Allow comparison

bench/build

@@ -18,6 +18,6 @@ for r in *; do
 done
 popd >/dev/null
 
-popd >/dev/null
-
 echo "All done!"
+
+popd >/dev/null

bench/run

@@ -29,6 +29,6 @@ for r in *; do
 done
 popd >/dev/null
 
-popd >/dev/null
-
 echo "All done!"
+
+popd >/dev/null

bench/runners/@minify-html%2Fjs/index.js

@@ -5,6 +5,7 @@ const path = require("path");
 const iterations = parseInt(process.env.MHB_ITERATIONS, 10);
 const inputDir = process.env.MHB_INPUT_DIR;
 const htmlOnly = process.env.MHB_HTML_ONLY === "1";
+const outputDir = process.env.MHB_OUTPUT_DIR;
 
 const minifyHtmlCfg = minifyHtml.createConfiguration({
   minify_css: !htmlOnly,
@@ -13,12 +14,19 @@ const minifyHtmlCfg = minifyHtml.createConfiguration({
 
 const results = fs.readdirSync(inputDir).map((name) => {
   const src = fs.readFileSync(path.join(inputDir, name));
+
+  const out = minifyHtml.minify(src, minifyHtmlCfg);
+  const len = out.byteLength;
+  if (outputDir) {
+    fs.writeFileSync(path.join(outputDir, name), out);
+  }
+
   const start = process.hrtime.bigint();
-  let len;
   for (let i = 0; i < iterations; i++) {
-    len = minifyHtml.minify(src, minifyHtmlCfg).byteLength;
+    minifyHtml.minify(src, minifyHtmlCfg);
   }
   const elapsed = process.hrtime.bigint() - start;
+
   return [name, len, iterations, Number(elapsed) / 1_000_000_000];
 });
 console.log(JSON.stringify(results));

bench/runners/README.md

@@ -5,5 +5,6 @@
   - `MHB_ITERATIONS`: times to run each input.
   - `MHB_INPUT_DIR`: path to directory containing inputs. Files should be read from this directory and used as the inputs.
   - `MHB_HTML_ONLY`: if set to `1`, `minify_css` and `minify_js` should be disabled.
-- The output should be a JSON array of tuples, where each tuples contains the input name, output size, iterations, and execution time in seconds (as a floating point value).
+  - `MHB_OUTPUT_DIR`: if set, output minified HTML for each input in this folder. 
+- The output should be a JSON array of tuples, where each tuples contains the input name, output UTF-8 byte length, iterations, and execution time in seconds (as a floating point value).
   - The execution time should be measured using high-precision monotonic system clocks where possible.

bench/runners/html-minifier/index.js

@@ -6,6 +6,7 @@ const path = require("path");
 const iterations = parseInt(process.env.MHB_ITERATIONS, 10);
 const inputDir = process.env.MHB_INPUT_DIR;
 const htmlOnly = process.env.MHB_HTML_ONLY === "1";
+const outputDir = process.env.MHB_OUTPUT_DIR;
 
 const esbuildCss = (code) =>
   esbuild.transformSync(code, {
@@ -47,12 +48,19 @@ const htmlMinifierCfg = {
 
 const results = fs.readdirSync(inputDir).map((name) => {
   const src = fs.readFileSync(path.join(inputDir, name), "utf8");
+
+  const out = htmlMinifier.minify(src, htmlMinifierCfg);
+  const len = Buffer.from(out, "utf8").length;
+  if (outputDir) {
+    fs.writeFileSync(path.join(outputDir, name), out);
+  }
+
   const start = process.hrtime.bigint();
-  let len;
   for (let i = 0; i < iterations; i++) {
-    len = htmlMinifier.minify(src, htmlMinifierCfg).length;
+    htmlMinifier.minify(src, htmlMinifierCfg);
   }
   const elapsed = process.hrtime.bigint() - start;
+
   return [name, len, iterations, Number(elapsed) / 1_000_000_000];
 });
 console.log(JSON.stringify(results));

bench/runners/minify-html-onepass/src/main.rs

@@ -14,8 +14,7 @@ fn main() {
         .ok()
         .filter(|v| v == "1")
         .is_some();
-
-    let tests = fs::read_dir(input_dir).unwrap().map(|d| d.unwrap());
+    let output_dir = env::var("MHB_OUTPUT_DIR").ok();
 
     let mut results: Vec<(String, usize, usize, f64)> = Vec::new();
     let cfg = Cfg {
@@ -23,21 +22,25 @@ fn main() {
         minify_js: !html_only,
     };
 
