html-to-markdown 2.24.4 → 2.24.5

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 7aba0e152474e36acd5ca4a9abb8cd891a6edd671086048415eec13bb25a69af
-  data.tar.gz: 55c7d8340aa73d115005745d8e93894c7494588a1458f55427b765ff392f1cd6
+  metadata.gz: 4d566a6fb7b9c675389f2f4998f64ccc15e5c10585f0ec70de610052787129d7
+  data.tar.gz: 8f32d24f5ff7fe7144772315f1524463199c063a1d561fc640d5077cbea2312f
 SHA512:
-  metadata.gz: 8a591d0bbbba4cc5682cc8d51e6f851199a70065294352251b4d6b7794ce53d257ab66c873b79da9c10f15b74da7670cf77caa3ddd1eec5d93593435aa6979b7
-  data.tar.gz: b979428207c4ebc1e7a7ea0f9d4e4859f11c50558bd11f9fae2e7e61d6901538c09594e7307bfce3910f4abfa3d9a4fe230f1a61312c592c86c5c30693ed3b5b
+  metadata.gz: b318d7be65796bad2aac9a4b466a76081e5db2ff86e53725ce43cb555da96224fe6a0502b523543197676ac649b9a8474fd043e59195bda9e5aef804828b742c
+  data.tar.gz: 322c526cb10c704bc7221885990f807a6500c3949cddefa2d0e45c92e5b89e9f08705a9cb62b71d25b768d95b613aa34d5722aea4a617da71fc71725f9f7963f

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    html-to-markdown (2.24.4)
+    html-to-markdown (2.24.5)
       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 (2.24.4)
+  html-to-markdown (2.24.5)
   i18n (1.14.8) sha256=285778639134865c5e0f6269e0b818256017e8cde89993fdfcbfb64d088824a5
   json (2.18.0) sha256=b10506aee4183f5cf49e0efc48073d7b75843ce3782c68dbeb763351c08fd505
   language_server-protocol (3.17.0.5) sha256=fd1e39a51a28bf3eec959379985a72e296e9f9acfce46f6a79d31ca8760803cc

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

@@ -384,7 +384,7 @@ dependencies = [
 
 [[package]]
 name = "html-to-markdown-rb"
-version = "2.24.4"
+version = "2.24.5"
 dependencies = [
  "html-to-markdown-rs",
  "magnus",
@@ -1387,9 +1387,9 @@ dependencies = [
 
 [[package]]
 name = "zmij"
-version = "1.0.18"
+version = "1.0.19"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "1966f8ac2c1f76987d69a74d0e0f929241c10e78136434e3be70ff7f58f64214"
+checksum = "3ff05f8caa9038894637571ae6b9e29466c1f4f829d26c9b28f869a29cbe3445"
 
 [[package]]
 name = "zune-core"

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

@@ -1,6 +1,6 @@
 [package]
 name = "html-to-markdown-rb"
-version ="2.24.4"
+version ="2.24.5"
 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 = '2.24.4'
+  VERSION = '2.24.5'
 end

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

@@ -52,7 +52,7 @@ pub fn handle_subscript(
     depth: usize,
     dom_ctx: &DomContext,
 ) {
-    use crate::converter::walk_node;
+    use crate::converter::{append_inline_suffix, chomp_inline, walk_node};
 
     let Some(node) = node_handle.get(parser) else { return };
 
@@ -61,29 +61,34 @@ pub fn handle_subscript(
         _ => return,
     };
 
-    if !ctx.in_code {
-        if options.output_format == OutputFormat::Djot {
-            output.push('~');
-        } else if !options.sub_symbol.is_empty() {
-            output.push_str(&options.sub_symbol);
-        }
-    }
-
+    let mut content = String::with_capacity(32);
     let children = tag.children();
     for child_handle in children.top().iter() {
-        walk_node(child_handle, parser, output, options, ctx, depth + 1, dom_ctx);
+        walk_node(child_handle, parser, &mut content, options, ctx, depth + 1, dom_ctx);
     }
 
-    if !ctx.in_code {
+    if ctx.in_code {
+        output.push_str(&content);
+        return;
+    }
+
+    let (prefix, suffix, trimmed) = chomp_inline(&content);
+    if !trimmed.is_empty() {
+        output.push_str(prefix);
         if options.output_format == OutputFormat::Djot {
             output.push('~');
+            output.push_str(trimmed);
+            output.push('~');
         } else if !options.sub_symbol.is_empty() {
+            output.push_str(&options.sub_symbol);
+            output.push_str(trimmed);
             if options.sub_symbol.starts_with('<') && !options.sub_symbol.starts_with("</") {
                 output.push_str(&options.sub_symbol.replace('<', "</"));
             } else {
                 output.push_str(&options.sub_symbol);
             }
         }
+        append_inline_suffix(output, suffix, !trimmed.is_empty(), node_handle, parser, dom_ctx);
     }
 }
 
