fix(parse): map sub/sup tokens to correct tags (#202)

Author: farnabaz

packages/comark/src/internal/parse/token-processor.ts

@@ -20,7 +20,8 @@ const INLINE_TAG_MAP: Record<string, string> = {
   strong_open: 'strong',
   em_open: 'em',
   s_open: 'del',
-  sub_open: 'del',
+  sub_open: 'sub',
+  sup_open: 'sup',
 }
 
 interface ProcessState {

packages/comark/test/parse/sub-sup.test.ts

@@ -0,0 +1,66 @@
+import { describe, expect, it } from 'vitest'
+import type { MarkdownItPlugin } from '../../src/types'
+import { defineComarkPlugin, parse } from '../../src/parse'
+
+// Minimal inline rules that mirror what `markdown-it-sub` / `markdown-it-sup`
+// emit: `~text~` → sub_open/text/sub_close, `^text^` → sup_open/text/sup_close.
+// This lets the test pin the bug from comarkdown/comark#201 without pulling in
+// the upstream packages.
+function makeDelimiterRule(tag: 'sub' | 'sup', char: number) {
+  return (state: any, silent: boolean) => {
+    if (silent) return false
+    if (state.src.charCodeAt(state.pos) !== char) return false
+
+    const start = state.pos + 1
+    const max = state.posMax
+    let pos = start
+    while (pos < max && state.src.charCodeAt(pos) !== char) {
+      if (state.src.charCodeAt(pos) === 0x20 /* space */) return false
+      pos++
+    }
+    if (pos === start || pos >= max) return false
+
+    state.push(`${tag}_open`, tag, 1)
+    const text = state.push('text', '', 0)
+    text.content = state.src.slice(start, pos)
+    state.push(`${tag}_close`, tag, -1)
+
+    state.pos = pos + 1
+    return true
+  }
+}
+
+const markdownItSub: MarkdownItPlugin = (md) => {
+  md.inline.ruler.after('emphasis', 'sub', makeDelimiterRule('sub', 0x7e /* ~ */))
+}
+
+const markdownItSup: MarkdownItPlugin = (md) => {
+  md.inline.ruler.after('emphasis', 'sup', makeDelimiterRule('sup', 0x5e /* ^ */))
+}
+
+const subscript = defineComarkPlugin(() => ({
+  name: 'subscript',
+  markdownItPlugins: [markdownItSub],
+}))
+
+const superscript = defineComarkPlugin(() => ({
+  name: 'superscript',
+  markdownItPlugins: [markdownItSup],
+}))
+
+describe('sub/sup token mapping (regression for #201)', () => {
+  it('maps sub_open tokens to a <sub> element (not <del>)', async () => {
+    const tree = await parse('H~2~O', { plugins: [subscript()] })
+    expect(tree.nodes).toEqual([['p', {}, 'H', ['sub', {}, '2'], 'O']])
+  })
+
+  it('maps sup_open tokens to a <sup> element (not dropped)', async () => {
+    const tree = await parse('x^2^', { plugins: [superscript()] })
+    expect(tree.nodes).toEqual([['p', {}, 'x', ['sup', {}, '2']]])
+  })
+
+  it('preserves both sub and sup in the same paragraph', async () => {
+    const tree = await parse('H~2~O and x^2^', { plugins: [subscript(), superscript()] })
+    expect(tree.nodes).toEqual([['p', {}, 'H', ['sub', {}, '2'], 'O and x', ['sup', {}, '2']]])
+  })
+})