html-to-markdown 3.0.2 → 3.2.0

data/vendor/html-to-markdown-rs/src/visitor_helpers.rs

@@ -58,7 +58,7 @@ use crate::visitor::{HtmlVisitor, NodeContext, NodeType, VisitResult};
 ///
 /// # Examples
 ///
-/// ```ignore
+/// ```text
 /// let ctx = build_node_context(
 ///     NodeType::Heading,
 ///     "h1",
@@ -129,7 +129,7 @@ pub fn build_node_context(
 ///
 /// # Examples
 ///
-/// ```ignore
+/// ```text
 /// let result = dispatch_visitor(
 ///     &visitor,
 ///     |v| v.visit_heading(&ctx, level, text, id),
@@ -252,7 +252,7 @@ impl VisitorDispatch {
 ///
 /// # Syntax
 ///
-/// ```ignore
+/// ```text
 /// try_visitor!(visitor_option, method_name, ctx, arg1, arg2, ...);
 /// ```
 ///
@@ -264,7 +264,7 @@ impl VisitorDispatch {
 ///
 /// # Examples
 ///
-/// ```ignore
+/// ```text
 /// // Before (verbose):
 /// let dispatch = dispatch_visitor(&visitor, |v| v.visit_heading(&ctx, level, text, id))?;
 /// match dispatch {
@@ -311,13 +311,13 @@ macro_rules! try_visitor {
 ///
 /// # Syntax
 ///
-/// ```ignore
+/// ```text
 /// try_visitor_element_start!(visitor_option, ctx);
 /// ```
 ///
 /// # Examples
 ///
-/// ```ignore
+/// ```text
 /// fn process_heading(...) -> Result<String> {
 ///     let ctx = build_node_context(...);
 ///     try_visitor_element_start!(visitor, &ctx)?;
@@ -340,13 +340,13 @@ macro_rules! try_visitor_element_start {
 ///
 /// # Syntax
 ///
-/// ```ignore
+/// ```text
 /// try_visitor_element_end!(visitor_option, ctx, default_output_string);
 /// ```
 ///
 /// # Examples
 ///
-/// ```ignore
+/// ```text
 /// fn process_heading(...) -> Result<String> {
 ///     let ctx = build_node_context(...);
 ///     let mut output = String::from("# Heading");

data/vendor/html-to-markdown-rs/tests/commonmark_compliance_test.rs

@@ -54,6 +54,12 @@ fn test_commonmark_compliance() {
             continue;
         }
 
+        // Example 231: expects indented code blocks but we default to backtick-fenced
+        if test.example == 231 {
+            skipped += 1;
+            continue;
+        }
+
         if test.section == "Tabs" {
             skipped += 1;
             continue;

data/vendor/html-to-markdown-rs/tests/integration_test.rs

@@ -55,7 +55,7 @@ fn test_inline_code() {
 fn test_code_block() {
     let html = "<pre><code>fn main() {\n    println!(\"Hello\");\n}</code></pre>";
     let result = convert(html, None).unwrap();
-    assert_eq!(result, "    fn main() {\n        println!(\"Hello\");\n    }\n");
+    assert_eq!(result, "```\nfn main() {\n    println!(\"Hello\");\n}\n```\n");
 }
 
 #[test]
@@ -141,8 +141,8 @@ fn test_nested_lists() {
     let html = "<ul><li>Item 1<ul><li>Nested 1</li><li>Nested 2</li></ul></li><li>Item 2</li></ul>";
     let result = convert(html, None).unwrap();
     assert!(result.contains("- Item 1"));
-    assert!(result.contains("  - Nested 1"));
-    assert!(result.contains("  - Nested 2"));
+    assert!(result.contains("Nested 1"), "Expected 'Nested 1' in result: {result:?}");
+    assert!(result.contains("Nested 2"), "Expected 'Nested 2' in result: {result:?}");
 }
 
 #[test]
@@ -591,6 +591,30 @@ fn q_element_produces_quotes() {
     assert!(result.contains(r#""hello""#), "q element should add quotes: {result}");
 }
 
