@readme/markdown 11.14.1 → 12.0.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/dist/main.node.js CHANGED
@@ -19024,7 +19024,6 @@ __webpack_require__.d(__webpack_exports__, {
19024
19024
  exports: () => (/* reexport */ lib_exports),
19025
19025
  gemojiRegex: () => (/* reexport */ gemoji_regex),
19026
19026
  hast: () => (/* reexport */ lib_hast),
19027
- isPlainText: () => (/* reexport */ isPlainText),
19028
19027
  mdast: () => (/* reexport */ lib_mdast),
19029
19028
  mdastV6: () => (/* reexport */ lib_mdastV6),
19030
19029
  mdx: () => (/* reexport */ lib_mdx),
@@ -73148,12 +73147,40 @@ const plain = (node, opts = {}) => {
73148
73147
  };
73149
73148
  /* harmony default export */ const lib_plain = (plain);
73150
73149
 
73150
+ ;// ./processor/transform/extract-text.ts
73151
+ /**
73152
+ * Extracts text content from a single AST node recursively.
73153
+ * Works with both MDAST and HAST-like node structures.
73154
+ *
73155
+ * Placed this outside of the utils.ts file to avoid circular dependencies.
73156
+ *
73157
+ * @param node - The node to extract text from (can be MDAST Node or HAST-like structure)
73158
+ * @returns The concatenated text content
73159
+ */
73160
+ const extractText = (node) => {
73161
+ if (node.type === 'text' && typeof node.value === 'string') {
73162
+ return node.value;
73163
+ }
73164
+ if (node.children && Array.isArray(node.children)) {
73165
+ return node.children
73166
+ .map(child => {
73167
+ if (child && typeof child === 'object' && 'type' in child) {
73168
+ return extractText(child);
73169
+ }
73170
+ return '';
73171
+ })
73172
+ .join('');
73173
+ }
73174
+ return '';
73175
+ };
73176
+
73151
73177
  ;// ./processor/transform/callouts.ts
73152
73178
 
