fix: Handle nested table with sibling elements

plone · Mar 16, 2023 · bc6a215 · bc6a215
1 parent 018a8d3
commit bc6a215
Show file tree

Hide file tree

Showing 2 changed files with 62 additions and 33 deletions.
diff --git a/src/converters/fromHtml.js b/src/converters/fromHtml.js
@@ -94,40 +94,26 @@ const isInline = (n) =>
 
 const extractElementsWithConverters = (el, defaultTextBlock, href) => {
   const result = [];
-  if (el.tagName === 'TABLE') {
-    for (const child of el.childNodes) {
-      const tmpResult = extractElementsWithConverters(
-        child,
-        defaultTextBlock,
-        href,
-      );
-      if (tmpResult.length > 0) {
-        result.push(...tmpResult);
-      }
-    }
-    result.push(blockFromElement(el, defaultTextBlock, href));
-  } else if (elementsWithConverters.includes(el.tagName)) {
-    result.push(blockFromElement(el, defaultTextBlock, href));
-  } else {
-    const children = el.childNodes;
-    if (el.tagName === 'A') {
-      href = el.getAttribute('href');
+  if (el.tagName === 'A') {
+    href = el.getAttribute('href');
+  }
+  // First, traverse all childNodes
+  for (const child of el.childNodes) {
+    const tmpResult = extractElementsWithConverters(
+      child,
+      defaultTextBlock,
+      href,
+    );
+    if (tmpResult.length > 0) {
+      result.push(...tmpResult);
     }
-    for (const child of children) {
-      if (elementsWithConverters.includes(child.tagName)) {
-        el.removeChild(child);
-        result.push(blockFromElement(child, defaultTextBlock, href));
-      } else {
-        const tmpResult = extractElementsWithConverters(
-          child,
-          defaultTextBlock,
-          href,
-        );
-        if (tmpResult.length > 0) {
-          result.push(...tmpResult);
-        }
-      }
+  }
+  if (elementsWithConverters.includes(el.tagName)) {
+    const parent = el.parentElement;
+    if (parent) {
+      parent.removeChild(el);
     }
+    result.push(blockFromElement(el, defaultTextBlock, href));
   }
 
   return result;
@@ -165,7 +151,6 @@ const convertFromHTML = (input, defaultTextBlock) => {
     for (const child of children) {
       // With children nodes, we keep the wrapper only
       // if at least one child is not  in elementsWithConverters
-      keepWrapper = keepWrapper || false;
       const tmpResult = extractElementsWithConverters(
         child,
         defaultTextBlock,

diff --git a/src/converters/fromHtml.test.js b/src/converters/fromHtml.test.js
@@ -543,4 +543,48 @@ describe('convertFromHTML parsing nested tags', () => {
     expect(result[0]['@type']).toBe('image');
     expect(result[1]['@type']).toBe('slateTable');
   });
+  describe('with paragraph, table and image blocks', () => {
+    const html = `<div>
+    <div>
+    <p>
+    <table>
+    <tbody>
+    <tr>
+    <td>
+    <div><img src="image.png"/></div>
+    </td>
+    </tr>
+    </tbody>
+    </table>
+    </p>
+    </div>
+    <p></p>
+    </div>`;
+    const result = convertFromHTML(html, 'slate');
+    expect(result).toHaveLength(2);
+    expect(result[0]['@type']).toBe('image');
+    expect(result[1]['@type']).toBe('slateTable');
+  });
+  describe('with paragraph, table, image blocks, and other paragraph', () => {
+    const html = `<div>
+    <p>
+    <table>
+    <tbody>
+    <tr>
+    <td>
+    <div><img src="image.png"/></div>
+    </td>
+    </tr>
+    </tbody>
+    </table>
+    </p>
+    <p></p>
+    <p>A text</p>
+    </div>`;
+    const result = convertFromHTML(html, 'slate');
+    expect(result).toHaveLength(3);
+    expect(result[0]['@type']).toBe('image');
+    expect(result[1]['@type']).toBe('slateTable');
+    expect(result[2]['@type']).toBe('slate');
+  });
 });