@link-assistant/web-capture 1.7.20 → 1.7.21

package/CHANGELOG.md

@@ -1,5 +1,11 @@
 # @link-assistant/web-capture
 
+## 1.7.21
+
+### Patch Changes
+
+- b8ad97d: Number consecutive top-level `<ol>`s continuously across the document (1, 2, 3, ... N) so JS and Rust HTML→Markdown converters agree. `<ol start="N">` resets the counter and is honoured by both implementations. Nested ordered lists keep their own per-list numbering.
+
 ## 1.7.20
 
 ### Patch Changes

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@link-assistant/web-capture",
-  "version": "1.7.20",
+  "version": "1.7.21",
   "description": "CLI and microservice to render web pages as HTML, Markdown, or PNG",
   "main": "src/index.js",
   "bin": {

package/src/gdocs-preprocess.js

@@ -245,6 +245,12 @@ function renderNestedListGroup($, group, classStyles) {
   const openTags = [];
   const itemOpen = [];
 
+  const topLevelStart =
+    group[0]?.tagName?.toLowerCase() === 'ol'
+      ? $(group[0]).attr('start')
+      : undefined;
+  let topLevelOpened = false;
+
   const closeItem = (level) => {
     if (itemOpen[level]) {
       html += '</li>';
@@ -261,7 +267,21 @@ function renderNestedListGroup($, group, classStyles) {
   const openList = (level, tagName) => {
     openTags[level] = tagName;
     itemOpen[level] = false;
-    html += `<${tagName}>`;
+    if (
+      level === 0 &&
+      !topLevelOpened &&
+      tagName === 'ol' &&
+      topLevelStart !== undefined &&
+      topLevelStart !== ''
+    ) {
+      html += `<ol start="${topLevelStart}">`;
+      topLevelOpened = true;
+    } else {
+      if (level === 0) {
+        topLevelOpened = true;
+      }
+      html += `<${tagName}>`;
+    }
   };
 
   for (const list of group) {

package/src/lib.js

@@ -59,6 +59,10 @@ export function convertHtmlToMarkdown(html, baseUrl) {
   //    Demoting avoids ATX heading prefix, leaving the sub-number on a clean line.
   preserveLeadingHeadingNumbering($);
 
+  // Number consecutive top-level <ol>s continuously across the document
+  // (1, 2, 3 ... N) so JS and Rust agree. <ol start="N"> resets the counter.
+  applyContinuousOrderedListNumbering($);
+
   // Remove <a> tags with no direct text content (including only whitespace or only child elements)
   $('a').each(function () {
     // Get all text nodes directly under this <a>
@@ -228,6 +232,28 @@ function preserveLeadingHeadingNumbering($) {
   });
 }
 
+// Walk every top-level <ol> in document order and assign a `start` attribute
+// so consecutive lists number continuously (1, 2, 3, ... N). An explicit
+// `start="N"` resets the running counter to N and is preserved.
+//
+// Top-level here means "not inside another <ol> or <ul>" — nested ordered
+// lists keep their own numbering and Turndown's per-list start handling.
+function applyContinuousOrderedListNumbering($) {
+  let counter = 1;
+  $('ol').each(function () {
+    if ($(this).parents('ol, ul').length > 0) {
+      return;
+    }
+    const explicitStart = parseInt($(this).attr('start') ?? '', 10);
+    if (Number.isFinite(explicitStart)) {
+      counter = explicitStart;
+    } else {
+      $(this).attr('start', String(counter));
+    }
+    counter += $(this).children('li').length;
+  });
+}
+
 function stripLeadingTextFromHeading($, $heading, prefixLen) {
   // Walk text nodes from the start of the heading and remove prefixLen characters.
   let remaining = prefixLen;
@@ -506,6 +532,9 @@ export function convertHtmlToMarkdownEnhanced(html, baseUrl, options = {}) {
   // Preserve hierarchical heading numbering (see convertHtmlToMarkdown).
   preserveLeadingHeadingNumbering($);
 
+  // Continuous numbering across consecutive top-level <ol>s (see convertHtmlToMarkdown).
+  applyContinuousOrderedListNumbering($);
+
   // Remove empty links (same logic as convertHtmlToMarkdown)
   $('a').each(function () {
     const directText = $(this)

package/tests/fixtures/ol-continuous-numbering.html

@@ -0,0 +1,12 @@
+<ol>
+  <li><strong>First</strong></li>
+  <li><strong>Second</strong></li>
+</ol>
+<p>Some prose between lists.</p>
+<ol>
+  <li><strong>Third</strong></li>
+  <li><strong>Fourth</strong></li>
+</ol>
+<ol start="13">
+  <li><strong>Top-level numbered heading.</strong></li>
+</ol>

package/tests/unit/html2md-ol-numbering.test.js

@@ -0,0 +1,20 @@
+import fs from 'fs';
+import path from 'path';
+import { fileURLToPath } from 'url';
+import { convertHtmlToMarkdown } from '../../src/lib.js';
+
+const __filename = fileURLToPath(import.meta.url);
+const __dirname = path.dirname(__filename);
+const HTML = fs.readFileSync(
+  path.join(__dirname, '..', 'fixtures', 'ol-continuous-numbering.html'),
+  'utf8'
+);
+
+it('numbers consecutive top-level ordered lists continuously', () => {
+  const md = convertHtmlToMarkdown(HTML);
+  expect(md).toMatch(/^\s*1\.\s+\*\*First\*\*/m);
+  expect(md).toMatch(/^\s*2\.\s+\*\*Second\*\*/m);
+  expect(md).toMatch(/^\s*3\.\s+\*\*Third\*\*/m);
+  expect(md).toMatch(/^\s*4\.\s+\*\*Fourth\*\*/m);
+  expect(md).toMatch(/^\s*13\.\s+\*\*Top-level/m);
+});