html-to-markdown 2.27.1 → 2.27.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 41d0b097b3f46c377ddadf6da05274a283efeddf1142795a1766e908d7c78290
-  data.tar.gz: 2e6585fb07a8e8cf3fc1f474cbadf6b2a3135f00ef4707029d3e35dce02726c5
+  metadata.gz: 4eaa23699bd7eae0f731da14c7c8a651f65734ae05d7f0e62b855ae885ee8bcb
+  data.tar.gz: 75e5050d21661d008d8ca8ee9d912666305b198f6ae55231c8b0394dab3a46a4
 SHA512:
-  metadata.gz: df64f0ec15405f15043aa6594aa0560885dfa9c957fcf40ec5b4de35457f06442efd0cd3bb868d06a2a53c67a3a096719ce026e260173114857f94ab5c6249cb
-  data.tar.gz: cbaa15dee930c2940b9aaf4f768d63cd95373cc4aea4c093d58a049e6612c3e533645cdbeaf9a64aca0571ad4a2995873ecb1b98a3034de83d5771439673989c
+  metadata.gz: d4bf930d4ddbaff7a94d6613f0b4f8a329d9cf9b49ed40c93bd8ca81499b2c6e4e85226d11abf91c83aa04689d3a711b003c24c886e961253de54a1c687a3acf
+  data.tar.gz: 7a9f8057810a314b6f7b363427dd8b6eff21cf6501b5899f7ea5a6b8d48710df28f8230aad58c073cbc68ad5c1969e8c7c4b4d701df17af7c20bc00141eedd56

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    html-to-markdown (2.27.1)
+    html-to-markdown (2.27.2)
       rb_sys (>= 0.9, < 1.0)
 
 GEM
@@ -172,7 +172,7 @@ CHECKSUMS
   ffi (1.17.3-x86_64-darwin) sha256=1f211811eb5cfaa25998322cdd92ab104bfbd26d1c4c08471599c511f2c00bb5
   ffi (1.17.3-x86_64-linux-gnu) sha256=3746b01f677aae7b16dc1acb7cb3cc17b3e35bdae7676a3f568153fb0e2c887f
   fileutils (1.8.0) sha256=8c6b1df54e2540bdb2f39258f08af78853aa70bad52b4d394bbc6424593c6e02
-  html-to-markdown (2.27.1)
+  html-to-markdown (2.27.2)
   i18n (1.14.8) sha256=285778639134865c5e0f6269e0b818256017e8cde89993fdfcbfb64d088824a5
   json (2.18.1) sha256=fe112755501b8d0466b5ada6cf50c8c3f41e897fa128ac5d263ec09eedc9f986
   json-schema (6.1.0) sha256=6bf70a2cfb6dfd5a06da28093fa8190f324c88eabd36a7f47097f227321dc702

data/ext/html-to-markdown-rb/native/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "html-to-markdown-rb"
-version ="2.27.1"
+version ="2.27.2"
 edition = "2024"
 authors = ["Na'aman Hirschfeld <naaman@kreuzberg.dev>"]
 license = "MIT"

data/lib/html_to_markdown/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module HtmlToMarkdown
-  VERSION = '2.27.1'
+  VERSION = '2.27.2'
 end

data/rust-vendor/html-to-markdown-rs/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "html-to-markdown-rs"
-version = "2.27.1"
+version = "2.27.2"
 edition = "2024"
 authors = ["Na'aman Hirschfeld <naaman@kreuzberg.dev>"]
 license = "MIT"

data/rust-vendor/html-to-markdown-rs/src/converter/plain_text.rs

@@ -4,9 +4,23 @@
 //! visible text content with structural whitespace, bypassing the full
 //! Markdown/Djot conversion pipeline.
 
