html-to-markdown 2.24.3 → 2.24.4

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 803d11193f6053eee9f8a73696e99148b74ecc73b237b7f587dcb56f206f7805
-  data.tar.gz: f7950a3e47d0c50bf5777ef4359ee45dbc794fe7dcff3306d9b97c8fc175e05b
+  metadata.gz: 7aba0e152474e36acd5ca4a9abb8cd891a6edd671086048415eec13bb25a69af
+  data.tar.gz: 55c7d8340aa73d115005745d8e93894c7494588a1458f55427b765ff392f1cd6
 SHA512:
-  metadata.gz: '049e92eb25488dc40d9ee107cc13de347123472ed9632667d8f3958399735589688d15c7001c636153af363c80fcac86b4b8ab953a75ab622cf346bc03e237e5'
-  data.tar.gz: c5fd6a3378383efe8a9581129e238ce292b6755607aebc8228d2183fa1adee1a84247d3559e00f6143e48503752e0b9517693110063ca32e4ff95bdc9c412591
+  metadata.gz: 8a591d0bbbba4cc5682cc8d51e6f851199a70065294352251b4d6b7794ce53d257ab66c873b79da9c10f15b74da7670cf77caa3ddd1eec5d93593435aa6979b7
+  data.tar.gz: b979428207c4ebc1e7a7ea0f9d4e4859f11c50558bd11f9fae2e7e61d6901538c09594e7307bfce3910f4abfa3d9a4fe230f1a61312c592c86c5c30693ed3b5b

data/.gitignore

@@ -1 +1,3 @@
 vendor/
+.cargo/
+rust-vendor/

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    html-to-markdown (2.24.3)
+    html-to-markdown (2.24.4)
       rb_sys (>= 0.9, < 1.0)
 
 GEM
@@ -161,7 +161,7 @@ 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 (2.24.3)
+  html-to-markdown (2.24.4)
   i18n (1.14.8) sha256=285778639134865c5e0f6269e0b818256017e8cde89993fdfcbfb64d088824a5
   json (2.18.0) sha256=b10506aee4183f5cf49e0efc48073d7b75843ce3782c68dbeb763351c08fd505
   language_server-protocol (3.17.0.5) sha256=fd1e39a51a28bf3eec959379985a72e296e9f9acfce46f6a79d31ca8760803cc

data/README.md

@@ -18,7 +18,7 @@
     <img src="https://img.shields.io/maven-central/v/dev.kreuzberg/html-to-markdown?label=Java&color=007ec6" alt="Java">
   </a>
   <a href="https://pkg.go.dev/github.com/kreuzberg-dev/html-to-markdown/packages/go/v2/htmltomarkdown">
-    <img src="https://img.shields.io/badge/Go-v2.24.1-007ec6" alt="Go">
+    <img src="https://img.shields.io/badge/Go-v2.24.4-007ec6" alt="Go">
   </a>
   <a href="https://www.nuget.org/packages/KreuzbergDev.HtmlToMarkdown/">
     <img src="https://img.shields.io/nuget/v/KreuzbergDev.HtmlToMarkdown?label=C%23&color=007ec6" alt="C#">

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

@@ -384,7 +384,7 @@ dependencies = [
 
 [[package]]
 name = "html-to-markdown-rb"
-version = "2.24.3"
+version = "2.24.4"
 dependencies = [
  "html-to-markdown-rs",
  "magnus",
@@ -396,6 +396,7 @@ dependencies = [
 name = "html-to-markdown-rs"
 version = "2.23.0"
 dependencies = [
+ "ahash",
  "astral-tl",
  "base64",
  "html-escape",

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

@@ -1,6 +1,6 @@
 [package]
 name = "html-to-markdown-rb"
-version ="2.24.3"
+version ="2.24.4"
 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 = '2.24.3'
+  VERSION = '2.24.4'
 end

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

@@ -31,6 +31,7 @@ regex = "1.12"
 once_cell = "1.21"
 thiserror = "2.0"
 base64 = "0.22"
+ahash = "0.8"
 html-escape = "0.2.13"
 image = { version = "0.25", default-features = false, features = ["gif", "jpeg", "png", "bmp", "webp"], optional = true }
 html5ever = "0.36"

data/rust-vendor/html-to-markdown-rs/src/convert_api.rs

@@ -567,6 +567,7 @@ fn fast_text_only(html: &str, options: &ConversionOptions) -> Option<String> {
                 options.escape_underscores,
                 options.escape_ascii,
             )
+            .into_owned()
         } else {
             normalized.into_owned()
         };