@@ -99,7 +104,7 @@ pub fn handle_superscript(
     depth: usize,
     dom_ctx: &DomContext,
 ) {
-    use crate::converter::walk_node;
+    use crate::converter::{append_inline_suffix, chomp_inline, walk_node};
 
     let Some(node) = node_handle.get(parser) else { return };
 
@@ -108,29 +113,34 @@ pub fn handle_superscript(
         _ => return,
     };
 
-    if !ctx.in_code {
-        if options.output_format == OutputFormat::Djot {
-            output.push('^');
-        } else if !options.sup_symbol.is_empty() {
-            output.push_str(&options.sup_symbol);
-        }
-    }
-
+    let mut content = String::with_capacity(32);
     let children = tag.children();
     for child_handle in children.top().iter() {
-        walk_node(child_handle, parser, output, options, ctx, depth + 1, dom_ctx);
+        walk_node(child_handle, parser, &mut content, options, ctx, depth + 1, dom_ctx);
     }
 
-    if !ctx.in_code {
+    if ctx.in_code {
+        output.push_str(&content);
+        return;
+    }
+
+    let (prefix, suffix, trimmed) = chomp_inline(&content);
+    if !trimmed.is_empty() {
+        output.push_str(prefix);
         if options.output_format == OutputFormat::Djot {
             output.push('^');
+            output.push_str(trimmed);
+            output.push('^');
         } else if !options.sup_symbol.is_empty() {
+            output.push_str(&options.sup_symbol);
+            output.push_str(trimmed);
             if options.sup_symbol.starts_with('<') && !options.sup_symbol.starts_with("</") {
                 output.push_str(&options.sup_symbol.replace('<', "</"));
             } else {
                 output.push_str(&options.sup_symbol);
             }
         }
+        append_inline_suffix(output, suffix, !trimmed.is_empty(), node_handle, parser, dom_ctx);
     }
 }
 

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

@@ -318,6 +318,28 @@ fn test_subscript() {
     assert_eq!(result, "H~2~O\n");
 }
 
+#[test]
+fn test_subscript_trailing_whitespace() {
+    let html = "<p><sub>hello </sub>world</p>";
+    let opts = ConversionOptions {
+        sub_symbol: "~".to_string(),
+        ..Default::default()
+    };
+    let result = convert(html, Some(opts)).unwrap();
+    assert_eq!(result, "~hello~ world\n");
+}
+
+#[test]
+fn test_subscript_leading_whitespace() {
+    let html = "<p>hello<sub> world</sub></p>";
+    let opts = ConversionOptions {
+        sub_symbol: "~".to_string(),
+        ..Default::default()
+    };
+    let result = convert(html, Some(opts)).unwrap();
+    assert_eq!(result, "hello ~world~\n");
+}
+
 #[test]
 fn test_superscript() {
     let html = "<p>x<sup>2</sup></p>";
@@ -329,6 +351,28 @@ fn test_superscript() {
     assert_eq!(result, "x^2^\n");
 }
 
+#[test]
+fn test_superscript_trailing_whitespace() {
+    let html = "<p><sup>hello </sup>world</p>";
+    let opts = ConversionOptions {
+        sup_symbol: "^".to_string(),
+        ..Default::default()
+    };
+    let result = convert(html, Some(opts)).unwrap();
+    assert_eq!(result, "^hello^ world\n");
+}
+
+#[test]
+fn test_superscript_leading_whitespace() {
+    let html = "<p>hello<sup> world</sup></p>";
+    let opts = ConversionOptions {
+        sup_symbol: "^".to_string(),
+        ..Default::default()
+    };
+    let result = convert(html, Some(opts)).unwrap();
+    assert_eq!(result, "hello ^world^\n");
+}
+
 #[test]
 fn test_autolink() {
     let html = "<p><a href=\"https://example.com\">https://example.com</a></p>";