+#[test]
+fn test_wikipedia_back_reference_caret_normalized() {
+    // Wikipedia back-references use <a href="#cite_ref-N">^</a>
+    // The caret should be normalized to ↑ to avoid confusion with markdown footnote syntax
+    let html = r##"<p>Some text<sup><a href="#cite_ref-1">^</a></sup> more text</p>"##;
+    let result = convert(html, None).unwrap();
+    assert!(
+        result.contains("[↑](#cite_ref-1)"),
+        "Back-reference caret should be normalized to ↑: {result}"
+    );
+    assert!(
+        !result.contains("[^]"),
+        "Should not produce [^] which looks like footnote syntax: {result}"
+    );
+}
+
+#[test]
+fn test_regular_caret_link_not_affected() {
+    // Regular links with ^ text but no # href should keep the ^
+    let html = r#"<a href="https://example.com">^</a>"#;
+    let result = convert(html, None).unwrap();
+    assert!(result.contains("[^]"), "Non-anchor caret links should keep ^: {result}");
+}
+
 fn convert(
     html: &str,
     opts: Option<html_to_markdown_rs::ConversionOptions>,

data/vendor/html-to-markdown-rs/tests/issue_140_regressions.rs

@@ -44,10 +44,16 @@ fn converts_should_not_escape_in_pre_or_code_fixture() {
     let pre_html = r"<pre>This pipe | should not be escaped.<pre/>";
 
     let pre_markdown_without_misc = convert(pre_html, Some(default_options())).expect("conversion should succeed");
-    assert_eq!(pre_markdown_without_misc.trim(), "This pipe | should not be escaped.");
+    assert_eq!(
+        pre_markdown_without_misc.trim(),
+        "```\nThis pipe | should not be escaped.\n```"
+    );
 
     let pre_markdown_with_misc = convert(pre_html, Some(escape_misc_options())).expect("conversion should succeed");
-    assert_eq!(pre_markdown_with_misc.trim(), "This pipe | should not be escaped.");
+    assert_eq!(
+        pre_markdown_with_misc.trim(),
+        "```\nThis pipe | should not be escaped.\n```"
+    );
 
     let code_html = r"<code>This pipe | should not be escaped.<code/>";
 

data/vendor/html-to-markdown-rs/tests/lists_test.rs

@@ -51,8 +51,8 @@ fn test_nested_lists() {
 
     let result = convert(html, None).unwrap();
     assert!(result.contains("- Item 1"));
-    assert!(result.contains("- Nested 1"));
-    assert!(result.contains("- Nested 2"));
+    assert!(result.contains("* Nested 1"));
+    assert!(result.contains("* Nested 2"));
     assert!(result.contains("- Item 2"));
 }
 
@@ -129,7 +129,7 @@ fn test_list_indent_spaces() {
 
     let result = convert(html, Some(options)).unwrap();
     assert!(result.contains("- Parent"));
-    assert!(result.contains("  - Child"));
+    assert!(result.contains("  * Child"));
 }
 
 #[test]
@@ -149,7 +149,7 @@ fn test_list_indent_tabs() {
 
     let result = convert(html, Some(options)).unwrap();
     assert!(result.contains("- Parent"));
-    assert!(result.contains("\t- Child"));
+    assert!(result.contains("\t* Child"));
 }
 
 #[test]