73153
73179
 
73154
73180
 
73155
73181
 
73156
73182
 
73183
+
73157
73184
  const callouts_regex = `^(${emoji_regex().source}|⚠)(\\s+|$)`;
73158
73185
  const findFirst = (node) => {
73159
73186
  if ('children' in node)
@@ -73174,42 +73201,76 @@ const wrapHeading = (node) => {
73174
73201
  },
73175
73202
  };
73176
73203
  };
73204
+ /**
73205
+ * Checks if a blockquote matches the expected callout structure:
73206
+ * blockquote > paragraph > text node
73207
+ */
73208
+ const isCalloutStructure = (node) => {
73209
+ const firstChild = node.children?.[0];
73210
+ if (!firstChild || firstChild.type !== 'paragraph')
73211
+ return false;
73212
+ if (!('children' in firstChild))
73213
+ return false;
73214
+ const firstTextChild = firstChild.children?.[0];
73215
+ return firstTextChild?.type === 'text';
73216
+ };
73217
+ const processBlockquote = (node, index, parent) => {
73218
+ if (!isCalloutStructure(node)) {
73219
+ // Only stringify empty blockquotes (no extractable text content)
73220
+ // Preserve blockquotes with actual content (e.g., headings, lists, etc.)
73221
+ const content = extractText(node);
73222
+ const isEmpty = !content || content.trim() === '';
73223
+ if (isEmpty && index !== undefined && parent) {
73224
+ const textNode = {
73225
+ type: 'text',
73226
+ value: '>',
73227
+ };
73228
+ const paragraphNode = {
73229
+ type: 'paragraph',
73230
+ children: [textNode],
73231
+ position: node.position,
73232
+ };
73233
+ parent.children.splice(index, 1, paragraphNode);
73234
+ }
73235
+ return;
73236
+ }
73237
+ // isCalloutStructure ensures node.children[0] is a Paragraph with children
73238
+ const firstParagraph = node.children[0];
73239
+ const startText = lib_plain(firstParagraph).toString();
73240
+ const [match, icon] = startText.match(callouts_regex) || [];
73241
+ if (icon && match) {
73242
+ const heading = startText.slice(match.length);
73243
+ const empty = !heading.length && firstParagraph.children.length === 1;
73244
+ const theme = themes[icon] || 'default';
73245
+ const firstChild = findFirst(node.children[0]);
73246
+ if (firstChild && 'value' in firstChild && typeof firstChild.value === 'string') {
73247
+ firstChild.value = firstChild.value.slice(match.length);
73248
+ }
73249
+ if (heading) {
73250
+ node.children[0] = wrapHeading(node);
73251
+ // @note: We add to the offset/column the length of the unicode
73252
+ // character that was stripped off, so that the start position of the
73253
+ // heading/text matches where it actually starts.
73254
+ node.children[0].position.start.offset += match.length;
73255
+ node.children[0].position.start.column += match.length;
73256
+ }
73257
+ Object.assign(node, {
73258
+ type: NodeTypes.callout,
73259
+ data: {
73260
+ hName: 'Callout',
73261
+ hProperties: {
73262
+ icon,
73263
+ ...(empty && { empty }),
73264
+ theme,
73265
+ },
73266
+ },
73267
+ });
73268
+ }
73269
+ };
73177
73270
  const calloutTransformer = () => {
73178
73271
  return (tree) => {
73179
- visit(tree, 'blockquote', (node) => {
73180
- if (!(node.children[0].type === 'paragraph' && node.children[0].children[0].type === 'text'))
73181
- return;
73182
- // @ts-expect-error -- @todo: update plain to accept mdast
73183
- const startText = lib_plain(node.children[0]).toString();
73184
- const [match, icon] = startText.match(callouts_regex) || [];
73185
- if (icon && match) {
73186
- const heading = startText.slice(match.length);
73187
- const empty = !heading.length && node.children[0].children.length === 1;
73188
- const theme = themes[icon] || 'default';
73189
- const firstChild = findFirst(node.children[0]);
73190
- if (firstChild && 'value' in firstChild && typeof firstChild.value === 'string') {
73191
- firstChild.value = firstChild.value.slice(match.length);
73192
- }
73193
- if (heading) {
73194
- node.children[0] = wrapHeading(node);
73195
- // @note: We add to the offset/column the length of the unicode
73196
- // character that was stripped off, so that the start position of the
73197
- // heading/text matches where it actually starts.
73198
- node.children[0].position.start.offset += match.length;
73199
- node.children[0].position.start.column += match.length;
73200
- }
73201
- Object.assign(node, {
73202
- type: NodeTypes.callout,
73203
- data: {
73204
- hName: 'Callout',
73205
- hProperties: {
73206
- icon,
73207
- ...(empty && { empty }),
73208
- theme,
73209
- },
73210
- },
73211
- });
73212
- }
73272
+ visit(tree, 'blockquote', (node, index, parent) => {
73273
+ processBlockquote(node, index, parent);
73213
73274
  });
73214
73275
  };
73215
73276
  };
@@ -90825,10 +90886,13 @@ const tagAttributePattern = /([a-zA-Z_:][-a-zA-Z0-9_:.]*)(?:\s*=\s*("[^"]*"|'[^'
90825
90886
  const MAX_LOOKAHEAD = 30;
90826
90887
  /**
90827
90888
  * Tags that have dedicated transformers and should NOT be handled by this plugin.
90828
