@adobe/helix-docx2md 1.6.4 → 1.6.6

package/CHANGELOG.md

@@ -1,3 +1,17 @@
+## [1.6.6](https://github.com/adobe/helix-docx2md/compare/v1.6.5...v1.6.6) (2024-08-31)
+
+
+### Bug Fixes
+
+* **deps:** update adobe fixes ([256c66d](https://github.com/adobe/helix-docx2md/commit/256c66de1413d0ea7958d15ac27f9c2523e5067f))
+
+## [1.6.5](https://github.com/adobe/helix-docx2md/compare/v1.6.4...v1.6.5) (2024-08-12)
+
+
+### Bug Fixes
+
+* ignore horizontal line marker in text ([#512](https://github.com/adobe/helix-docx2md/issues/512)) ([cf5defd](https://github.com/adobe/helix-docx2md/commit/cf5defde68f778f72b59906de4267105392285f1)), closes [#508](https://github.com/adobe/helix-docx2md/issues/508)
+
 ## [1.6.4](https://github.com/adobe/helix-docx2md/compare/v1.6.3...v1.6.4) (2024-07-13)
 
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@adobe/helix-docx2md",
-  "version": "1.6.4",
+  "version": "1.6.6",
   "description": "Helix library that converts word documents to markdown",
   "type": "module",
   "main": "./src/index.js",
@@ -34,7 +34,7 @@
   },
   "homepage": "https://github.com/adobe/helix-docx2md#readme",
   "dependencies": {
-    "@adobe/helix-markdown-support": "7.1.4",
+    "@adobe/helix-markdown-support": "7.1.5",
     "@adobe/helix-shared-process-queue": "3.0.4",
     "@adobe/mammoth": "1.7.1-bleeding.2",
     "@adobe/mdast-util-gridtables": "4.0.6",
@@ -53,23 +53,23 @@
   },
   "devDependencies": {
     "@adobe/eslint-config-helix": "2.0.6",
-    "@adobe/helix-admin-support": "2.8.22",
-    "@adobe/helix-mediahandler": "2.5.13",
-    "@adobe/helix-onedrive-support": "11.5.5",
-    "@adobe/helix-shared-tokencache": "1.4.22",
+    "@adobe/helix-admin-support": "2.9.3",
+    "@adobe/helix-mediahandler": "2.5.19",
+    "@adobe/helix-onedrive-support": "11.5.13",
+    "@adobe/helix-shared-tokencache": "1.4.29",
     "@semantic-release/changelog": "6.0.3",
     "@semantic-release/exec": "6.0.3",
     "@semantic-release/git": "10.0.1",
     "c8": "10.1.2",
     "dotenv": "16.4.5",
     "eslint": "8.57.0",
-    "husky": "9.0.11",
-    "junit-report-builder": "3.2.1",
-    "lint-staged": "15.2.7",
-    "mocha": "10.6.0",
+    "husky": "9.1.5",
+    "junit-report-builder": "5.0.0",
+    "lint-staged": "15.2.9",
+    "mocha": "10.7.3",
     "mocha-multi-reporters": "1.5.1",
-    "semantic-release": "24.0.0",
-    "unist-util-inspect": "8.0.0"
+    "semantic-release": "24.1.0",
+    "unist-util-inspect": "8.1.0"
   },
   "lint-staged": {
     "*.js": "eslint"

package/src/dast2mdast/handlers/paragraph.js

@@ -39,9 +39,15 @@ const HEADING_MAP = {
   'heading 8': 8,
 };
 
+/**
+ * Checks if the given nodes represent a horizontal line '---' or em dash character
+ *
+ * @param {Array} nodes - The nodes to check.
+ * @returns {boolean} - True if the nodes represent a horizontal line, false otherwise.
+ */
 function isHorizontalLine(nodes) {
   const value = toString(nodes).trim();
-  return (value === '---' || value === '\u2014');
+  return value === '---' || value === '\u2014';
 }
 
 function getHeadingDepth(node) {
@@ -265,6 +271,12 @@ export default function paragraph(h, node, parent, siblings) {
   // eslint-disable-next-line no-param-reassign
   h.listContainers[0] = [];
 
+  // check if no horizontal line in heading
+  if (isHorizontalLine(nodes)) {
+    ret.push(h('thematicBreak'));
+    return ret;
+  }
+
   // check for heading
   let depth = getHeadingDepth(node);
   if (!depth) {
@@ -277,23 +289,18 @@ export default function paragraph(h, node, parent, siblings) {
   }
 
   if (depth) {
-    // check if no horizontal line in heading
-    if (isHorizontalLine(nodes)) {
-      ret.push(h('thematicBreak'));
-    } else {
-      const heading = h('heading', { depth }, nodes);
-      // check bookmark children (could have multiple)
-      for (let idx = 0; idx < nodes.length; idx += 1) {
-        const child = nodes[idx];
-        if (child.type === 'bookmark') {
-          // set the bookmark target to this heading and remove the child
-          child.bookmark.target = heading;
-          nodes.splice(idx, 1);
-          idx -= 1;
-        }
+    const heading = h('heading', { depth }, nodes);
+    // check bookmark children (could have multiple)
+    for (let idx = 0; idx < nodes.length; idx += 1) {
+      const child = nodes[idx];
+      if (child.type === 'bookmark') {
+        // set the bookmark target to this heading and remove the child
+        child.bookmark.target = heading;
+        nodes.splice(idx, 1);
+        idx -= 1;
       }
-      ret.push(heading);
     }
+    ret.push(heading);
     handleBorder(node.border?.bottom, ret);
     return ret;
   }
@@ -323,47 +330,6 @@ export default function paragraph(h, node, parent, siblings) {
     }
   }
 
-  // check for thematicBreaks and frontmatter. they need to be block elements
-  let prevBreak;
-  let idx = nodes.findIndex(isHorizontalLine);
-  while (idx >= 0) {
-    const brk = h('thematicBreak');
-    // remove the nodes up until the horizontal line
-    const removed = nodes.splice(0, idx);
-    // if last element is a break, ignore it.
-    if (removed.length && removed[removed.length - 1].type === 'break') {
-      brk.breakBefore = true;
-      removed.pop();
-    }
-
-    // remove the horizontal line
-    nodes.splice(0, 1);
-
-    // also check if element following is a break
-    if (nodes.length && nodes[0].type === 'break') {
-      brk.breakAfter = true;
-      nodes.splice(0, 1);
-    }
-
-    // if we had a previous thematic break, and all the text in between looks like yaml, then
-    // create a frontmatter block
-    if (prevBreak && prevBreak.breakAfter && brk.breakBefore) {
-      const src = toString({ children: removed });
-      prevBreak.type = 'yaml';
-      prevBreak.value = src;
-      prevBreak = null;
-    } else {
-      // else add the removed as a paragraph
-      if (removed.length) {
-        ret.push(h('paragraph', removed));
-      }
-      // and append the thematic break
-      ret.push(brk);
-      prevBreak = brk;
-    }
-    idx = nodes.findIndex(isHorizontalLine);
-  }
-
   // handle remaining nodes
   if (nodes.length) {
     // avoid paragraphs with just a break