Merge pull request #601 from kivikakk/push-lsmztupslltn

Replace io::Write with fmt::Write to avoid revalidating UTF-8.

Author: kivikakk

.gitignore

@@ -9,3 +9,4 @@ benches/comrak-*
 benches/pulldown-cmark
 benches/markdown-it
 /result
+/bench-output.md

Cargo.lock

@@ -45,6 +45,7 @@ emojis = { version = "0.6.2", optional = true }
 arbitrary = { version = "1", optional = true, features = ["derive"] }
 bon = { version = "3", optional = true }
 caseless = "0.2.1"
+fmt2io = { version = "1.0.0", optional = true }
 
 [dev-dependencies]
 ntest = "0.9"
@@ -54,7 +55,7 @@ toml = "0.7.3"
 
 [features]
 default = ["cli", "syntect", "bon"]
-cli = ["clap", "shell-words", "xdg"]
+cli = ["clap", "shell-words", "xdg", "fmt2io"]
 shortcodes = ["emojis"]
 bon = ["dep:bon"]
 

Cargo.toml

@@ -10,13 +10,13 @@ bench:
 	cargo build --release
 	(cd vendor/cmark-gfm/; make bench PROG=../../target/release/comrak)
 
-binaries: build-comrak-branch build-comrak-master build-cmark-gfm build-pulldown-cmark build-markdown-it
+binaries: build-comrak-branch build-comrak-main build-cmark-gfm build-pulldown-cmark build-markdown-it
 
 build-comrak-branch:
 	cargo build --release
 	cp ${ROOT}/target/release/comrak ${ROOT}/benches/comrak-${COMMIT}
 
-build-comrak-master:
+build-comrak-main:
 	git clone https://github.com/kivikakk/comrak.git --depth 1 --single-branch ${ROOT}/vendor/comrak || true
 	cd ${ROOT}/vendor/comrak && \
 	cargo build --release && \
@@ -40,11 +40,17 @@ build-pulldown-cmark:
 bench-comrak: build-comrak-branch
 	git clone https://github.com/progit/progit.git ${ROOT}/vendor/progit || true > /dev/null
 	cd benches && \
-	hyperfine --warmup 3 --min-runs ${MIN_RUNS}  -L binary comrak-${COMMIT} './bench.sh ./{binary}'
+	hyperfine --warmup 3 --min-runs ${MIN_RUNS} -L binary comrak-${COMMIT} './bench.sh ./{binary}'
+
+bench-comrak-vs-main: build-comrak-branch build-comrak-main
+	git clone https://github.com/progit/progit.git ${ROOT}/vendor/progit || true > /dev/null
+	cd benches && \
+	hyperfine --warmup 10 --min-runs ${MIN_RUNS} -L binary comrak-${COMMIT},comrak-main './bench.sh ./{binary}' --export-markdown ${ROOT}/bench-output.md &&\
+	echo "\n\nRun on" `date -u` >> ${ROOT}/bench-output.md
 
 bench-all: binaries
 	git clone https://github.com/progit/progit.git ${ROOT}/vendor/progit || true > /dev/null
 	cd benches && \
-	hyperfine --warmup 10 --min-runs ${MIN_RUNS}  -L binary comrak-${COMMIT},comrak-main,pulldown-cmark,cmark-gfm,markdown-it './bench.sh ./{binary}' --export-markdown ${ROOT}/bench-output.md &&\
+	hyperfine --warmup 10 --min-runs ${MIN_RUNS} -L binary comrak-${COMMIT},comrak-main,pulldown-cmark,cmark-gfm,markdown-it './bench.sh ./{binary}' --export-markdown ${ROOT}/bench-output.md &&\
 	echo "\n\nRun on" `date -u` >> ${ROOT}/bench-output.md
 

Makefile

@@ -16,7 +16,7 @@ fn bench_progit(b: &mut Bencher) {
     b.iter(|| {
         let arena = Arena::new();
         let root = parse_document(&arena, &s, &Options::default());
-        let mut output = vec![];
+        let mut output = String::new();
         format_html(root, &Options::default(), &mut output).unwrap()
     });
 }

benches/progit.rs

@@ -1,3 +1,10 @@
+# [v0.43.0] - unreleased
+
+* `html::format_document`, `xml::format_document`, `cm::format_document` and
+  friends now take an `std::fmt::Write` as their `output` argument, instead of
+  an `std::io::Write`.
+
+
 # [v0.42.0] - 2025-09-24
 
 New APIs:

changelog.txt

@@ -1,22 +1,22 @@
 use comrak::html::ChildRendering;
 use comrak::{create_formatter, nodes::NodeValue};
-use std::io::Write;
+use std::fmt::Write;
 
 create_formatter!(CustomFormatter, {
     NodeValue::Emph => |context, entering| {
         if entering {
-            context.write_all(b"<i>")?;
+            context.write_str("<i>")?;
         } else {
-            context.write_all(b"</i>")?;
+            context.write_str("</i>")?;
         }
     },
     NodeValue::Strong => |context, entering| {
-        context.write_all(if entering { b"<b>" } else { b"</b>" })?;
+        context.write_str(if entering { "<b>" } else { "</b>" })?;
     },
     NodeValue::Image(ref nl) => |context, node, entering| {
         assert!(node.data.borrow().sourcepos == (3, 1, 3, 18).into());
         if entering {
-            context.write_all(nl.url.to_uppercase().as_bytes())?;
+            context.write_str(&nl.url.to_uppercase())?;
         }
         return Ok(ChildRendering::Skip);
     },
@@ -33,11 +33,8 @@ fn main() {
         &options,
     );
 