data/rust-vendor/zmij/.cargo-checksum.json

@@ -1 +1 @@
-{"files":{".cargo_vcs_info.json":"fc4cfe0823e2b7f29e3cbf71460386760fe0e52e23fd2d654b588bd60b29a99d",".github/workflows/ci.yml":"c6a8fe8d53e0556a4ebaefae94a0b2ffa74288e1652093667b135ed37f4374f2","Cargo.lock":"d4964a22db25c234c84332212016fdab413d4c8cabde24bcd5d57f8f4b6da042","Cargo.toml":"206bff1b72ce6a6d5123d53a6a4930ec103a634efd03deec287f437fc0c54e44","Cargo.toml.orig":"a65a097cbb8779590a2243773290f7dddb2bcaa904255485428dd4c1be39272f","LICENSE-MIT":"23f18e03dc49df91622fe2a76176497404e46ced8a715d9d2b67a7446571cca3","README.md":"cb4094e229373958e8de8b88048f7f09ce41b403b14543469354ba4049075eda","benches/bench.rs":"28d23a29046c4d50c73a19fd81ad80cc97e474af6d8c30f81dcc8c9621065cca","build.rs":"13201b550236a9ff2186b5303c77341d3b0874989fdaf39613a7452e4ce57817","src/hint.rs":"7e03e3c031b7c20127517b794a4b641b3efb6ccd4c7b1aae1b7b6f28b9968052","src/lib.rs":"15a8f3dda4a51320e1da93a052f015b2a9cc720306d3967dce854feab44f96de","src/stdarch_x86.rs":"b19b4c54f0c08c92f96d18c152f796d85cc17f9e6195f3e7d5612f05f28836fb","src/tests.rs":"04ba0d4a2b0e46e0c4f294af3bc1a1c938e72b7f9d0435b2eac0848ce5d4866a","src/traits.rs":"d289ade8d66f03096349f1cf2d4e68fad792e7b089bc956410cf700d9aad67e1","tests/exhaustive.rs":"3d5c1036dcb391bfd038fcb9feb03380b467a184ed9e0463d52f35b521684029","tests/ryu_comparison.rs":"5874eaff05486484029b333a8e7104db62b2706c8e64aecd69d8ccffad0880c3","tests/test.rs":"0f5efa4354a932d767ed561ddee411a10dd71faafa202372f3c00d81b88d1381"},"package":"1966f8ac2c1f76987d69a74d0e0f929241c10e78136434e3be70ff7f58f64214"}
+{"files":{".cargo_vcs_info.json":"43d54f8649f2696fa0547ab60188ba59d1f89d875de1c168e0c36cf1841ebc75",".github/workflows/ci.yml":"c6a8fe8d53e0556a4ebaefae94a0b2ffa74288e1652093667b135ed37f4374f2","Cargo.lock":"25584c87076d00dcf3e855ab05536cc3c61bdbc8eb838582f7129fb05634389e","Cargo.toml":"51d5a637a4a00f6d6d004d7c9b68d49ad6f88082176a0d2d05b9d1c0e12f4f82","Cargo.toml.orig":"45c6ba574b101f3ab87e0a19513589c78fcb3fb45c9b4c4195a145244c251de6","LICENSE-MIT":"23f18e03dc49df91622fe2a76176497404e46ced8a715d9d2b67a7446571cca3","README.md":"21305a2d248c7375059fa0e63402dd702bc18d1801fddd63df9b5f8ba39efe51","benches/bench.rs":"28d23a29046c4d50c73a19fd81ad80cc97e474af6d8c30f81dcc8c9621065cca","build.rs":"13201b550236a9ff2186b5303c77341d3b0874989fdaf39613a7452e4ce57817","src/hint.rs":"7e03e3c031b7c20127517b794a4b641b3efb6ccd4c7b1aae1b7b6f28b9968052","src/lib.rs":"69f08c4f84b63b5fc3d772fa7c0d23094fb9d0f2a90c0c1037275863a510c628","src/stdarch_x86.rs":"b19b4c54f0c08c92f96d18c152f796d85cc17f9e6195f3e7d5612f05f28836fb","src/tests.rs":"04ba0d4a2b0e46e0c4f294af3bc1a1c938e72b7f9d0435b2eac0848ce5d4866a","src/traits.rs":"d289ade8d66f03096349f1cf2d4e68fad792e7b089bc956410cf700d9aad67e1","tests/exhaustive.rs":"3d5c1036dcb391bfd038fcb9feb03380b467a184ed9e0463d52f35b521684029","tests/ryu_comparison.rs":"5874eaff05486484029b333a8e7104db62b2706c8e64aecd69d8ccffad0880c3","tests/test.rs":"0f5efa4354a932d767ed561ddee411a10dd71faafa202372f3c00d81b88d1381"},"package":"3ff05f8caa9038894637571ae6b9e29466c1f4f829d26c9b28f869a29cbe3445"}

data/rust-vendor/zmij/.cargo_vcs_info.json

@@ -1,6 +1,6 @@
 {
   "git": {
-    "sha1": "3a9446327c05de8446d7423854bbb03d0b243968"
+    "sha1": "990ad7ec835af798afdebc4361a346775227953b"
   },
   "path_in_vcs": ""
 }

data/rust-vendor/zmij/Cargo.lock

@@ -436,7 +436,7 @@ dependencies = [
  "memchr",
  "serde",
  "serde_core",
- "zmij 1.0.17",
+ "zmij 1.0.18",
 ]
 
