
 
+async function md2jcr(url, document, transformCfg, config, params = {}) {
+  let doc = document;
+  if (typeof doc === 'string') {
+    doc = parseStringDocument(document, config);
+  }
+  return html2x(url, doc, transformCfg, {
+    ...config, toMd: true, toDocx: false, toJcr: true,
+  }, params);
+}
+
 export {
+  md2jcr,
   html2md,
   html2docx,
   defaultTransformDOM,

src/importer/PageImporter.js

@@ -19,6 +19,7 @@ import { defaultHandlers, toMdast } from 'hast-util-to-mdast';
 import stringify from 'remark-stringify';
 import fs from 'fs-extra';
 import { md2docx } from '@adobe/helix-md2docx';
+import { md2jcr } from '@adobe/helix-md2jcr';
 import remarkGridTable from '@adobe/remark-gridtables';
 import {
   imageReferences,
@@ -73,6 +74,11 @@ export default class PageImporter {
     return this.params.storageHandler.put(docxPath, buffer);
   }
 
+  async convertToJcr(jcrPath, content) {
+    const buffer = await md2jcr(content, this.params.components);
+    return this.params.storageHandler.put(jcrPath, buffer);
+  }
+
   async createMarkdown(resource, url) {
     const { name } = resource;
     const { directory } = resource;
@@ -358,6 +364,13 @@ export default class PageImporter {
               // eslint-disable-next-line no-param-reassign
               entry.docx = docxPath;
             }
+
+            if (!this.params.skipJcrFileCreation) {
+              const jcrPath = `${res.path}.xml`;
+              await this.convertToJcr(jcrPath, res.content);
+              // eslint-disable-next-line no-param-reassign
+              entry.jcr = jcrPath;
+            }
           }
 
           results.push(entry);

src/index.js

@@ -25,7 +25,7 @@ import Utils from './utils/Utils.js';
 
 import WPUtils from './wp/WPUtils.js';
 
-import { html2md, html2docx } from './importer/HTML2x.js';
+import { html2md, html2docx, md2jcr } from './importer/HTML2x.js';
 
 import createMetadata from './importer/defaults/rules/createMetadata.js';
 import adjustImageUrls from './importer/defaults/rules/adjustImageUrls.js';
@@ -54,5 +54,6 @@ export {
   WPUtils,
   html2md,
   html2docx,
+  md2jcr,
   rules,
 };