Merge pull request #15 from dplabs/fix/issue-4-property-extraction

fix: only extract properties from leading front-matter block

Author: dpecos

src/__tests__/parser.test.ts

@@ -197,7 +197,36 @@ describe('Parser', () => {
     });
 
     it('removes extracted properties from content', () => {
-      expect(parser.parse('name: a\n1')[0].content).toEqual(['\n1']);
+      expect(parser.parse('name: a\n1')[0].content).toEqual(['1']);
+    });
+
+    it('does not consume a word:value line that appears after other content (issue #4)', () => {
+      // "Example: this line disappears" appears after real content — must stay as content.
+      const slide = parser.parse('# Heading\n\nExample: this line disappears\n\nMore content.')[0];
+      expect(slide.properties['Example']).toBeUndefined();
+      expect(slide.content.join('')).toContain('Example: this line disappears');
+    });
+
+    it('does not consume a word:value line mixed in with regular content', () => {
+      const slide = parser.parse('Some intro\n\nNote: this should stay\n\nMore text')[0];
+      expect(slide.properties['Note']).toBeUndefined();
+      expect(slide.content.join('')).toContain('Note: this should stay');
+    });
+
+    it('stops extracting properties at the first non-property line', () => {
+      // "name" is a valid property but the next line is not, so class:b further down stays as content.
+      const slide = parser.parse('name: a\nNot valid prop\nclass: b')[0];
+      expect(slide.properties.name).toBe('a');
+      expect(slide.properties.class).toBeUndefined();
+      expect(slide.content.join('')).toContain('class: b');
+    });
+
+    it('stops extracting properties at a blank line separating front-matter from content', () => {
+      const slide = parser.parse('name: a\n\nclass: b')[0];
+      expect(slide.properties.name).toBe('a');
+      // class: b is after a blank line — it is content, not a property
+      expect(slide.properties.class).toBeUndefined();
+      expect(slide.content.join('')).toContain('class: b');
     });
   });
 

src/mdeck/parser.ts

@@ -114,18 +114,39 @@ function appendTo(element: ParsedSlide | ContentClass, content: ContentItem): vo
 }
 
 function extractProperties(source: string, properties: Record<string, string>): string {
-  const propertyFinder = /^\n*([-\w]+):([^$\n]*)|\n*(?:<!--\s*)([-\w]+):([^$\n]*?)(?:\s*-->)/i;
+  if (typeof source !== 'string') return source as unknown as string;
+
+  // Extract inline HTML comment properties anywhere in the source (they are invisible markers).
+  const commentFinder = /\n*(?:<!--\s*)([-\w]+):([^$\n]*?)(?:\s*-->)/gi;
   let match: RegExpExecArray | null;
-  while ((match = propertyFinder.exec(source)) !== null) {
+  while ((match = commentFinder.exec(source)) !== null) {
     source = source.slice(0, match.index) + source.slice(match.index + match[0].length);
-    if (match[1] !== undefined) {
-      properties[match[1].trim()] = match[2].trim();
-    } else {
-      properties[match[3].trim()] = match[4].trim();
-    }
-    propertyFinder.lastIndex = match.index;
+    properties[match[1].trim()] = match[2].trim();
+    commentFinder.lastIndex = match.index;
+  }
+
+  // Extract plain `key: value` properties only from the leading front-matter block.
+  // Stop as soon as a line that is not a property (or blank) is encountered so that
+  // content lines like "Example: this disappears" are never consumed as properties.
+  const lines = source.split('\n');
+  const consumed: number[] = [];
+  const propertyLine = /^([-\w]+):([^$\n]*)$/i;
+  let i = 0;
+  // Skip leading blank lines.
+  while (i < lines.length && lines[i].trim() === '') { i++; }
+  // Consume contiguous property lines.
+  while (i < lines.length) {
+    const line = lines[i];
+    if (line.trim() === '') break; // blank line ends the front-matter block
+    const m = propertyLine.exec(line);
+    if (!m) break; // non-property line ends the block
+    properties[m[1].trim()] = m[2].trim();
+    consumed.push(i);
+    i++;
   }
-  return source;
+  // Remove consumed lines from source.
+  const remaining = lines.filter((_, idx) => !consumed.includes(idx));
+  return remaining.join('\n');
 }
 
 function cleanInput(source: string): string {