-    let mut buf: Vec<u8> = vec![];
-    CustomFormatter::format_document(doc, &options, &mut buf).unwrap();
+    let mut out = String::new();
+    CustomFormatter::format_document(doc, &options, &mut out).unwrap();
 
-    assert_eq!(
-        std::str::from_utf8(&buf).unwrap(),
-        "<p><i>Hello</i>, <b>world</b>.</p>\n<p>/IMG.PNG</p>\n"
-    );
+    assert_eq!(out, "<p><i>Hello</i>, <b>world</b>.</p>\n<p>/IMG.PNG</p>\n");
 }

examples/custom_formatter.rs

@@ -34,7 +34,7 @@ fn formatter<'a>(
     context: &mut comrak::html::Context,
     node: &'a comrak::nodes::AstNode<'a>,
     entering: bool,
-) -> std::io::Result<comrak::html::ChildRendering> {
+) -> Result<comrak::html::ChildRendering, std::fmt::Error> {
     let mut borrow = node.data.borrow_mut();
     if let NodeValue::Image(ref mut nl) = borrow.value {
         autotitle_images(nl, context, node, entering);
@@ -47,7 +47,7 @@ fn main() {
     let arena = Arena::new();
     let parsed = parse_document(&arena, "![my epic image](/img.png)", &Default::default());
 
-    let mut out = Vec::new();
+    let mut out = String::new();
     comrak::html::format_document_with_formatter(
         parsed,
         &Default::default(),
@@ -58,6 +58,5 @@ fn main() {
     )
     .unwrap_or_else(|_| unreachable!("writing to Vec<u8> cannot fail"));
 
-    // SAFETY: the formatter always emits valid UTF-8
-    println!("{}", unsafe { String::from_utf8_unchecked(out) });
+    println!("{out}");
 }

examples/custom_formatter_alt_text.rs

@@ -1,24 +1,24 @@
 use comrak::html::ChildRendering;
 use comrak::{create_formatter, nodes::NodeValue};
-use std::io::Write;
+use std::fmt::Write;
 
 create_formatter!(CustomFormatter<usize>, {
     NodeValue::Emph => |context, entering| {
         context.user += 1;
         if entering {
-            context.write_all(b"<i>")?;
+            context.write_str("<i>")?;
         } else {
-            context.write_all(b"</i>")?;
+            context.write_str("</i>")?;
         }
     },
     NodeValue::Strong => |context, entering| {
         context.user += 1;
-        context.write_all(if entering { b"<b>" } else { b"</b>" })?;
+        context.write_str(if entering { "<b>" } else { "</b>" })?;
     },
     NodeValue::Image(ref nl) => |context, node, entering| {
         assert!(node.data.borrow().sourcepos == (3, 1, 3, 18).into());
         if entering {
-            context.write_all(nl.url.to_uppercase().as_bytes())?;
+            context.write_str(&nl.url.to_uppercase())?;
         }
         return Ok(ChildRendering::Skip);
     },
@@ -35,13 +35,10 @@ fn main() {
         &options,
     );
 
-    let mut buf: Vec<u8> = vec![];
-    let converted_count = CustomFormatter::format_document(doc, &options, &mut buf, 0).unwrap();
+    let mut out = String::new();
+    let converted_count = CustomFormatter::format_document(doc, &options, &mut out, 0).unwrap();
 
-    assert_eq!(
-        std::str::from_utf8(&buf).unwrap(),
-        "<p><i>Hello</i>, <b>world</b>.</p>\n<p>/IMG.PNG</p>\n"
-    );
+    assert_eq!(out, "<p><i>Hello</i>, <b>world</b>.</p>\n<p>/IMG.PNG</p>\n");
 
     assert_eq!(converted_count, 4);
 }

examples/custom_formatter_user.rs

@@ -4,7 +4,7 @@ use comrak::{
     nodes::Sourcepos,
     Options, Plugins,
 };
-use std::io::{self, Write};
+use std::fmt::{self, Write};
 
 fn main() {
     let adapter = CustomHeadingAdapter;
@@ -39,7 +39,7 @@ impl HeadingAdapter for CustomHeadingAdapter {
         output: &mut dyn Write,
         heading: &HeadingMeta,
         sourcepos: Option<Sourcepos>,
-    ) -> io::Result<()> {
+    ) -> fmt::Result {
         let id = slug::slugify(&heading.content);
 
         let search_include = !&heading.content.contains("hide");
@@ -57,7 +57,7 @@ impl HeadingAdapter for CustomHeadingAdapter {
         )
     }
 
-    fn exit(&self, output: &mut dyn Write, heading: &HeadingMeta) -> io::Result<()> {
+    fn exit(&self, output: &mut dyn Write, heading: &HeadingMeta) -> fmt::Result {
         write!(output, "</h{}>", heading.level)
     }
 }