 [[package]]
@@ -565,13 +565,13 @@ dependencies = [
 
 [[package]]
 name = "zmij"
-version = "1.0.17"
+version = "1.0.18"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "02aae0f83f69aafc94776e879363e9771d7ecbffe2c7fbb6c14c5e00dfe88439"
+checksum = "1966f8ac2c1f76987d69a74d0e0f929241c10e78136434e3be70ff7f58f64214"
 
 [[package]]
 name = "zmij"
-version = "1.0.18"
+version = "1.0.19"
 dependencies = [
  "criterion",
  "no-panic",

data/rust-vendor/zmij/Cargo.toml

@@ -13,7 +13,7 @@
 edition = "2021"
 rust-version = "1.68"
 name = "zmij"
-version = "1.0.18"
+version = "1.0.19"
 authors = ["David Tolnay <dtolnay@gmail.com>"]
 build = "build.rs"
 exclude = ["*.png"]

data/rust-vendor/zmij/Cargo.toml.orig

@@ -1,6 +1,6 @@
 [package]
 name = "zmij"
-version = "1.0.18"
+version = "1.0.19"
 authors = ["David Tolnay <dtolnay@gmail.com>"]
 categories = ["value-formatting", "no-std", "no-std::no-alloc"]
 description = "A double-to-string conversion algorithm based on Schubfach and yy"

data/rust-vendor/zmij/README.md

@@ -11,7 +11,7 @@ numbers to decimal strings.
 This Rust implementation is a line-by-line port of Victor Zverovich's
 implementation in C++, [https://github.com/vitaut/zmij][upstream].
 
-[upstream]: https://github.com/vitaut/zmij/tree/fb4fafadd8ad318df51ee00140699062db8d5b3e
+[upstream]: https://github.com/vitaut/zmij/tree/d8cb94516d0c480d6d70851ca677beae6ba721fc
 
 ## Example
 

data/rust-vendor/zmij/src/lib.rs

@@ -41,7 +41,7 @@
 //! ![performance](https://raw.githubusercontent.com/dtolnay/zmij/master/dtoa-benchmark.png)
 
 #![no_std]
-#![doc(html_root_url = "https://docs.rs/zmij/1.0.18")]
+#![doc(html_root_url = "https://docs.rs/zmij/1.0.19")]
 #![deny(unsafe_op_in_unsafe_fn)]
 #![allow(non_camel_case_types, non_snake_case)]
 #![allow(
@@ -806,8 +806,7 @@ where
 
     // The idea of using a single shorter candidate is by Cassio Neri.
     // It is less or equal to the upper bound by construction.
-    let div10 = (upper >> BOUND_SHIFT) / UInt::from(10);
-    let shorter = div10 * UInt::from(10);
+    let shorter = (upper >> BOUND_SHIFT) / UInt::from(10) * UInt::from(10);
     if (shorter << BOUND_SHIFT) >= lower {
         return ToDecimalResult {
             sig: shorter.into() as i64,
@@ -929,30 +928,20 @@ where
 
         // Check for boundary case when rounding down to nearest 10 and
         // near-boundary case when rounding up to nearest 10.
-        if scaled_sig_mod10 == scaled_half_ulp
-            // Case where upper == ten is insufficient: 1.342178e+08f.
-            // upper == ten || upper == ten - 1
-            || ten.wrapping_sub(upper) <= 1
+        // Case where upper == ten is insufficient: 1.342178e+08f.
+        if ten.wrapping_sub(upper) <= 1 // upper == ten || upper == ten - 1
+            || scaled_sig_mod10 == scaled_half_ulp
         {
             break;
         }
 
         let round_up = upper >= ten;
-        let mut shorter = (integral.into() - digit) as i64;
+        let shorter = (integral.into() - digit) as i64;
         let longer = (integral.into() + u64::from(cmp >= 0)) as i64;
-        if cfg!(target_arch = "aarch64") {
-            // Faster version without ccmp.
-            let dec_sig =
-                hint::select_unpredictable(scaled_sig_mod10 < scaled_half_ulp, shorter, longer);
-            return ToDecimalResult {
-                sig: hint::select_unpredictable(round_up, shorter + 10, dec_sig),
-                exp: dec_exp,
-            };
-        }
-        shorter += i64::from(round_up) * 10;
-        let use_shorter = scaled_sig_mod10 <= scaled_half_ulp || round_up;
+        let dec_sig =
+            hint::select_unpredictable(scaled_sig_mod10 < scaled_half_ulp, shorter, longer);
         return ToDecimalResult {
-            sig: hint::select_unpredictable(use_shorter, shorter, longer),
+            sig: hint::select_unpredictable(round_up, shorter + 10, dec_sig),
             exp: dec_exp,
         };
     }

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: html-to-markdown
 version: !ruby/object:Gem::Version
-  version: 2.24.4
+  version: 2.24.5
 platform: ruby
 authors:
 - Na'aman Hirschfeld