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

@@ -149,6 +149,7 @@ fn add_list_continuation_indent(
         output.push('\n');
     }
 
-    let indent = "  ".repeat(list_depth);
-    output.push_str(&indent);
+    for _ in 0..list_depth {
+        output.push_str("  ");
+    }
 }

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

@@ -216,28 +216,40 @@ pub(crate) fn push_heading(output: &mut String, ctx: &Context, options: &Convers
             if level == 1 {
                 output.push_str(text);
                 output.push('\n');
-                output.push_str(&"=".repeat(text.len()));
+                for _ in 0..text.len() {
+                    output.push('=');
+                }
             } else if level == 2 {
                 output.push_str(text);
                 output.push('\n');
-                output.push_str(&"-".repeat(text.len()));
+                for _ in 0..text.len() {
+                    output.push('-');
+                }
             } else {
-                output.push_str(&"#".repeat(level));
+                for _ in 0..level {
+                    output.push('#');
+                }
                 output.push(' ');
                 output.push_str(text);
             }
         }
         HeadingStyle::Atx => {
-            output.push_str(&"#".repeat(level));
+            for _ in 0..level {
+                output.push('#');
+            }
             output.push(' ');
             output.push_str(text);
         }
         HeadingStyle::AtxClosed => {
-            output.push_str(&"#".repeat(level));
+            for _ in 0..level {
+                output.push('#');
+            }
             output.push(' ');
             output.push_str(text);
             output.push(' ');
-            output.push_str(&"#".repeat(level));
+            for _ in 0..level {
+                output.push('#');
+            }
         }
     }
     output.push_str(heading_suffix);
@@ -248,7 +260,11 @@ fn continuation_indent_string(list_depth: usize, _options: &ConversionOptions) -
     if list_depth == 0 {
         return None;
     }
-    Some(" ".repeat(4 * list_depth))
+    let mut indent = String::new();
+    for _ in 0..(4 * list_depth) {
+        indent.push(' ');
+    }
+    Some(indent)
 }
 
 /// Process heading with visitor callback if available.

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

@@ -107,8 +107,9 @@ fn add_list_continuation_indent(
     if needs_space && !output.ends_with(' ') && !output.ends_with('\n') {
         output.push(' ');
     }
-    let indent = " ".repeat(4 * list_depth);
-    output.push_str(&indent);
+    for _ in 0..(4 * list_depth) {
+        output.push(' ');
+    }
 }
 
 /// Check if an element is empty (has no text content).

data/rust-vendor/html-to-markdown-rs/src/converter/block/table/cell.rs

@@ -153,7 +153,7 @@ pub fn convert_table_cell(
                 options.escape_ascii,
             );
             if options.escape_misc {
-                text = escaped;
+                text = escaped.into_owned();
             } else {
                 text = escaped.replace('|', r"\|");
             }
@@ -175,5 +175,7 @@ pub fn convert_table_cell(
 
     output.push(' ');
     output.push_str(&text);
-    output.push_str(&" |".repeat(colspan));
+    for _ in 0..colspan {
+        output.push_str(" |");
+    }
 }

data/rust-vendor/html-to-markdown-rs/src/converter/block/table/layout.rs

@@ -33,7 +33,9 @@ pub(crate) fn indent_table_for_list(
     if matches!(options.list_indent_type, ListIndentType::Spaces) {
         let space_count = indent.chars().filter(|c| *c == ' ').count();
         if space_count < 4 {
-            indent.push_str(&" ".repeat(4 - space_count));
+            for _ in 0..(4 - space_count) {
+                indent.push(' ');
+            }
         }
     }
 