- * These components have special parsing requirements that the generic component
90829
- * block transformer cannot handle correctly.
90889
+ * These components either have special parsing requirements that the generic component
90890
+ * block transformer cannot handle correctly, or are inline components that we don't
90891
+ * want to convert to mdxJsxFlowElement which is a block level element.
90892
+ *
90893
+ * Glossary and Anchor are inline components.
90830
90894
  */
90831
- const EXCLUDED_TAGS = new Set(['HTMLBlock', 'Table']);
90895
+ const EXCLUDED_TAGS = new Set(['HTMLBlock', 'Table', 'Glossary', 'Anchor']);
90832
90896
  const inlineMdProcessor = unified().use(remarkParse);
90833
90897
  const isClosingTag = (value, tag) => value.trim() === `</${tag}>`;
90834
90898
  /**
@@ -90872,6 +90936,17 @@ const parseTag = (value) => {
90872
90936
  contentAfterTag,
90873
90937
  };
90874
90938
  };
90939
+ /**
90940
+ * Parse substring content of a node and update the parent's children to include the new nodes.
90941
+ */
90942
+ const parseSibling = (stack, parent, index, sibling) => {
90943
+ const siblingNodes = parseMdChildren(sibling);
90944
+ // The new sibling nodes might contain new components to be processed
90945
+ if (siblingNodes.length > 0) {
90946
+ parent.children.splice(index + 1, 0, ...siblingNodes);
90947
+ stack.push(parent);
90948
+ }
90949
+ };
90875
90950
  /**
90876
90951
  * Create an MdxJsxFlowElement node from component data.
90877
90952
  */
@@ -91014,11 +91089,12 @@ const mdxishComponentBlocks = () => tree => {
91014
91089
  if ('children' in node && Array.isArray(node.children)) {
91015
91090
  stack.push(node);
91016
91091
  }
91017
- // Only visit HTML nodes with an actual html tag
91092
+ // Only visit HTML nodes with an actual html tag,
91093
+ // which means a potential unparsed MDX component
91018
91094
  const value = node.value;
91019
91095
  if (node.type !== 'html' || typeof value !== 'string')
91020
91096
  return;
91021
- const parsed = parseTag(value);
91097
+ const parsed = parseTag(value.trim());
91022
91098
  if (!parsed)
91023
91099
  return;
91024
91100
  const { tag, attributes, selfClosing, contentAfterTag = '' } = parsed;
@@ -91035,11 +91111,19 @@ const mdxishComponentBlocks = () => tree => {
91035
91111
  startPosition: node.position,
91036
91112
  });
91037
91113
  substituteNodeWithMdxNode(parent, index, componentNode);
91114
+ // Check and parse if there's relevant content after the current closing tag
91115
+ const remainingContent = contentAfterTag.trim();
91116
+ if (remainingContent) {
91117
+ parseSibling(stack, parent, index, remainingContent);
91118
+ }
91038
91119
  return;
91039
91120
  }
91040
91121
  // Case 2: Self-contained block (closing tag in content)
91041
91122
  if (contentAfterTag.includes(closingTagStr)) {
91042
- const componentInnerContent = contentAfterTag.substring(0, contentAfterTag.lastIndexOf(closingTagStr)).trim();
91123
+ // Find the first closing tag
91124
+ const closingTagIndex = contentAfterTag.indexOf(closingTagStr);
91125
+ const componentInnerContent = contentAfterTag.substring(0, closingTagIndex).trim();
91126
+ const contentAfterClose = contentAfterTag.substring(closingTagIndex + closingTagStr.length).trim();
91043
91127
  const componentNode = createComponentNode({
91044
91128
  tag,
91045
91129
  attributes,
@@ -91047,6 +91131,14 @@ const mdxishComponentBlocks = () => tree => {
91047
91131
  startPosition: node.position,
91048
91132
  });
91049
91133
  substituteNodeWithMdxNode(parent, index, componentNode);
91134
+ // After the closing tag, there might be more content to be processed
91135
+ if (contentAfterClose) {
91136
+ parseSibling(stack, parent, index, contentAfterClose);
91137
+ }
91138
+ else if (componentNode.children.length > 0) {
91139
+ // The content inside the component block might contain new components to be processed
91140
+ stack.push(componentNode);
91141
+ }
91050
91142
  return;
91051
91143
  }
91052
91144
  // Case 3: Multi-sibling component (closing tag in a following sibling)
@@ -91072,8 +91164,13 @@ const mdxishComponentBlocks = () => tree => {
91072
91164
  });
91073
91165
  // Remove all nodes from opening tag to closing tag (inclusive) and replace with component node
91074
91166
  parent.children.splice(index, closingIndex - index + 1, componentNode);
91167
+ // Since we might be merging sibling nodes together and combining content,
91168
+ // there might be new components to process
91169
+ if (componentNode.children.length > 0) {
91170
+ stack.push(componentNode);
91171
+ }
91075
91172
  };
91076
- // Travel the tree depth-first
91173
+ // Process the nodes with the components depth-first to maintain the correct order of the nodes
91077
91174
  while (stack.length) {
91078
91175
  const parent = stack.pop();
91079
91176
  if (parent?.children) {
@@ -91094,6 +91191,7 @@ const mdxishComponentBlocks = () => tree => {
91094
91191
 