data/vendor/html-to-markdown-rs/tests/reference_links_test.rs

@@ -0,0 +1,169 @@
+#![allow(missing_docs)]
+
+use html_to_markdown_rs::{ConversionOptions, LinkStyle};
+
+fn convert(html: &str, options: Option<ConversionOptions>) -> String {
+    html_to_markdown_rs::convert(html, options)
+        .unwrap()
+        .content
+        .unwrap_or_default()
+}
+
+fn ref_options() -> ConversionOptions {
+    ConversionOptions {
+        link_style: LinkStyle::Reference,
+        ..Default::default()
+    }
+}
+
+#[test]
+fn basic_reference_link() {
+    let html = r#"<a href="https://example.com">Click here</a>"#;
+    let result = convert(html, Some(ref_options()));
+    assert!(
+        result.contains("[Click here][1]"),
+        "Expected reference-style link, got: {result}"
+    );
+    assert!(
+        result.contains("[1]: https://example.com"),
+        "Expected reference definition, got: {result}"
+    );
+}
+
+#[test]
+fn reference_link_with_title() {
+    let html = r#"<a href="https://example.com" title="Example">Click</a>"#;
+    let result = convert(html, Some(ref_options()));
+    assert!(
+        result.contains("[Click][1]"),
+        "Expected reference-style link, got: {result}"
+    );
+    assert!(
+        result.contains(r#"[1]: https://example.com "Example""#),
+        "Expected reference definition with title, got: {result}"
+    );
+}
+
+#[test]
+fn url_deduplication() {
+    let html = r#"<a href="https://example.com">First</a> <a href="https://example.com">Second</a>"#;
+    let result = convert(html, Some(ref_options()));
+    assert!(
+        result.contains("[First][1]"),
+        "Expected first link with ref 1, got: {result}"
+    );
+    assert!(
+        result.contains("[Second][1]"),
+        "Expected second link reusing ref 1, got: {result}"
+    );
+    // Should only have one definition
+    let count = result.matches("[1]: https://example.com").count();
+    assert_eq!(count, 1, "Expected exactly one definition, got: {result}");
+}
+
+#[test]
+fn different_titles_different_refs() {
+    let html =
+        r#"<a href="https://example.com" title="A">First</a> <a href="https://example.com" title="B">Second</a>"#;
+    let result = convert(html, Some(ref_options()));
+    assert!(
+        result.contains("[First][1]"),
+        "Expected first link ref 1, got: {result}"
+    );
+    assert!(
+        result.contains("[Second][2]"),
+        "Expected second link ref 2 (different title), got: {result}"
+    );
+}
+
+#[test]
+fn image_reference_style() {
+    let html = r#"<img src="https://example.com/img.png" alt="A photo">"#;
+    let result = convert(html, Some(ref_options()));
+    assert!(
+        result.contains("![A photo][1]"),
+        "Expected reference-style image, got: {result}"
+    );
+    assert!(
+        result.contains("[1]: https://example.com/img.png"),
+        "Expected image reference definition, got: {result}"
+    );
+}
+
+#[test]
+fn mixed_links_and_images_share_numbering() {
+    let html = r#"<a href="https://a.com">Link</a><img src="https://b.com/img.png" alt="Img">"#;
+    let result = convert(html, Some(ref_options()));
+    assert!(result.contains("[Link][1]"), "Expected link as ref 1, got: {result}");
+    assert!(result.contains("![Img][2]"), "Expected image as ref 2, got: {result}");
+}
+
+#[test]
+fn autolinks_unaffected() {
+    let html = r#"<a href="https://example.com">https://example.com</a>"#;
+    let options = ConversionOptions {
+        link_style: LinkStyle::Reference,
+        autolinks: true,
+        ..Default::default()
+    };
+    let result = convert(html, Some(options));
+    // Autolinks should still render as <url>
+    assert!(
+        result.contains("<https://example.com>"),
+        "Autolinks should not be affected by reference style, got: {result}"
+    );
+}
+
+#[test]
+fn default_inline_unchanged() {
+    let html = r#"<a href="https://example.com">Click</a>"#;
+    let result = convert(html, None);
+    assert!(
+        result.contains("[Click](https://example.com)"),
+        "Default should use inline style, got: {result}"
+    );
+}
+
+#[test]
+fn multiple_paragraphs_references_at_end() {
+    let html = r#"<p><a href="https://a.com">A</a></p><p><a href="https://b.com">B</a></p>"#;
+    let result = convert(html, Some(ref_options()));
+    // References should be at the very end
+    let ref_section_start = result.find("[1]:").expect("Should have ref section");
+    let content_end = result.find("[A][1]").expect("Should have inline ref");
+    assert!(
+        ref_section_start > content_end,
+        "Reference section should be after content"
+    );
+}
+
+#[test]
+fn empty_href_no_reference() {
+    let html = r#"<a href="">Empty</a>"#;
+    let result = convert(html, Some(ref_options()));
+    // Empty href should not create a reference
+    assert!(
+        !result.contains("[1]:"),
+        "Empty href should not create reference, got: {result}"
+    );
+}
+
+#[test]
+fn title_with_quotes_escaped() {
+    let html = r#"<a href="https://example.com" title='Say "hello"'>Link</a>"#;
+    let result = convert(html, Some(ref_options()));
+    assert!(
+        result.contains(r#"[1]: https://example.com "Say \"hello\"""#),
+        "Quotes in title should be escaped, got: {result}"
+    );
+}
+
+#[test]
+fn media_elements_reference_style() {
+    let html = r#"<video src="https://example.com/video.mp4"></video>"#;
+    let result = convert(html, Some(ref_options()));
+    assert!(
+        result.contains("[1]: https://example.com/video.mp4"),
+        "Video should use reference style, got: {result}"
+    );
+}

