Implement JS and CSS minification; add tests

Cargo.toml

@@ -22,6 +22,6 @@ js-esbuild = ["crossbeam", "esbuild-rs"]
 [dependencies]
 aho-corasick = "0.7"
 crossbeam = { version = "0.7", optional = true }
-esbuild-rs = { version = "0.8.30", optional = true }
+esbuild-rs = { version = "0.12.18", optional = true }
 lazy_static = "1.4"
 memchr = "2"

src/minify/css.rs

@@ -1,6 +1,39 @@
+#[cfg(feature = "js-esbuild")]
+use {
+    crate::minify::esbuild::minify_using_esbuild,
+    aho_corasick::{AhoCorasick, AhoCorasickBuilder},
+    esbuild_rs::{Loader, TransformOptions, TransformOptionsBuilder},
+    lazy_static::lazy_static,
+    std::sync::Arc,
+};
+
 use crate::cfg::Cfg;
 
-pub fn minify_css(_cfg: &Cfg, out: &mut Vec<u8>, code: &[u8]) {
-    // TODO
-    out.extend_from_slice(code);
+#[cfg(feature = "js-esbuild")]
+lazy_static! {
+    static ref STYLE_END: AhoCorasick = AhoCorasickBuilder::new()
+        .ascii_case_insensitive(true)
+        .build(&["</style"]);
+    static ref TRANSFORM_OPTIONS: Arc<TransformOptions> = {
+        let mut builder = TransformOptionsBuilder::new();
+        builder.loader = Loader::CSS;
+        builder.minify_identifiers = true;
+        builder.minify_syntax = true;
+        builder.minify_whitespace = true;
+        builder.build()
+    };
+}
+
+#[cfg(not(feature = "js-esbuild"))]
+pub fn minify_css(_cfg: &Cfg, out: &mut Vec<u8>, code: &[u8]) {
+    out.extend_from_slice(&code);
+}
+
+#[cfg(feature = "js-esbuild")]
+pub fn minify_css(cfg: &Cfg, out: &mut Vec<u8>, code: &[u8]) {
+    if !cfg.minify_css {
+        out.extend_from_slice(&code);
+    } else {
+        minify_using_esbuild(out, code, &TRANSFORM_OPTIONS.clone(), &STYLE_END);
+    }
 }

src/minify/esbuild.rs

@@ -0,0 +1,42 @@
+#[cfg(feature = "js-esbuild")]
+use {aho_corasick::AhoCorasick, crossbeam::sync::WaitGroup, esbuild_rs::TransformOptions};
+
+#[cfg(feature = "js-esbuild")]
+// TODO The use of WG is ugly and we don't want to be multi-threaded; wait for Rust port esbuild-transform-rs.
+// `tag_to_escape` must be case insensitive.
+pub fn minify_using_esbuild(
+    out: &mut Vec<u8>,
+    code: &[u8],
+    transform_options: &TransformOptions,
+    tag_to_escape: &'static AhoCorasick,
+) {
+    let wg = WaitGroup::new();
+    unsafe {
+        let wg = wg.clone();
+        esbuild_rs::transform_direct_unmanaged(code, transform_options, move |result| {
+            // TODO (JS) Handle other forms:
+            // 1 < /script/.exec(a).length
+            // `  ${`  ${a</script/}  `}  `
+            // // </script>
+            // /* </script>
+            // Considerations:
+            // - Need to parse strings (e.g. "", '', ``) so syntax within strings aren't mistakenly interpreted as code.
+            // - Need to be able to parse regex literals to determine string delimiters aren't actually characters in the regex.
+            // - Determining whether a slash is division or regex requires a full-blown JS parser to handle all cases (this is a well-known JS parsing problem).
+            // - `/</script` or `/</ script` are not valid JS so don't need to be handled.
+            // TODO (CSS) Are there other places that can have unintentional closing tags?
+            tag_to_escape.replace_all_with_bytes(
+                result.code.as_str().trim().as_bytes(),
+                out,
+                |_, orig, dst| {
+                    dst.extend(b"<\\/");
+                    // Keep original case.
+                    dst.extend(&orig[2..]);
+                    true
+                },
+            );
+            drop(wg);
+        });
+    };
+    wg.wait();
+}

src/minify/js.rs

@@ -1,6 +1,38 @@
-use crate::cfg::Cfg;
+#[cfg(feature = "js-esbuild")]
+use {
+    crate::minify::esbuild::minify_using_esbuild,
+    aho_corasick::{AhoCorasick, AhoCorasickBuilder},
+    esbuild_rs::{TransformOptions, TransformOptionsBuilder},
+    lazy_static::lazy_static,
+    std::sync::Arc,
+};
 