91095
91192
 
91096
91193
 
91194
+
91097
91195
  const isTableCell = (node) => isMDXElement(node) && ['th', 'td'].includes(node.name);
91098
91196
  const tableTypes = {
91099
91197
  tr: 'tableRow',
@@ -91118,18 +91216,13 @@ const isTextOnly = (children) => {
91118
91216
  });
91119
91217
  };
91120
91218
  /**
91121
- * Extract text content from children nodes
91219
+ * Convenience wrapper that extracts text content from an array of children nodes.
91122
91220
  */
91123
- const extractText = (children) => {
91221
+ const extractTextFromChildren = (children) => {
91124
91222
  return children
91125
91223
  .map(child => {
91126
91224
  if (child && typeof child === 'object' && 'type' in child) {
91127
- if (child.type === 'text' && 'value' in child && typeof child.value === 'string') {
91128
- return child.value;
91129
- }
91130
- if (child.type === 'mdxJsxTextElement' && 'children' in child && Array.isArray(child.children)) {
91131
- return extractText(child.children);
91132
- }
91225
+ return extractText(child);
91133
91226
  }
91134
91227
  return '';
91135
91228
  })
@@ -91160,7 +91253,7 @@ const processTableNode = (node, index, parent) => {
91160
91253
  let parsedChildren = cellChildren;
91161
91254
  // If cell contains only text nodes, try to re-parse as markdown
91162
91255
  if (isTextOnly(cellChildren)) {
91163
- const textContent = extractText(cellChildren);
91256
+ const textContent = extractTextFromChildren(cellChildren);
91164
91257
  if (textContent.trim()) {
91165
91258
  try {
91166
91259
  const parsed = parseMarkdown(textContent);
@@ -106575,7 +106668,7 @@ const CUSTOM_PROP_BOUNDARIES = [
106575
106668
  /**
106576
106669
  * Tags that should be passed through and handled at runtime (not by the mdxish plugin)
106577
106670
  */
106578
- const RUNTIME_COMPONENT_TAGS = new Set(['Variable', 'variable']);
106671
+ const RUNTIME_COMPONENT_TAGS = new Set(['Variable', 'variable', 'rdme-pin']);
106579
106672
  /**
106580
106673
  * Standard HTML tags that should never be treated as custom components.
106581
106674
  * Uses the html-tags package, converted to a Set<string> for efficient lookups.
@@ -113206,6 +113299,33 @@ const INLINE_COMPONENT_TAGS = new Set(['anchor', 'glossary']);
113206
113299
  function isElementContentNode(node) {
113207
113300
  return node.type === 'element' || node.type === 'text' || node.type === 'comment';
113208
113301
  }
113302
+ /**
113303
+ * Components are assumed to be block-level, so whitespace between them can be removed
113304
+ * We want to remove them because it can be treated as actual children of the component
113305
+ * when it doesn't need to.
113306
+ *
113307
+ * It can be a problem because it can be passed as args to the components, like the ones
113308
+ * defined in /components, which can break the type assumptions of the components and cause
113309
+ * type errors, accessing properties that don't exist.
113310
+ */
113311
+ function areAllChildrenComponents(children) {
113312
+ return children.every(child => {
113313
+ // Whitespace-only text nodes don't affect the check
113314
+ if (child.type === 'text' && !child.value.trim())
113315
+ return true;
113316
+ // Text with actual content means we have mixed content
113317
+ if (child.type === 'text')
113318
+ return false;
113319
+ // Comments don't affect the check
113320
+ if (child.type === 'comment')
113321
+ return true;
113322
+ // Standard HTML tags are not considered components
113323
+ if (child.type === 'element' && 'tagName' in child) {
113324
+ return !STANDARD_HTML_TAGS.has(child.tagName.toLowerCase());
113325
+ }
113326
+ return false;
113327
+ });
113328
+ }
113209
113329
  /** Check if nodes represent a single paragraph with only text (no markdown formatting) */
113210
113330
  function isSingleParagraphTextNode(nodes) {
113211
113331
  return (nodes.length === 1 &&
@@ -113250,6 +113370,7 @@ function isActualHtmlTag(tagName, originalExcerpt) {
113250
113370
  function parseTextChildren(node, processMarkdown) {
113251
113371
  if (!node.children?.length)
113252
113372
  return;
113373
+ // First pass: Recursively process text children as they may contain stringified markdown / mdx content
113253
113374
  node.children = node.children.flatMap(child => {
113254
113375
  if (child.type !== 'text' || !child.value.trim())
113255
113376
  return [child];
@@ -113261,6 +113382,15 @@ function parseTextChildren(node, processMarkdown) {
113261
113382
  }
113262
113383
  return children;
113263
113384
  });
113385
+ // Post-processing: remove whitespace-only text nodes if all siblings are components
113386
+ // This prevents whitespace between component children from being counted as extra children
113387
+ if (areAllChildrenComponents(node.children)) {
113388
+ node.children = node.children.filter(child => {
113389
+ if (child.type === 'text' && !child.value.trim())
113390
+ return false;
113391
+ return true;
113392
+ });
113393
+ }
113264
113394
  }
113265
113395
  /** Convert node properties from kebab-case/lowercase to camelCase */
113266
113396
  function normalizeProperties(node) {
@@ -113323,6 +113453,15 @@ const mdxExpressionHandler = (_state, node) => ({
113323
113453
  type: 'text',
113324
113454
  value: node.value || '',
113325
113455
  });
113456
+ // Since we serialize component / html tag attributes
113457
+ function decodeHtmlEntities(value) {
113458
+ return value
113459
+ .replace(/&quot;/g, '"')
113460
+ .replace(/&lt;/g, '<')
113461
+ .replace(/&gt;/g, '>')
113462
+ .replace(/&#10;/g, '\n')
113463
+ .replace(/&amp;/g, '&');
113464
+ }
113326
113465
  // Convert MDX JSX elements to HAST elements, preserving attributes and children
113327
113466
  const mdxJsxElementHandler = (state, node) => {
113328
113467
  const { attributes = [], name } = node;
@@ -113334,7 +113473,7 @@ const mdxJsxElementHandler = (state, node) => {
113334
113473
  properties[attribute.name] = true;
113335
113474
  }
113336
113475
  else if (typeof attribute.value === 'string') {
113337
- properties[attribute.name] = attribute.value;
113476
+ properties[attribute.name] = decodeHtmlEntities(attribute.value);
