feat: allow to transform and report back (#59)

adobe · Jan 5, 2023 · bfeea5d · bfeea5d
1 parent e2d6e50
commit bfeea5d
Show file tree

Hide file tree

Showing 2 changed files with 68 additions and 5 deletions.
diff --git a/src/importer/HTML2x.js b/src/importer/HTML2x.js
@@ -116,10 +116,15 @@ async function html2x(
         results.forEach((result) => {
           const name = path.basename(result.path);
           const dirname = path.dirname(result.path);
-
-          const pir = new PageImporterResource(name, dirname, result.element, null, {
+          const extra = {
             html: result.element.outerHTML,
-          });
+          };
+
+          if (result.report) {
+            extra.report = result.report;
+          }
+
+          const pir = new PageImporterResource(name, dirname, result.element, null, extra);
           pirs.push(pir);
         });
         return pirs;
@@ -181,6 +186,10 @@ async function html2x(
       html: pir.extra.html,
     };
 
+    if (pir.extra.report) {
+      res.report = pir.extra.report;
+    }
+
     res.path = path.resolve(pir.directory, pir.name);
 
     if (config.toMd) {

diff --git a/test/importers/HTML2x.spec.js b/test/importers/HTML2x.spec.js
@@ -10,7 +10,7 @@
  * governing permissions and limitations under the License.
  */
 
-import { ok, strictEqual } from 'assert';
+import { deepStrictEqual, ok, strictEqual } from 'assert';
 import { describe, it } from 'mocha';
 import { JSDOM } from 'jsdom';
 import { docx2md } from '@adobe/helix-docx2md';
@@ -155,7 +155,7 @@ describe('html2md tests', () => {
     strictEqual(out2.path, '/folder/my-custom-path-p2');
   });
 
-  it('html2md handles multiple transform', async () => {
+  it('html2md handles multiple transform (but single output)', async () => {
     const out = await html2md('https://www.sample.com/page.html', '<html><body><h1>Hello World</h1></body></html>', {
       transform: ({ document }) => {
         const p1 = document.createElement('p');
@@ -176,6 +176,60 @@ describe('html2md tests', () => {
     strictEqual(out.path, '/my-custom-path-p1');
   });
 
+  it('html2md allows to report when using transform', async () => {
+    const out = await html2md('https://www.sample.com/page.html', '<html><body><h1>Hello World</h1></body></html>', {
+      transform: ({ document }) => {
+        const p1 = document.createElement('p');
+        p1.innerHTML = 'My Hello to the World 1';
+
+        const p2 = document.createElement('p');
+        p2.innerHTML = 'My Hello to the World 2';
+
+        return [{
+          element: p1,
+          path: '/my-custom-path-p1',
+          report: {
+            custom: 'A custom property',
+            customArray: ['a', 'b', 'c'],
+            customObject: {
+              a: 1,
+              b: true,
+              c: {
+                d: 'e',
+              },
+            },
+          },
+        }, {
+          element: p2,
+          path: '/folder/my-custom-path-p2',
+          report: {
+            custom: 'Another value',
+            customArray: ['a', 'b', 'c'],
+            somethingElse: 'something else',
+          },
+        }];
+      },
+    });
+
+    const out1 = out[0];
+    strictEqual(out1.html.trim(), '<p>My Hello to the World 1</p>');
+    strictEqual(out1.md.trim(), 'My Hello to the World 1');
+    strictEqual(out1.path, '/my-custom-path-p1');
+    ok(out1.report);
+    strictEqual(out1.report.custom, 'A custom property');
+    deepStrictEqual(out1.report.customArray, ['a', 'b', 'c']);
+    deepStrictEqual(out1.report.customObject, { a: 1, b: true, c: { d: 'e' } });
+
+    const out2 = out[1];
+    strictEqual(out2.html.trim(), '<p>My Hello to the World 2</p>');
+    strictEqual(out2.md.trim(), 'My Hello to the World 2');
+    strictEqual(out2.path, '/folder/my-custom-path-p2');
+    ok(out2.report);
+    strictEqual(out2.report.custom, 'Another value');
+    deepStrictEqual(out2.report.customArray, ['a', 'b', 'c']);
+    strictEqual(out2.report.somethingElse, 'something else');
+  });
+
   it('html2md does not crash if transform returns null', async () => {
     const out = await html2md('https://www.sample.com/page.html', '<html><body><h1>Hello World</h1></body></html>', {
       transform: () => null,