-pub fn minify_js(_cfg: &Cfg, out: &mut Vec<u8>, code: &[u8]) {
-    // TODO
-    out.extend_from_slice(code);
+use crate::Cfg;
+
+#[cfg(feature = "js-esbuild")]
+lazy_static! {
+    static ref SCRIPT_END: AhoCorasick = AhoCorasickBuilder::new()
+        .ascii_case_insensitive(true)
+        .build(&["</script"]);
+    static ref TRANSFORM_OPTIONS: Arc<TransformOptions> = {
+        let mut builder = TransformOptionsBuilder::new();
+        builder.minify_identifiers = true;
+        builder.minify_syntax = true;
+        builder.minify_whitespace = true;
+        builder.build()
+    };
+}
+
+#[cfg(not(feature = "js-esbuild"))]
+pub fn minify_js(_cfg: &Cfg, out: &mut Vec<u8>, code: &[u8]) {
+    out.extend_from_slice(&code);
+}
+
+#[cfg(feature = "js-esbuild")]
+pub fn minify_js(cfg: &Cfg, out: &mut Vec<u8>, code: &[u8]) {
+    if !cfg.minify_js {
+        out.extend_from_slice(&code);
+    } else {
+        minify_using_esbuild(out, code, &TRANSFORM_OPTIONS.clone(), &SCRIPT_END);
+    }
 }

src/minify/mod.rs

@@ -4,6 +4,7 @@ pub mod comment;
 pub mod content;
 pub mod css;
 pub mod element;
+pub mod esbuild;
 pub mod instruction;
 pub mod js;
 #[cfg(test)]

src/pattern.rs

@@ -15,6 +15,7 @@ pub enum TrieNodeMatch<V: 'static + Copy> {
     NotFound { reached: usize },
 }
 
+#[allow(dead_code)]
 impl<V: 'static + Copy> TrieNode<V> {
     // Find the node that matches the shortest prefix of {@param text} that:
     // - has a value (except the start node if it has a value);
@@ -32,7 +33,7 @@ impl<V: 'static + Copy> TrieNode<V> {
     // - "&amx" will return node `m`.
     // - "&ax" will return node `a`.
     // - "+ax" will return itself.
-    // - "" will return the itself.
+    // - "" will return itself.
     pub fn shortest_matching_prefix(&self, text: &[u8], from: usize) -> (&TrieNode<V>, usize) {
         let mut node: &TrieNode<V> = self;
         let mut pos = from;

src/tests/mod.rs

@@ -1,4 +1,4 @@
-fn _eval(src: &'static [u8], expected: &'static [u8], cfg: &super::Cfg) {
+fn eval_with_cfg(src: &'static [u8], expected: &'static [u8], cfg: &super::Cfg) {
     let mut code = src.to_vec();
     let min = super::minify(&mut code, cfg);
     assert_eq!(
@@ -8,21 +8,27 @@ fn _eval(src: &'static [u8], expected: &'static [u8], cfg: &super::Cfg) {
 }
 
 fn eval(src: &'static [u8], expected: &'static [u8]) {
-    _eval(src, expected, &super::Cfg::new());
+    eval_with_cfg(src, expected, &super::Cfg::new());
+}
+
+fn eval_with_keep_html_head(src: &'static [u8], expected: &'static [u8]) -> () {
+    let mut cfg = super::Cfg::new();
+    cfg.keep_html_and_head_opening_tags = true;
+    eval_with_cfg(src, expected, &cfg);
 }
 
 #[cfg(feature = "js-esbuild")]
 fn eval_with_js_min(src: &'static [u8], expected: &'static [u8]) -> () {
     let mut cfg = super::Cfg::new();
     cfg.minify_js = true;
-    _eval(src, expected, &cfg);
+    eval_with_cfg(src, expected, &cfg);
 }
 
 #[cfg(feature = "js-esbuild")]
 fn eval_with_css_min(src: &'static [u8], expected: &'static [u8]) -> () {
     let mut cfg = super::Cfg::new();
     cfg.minify_css = true;
-    _eval(src, expected, &cfg);
+    eval_with_cfg(src, expected, &cfg);
 }
 
 #[test]
@@ -97,17 +103,17 @@ fn test_no_whitespace_minification() {
 #[test]
 fn test_parsing_extra_head_tag() {
     // Extra `<head>` in `<label>` should be dropped, so whitespace around `<head>` should be joined and therefore trimmed due to `<label>` whitespace rules.
-    eval(
+    eval_with_keep_html_head(
         b"<html><head><meta><head><link><head><body><label>  <pre> </pre> <head>  </label>",
         b"<html><head><meta><link><body><label><pre> </pre></label>",
     );
     // Same as above except it's a `</head>`, which should get reinterpreted as a `<head>`.
-    eval(
+    eval_with_keep_html_head(
         b"<html><head><meta><head><link><head><body><label>  <pre> </pre> </head>  </label>",
         b"<html><head><meta><link><body><label><pre> </pre></label>",
     );
     // `<head>` gets implicitly closed by `<body>`, so any following `</head>` should be ignored. (They should be anyway, since `</head>` would not be a valid closing tag.)
-    eval(
+    eval_with_keep_html_head(
         b"<html><head><body><label> </head> </label>",
         b"<html><head><body><label></label>",
     );
@@ -115,10 +121,10 @@ fn test_parsing_extra_head_tag() {
 