113338
113477
  }
113339
113478
  else {
113340
113479
  properties[attribute.name] = attribute.value.value;
@@ -113396,6 +113535,17 @@ function base64Decode(str) {
113396
113535
  }
113397
113536
  return decodeURIComponent(escape(atob(str)));
113398
113537
  }
113538
+ function escapeHtmlAttribute(value) {
113539
+ return value
113540
+ .replace(/&/g, '&amp;')
113541
+ .replace(/"/g, '&quot;')
113542
+ .replace(/</g, '&lt;')
113543
+ .replace(/>/g, '&gt;')
113544
+ .replace(/\n/g, '&#10;');
113545
+ }
113546
+ // Marker prefix for JSON-serialized complex values (arrays/objects)
113547
+ // Using a prefix that won't conflict with regular string values
113548
+ const JSON_VALUE_MARKER = '__MDXISH_JSON__';
113399
113549
  // Markers for protected HTMLBlock content (HTML comments avoid markdown parsing issues)
113400
113550
  const HTML_BLOCK_CONTENT_START = '<!--RDMX_HTMLBLOCK:';
113401
113551
  const HTML_BLOCK_CONTENT_END = ':RDMX_HTMLBLOCK-->';
@@ -113525,6 +113675,16 @@ function extractBalancedBraces(content, start) {
113525
113675
  return null;
113526
113676
  return { content: content.slice(start, pos - 1), end: pos };
113527
113677
  }
113678
+ function restoreInlineCode(content, protectedCode) {
113679
+ return content.replace(/___INLINE_CODE_(\d+)___/g, (_m, idx) => {
113680
+ return protectedCode.inlineCode[parseInt(idx, 10)];
113681
+ });
113682
+ }
113683
+ function restoreCodeBlocks(content, protectedCode) {
113684
+ return content.replace(/___CODE_BLOCK_(\d+)___/g, (_m, idx) => {
113685
+ return protectedCode.codeBlocks[parseInt(idx, 10)];
113686
+ });
113687
+ }
113528
113688
  /**
113529
113689
  * Converts JSX attribute expressions (attribute={expression}) to HTML attributes (attribute="value").
113530
113690
  * Handles style objects (camelCase → kebab-case), className → class, and JSON stringifies objects.
@@ -113540,7 +113700,7 @@ function extractBalancedBraces(content, start) {
113540
113700
  * // Returns: '<a href="https://example.com">Link</a>'
113541
113701
  * ```
113542
113702
  */
113543
- function evaluateAttributeExpressions(content, context) {
113703
+ function evaluateAttributeExpressions(content, context, protectedCode) {
113544
113704
  const attrStartRegex = /(\w+)=\{/g;
113545
113705
  let result = '';
113546
113706
  let lastEnd = 0;
@@ -113550,7 +113710,14 @@ function evaluateAttributeExpressions(content, context) {
113550
113710
  const braceStart = match.index + match[0].length;
113551
113711
  const extracted = extractBalancedBraces(content, braceStart);
113552
113712
  if (extracted) {
113553
- const expression = extracted.content;
113713
+ // The expression might contain template literals in MDX component tag props
113714
+ // E.g. <Component greeting={`Hello World!`} />
113715
+ // that is marked as inline code. So we need to restore the inline codes
113716
+ // in the expression to evaluate it
113717
+ let expression = extracted.content;
113718
+ if (protectedCode) {
113719
+ expression = restoreInlineCode(expression, protectedCode);
113720
+ }
113554
113721
  const fullMatchEnd = extracted.end;
113555
113722
  result += content.slice(lastEnd, match.index);
113556
113723
  try {
@@ -113566,14 +113733,20 @@ function evaluateAttributeExpressions(content, context) {
113566
113733
  result += `style="${cssString}"`;
113567
113734
  }
113568
113735
  else {
113569
- result += `${attributeName}='${JSON.stringify(evalResult)}'`;
113736
+ // These are arrays / objects attribute values
113737
+ // Mark JSON-serialized values with a prefix so they can be parsed back correctly
113738
+ const jsonValue = escapeHtmlAttribute(JSON_VALUE_MARKER + JSON.stringify(evalResult));
113739
+ // Use double quotes so that multi-paragraph values are not split into multiple attributes by the processors
113740
+ result += `${attributeName}="${jsonValue}"`;
113570
113741
  }
113571
113742
  }
113572
113743
  else if (attributeName === 'className') {
113573
- result += `class="${evalResult}"`;
113744
+ // Escape special characters so that it doesn't break and split the attribute value to nodes
113745
+ // This will be restored later in the pipeline
113746
+ result += `class="${escapeHtmlAttribute(String(evalResult))}"`;
113574
113747
  }
113575
113748
  else {
113576
- result += `${attributeName}="${evalResult}"`;
113749
+ result += `${attributeName}="${escapeHtmlAttribute(String(evalResult))}"`;
113577
113750
  }
113578
113751
  }
113579
113752
  catch (_error) {
@@ -113604,13 +113777,9 @@ function evaluateAttributeExpressions(content, context) {
113604
113777
  * // Returns: 'Text with `inline` and ```js\ncode\n```'
113605
113778
  * ```
113606
113779
  */
113607
- function restoreCodeBlocks(content, protectedCode) {
113608
- let restored = content.replace(/___CODE_BLOCK_(\d+)___/g, (_match, index) => {
113609
- return protectedCode.codeBlocks[parseInt(index, 10)];
113610
- });
113611
- restored = restored.replace(/___INLINE_CODE_(\d+)___/g, (_match, index) => {
113612
- return protectedCode.inlineCode[parseInt(index, 10)];
113613
- });
113780
+ function restoreProtectedCodes(content, protectedCode) {
113781
+ let restored = restoreCodeBlocks(content, protectedCode);
113782
+ restored = restoreInlineCode(restored, protectedCode);
113614
113783
  return restored;
113615
113784
  }
113616
113785
  /**
@@ -113631,9 +113800,9 @@ function preprocessJSXExpressions(content, context = {}) {
113631
113800
  // Step 3: Evaluate attribute expressions (JSX attribute syntax: href={baseUrl})
113632
113801
  // For inline expressions, we use a library to parse the expression & evaluate it later
113633
113802
  // For attribute expressions, it was difficult to use a library to parse them, so do it manually
113634
- processed = evaluateAttributeExpressions(processed, context);
113803
+ processed = evaluateAttributeExpressions(processed, context, protectedCode);
113635
113804
  // Step 4: Restore protected code blocks
113636
- processed = restoreCodeBlocks(processed, protectedCode);
113805
+ processed = restoreProtectedCodes(processed, protectedCode);
113637
113806
  return processed;
113638
113807
  }
113639
113808
 