-    for t in tests {
+    for t in fs::read_dir(input_dir).unwrap().map(|d| d.unwrap()) {
         let source = fs::read(t.path()).unwrap();
+        let input_name = t.file_name().into_string().unwrap();
+
+        let mut output = source.to_vec();
+        let len = in_place(&mut output, &cfg).expect("failed to minify");
+        output.truncate(len);
+        if let Some(output_dir) = &output_dir {
+            fs::write(format!("{}/{}", output_dir, input_name), output).unwrap();
+        };
+
         let start = Instant::now();
-        let mut len = 0;
         for _ in 0..iterations {
             let mut data = source.to_vec();
-            len = in_place(&mut data, &cfg).expect("failed to minify");
+            let _ = in_place(&mut data, &cfg).expect("failed to minify");
         }
         let elapsed = start.elapsed().as_secs_f64();
-        results.push((
-            t.file_name().into_string().unwrap(),
-            len,
-            iterations,
-            elapsed,
-        ));
+
+        results.push((input_name, len, iterations, elapsed));
     }
 
     serde_json::to_writer(stdout(), &results).unwrap();

bench/runners/minify-html/src/main.rs

@@ -14,8 +14,7 @@ fn main() {
         .ok()
         .filter(|v| v == "1")
         .is_some();
-
-    let tests = fs::read_dir(input_dir).unwrap().map(|d| d.unwrap());
+    let output_dir = env::var("MHB_OUTPUT_DIR").ok();
 
     let mut results: Vec<(String, usize, usize, f64)> = Vec::new();
     let mut cfg = Cfg::new();
@@ -24,20 +23,23 @@ fn main() {
         cfg.minify_js = true;
     };
 
-    for t in tests {
+    for t in fs::read_dir(input_dir).unwrap().map(|d| d.unwrap()) {
         let source = fs::read(t.path()).unwrap();
+        let input_name = t.file_name().into_string().unwrap();
+
+        let output = minify(&source, &cfg);
+        let len = output.len();
+        if let Some(output_dir) = &output_dir {
+            fs::write(format!("{}/{}", output_dir, input_name), output).unwrap();
+        };
+
         let start = Instant::now();
-        let mut len = 0;
         for _ in 0..iterations {
-            len = minify(&source, &cfg).len();
+            let _ = minify(&source, &cfg);
         }
         let elapsed = start.elapsed().as_secs_f64();
-        results.push((
-            t.file_name().into_string().unwrap(),
-            len,
-            iterations,
-            elapsed,
-        ));
+
+        results.push((input_name, len, iterations, elapsed));
     }
 
     serde_json::to_writer(stdout(), &results).unwrap();

bench/runners/minimize/index.js

@@ -6,6 +6,7 @@ const path = require("path");
 const iterations = parseInt(process.env.MHB_ITERATIONS, 10);
 const inputDir = process.env.MHB_INPUT_DIR;
 const htmlOnly = process.env.MHB_HTML_ONLY === "1";
+const outputDir = process.env.MHB_OUTPUT_DIR;
 
 const jsMime = new Set([
   undefined,
@@ -58,14 +59,23 @@ const jsCssPlugin = {
 
 const plugins = htmlOnly ? [] : [jsCssPlugin];
 
+const minifier = new minimize({ plugins });
+
 const results = fs.readdirSync(inputDir).map((name) => {
   const src = fs.readFileSync(path.join(inputDir, name), "utf8");
+
+  const out = minifier.parse(src);
+  const len = Buffer.from(out, "utf8").length;
+  if (outputDir) {
+    fs.writeFileSync(path.join(outputDir, name), out);
+  }
+
   const start = process.hrtime.bigint();
-  let len;
   for (let i = 0; i < iterations; i++) {
-    len = new minimize({ plugins }).parse(src).length;
+    minifier.parse(src);
   }
   const elapsed = process.hrtime.bigint() - start;
+
   return [name, len, iterations, Number(elapsed) / 1_000_000_000];
 });
 console.log(JSON.stringify(results));

debug/diff/.gitignore

@@ -0,0 +1 @@
+/outputs/

debug/diff/README.md

@@ -1 +1 @@
-[compare.sh](./compare.sh) is a useful script for viewing a character-by-character diff between the minified outputs of minify-html and html-minifier for a specific test. Pass the test's file name as the first argument.
+[compare](./compare) is a useful script for viewing a character-by-character diff between the minified outputs of minify-html and html-minifier for a specific input. Pass the input's file name as the first argument.

debug/diff/compare

@@ -0,0 +1,9 @@
+#!/usr/bin/env bash
+
+set -Eeuxo pipefail
+
+pushd "$(dirname "$0")" >/dev/null
+
+git --no-pager diff --no-index --word-diff=color --word-diff-regex=. "outputs/html-minifier/$1" "outputs/minify-html/$1" | less
+
+popd >/dev/null

debug/diff/compare.sh

@@ -1,3 +0,0 @@
-#!/usr/bin/env bash
-
-git --no-pager diff --no-index --word-diff=color --word-diff-regex=. "min/html-minifier/$1.html" "min/@minify-html/js/$1.html" | less

debug/diff/run

@@ -0,0 +1,30 @@
+#!/usr/bin/env bash
+
+set -Eeuo pipefail
+
+shopt -s nullglob
+
+pushd "$(dirname "$0")" >/dev/null
+
+input_dir="$PWD/../../bench/inputs"
+output_dir="$PWD/outputs"
+
+rm -rf "$output_dir"
+mkdir -p "$output_dir"
+
+pushd "../../bench/runners" >/dev/null
+for r in *; do
+  if [ ! -d "$r" ]; then
+    continue
+  fi
+  mkdir -p "$output_dir/$r"
+  echo "Running $r..."
+  pushd "$r" >/dev/null
+  MHB_ITERATIONS=1 MHB_INPUT_DIR="$input_dir" MHB_OUTPUT_DIR="$output_dir/$r" RUST_BACKTRACE=1 ./run >/dev/null
+  popd >/dev/null
+done
+popd >/dev/null
+
+echo "All done!"
+
+popd >/dev/null