+use std::fmt::Write;
+
 use crate::options::ConversionOptions;
 use crate::text;
 
+/// Tracks list context for proper marker emission on `<li>` elements.
+#[derive(Clone, Debug)]
+enum ListContext {
+    /// Not inside any list.
+    None,
+    /// Inside `<ul>` — each `<li>` gets a `- ` prefix.
+    Unordered,
+    /// Inside `<ol>` — each `<li>` gets a sequential `N. ` prefix.
+    /// The `next_index` is incremented after each `<li>`.
+    Ordered { next_index: u32 },
+}
+
 /// Tags whose content should be skipped entirely.
 const SKIP_TAGS: &[&str] = &["script", "style", "head", "template", "noscript", "svg", "math"];
 
@@ -49,9 +63,10 @@ const BLOCK_TAGS: &[&str] = &[
 /// - Inline elements are recursed without markers
 pub fn extract_plain_text(dom: &tl::VDom, parser: &tl::Parser, options: &ConversionOptions) -> String {
     let mut buf = String::with_capacity(1024);
+    let mut list_ctx = ListContext::None;
 
     for child_handle in dom.children() {
-        walk_plain(child_handle, parser, &mut buf, options, false);
+        walk_plain(child_handle, parser, &mut buf, options, false, &mut list_ctx);
     }
 
     post_process(&mut buf);
@@ -65,6 +80,7 @@ fn walk_plain(
     buf: &mut String,
     options: &ConversionOptions,
     in_pre: bool,
+    list_ctx: &mut ListContext,
 ) {
     let Some(node) = node_handle.get(parser) else {
         return;
@@ -105,7 +121,7 @@ fn walk_plain(
                 }
                 "pre" => {
                     ensure_blank_line(buf);
-                    walk_children(tag, parser, buf, options, true);
+                    walk_children(tag, parser, buf, options, true, list_ctx);
                     ensure_blank_line(buf);
                 }
                 "img" => {
@@ -123,19 +139,50 @@ fn walk_plain(
                     walk_table(tag, parser, buf, options);
                     ensure_blank_line(buf);
                 }
+                "ul" => {
+                    ensure_newline(buf);
+                    let mut child_ctx = ListContext::Unordered;
+                    walk_children(tag, parser, buf, options, false, &mut child_ctx);
+                    ensure_newline(buf);
+                }
+                "ol" => {
+                    let start = tag
+                        .attributes()
+                        .get("start")
+                        .flatten()
+                        .and_then(|v| v.as_utf8_str().parse::<u32>().ok())
+                        .unwrap_or(1);
+                    ensure_newline(buf);
+                    let mut child_ctx = ListContext::Ordered { next_index: start };
+                    walk_children(tag, parser, buf, options, false, &mut child_ctx);
+                    ensure_newline(buf);
+                }
                 "li" => {
                     ensure_newline(buf);
-                    walk_children(tag, parser, buf, options, false);
+                    match list_ctx {
+                        ListContext::Unordered => {
+                            buf.push_str("- ");
+                        }
+                        ListContext::Ordered { next_index } => {
+                            let _ = write!(buf, "{}. ", next_index);
+                            *next_index += 1;
+                        }
+                        ListContext::None => {
+                            // <li> outside a list — emit with bullet as fallback
+                            buf.push_str("- ");
+                        }
+                    }
+                    walk_children(tag, parser, buf, options, false, list_ctx);
                     ensure_newline(buf);
                 }
                 _ if BLOCK_TAGS.contains(&tag_str) => {
                     ensure_blank_line(buf);
-                    walk_children(tag, parser, buf, options, in_pre);
+                    walk_children(tag, parser, buf, options, in_pre, list_ctx);
                     ensure_blank_line(buf);
                 }
                 _ => {
-                    // Inline elements and structural containers (html, body, ul, ol, etc.)
-                    walk_children(tag, parser, buf, options, in_pre);
+                    // Inline elements and structural containers (html, body, etc.)
+                    walk_children(tag, parser, buf, options, in_pre, list_ctx);
                 }
             }
         }
@@ -144,11 +191,18 @@ fn walk_plain(
 }
 
 /// Walk all children of a tag.
-fn walk_children(tag: &tl::HTMLTag, parser: &tl::Parser, buf: &mut String, options: &ConversionOptions, in_pre: bool) {
+fn walk_children(
+    tag: &tl::HTMLTag,
+    parser: &tl::Parser,
+    buf: &mut String,
+    options: &ConversionOptions,
+    in_pre: bool,
+    list_ctx: &mut ListContext,
+) {
     let children = tag.children();
     let top = children.top();
     for child in top.iter() {
-        walk_plain(child, parser, buf, options, in_pre);
+        walk_plain(child, parser, buf, options, in_pre, list_ctx);
     }
 }
 