@@ -114045,7 +114214,7 @@ const wrapPinnedBlocks = (node, json) => {
114045
114214
  return node;
114046
114215
  return {
114047
114216
  children: [node],
114048
- data: { className: 'pin', hName: 'rdme-pin' },
114217
+ data: { hName: 'rdme-pin', hProperties: { className: 'pin' } },
114049
114218
  type: 'rdme-pin',
114050
114219
  };
114051
114220
  };
@@ -114133,15 +114302,16 @@ function parseMagicBlock(raw, options = {}) {
114133
114302
  type: 'code',
114134
114303
  value: obj.code.trim(),
114135
114304
  }));
114136
- // Single code block without a tab name renders as a plain code block
114305
+ // Single code block without a tab name (meta or language) renders as a plain code block
114306
+ // Otherwise, we want to render it as a code tabs block
114137
114307
  if (children.length === 1) {
114138
114308
  if (!children[0].value)
114139
114309
  return [];
114140
- if (children[0].meta)
114310
+ if (!(children[0].meta || children[0].lang))
114141
114311
  return [wrapPinnedBlocks(children[0], json)];
114142
114312
  }
114143
- // Multiple code blocks render as tabbed code blocks
114144
- return [wrapPinnedBlocks({ children, className: 'tabs', data: { hName: 'code-tabs' }, type: 'code-tabs' }, json)];
114313
+ // Multiple code blocks or a single code block with a tab name (meta or language) renders as a code tabs block
114314
+ return [wrapPinnedBlocks({ children, className: 'tabs', data: { hName: 'CodeTabs' }, type: 'code-tabs' }, json)];
114145
114315
  }
114146
114316
  // API header: renders as a heading element (h1-h6)
114147
114317
  case 'api-header': {
@@ -114342,11 +114512,31 @@ function parseMagicBlock(raw, options = {}) {
114342
114512
  }
114343
114513
  }
114344
114514
  /**
114345
- * Check if a child node is a flow element that needs unwrapping (mdxJsxFlowElement, etc.)
114515
+ * Block-level node types that cannot be nested inside paragraphs.
114346
114516
  */
114347
- const needsUnwrapping = (child) => {
114348
- return child.type === 'mdxJsxFlowElement';
114349
- };
114517
+ const blockTypes = [
114518
+ 'heading',
114519
+ 'code',
114520
+ 'code-tabs',
114521
+ 'paragraph',
114522
+ 'blockquote',
114523
+ 'list',
114524
+ 'table',
114525
+ 'thematicBreak',
114526
+ 'html',
114527
+ 'yaml',
114528
+ 'toml',
114529
+ 'rdme-pin',
114530
+ 'rdme-callout',
114531
+ 'html-block',
114532
+ 'embed',
114533
+ 'figure',
114534
+ 'mdxJsxFlowElement',
114535
+ ];
114536
+ /**
114537
+ * Check if a node is a block-level node (cannot be inside a paragraph)
114538
+ */
114539
+ const isBlockNode = (node) => blockTypes.includes(node.type);
114350
114540
  /**
114351
114541
  * Unified plugin that restores magic blocks from placeholder tokens.
114352
114542
  *
@@ -114359,8 +114549,9 @@ const magicBlockRestorer = ({ blocks }) => tree => {
114359
114549
  return;
114360
114550
  // Map: key → original raw magic block content
114361
114551
  const magicBlockKeys = new Map(blocks.map(({ key, raw }) => [key, raw]));
114362
- // Find inlineCode nodes that match our placeholder tokens
114363
- const modifications = [];
114552
+ // Collect replacements to apply (we need to visit in reverse to maintain indices)
114553
+ const replacements = [];
114554
+ // First pass: collect all replacements
114364
114555
  visit(tree, 'inlineCode', (node, index, parent) => {
114365
114556
  if (!parent || index == null)
114366
114557
  return undefined;
@@ -114370,31 +114561,73 @@ const magicBlockRestorer = ({ blocks }) => tree => {
114370
114561
  const children = parseMagicBlock(raw);
114371
114562
  if (!children.length)
114372
114563
  return undefined;
114373
- if (children[0] && needsUnwrapping(children[0]) && parent.type === 'paragraph') {
114374
- // Find paragraph's parent and unwrap
114375
- let paragraphParent;
114376
- visit(tree, 'paragraph', (p, pIndex, pParent) => {
114377
- if (p === parent && pParent && 'children' in pParent) {
114378
- paragraphParent = pParent;
114379
- return false;
114564
+ // If parent is a paragraph and we're inserting block nodes (which must not be in paragraphs), lift them out
114565
+ if (parent.type === 'paragraph' && children.some(child => isBlockNode(child))) {
114566
+ const blockNodes = [];
114567
+ const inlineNodes = [];
114568
+ // Separate block and inline nodes
114569
+ children.forEach(child => {
114570
+ if (isBlockNode(child)) {
114571
+ blockNodes.push(child);
114380
114572
  }
114381
- return undefined;
114382
- });
114383
- if (paragraphParent) {
114384
- const paragraphIndex = paragraphParent.children.indexOf(parent);
114385
- if (paragraphIndex !== -1) {
114386
- modifications.push({ children, index: paragraphIndex, parent: paragraphParent });
114573
+ else {
114574
+ inlineNodes.push(child);
114387
114575
  }
114388
- }
114389
- return SKIP;
114576
+ });
114577
+ const before = parent.children.slice(0, index);
114578
+ const after = parent.children.slice(index + 1);
114579
+ replacements.push({
114580
+ parent,
114581
+ blockNodes,
114582
+ inlineNodes,
114583
+ before,
114584
+ after,
114585
+ });
114390
114586
  }
114391
- parent.children.splice(index, 1, ...children);
114392
- return [SKIP, index + children.length];
114393
- });
114394
- // Apply modifications in reverse order to avoid index shifting
114395
- modifications.reverse().forEach(({ children, index, parent }) => {
114396
- parent.children.splice(index, 1, ...children);
114587
+ else {
114588
+ // Normal case: just replace the inlineCode with the children
114589
+ parent.children.splice(index, 1, ...children);
114590
+ }
114591
+ return undefined;
114397
114592
  });
114593
+ // Second pass: apply replacements that require lifting block nodes out of paragraphs
114594
+ // Process in reverse order to maintain correct indices
114595
+ for (let i = replacements.length - 1; i >= 0; i -= 1) {
114596
+ const { after, before, blockNodes, inlineNodes, parent } = replacements[i];
114597
+ // Find the paragraph's position in the root
114598
+ const rootChildren = tree.children;
114599
+ const paraIndex = rootChildren.findIndex(child => child === parent);
114600
+ if (paraIndex === -1) {
114601
+ // Paragraph not found in root - fall back to normal replacement
114602
+ // This shouldn't happen normally, but handle it gracefully
114603
+ // Reconstruct the original index from before.length
114604
+ const originalIndex = before.length;
114605
+ parent.children.splice(originalIndex, 1, ...blockNodes, ...inlineNodes);
114606
+ // eslint-disable-next-line no-continue
114607
+ continue;
114608
+ }
114609
+ // Update or remove the paragraph
114610
+ if (inlineNodes.length > 0) {
114611
+ // Keep paragraph with inline nodes
114612
+ parent.children = [...before, ...inlineNodes, ...after];
114613
+ // Insert block nodes after the paragraph
114614
+ if (blockNodes.length > 0) {
114615
+ rootChildren.splice(paraIndex + 1, 0, ...blockNodes);
114616
+ }
114617
+ }
114618
+ else if (before.length === 0 && after.length === 0) {
114619
+ // Remove empty paragraph and replace with block nodes
114620
+ rootChildren.splice(paraIndex, 1, ...blockNodes);
114621
+ }
114622
+ else {
114623
+ // Keep paragraph with remaining content
114624
+ parent.children = [...before, ...after];
114625
+ // Insert block nodes after the paragraph
114626
+ if (blockNodes.length > 0) {
114627
+ rootChildren.splice(paraIndex + 1, 0, ...blockNodes);
114628
+ }
114629
+ }
114630
+ }
114398
114631
  };
114399
114632
  /* harmony default export */ const mdxish_magic_blocks = (magicBlockRestorer);
114400
114633
 
