@adobe/helix-markdown-support 6.2.1 → 6.3.1

package/.circleci/config.yml

@@ -2,7 +2,7 @@ version: 2.1
 executors:
   node18:
     docker:
-      - image: cimg/node:18.16
+      - image: cimg/node:18.17
 
 orbs:
   codecov: codecov/codecov@3.2.5

package/CHANGELOG.md

@@ -1,3 +1,17 @@
+## [6.3.1](https://github.com/adobe/helix-markdown-support/compare/v6.3.0...v6.3.1) (2023-08-24)
+
+
+### Bug Fixes
+
+* properly handle whitespace after sort ([#211](https://github.com/adobe/helix-markdown-support/issues/211)) ([03bf02a](https://github.com/adobe/helix-markdown-support/commit/03bf02a249f594e03c5d523d876e0b6522ad6ed5))
+
+# [6.3.0](https://github.com/adobe/helix-markdown-support/compare/v6.2.1...v6.3.0) (2023-07-27)
+
+
+### Features
+
+* sort formats ([#206](https://github.com/adobe/helix-markdown-support/issues/206)) ([b734a35](https://github.com/adobe/helix-markdown-support/commit/b734a359e764ac4630b7b2418c0e71e342ad52b6))
+
 ## [6.2.1](https://github.com/adobe/helix-markdown-support/compare/v6.2.0...v6.2.1) (2023-07-18)
 
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@adobe/helix-markdown-support",
-  "version": "6.2.1",
+  "version": "6.3.1",
   "description": "Helix Markdown Support",
   "type": "module",
   "scripts": {
@@ -58,20 +58,20 @@
     "node": ">=14"
   },
   "devDependencies": {
-    "@adobe/eslint-config-helix": "2.0.2",
+    "@adobe/eslint-config-helix": "2.0.3",
     "@adobe/remark-gridtables": "1.0.4",
     "@semantic-release/changelog": "6.0.3",
     "@semantic-release/git": "10.0.1",
-    "c8": "8.0.0",
-    "eslint": "8.45.0",
+    "c8": "8.0.1",
+    "eslint": "8.47.0",
     "husky": "8.0.3",
     "junit-report-builder": "3.0.1",
-    "lint-staged": "13.2.3",
+    "lint-staged": "13.3.0",
     "mdast-builder": "1.1.1",
     "mocha": "10.2.0",
     "mocha-multi-reporters": "1.5.1",
     "rehype-format": "4.0.1",
-    "rehype-stringify": "9.0.3",
+    "rehype-stringify": "9.0.4",
     "remark-gfm": "3.0.1",
     "remark-parse": "10.0.2",
     "remark-rehype": "10.1.0",

package/src/mdast-sanitize-text-and-formats.js

@@ -25,22 +25,56 @@ function isSnug(type) {
   return type === 'superscript' || type === 'subscript';
 }
 