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: html-to-markdown
 version: !ruby/object:Gem::Version
-  version: 3.0.2
+  version: 3.2.0
 platform: ruby
 authors:
 - Na'aman Hirschfeld
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2026-04-01 00:00:00.000000000 Z
+date: 2026-04-16 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rb_sys
@@ -38,7 +38,7 @@ email:
 executables:
 - html-to-markdown
 extensions:
-- ext/html-to-markdown-rb/extconf.rb
+- ext/html_to_markdown_rb/extconf.rb
 extra_rdoc_files:
 - README.md
 files:
@@ -51,33 +51,26 @@ files:
 - Rakefile
 - Steepfile
 - exe/html-to-markdown
-- ext/html-to-markdown-rb/extconf.rb
-- ext/html-to-markdown-rb/native/Cargo.lock
-- ext/html-to-markdown-rb/native/Cargo.toml
-- ext/html-to-markdown-rb/native/README.md
-- ext/html-to-markdown-rb/native/extconf.rb
-- ext/html-to-markdown-rb/native/src/conversion/inline_images.rs
-- ext/html-to-markdown-rb/native/src/conversion/metadata.rs
-- ext/html-to-markdown-rb/native/src/conversion/mod.rs
-- ext/html-to-markdown-rb/native/src/lib.rs
-- ext/html-to-markdown-rb/native/src/options.rs
-- ext/html-to-markdown-rb/native/src/types.rs
+- ext/html-to-markdown_rb/Cargo.toml
+- ext/html_to_markdown_rb/Cargo.toml
+- ext/html_to_markdown_rb/extconf.rb
+- ext/html_to_markdown_rb/src/html_to_markdown_rs.rb
+- ext/html_to_markdown_rb/src/html_to_markdown_rs/version.rb
+- ext/html_to_markdown_rb/src/lib.rs
 - html-to-markdown-rb.gemspec
 - lib/html_to_markdown.rb
-- lib/html_to_markdown/cli.rb
-- lib/html_to_markdown/cli_proxy.rb
 - lib/html_to_markdown/version.rb
+- lib/html_to_markdown_rs.rb
 - sig/html_to_markdown.rbs
 - sig/html_to_markdown/cli.rbs
 - sig/html_to_markdown/cli_proxy.rbs
 - sig/open3.rbs
-- spec/cli_proxy_spec.rb
-- spec/spec_helper.rb
 - vendor/Cargo.toml
 - vendor/html-to-markdown-rs/Cargo.toml
 - vendor/html-to-markdown-rs/README.md
 - vendor/html-to-markdown-rs/examples/basic.rs
 - vendor/html-to-markdown-rs/examples/table.rs
+- vendor/html-to-markdown-rs/examples/test_deser.rs
 - vendor/html-to-markdown-rs/examples/test_escape.rs
 - vendor/html-to-markdown-rs/examples/test_inline_formatting.rs
 - vendor/html-to-markdown-rs/examples/test_lists.rs
@@ -142,6 +135,7 @@ files:
 - vendor/html-to-markdown-rs/src/converter/mod.rs
 - vendor/html-to-markdown-rs/src/converter/plain_text.rs
 - vendor/html-to-markdown-rs/src/converter/preprocessing_helpers.rs
+- vendor/html-to-markdown-rs/src/converter/reference_collector.rs
 - vendor/html-to-markdown-rs/src/converter/semantic/attributes.rs
 - vendor/html-to-markdown-rs/src/converter/semantic/definition_list.rs
 - vendor/html-to-markdown-rs/src/converter/semantic/figure.rs
@@ -224,6 +218,7 @@ files:
 - vendor/html-to-markdown-rs/tests/lists_test.rs
 - vendor/html-to-markdown-rs/tests/plain_output_test.rs
 - vendor/html-to-markdown-rs/tests/preprocessing_tests.rs
+- vendor/html-to-markdown-rs/tests/reference_links_test.rs
 - vendor/html-to-markdown-rs/tests/skip_images_test.rs
 - vendor/html-to-markdown-rs/tests/tables_test.rs
 - vendor/html-to-markdown-rs/tests/test_custom_elements.rs

data/ext/html-to-markdown-rb/extconf.rb

@@ -1,41 +0,0 @@
-# frozen_string_literal: true
-
-require 'mkmf'
-require 'rb_sys/mkmf'
-require 'rbconfig'
-require 'pathname'
-
-if RbConfig::CONFIG['host_os'] =~ /mswin|mingw/
-  devkit = ENV.fetch('RI_DEVKIT', nil)
-  prefix = ENV['MSYSTEM_PREFIX'] || '/ucrt64'
-
-  if devkit
-    sysroot = "#{devkit}#{prefix}".tr('\\\\', '/')
-    extra_args = [
-      '--target=x86_64-pc-windows-gnu',
-      "--sysroot=#{sysroot}"
-    ]
-
-    existing = ENV['BINDGEN_EXTRA_CLANG_ARGS'].to_s.split(/\s+/)
-    ENV['BINDGEN_EXTRA_CLANG_ARGS'] = (existing + extra_args).uniq.join(' ')
-  end
-end
-
-default_profile = ENV.fetch('CARGO_PROFILE', 'release')
-
-create_rust_makefile('html_to_markdown_rb') do |config|
-  config.profile = default_profile.to_sym
-  features_env = ENV.fetch('HTML_TO_MARKDOWN_CARGO_FEATURES', '')
-  features = features_env.split(',').map(&:strip).reject(&:empty?)
-  config.features = features unless features.empty?
-
-  native_dir = File.expand_path('native', __dir__)
-  relative_native =
-    begin
-      Pathname.new(native_dir).relative_path_from(Pathname.new(__dir__)).to_s
-    rescue ArgumentError
-      native_dir
-    end
-
-  config.ext_dir = relative_native
-end