@helping-ai-workflow/md2doc 2.1.1 → 2.2.0

package/lib/md2doc.js

@@ -415,15 +415,52 @@ ${itemsHtml}
 
   marked.setOptions({ gfm: true, breaks: false, renderer });
 
-  // Pre-process non-standard inline syntax before marked parses
+  // Code-aware subscript / superscript as marked inline extensions.
+  // The old raw-text pre-pass (mdPre.replace(/~([^~]+)~/...)) ran BEFORE marked
+  // tokenised, so it rewrote ~NOT / ^XOR operators inside fenced, indented and
+  // inline code into <sub>/<sup> (96 such mangles in one RTL spec). As inline
+  // extensions marked tokenises code first, so these never fire inside code.
+  // The tokenizer also requires a single whitespace-free token (~x~ / ^x^), so
+  // spaced operator expressions (~a & ~b, a ^ b) and lone operators (2^24, ~rst)
+  // stay literal even in prose — only a genuine subscript/superscript converts.
+  marked.use({
+    extensions: [
+      {
+        name: 'subscript',
+        level: 'inline',
+        start(src) { const i = src.indexOf('~'); return i < 0 ? undefined : i; },
+        tokenizer(src) {
+          const m = /^~(?=\S)([^~\s\n]+)~/.exec(src);
+          if (m) {
+            return { type: 'subscript', raw: m[0], text: m[1],
+                     tokens: this.lexer.inlineTokens(m[1]) };
+          }
+        },
+        renderer(token) { return `<sub>${this.parser.parseInline(token.tokens)}</sub>`; },
+      },
+      {
+        name: 'superscript',
+        level: 'inline',
+        start(src) { const i = src.indexOf('^'); return i < 0 ? undefined : i; },
+        tokenizer(src) {
+          const m = /^\^(?=\S)([^^\s\n]+)\^/.exec(src);
+          if (m) {
+            return { type: 'superscript', raw: m[0], text: m[1],
+                     tokens: this.lexer.inlineTokens(m[1]) };
+          }
+        },
+        renderer(token) { return `<sup>${this.parser.parseInline(token.tokens)}</sup>`; },
+      },
+    ],
+  });
+
+  // Pre-process the remaining non-standard inline syntax before marked parses
   const escAttr = (s) => String(s)
     .replace(/&/g, '&amp;')
     .replace(/</g, '&lt;')
     .replace(/>/g, '&gt;')
     .replace(/"/g, '&quot;');
   const mdPre = md
-    .replace(/\^([^^]+)\^/g, '<sup>$1</sup>')   // ^a^  → <sup>a</sup>
-    .replace(/~([^~]+)~/g,   '<sub>$1</sub>')   // ~a~  → <sub>a</sub>
     .replace(/\[\[([^\]\n]+)\]\]/g, (_, inner) => {  // [[ref-id, §sub]] → clickable citation
       const body = inner.trim();
       const slug = body.split(',', 1)[0].trim();

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@helping-ai-workflow/md2doc",
-  "version": "2.1.1",
+  "version": "2.2.0",
   "description": "Markdown → HTML / PDF renderer with WaveDrom, Mermaid, and Graphviz support",
   "keywords": [
     "markdown",
@@ -31,7 +31,7 @@
   },
   "scripts": {
     "preinstall": "node scripts/preinstall.js",
-    "test": "node test/md2doc.test.js && node test/cli.test.js"
+    "test": "node test/md2doc.test.js && node test/cli.test.js && node test/code-operator.test.js"
   },
   "repository": {
     "type": "git",