-/**
- * Sanitizes text:
- * - collapses consecutive formats
- * - collapses consecutive text blocks
- * - trims ends of texts before break
- * - trims ends of texts at the end
- * - moves leading and trailing whitespaces out of formats
- * - ensures spaces after formats
- * - removes trailing breaks in containers
- *   see https://github.com/micromark/micromark/issues/118#issuecomment-1238225086
- * - removes empty text blocks, formats, paragraphs
- *
- * @param {object} tree
- * @returns {object} The modified (original) tree.
- */
-export default function sanitizeTextAndFormats(tree) {
+const FLOW_SORT_ORDER = [
+  'delete',
+  'strong',
+  'emphasis',
+  'link',
+  'underline',
+  'subscript',
+  'superscript',
+];
+
+export function sort(tree) {
+  if (!tree.children) {
+    return;
+  }
+  for (let i = 0; i < tree.children.length; i += 1) {
+    let node = tree.children[i];
+    let key = FLOW_SORT_ORDER.indexOf(node.type);
+    if (key >= 0) {
+      // find the longest chain of formats
+      const chain = [];
+      while (key >= 0) {
+        chain.push({ node, key });
+        node = node.children?.length === 1 ? node.children[0] : null;
+        key = node ? FLOW_SORT_ORDER.indexOf(node?.type) : -1;
+      }
+      if (chain.length > 1) {
+        // remember children of last node in chain
+        const lastChildren = chain[chain.length - 1].node.children;
+
+        // sort chain
+        chain.sort((n0, n1) => (n0.key - n1.key));
+
+        // relink chain
+        for (let j = 0; j < chain.length - 1; j += 1) {
+          chain[j].node.children = [chain[j + 1].node];
+        }
+        chain[chain.length - 1].node.children = lastChildren;
+
+        // eslint-disable-next-line no-param-reassign
+        tree.children[i] = chain[0].node;
+      }
+      // continue on last node
+      sort(chain[chain.length - 1].node);
+    } else {
+      sort(node);
+    }
+  }
+}
+
+function collapse(tree) {
   visit(tree, (node, index, parent) => {
     const { children: siblings = [] } = parent || {};
     const { children = [] } = node;
@@ -56,7 +90,17 @@ export default function sanitizeTextAndFormats(tree) {
         siblings.splice(index, 1);
         return index - 1;
       }
+    }
+    return CONTINUE;
+  });
+}
 
+function whitespace(tree) {
+  visit(tree, (node, index, parent) => {
+    const { children: siblings = [] } = parent || {};
+    const { children = [] } = node;
+
+    if (isFormat(node.type)) {
       // check if last text block has trailing whitespace
       const last = children[children.length - 1];
       if (last?.type === 'text') {
@@ -127,7 +171,9 @@ export default function sanitizeTextAndFormats(tree) {
     }
     return CONTINUE;
   });
+}
 
+function cleanup(tree) {
   visit(tree, (node, index, parent) => {
     const { children: siblings = [] } = parent || {};
 
@@ -177,28 +223,56 @@ export default function sanitizeTextAndFormats(tree) {
 
     return CONTINUE;
   });
+}
 
-  // remove text, formats and paragraphs
-  function prune(node) {
-    const { children, type } = node;
-    if (type === 'text') {
-      return !node.value;
-    }
-    if (!children) {
-      return false;
-    }
-    for (let i = 0; i < children.length; i += 1) {
-      if (prune(children[i])) {
-        children.splice(i, 1);
-        i -= 1;
-      }
-    }
-    if (type === 'paragraph' || isFormat(type)) {
-      return children.length === 0;
-    }
+/**
+ * remove empty text, formats, paragraphs
+ * @param node
+ * @return {boolean}
+ */
+function prune(node) {
+  const { children, type } = node;
+  if (type === 'text') {
+    return !node.value;
+  }
+  if (!children) {
     return false;
   }
-  prune(tree);
+  for (let i = 0; i < children.length; i += 1) {
+    if (prune(children[i])) {
+      children.splice(i, 1);
+      i -= 1;
+    }
+  }
+  if (type === 'paragraph' || isFormat(type)) {
+    return children.length === 0;
+  }
+  return false;
+}
 
+/**
+ * Sanitizes text:
+ * - collapses consecutive formats
+ * - collapses consecutive text blocks
+ * - trims ends of texts before break
+ * - trims ends of texts at the end
+ * - moves leading and trailing whitespaces out of formats
+ * - ensures spaces after formats
+ * - removes trailing breaks in containers
+ *   see https://github.com/micromark/micromark/issues/118#issuecomment-1238225086
+ * - removes empty text blocks, formats, paragraphs
+ *
+ * @param {object} tree
+ * @returns {object} The modified (original) tree.
+ */
+export default function sanitizeTextAndFormats(tree) {
+  collapse(tree);
+  prune(tree);
+  sort(tree);
+  // collapse again, because sorting the nodes might have produce new collapsable siblings
+  collapse(tree);
+  whitespace(tree);
+  cleanup(tree);
+  prune(tree);
   return tree;
 }