@@ -185,7 +239,8 @@ fn walk_table(table_tag: &tl::HTMLTag, parser: &tl::Parser, buf: &mut String, op
             }
             let mut cell_buf = String::new();
             if let Some(tl::Node::Tag(cell_tag)) = cell_handle.get(parser) {
-                walk_children(cell_tag, parser, &mut cell_buf, options, false);
+                let mut cell_list_ctx = ListContext::None;
+                walk_children(cell_tag, parser, &mut cell_buf, options, false, &mut cell_list_ctx);
             }
             buf.push_str(cell_buf.trim());
         }

data/rust-vendor/html-to-markdown-rs/tests/plain_output_test.rs

@@ -75,12 +75,7 @@ fn test_plain_blockquote_no_prefix() {
 fn test_plain_list_items_on_separate_lines() {
     let html = "<ul><li>First</li><li>Second</li><li>Third</li></ul>";
     let result = convert(html, Some(plain_options())).unwrap();
-    assert!(result.contains("First"));
-    assert!(result.contains("Second"));
-    assert!(result.contains("Third"));
-    // Items should be on separate lines
-    let lines: Vec<&str> = result.lines().filter(|l| !l.is_empty()).collect();
-    assert!(lines.len() >= 3, "Expected at least 3 lines, got: {result}");
+    assert_eq!(result, "- First\n- Second\n- Third\n");
 }
 
 #[test]
@@ -212,3 +207,61 @@ fn test_plain_pre_preserves_whitespace() {
         "Pre blocks should preserve whitespace, got: {result}"
     );
 }
+
+#[test]
+fn test_plain_unordered_list_markers() {
+    let html = "<ul><li>Alpha</li><li>Beta</li><li>Gamma</li></ul>";
+    let result = convert(html, Some(plain_options())).unwrap();
+    assert_eq!(result, "- Alpha\n- Beta\n- Gamma\n");
+}
+
+#[test]
+fn test_plain_ordered_list_markers() {
+    let html = "<ol><li>First</li><li>Second</li><li>Third</li></ol>";
+    let result = convert(html, Some(plain_options())).unwrap();
+    assert_eq!(result, "1. First\n2. Second\n3. Third\n");
+}
+
+#[test]
+fn test_plain_ordered_list_custom_start() {
+    let html = r#"<ol start="42"><li>First item starting at 42</li><li>Second item</li></ol>"#;
+    let result = convert(html, Some(plain_options())).unwrap();
+    assert_eq!(result, "42. First item starting at 42\n43. Second item\n");
+}
+
+#[test]
+fn test_plain_nested_lists() {
+    let html = "<ul><li>Outer 1<ul><li>Inner A</li><li>Inner B</li></ul></li><li>Outer 2</li></ul>";
+    let result = convert(html, Some(plain_options())).unwrap();
+    // The outer items should have `- ` prefix and inner items should also have `- ` prefix
+    assert!(
+        result.contains("- Outer 1"),
+        "Expected '- Outer 1' in output, got: {result}"
+    );
+    assert!(
+        result.contains("- Inner A"),
+        "Expected '- Inner A' in output, got: {result}"
+    );
+    assert!(
+        result.contains("- Inner B"),
+        "Expected '- Inner B' in output, got: {result}"
+    );
+    assert!(
+        result.contains("- Outer 2"),
+        "Expected '- Outer 2' in output, got: {result}"
+    );
+}
+
+#[test]
+fn test_plain_ordered_list_inside_unordered() {
+    let html = "<ul><li>Bullet<ol><li>Numbered</li></ol></li></ul>";
+    let result = convert(html, Some(plain_options())).unwrap();
+    assert!(
+        result.contains("- Bullet"),
+        "Expected '- Bullet' in output, got: {result}"
+    );
+    assert!(
+        result.contains("1. Numbered"),
+        "Expected '1. Numbered' in output, got: {result}"
+    );
+}

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: html-to-markdown
 version: !ruby/object:Gem::Version
-  version: 2.27.1
+  version: 2.27.2
 platform: ruby
 authors:
 - Na'aman Hirschfeld
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2026-03-01 00:00:00.000000000 Z
+date: 2026-03-02 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rb_sys