@@ -114618,6 +114851,92 @@ const normalizeEmphasisAST = () => (tree) => {
114618
114851
  };
114619
114852
  /* harmony default export */ const normalize_malformed_md_syntax = (normalizeEmphasisAST);
114620
114853
 
114854
+ ;// ./processor/transform/mdxish/normalize-table-separator.ts
114855
+ /**
114856
+ * Preprocessor to normalize malformed GFM table separator syntax.
114857
+ *
114858
+ * Fixes the common mistake where the alignment colon is placed after the pipe
114859
+ * instead of before the dashes:
114860
+ *
114861
+ * Invalid: `|: ---` or `|:---` (colon after pipe)
114862
+ * Valid: `| :---` (colon before dashes)
114863
+ *
114864
+ * Also handles right alignment:
114865
+ * Invalid: `| ---:| ` with space before pipe
114866
+ * Valid: `| ---:|` (no space before closing pipe)
114867
+ *
114868
+ * This runs before remark-parse to ensure the table is recognized as a valid GFM table.
114869
+ */
114870
+ /**
114871
+ * Pattern to match a table separator row.
114872
+ * A separator row consists of cells that contain only dashes, colons (for alignment), and spaces.
114873
+ *
114874
+ * Valid GFM separator formats:
114875
+ * - `---` or `----` (no alignment, defaults to left)
114876
+ * - `:---` (left aligned)
114877
+ * - `---:` (right aligned)
114878
+ * - `:---:` (center aligned)
114879
+ *
114880
+ * Invalid formats this fixes:
114881
+ * - `|: ---` → `| :---` (colon wrongly placed after pipe)
114882
+ * - `|:---` → `| :---` (colon directly after pipe, missing space)
114883
+ * - `|::---` or `| ::---` → `| :---` (double colon typo)
114884
+ */
114885
+ // Match a line that looks like a table separator row
114886
+ // This regex captures the whole line if it contains only pipe-separated cells with dashes/colons
114887
+ const TABLE_SEPARATOR_LINE_REGEX = /^(\|[:\s-]+)+\|?\s*$/;
114888
+ // Match malformed left-alignment: `|: ` or `|:` followed by dashes
114889
+ // Captures: group 1 = pipe, group 2 = spaces after colon, group 3 = dashes
114890
+ const MALFORMED_LEFT_ALIGN_REGEX = /\|:(\s*)(-+)/g;
114891
+ // Match malformed double colon: `|::---` or `| ::---` → `| :---`
114892
+ const MALFORMED_DOUBLE_COLON_REGEX = /\|\s*::(\s*)(-+)/g;
114893
+ // Match malformed patterns with spaces before closing colons: `| --- : |` → `| ---: |`
114894
+ const MALFORMED_RIGHT_ALIGN_SPACE_REGEX = /(-+)\s+:(\s*\|)/g;
114895
+ // Match malformed center alignment with spaces: `| : --- : |` → `| :---: |`
114896
+ const MALFORMED_CENTER_ALIGN_REGEX = /\|:(\s+)(-+)(\s+):/g;
114897
+ /**
114898
+ * Normalizes a single table separator line.
114899
+ */
114900
+ function normalizeTableSeparatorLine(line) {
114901
+ // Check if this line looks like a table separator
114902
+ if (!TABLE_SEPARATOR_LINE_REGEX.test(line)) {
114903
+ return line;
114904
+ }
114905
+ let normalized = line;
114906
+ // Fix `|::---` → `| :---` (double colon typo)
114907
+ // Must run before single colon fix to avoid partial replacement
114908
+ normalized = normalized.replace(MALFORMED_DOUBLE_COLON_REGEX, '| :$2');
114909
+ // Fix `|: ---` or `|:---` → `| :---`
114910
+ // The colon should be adjacent to the dashes, not the pipe
114911
+ normalized = normalized.replace(MALFORMED_LEFT_ALIGN_REGEX, '| :$2');
114912
+ // Fix `| --- : |` → `| ---: |`
114913
+ // Remove space before right-alignment colon
114914
+ normalized = normalized.replace(MALFORMED_RIGHT_ALIGN_SPACE_REGEX, '$1:$2');
114915
+ // Fix `| : --- : |` → `| :---: |`
114916
+ // Remove spaces around center-aligned dashes
114917
+ normalized = normalized.replace(MALFORMED_CENTER_ALIGN_REGEX, '| :$2:');
114918
+ return normalized;
114919
+ }
114920
+ /**
114921
+ * Preprocesses markdown content to normalize malformed table separator syntax.
114922
+ *
114923
+ * @param content - The raw markdown content
114924
+ * @returns The content with normalized table separators
114925
+ */
114926
+ function normalizeTableSeparator(content) {
114927
+ const lines = content.split('\n');
114928
+ const normalizedLines = lines.map((line, index) => {
114929
+ const prevLine = index > 0 ? lines[index - 1] : '';
114930
+ const isPrevLineTableRow = prevLine.trim().startsWith('|');
114931
+ if (isPrevLineTableRow) {
114932
+ return normalizeTableSeparatorLine(line);
114933
+ }
114934
+ return line;
114935
+ });
114936
+ return normalizedLines.join('\n');
114937
+ }
114938
+ /* harmony default export */ const normalize_table_separator = ((/* unused pure expression or super */ null && (normalizeTableSeparator)));
114939
+
114621
114940
  ;// ./processor/transform/mdxish/restore-snake-case-component-name.ts
114622
114941
 
114623
114942
 
@@ -114881,6 +115200,7 @@ function loadComponents() {
114881
115200
 
114882
115201
 
114883
115202
 
115203
+
114884
115204
 
114885
115205
 
114886
115206
  const defaultTransformers = [callouts, code_tabs, gemoji_, transform_embeds];
@@ -114893,9 +115213,11 @@ function mdxishAstProcessor(mdContent, opts = {}) {
114893
115213
  // Preprocessing pipeline: Transform content to be parser-ready
114894
115214
  // Step 1: Extract legacy magic blocks
114895
115215
  const { replaced: contentAfterMagicBlocks, blocks } = extractMagicBlocks(mdContent);
114896
- // Step 2: Evaluate JSX expressions in attributes
114897
- const contentAfterJSXEvaluation = preprocessJSXExpressions(contentAfterMagicBlocks, jsxContext);
114898
- // Step 3: Replace snake_case component names with parser-safe placeholders
115216
+ // Step 2: Normalize malformed table separator syntax (e.g., `|: ---` → `| :---`)
115217
+ const contentAfterTableNormalization = normalizeTableSeparator(contentAfterMagicBlocks);
115218
+ // Step 3: Evaluate JSX expressions in attributes
115219
+ const contentAfterJSXEvaluation = preprocessJSXExpressions(contentAfterTableNormalization, jsxContext);
115220
+ // Step 4: Replace snake_case component names with parser-safe placeholders
114899
115221
  // (e.g., <Snake_case /> → <MDXishSnakeCase0 /> which will be restored after parsing)
114900
115222
  const { content: parserReadyContent, mapping: snakeCaseMapping } = processSnakeCaseComponent(contentAfterJSXEvaluation);
114901
115223
  // Create string map for tailwind transformer
@@ -115159,6 +115481,40 @@ const makeUseMDXComponents = (more = {}) => {
115159
115481
 
115160
115482
 
115161
115483
 
115484
+
115485
+ /**
115486
+ * Parse JSON-marked string values in props back to their original types.
115487
+ * This handles arrays and objects that were serialized during JSX preprocessing.
115488
+ */
115489
+ function parseJsonProps(props) {
115490
+ if (!props)
115491
+ return props;
115492
+ const parsed = {};
115493
+ Object.entries(props).forEach(([key, value]) => {
115494
+ if (typeof value === 'string' && value.startsWith(JSON_VALUE_MARKER)) {
115495
+ try {
115496
+ parsed[key] = JSON.parse(value.slice(JSON_VALUE_MARKER.length));
115497
+ }
115498
+ catch {
115499
+ // If parsing fails, use the value without the marker
115500
+ parsed[key] = value.slice(JSON_VALUE_MARKER.length);
115501
+ }
115502
+ }
115503
+ else {
115504
+ parsed[key] = value;
115505
+ }
115506
+ });
115507
+ return parsed;
115508
+ }
115509
+ /**
115510
+ * Custom createElement wrapper that parses JSON-marked string props.
115511
+ * This is needed because rehype-react converts HAST to React, but complex
115512
+ * types (arrays/objects) get serialized to strings during markdown parsing.
115513
+ */
115514
+ function createElementWithJsonProps(type, props, ...children) {
115515
+ const parsedProps = parseJsonProps(props);
115516
+ return external_react_default().createElement(type, parsedProps, ...children);
115517
+ }
115162
115518
  /** Flatten CustomComponents into a component map for rehype-react */
115163
115519
  function exportComponentsForRehype(components) {
115164
115520
  const exported = Object.entries(components).reduce((memo, [tag, mod]) => {
@@ -115190,7 +115546,7 @@ function exportComponentsForRehype(components) {
115190
115546
  function createRehypeReactProcessor(components) {
115191
115547
  // @ts-expect-error - rehype-react types are incompatible with React.Fragment return type
115192
115548
  return unified().use((rehype_react_default()), {
115193
- createElement: (external_react_default()).createElement,
115549
+ createElement: createElementWithJsonProps,
115194
115550
  Fragment: (external_react_default()).Fragment,
115195
115551
  components,
115196
115552
  });
@@ -115499,89 +115855,6 @@ async function stripComments(doc, { mdx } = {}) {
115499
115855
  }
115500
115856
  /* harmony default export */ const lib_stripComments = (stripComments);
115501
115857
 