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

@@ -176,7 +176,9 @@ pub fn handle_legend(
             if ctx.convert_as_inline {
                 output.push_str(trimmed);
             } else {
-                let symbol = options.strong_em_symbol.to_string().repeat(2);
+                let mut symbol = String::with_capacity(2);
+                symbol.push(options.strong_em_symbol);
+                symbol.push(options.strong_em_symbol);
                 output.push_str(&symbol);
                 output.push_str(trimmed);
                 output.push_str(&symbol);
@@ -381,7 +383,9 @@ pub fn handle_optgroup(
             .map_or(Cow::Borrowed(""), |v| v.as_utf8_str());
 
         if !label.is_empty() {
-            let symbol = options.strong_em_symbol.to_string().repeat(2);
+            let mut symbol = String::with_capacity(2);
+            symbol.push(options.strong_em_symbol);
+            symbol.push(options.strong_em_symbol);
             output.push_str(&symbol);
             output.push_str(&label);
             output.push_str(&symbol);

data/rust-vendor/html-to-markdown-rs/src/converter/inline/semantic/marks.rs

@@ -70,7 +70,9 @@ pub fn handle_mark(
                 output.push_str("</mark>");
             }
             HighlightStyle::Bold => {
-                let symbol = options.strong_em_symbol.to_string().repeat(2);
+                let mut symbol = String::with_capacity(2);
+                symbol.push(options.strong_em_symbol);
+                symbol.push(options.strong_em_symbol);
                 output.push_str(&symbol);
                 let bold_ctx = Context {
                     in_strong: true,

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

@@ -157,9 +157,16 @@ pub fn add_list_continuation_indent(
     let indent_level = calculate_list_continuation_indent(list_depth);
     let indent_char = match options.list_indent_type {
         ListIndentType::Tabs => "\t",
-        ListIndentType::Spaces => &" ".repeat(options.list_indent_width),
+        ListIndentType::Spaces => {
+            for _ in 0..options.list_indent_width {
+                output.push(' ');
+            }
+            return;
+        }
     };
-    output.push_str(&indent_char.repeat(indent_level));
+    for _ in 0..indent_level {
+        output.push_str(indent_char);
+    }
 }
 
 /// Calculate the indentation string for list continuations based on depth and options.
@@ -169,10 +176,19 @@ pub fn continuation_indent_string(list_depth: usize, options: &ConversionOptions
         return None;
     }
 
-    let indent = match options.list_indent_type {
-        ListIndentType::Tabs => "\t".repeat(indent_level),
-        ListIndentType::Spaces => " ".repeat(options.list_indent_width * indent_level),
-    };
+    let mut indent = String::new();
+    match options.list_indent_type {
+        ListIndentType::Tabs => {
+            for _ in 0..indent_level {
+                indent.push('\t');
+            }
+        }
+        ListIndentType::Spaces => {
+            for _ in 0..(options.list_indent_width * indent_level) {
+                indent.push(' ');
+            }
+        }
+    }
     Some(indent)
 }
 

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

@@ -126,7 +126,9 @@ pub fn handle_summary(
                 output.push_str(trimmed);
             } else {
                 // Block mode: output with strong markers
-                let symbol = options.strong_em_symbol.to_string().repeat(2);
+                let mut symbol = String::with_capacity(2);
+                symbol.push(options.strong_em_symbol);
+                symbol.push(options.strong_em_symbol);
                 output.push_str(&symbol);
                 output.push_str(trimmed);
                 output.push_str(&symbol);

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

@@ -28,28 +28,29 @@ pub fn dedent_code_block(content: &str) -> String {
         .min()
         .unwrap_or(0);
 
-    lines
-        .iter()
-        .map(|line| {
-            if line.trim().is_empty() {
-                *line
-            } else {
-                let mut remaining = min_indent;
-                let mut cut = 0;
-                for (idx, ch) in line.char_indices() {
-                    if remaining == 0 {
-                        break;
-                    }
-                    if ch.is_whitespace() {
-                        remaining -= 1;
-                        cut = idx + ch.len_utf8();
-                    } else {
-                        break;
-                    }
+    lines.iter().fold(String::new(), |mut acc, line| {
+        if !acc.is_empty() {
+            acc.push('\n');
+        }
+        let processed = if line.trim().is_empty() {
+            *line
+        } else {
+            let mut remaining = min_indent;
+            let mut cut = 0;
+            for (idx, ch) in line.char_indices() {
+                if remaining == 0 {
+                    break;
+                }
+                if ch.is_whitespace() {
+                    remaining -= 1;
+                    cut = idx + ch.len_utf8();
+                } else {
+                    break;
                 }
-                &line[cut..]
             }
-        })
-        .collect::<Vec<_>>()
-        .join("\n")
+            &line[cut..]
+        };
+        acc.push_str(processed);
+        acc
+    })
 }