 #[test]
 fn test_parsing_omitted_closing_tag() {
-    eval(b"<html>", b"<html>");
-    eval(b" <html>\n", b"<html>");
-    eval(b" <!doctype html> <html>\n", b"<!doctype html><html>");
-    eval(
+    eval_with_keep_html_head(b"<html>", b"<html>");
+    eval_with_keep_html_head(b" <html>\n", b"<html>");
+    eval_with_keep_html_head(b" <!doctype html> <html>\n", b"<!doctype html><html>");
+    eval_with_keep_html_head(
         b"<!doctype html><html><div> <p>Foo</div></html>",
         b"<!doctype html><html><div><p>Foo</div>",
     );
@@ -136,24 +142,24 @@ fn test_self_closing_svg_tag_whitespace_removal() {
 
 #[test]
 fn test_parsing_with_omitted_tags() {
-    eval(b"<ul><li>1<li>2<li>3</ul>", b"<ul><li>1<li>2<li>3</ul>");
-    eval(b"<rt>", b"<rt>");
-    eval(b"<rt><rp>1</rp><div></div>", b"<rt><rp>1</rp><div></div>");
-    eval(b"<div><rt></div>", b"<div><rt></div>");
-    eval(b"<html><head><body>", b"<html><head><body>");
-    eval(b"<html><head><body>", b"<html><head><body>");
+    eval_with_keep_html_head(b"<ul><li>1<li>2<li>3</ul>", b"<ul><li>1<li>2<li>3</ul>");
+    eval_with_keep_html_head(b"<rt>", b"<rt>");
+    eval_with_keep_html_head(b"<rt><rp>1</rp><div></div>", b"<rt><rp>1</rp><div></div>");
+    eval_with_keep_html_head(b"<div><rt></div>", b"<div><rt></div>");
+    eval_with_keep_html_head(b"<html><head><body>", b"<html><head><body>");
+    eval_with_keep_html_head(b"<html><head><body>", b"<html><head><body>");
     // Tag names should be case insensitive.
-    eval(b"<rt>", b"<rt>");
+    eval_with_keep_html_head(b"<rt>", b"<rt>");
 }
 
 #[test]
 fn test_unmatched_closing_tag() {
-    eval(b"Hello</p>Goodbye", b"Hello<p>Goodbye");
-    eval(b"Hello<br></br>Goodbye", b"Hello<br>Goodbye");
-    eval(b"<div>Hello</p>Goodbye", b"<div>Hello<p>Goodbye");
-    eval(b"<ul><li>a</p>", b"<ul><li>a<p>");
-    eval(b"<ul><li><rt>a</p>", b"<ul><li><rt>a<p>");
-    eval(
+    eval_with_keep_html_head(b"Hello</p>Goodbye", b"Hello<p>Goodbye");
+    eval_with_keep_html_head(b"Hello<br></br>Goodbye", b"Hello<br>Goodbye");
+    eval_with_keep_html_head(b"<div>Hello</p>Goodbye", b"<div>Hello<p>Goodbye");
+    eval_with_keep_html_head(b"<ul><li>a</p>", b"<ul><li>a<p>");
+    eval_with_keep_html_head(b"<ul><li><rt>a</p>", b"<ul><li><rt>a<p>");
+    eval_with_keep_html_head(
         b"<html><head><body><ul><li><rt>a</p>",
         b"<html><head><body><ul><li><rt>a<p>",
     );
@@ -175,17 +181,17 @@ fn test_removal_of_html_and_head_opening_tags() {
 
 #[test]
 fn test_removal_of_optional_tags() {
-    eval(
+    eval_with_keep_html_head(
         b"<ul><li>1</li><li>2</li><li>3</li></ul>",
         b"<ul><li>1<li>2<li>3</ul>",
     );
-    eval(b"<rt></rt>", b"<rt>");
-    eval(
+    eval_with_keep_html_head(b"<rt></rt>", b"<rt>");
+    eval_with_keep_html_head(
         b"<rt></rt><rp>1</rp><div></div>",
         b"<rt><rp>1</rp><div></div>",
     );
-    eval(b"<div><rt></rt></div>", b"<div><rt></div>");
-    eval(
+    eval_with_keep_html_head(b"<div><rt></rt></div>", b"<div><rt></div>");
+    eval_with_keep_html_head(
         br#"
         <html>
             <head>
@@ -198,7 +204,7 @@ fn test_removal_of_optional_tags() {
         b"<html><head><body>",
     );
     // Tag names should be case insensitive.
-    eval(b"<RT></rt>", b"<rt>");
+    eval_with_keep_html_head(b"<RT></rt>", b"<rt>");
 }
 
 #[test]