115502
- ;// ./lib/utils/isPlainText.ts
115503
-
115504
- /**
115505
- * Detects if content contains HTML, magic blocks, or MDX syntax.
115506
- *
115507
- * We can use this in some pipelines to determine if we should have to parse content through
115508
- * `.plain() or if it is already plain text and it should be able to detect everything that would
115509
- * be stripped or sanitized by `.plain()`.
115510
- *
115511
- */
115512
- function isPlainText(content) {
115513
- if (!content || typeof content !== 'string') {
115514
- return true;
115515
- }
115516
- // Exclude markdown code blocks and inline code to avoid false positives
115517
- // Match code blocks with optional language identifier: ```lang\n...\n```
115518
- const codeBlockRegex = /```[^\n]*\n[\s\S]*?```/g;
115519
- // Match inline code: `code` (but not escaped backticks)
115520
- const inlineCodeRegex = /`[^`\n]+`/g;
115521
- // Remove code blocks and inline code to avoid false positives
115522
- let contentWithoutCode = structuredClone(content);
115523
- contentWithoutCode = contentWithoutCode.replace(codeBlockRegex, '');
115524
- contentWithoutCode = contentWithoutCode.replace(inlineCodeRegex, '');
115525
- // Check for magic blocks: `[block:TYPE]...[/block]`
115526
- // Only check after removing code blocks to avoid detecting magic blocks in code
115527
- if (contentWithoutCode.match(MAGIC_BLOCK_REGEX) !== null) {
115528
- return false;
115529
- }
115530
- // Check for markdown links: [text](url) or [text][reference]
115531
- // Pattern matches inline links and reference-style links
115532
- // Exclude images which start with ! before the bracket
115533
- // Only check after removing code blocks
115534
- const markdownLinkPattern = /(?<!!)\[([^\]]+)\]\(([^)]+)\)|(?<!!)\[([^\]]+)\]\[([^\]]*)\]/;
115535
- if (markdownLinkPattern.test(contentWithoutCode)) {
115536
- return false;
115537
- }
115538
- // Check for JSX elements (PascalCase components) in the original content
115539
- // This includes code blocks since JSX code examples should be detected
115540
- // Pattern matches:
115541
- // - Self-closing: <Component /> or <Component/>
115542
- // - With attributes: <Component prop="value" />
115543
- // - With children: <Component>...</Component>
115544
- // Use simpler, safer patterns to avoid ReDoS from backtracking
115545
- // Match self-closing tags with bounded attribute length to prevent excessive backtracking
115546
- const jsxSelfClosingPattern = /<[A-Z][a-zA-Z0-9]*(?:\s[^>]{0,50})?\/>/;
115547
- if (jsxSelfClosingPattern.test(content)) {
115548
- return false;
115549
- }
115550
- // For components with children, use a safer pattern that limits backtracking
115551
- // Match opening tag with bounded attributes, then look for closing tag with same name
115552
- const jsxWithChildrenPattern = /<([A-Z][a-zA-Z0-9]*)(?:\s[^>]{0,50})?>[\s\S]{0,50}<\/\1>/;
115553
- if (jsxWithChildrenPattern.test(content)) {
115554
- return false;
115555
- }
115556
- // Check for MDX expressions and HTML tags in the original content
115557
- // HTML/JSX/MDX in code blocks should be detected (as per test requirements)
115558
- // But exclude inline code that contains magic block patterns to avoid false positives
115559
- let contentForHtmlMdx = content;
115560
- // Find inline code blocks and check if they contain [block: pattern
115561
- // Exclude these from HTML/MDX detection to avoid false positives
115562
- const inlineCodePattern = /`[^`\n]+`/g;
115563
- let inlineCodeMatch;
115564
- inlineCodePattern.lastIndex = 0;
115565
- while ((inlineCodeMatch = inlineCodePattern.exec(content)) !== null) {
115566
- if (inlineCodeMatch[0].includes('[block:')) {
115567
- contentForHtmlMdx = contentForHtmlMdx.replace(inlineCodeMatch[0], '');
115568
- }
115569
- }
115570
- // Match simple MDX variable expressions like {variable}, {user.name}, {getValue()}, {}
115571
- // Use bounded quantifier to prevent ReDoS - limit to reasonable variable name length
115572
- // Allow empty braces {} to be detected as well
115573
- const jsxExpressionPattern = /\{[^}"]{0,50}\}/;
115574
- if (jsxExpressionPattern.test(contentForHtmlMdx)) {
115575
- return false;
115576
- }
115577
- // Match HTML tags with bounded attribute length to prevent ReDoS
115578
- const htmlTagPattern = /<[a-z][a-z0-9]*(?:\s[^>]{0,50})?(?:\/>|>)/i;
115579
- if (htmlTagPattern.test(contentForHtmlMdx)) {
115580
- return false;
115581
- }
115582
- return true;
115583
- }
115584
-
115585
115858
  ;// ./lib/index.ts
115586
115859
 
115587
115860
 
@@ -115600,7 +115873,6 @@ function isPlainText(content) {
115600
115873
 
115601
115874
 
115602
115875
 
115603
-
115604
115876
  ;// ./index.tsx
115605
115877
 
115606
115878