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

@@ -110,13 +110,14 @@ pub fn process_text_node(
     } else if ctx.in_table_cell {
         let escaped = if options.whitespace_mode == crate::options::WhitespaceMode::Normalized {
             let normalized_text = text::normalize_whitespace_cow(text.as_ref());
-            text::escape(
+            let escaped_result = text::escape(
                 normalized_text.as_ref(),
                 options.escape_misc,
                 options.escape_asterisks,
                 options.escape_underscores,
                 options.escape_ascii,
-            )
+            );
+            escaped_result.into_owned()
         } else {
             text::escape(
                 text.as_ref(),
@@ -125,6 +126,7 @@ pub fn process_text_node(
                 options.escape_underscores,
                 options.escape_ascii,
             )
+            .into_owned()
         };
         if options.escape_misc {
             escaped
@@ -139,6 +141,7 @@ pub fn process_text_node(
             options.escape_underscores,
             options.escape_ascii,
         )
+        .into_owned()
     } else {
         let has_double_newline = text.contains("\n\n") || text.contains("\r\n\r\n");
         let has_trailing_single_newline =

data/rust-vendor/html-to-markdown-rs/src/hocr/types.rs

@@ -3,7 +3,7 @@
 //!
 //! Complete type system for hOCR 1.2 specification elements and properties.
 
-use std::collections::HashMap;
+use ahash::AHashMap as HashMap;
 
 /// All hOCR 1.2 element types
 #[derive(Debug, Clone, Copy, PartialEq, Eq)]

data/rust-vendor/html-to-markdown-rs/src/text.rs

@@ -36,13 +36,13 @@ pub fn escape(
     escape_asterisks: bool,
     escape_underscores: bool,
     escape_ascii: bool,
-) -> String {
+) -> Cow<'_, str> {
     if text.is_empty() {
-        return String::new();
+        return Cow::Borrowed("");
     }
 
     if !escape_misc && !escape_asterisks && !escape_underscores && !escape_ascii {
-        return text.to_string();
+        return Cow::Borrowed(text);
     }
 
     if escape_ascii
@@ -83,7 +83,7 @@ pub fn escape(
             )
         })
     {
-        return text.to_string();
+        return Cow::Borrowed(text);
     }
 
     if !escape_ascii && escape_misc && !escape_asterisks && !escape_underscores {
@@ -95,7 +95,7 @@ pub fn escape(
         });
         let needs_numbered = text.as_bytes().iter().any(|b| matches!(b, b'.' | b')'));
         if !needs_misc && !needs_numbered {
-            return text.to_string();
+            return Cow::Borrowed(text);
         }
     }
 
@@ -103,7 +103,7 @@ pub fn escape(
 
     if escape_ascii {
         result = ESCAPE_ASCII_RE.replace_all(&result, r"\$1").to_string();
-        return result;
+        return Cow::Owned(result);
     }
 
     if escape_misc {
@@ -120,7 +120,7 @@ pub fn escape(
         result = result.replace('_', r"\_");
     }
 
-    result
+    Cow::Owned(result)
 }
 
 /// Extract boundary whitespace from text (chomp).

data/rust-vendor/html-to-markdown-rs/tests/issue_139_regressions.rs

@@ -10,10 +10,10 @@ fn long_multibyte_link_label_does_not_panic() {
     html.push_str("</a>");
 
     let markdown = convert(&html, Some(ConversionOptions::default())).unwrap();
-    let expected_label = format!("{}…", "a".repeat(511));
+    let expected_label = format!("{}👍", "a".repeat(511));
 
     assert!(
         markdown.contains(&format!("[{}]", expected_label)),
-        "expected truncated label to appear in markdown output; got: {markdown}"
+        "expected full label to appear in markdown output; got: {markdown}"
     );
 }

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: html-to-markdown
 version: !ruby/object:Gem::Version
-  version: 2.24.3
+  version: 2.24.4
 platform: ruby
 authors:
 - Na'aman Hirschfeld
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2026-01-31 00:00:00.000000000 Z
+date: 2026-02-01 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rb_sys