From cdd56e866738d354ca295b87ab5bdb1158c9751e Mon Sep 17 00:00:00 2001
From: Wilson Lin <code@wilsonl.in>
Date: Thu, 5 Jan 2023 14:28:16 +1100
Subject: [PATCH] Do not encode entities in RCDATA text content

---
 CHANGELOG.md                    |  4 ++++
 rust/main/src/ast/c14n.rs       |  3 +++
 rust/main/src/ast/mod.rs        |  8 ++++++++
 rust/main/src/lib.rs            |  1 +
 rust/main/src/minify/content.rs |  3 +++
 rust/main/src/minify/mod.rs     |  1 +
 rust/main/src/minify/rcdata.rs  |  5 +++++
 rust/main/src/parse/element.rs  | 10 +++++-----
 rust/main/src/parse/textarea.rs | 17 ++++-------------
 rust/main/src/parse/title.rs    | 17 ++++-------------
 rust/main/src/tag/mod.rs        | 12 ++++++++++++
 11 files changed, 50 insertions(+), 31 deletions(-)
 create mode 100644 rust/main/src/minify/rcdata.rs
 create mode 100644 rust/main/src/tag/mod.rs

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 9fd68dd..2004abe 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -1,5 +1,9 @@
 # minify-html changelog
 
+## Pending
+
+- Do not encode entities in RCDATA text content (e.g. contents of `<textarea>` and `<title>`).
+
 ## 0.10.4
 
 - Use FxHasher for internal hash-based data structures.
diff --git a/rust/main/src/ast/c14n.rs b/rust/main/src/ast/c14n.rs
index ac8b850..4dc6d7c 100644
--- a/rust/main/src/ast/c14n.rs
+++ b/rust/main/src/ast/c14n.rs
@@ -129,6 +129,9 @@ pub fn c14n_serialise_ast<T: Write>(out: &mut T, node: &NodeData) -> std::io::Re
             out.write_all(code)?;
             out.write_all(b"?>")?;
         }
+        NodeData::RcdataContent { content } => {
+            out.write_all(content)?;
+        }
         NodeData::ScriptOrStyleContent { code, .. } => {
             out.write_all(code)?;
         }
diff --git a/rust/main/src/ast/mod.rs b/rust/main/src/ast/mod.rs
index d1c0203..b988ef1 100644
--- a/rust/main/src/ast/mod.rs
+++ b/rust/main/src/ast/mod.rs
@@ -83,6 +83,10 @@ pub enum NodeData {
         // If the source unexpectedly ended before `?>`, we can't add it, as otherwise output could be longer than source.
         ended: bool,
     },
+    // <title> or <textarea> element contents.
+    RcdataContent {
+        content: Vec<u8>,
+    },
     // Entities should not be decoded in ScriptOrStyleContent.
     ScriptOrStyleContent {
         code: Vec<u8>,
@@ -139,6 +143,10 @@ impl Debug for NodeData {
                 .field("code", &from_utf8(code).unwrap().to_string())
                 .field("ended", ended)
                 .finish(),
+            NodeData::RcdataContent { content } => f
+                .debug_struct("RcdataContent")
+                .field("content", &from_utf8(content).unwrap().to_string())
+                .finish(),
             NodeData::ScriptOrStyleContent { code, lang } => f
                 .debug_struct("ScriptOrStyleContent")
                 .field("code", &from_utf8(code).unwrap().to_string())
diff --git a/rust/main/src/lib.rs b/rust/main/src/lib.rs
index 1e8f271..9f00c2c 100644
--- a/rust/main/src/lib.rs
+++ b/rust/main/src/lib.rs
@@ -14,6 +14,7 @@ mod common;
 mod entity;
 mod minify;
 mod parse;
+mod tag;
 #[cfg(test)]
 mod tests;
 
diff --git a/rust/main/src/minify/content.rs b/rust/main/src/minify/content.rs
index 6539063..15cf6a1 100644
--- a/rust/main/src/minify/content.rs
+++ b/rust/main/src/minify/content.rs
@@ -19,6 +19,8 @@ use crate::minify::element::minify_element;
 use crate::minify::instruction::minify_instruction;
 use crate::minify::js::minify_js;
 
+use super::rcdata::minify_rcdata;
+
 fn build_chevron_replacer() -> Replacer {
     let mut patterns = Vec::<Vec<u8>>::new();
     let mut replacements = Vec::<Vec<u8>>::new();
@@ -142,6 +144,7 @@ pub fn minify_content(
                 children,
             ),
             NodeData::Instruction { code, ended } => minify_instruction(cfg, out, &code, ended),
+            NodeData::RcdataContent { content } => minify_rcdata(cfg, out, &content),
             NodeData::ScriptOrStyleContent { code, lang } => match lang {
                 ScriptOrStyleLang::CSS => minify_css(cfg, out, &code),
                 ScriptOrStyleLang::Data => out.extend_from_slice(&code),
diff --git a/rust/main/src/minify/mod.rs b/rust/main/src/minify/mod.rs
index 2b4db78..42227c7 100644
--- a/rust/main/src/minify/mod.rs
+++ b/rust/main/src/minify/mod.rs
@@ -7,5 +7,6 @@ pub mod doctype;
 pub mod element;
 pub mod instruction;
 pub mod js;
+pub mod rcdata;
 #[cfg(test)]
 mod tests;
diff --git a/rust/main/src/minify/rcdata.rs b/rust/main/src/minify/rcdata.rs
new file mode 100644
index 0000000..efca760
--- /dev/null
+++ b/rust/main/src/minify/rcdata.rs
@@ -0,0 +1,5 @@
+use crate::cfg::Cfg;
+
+pub fn minify_rcdata(cfg: &Cfg, out: &mut Vec<u8>, content: &[u8]) {
+    out.extend_from_slice(content);
+}
diff --git a/rust/main/src/parse/element.rs b/rust/main/src/parse/element.rs
index 0a81dad..8d32802 100644
--- a/rust/main/src/parse/element.rs
+++ b/rust/main/src/parse/element.rs
@@ -164,8 +164,8 @@ pub fn parse_element(code: &mut Code, ns: Namespace, parent: &[u8]) -> NodeData
     let ParsedContent {
         closing_tag_omitted,
         children,
-    } = match elem_name.as_slice() {
-        b"script" => match attributes.get(b"type".as_ref()) {
+    } = match (ns, elem_name.as_slice()) {
+        (_, b"script") => match attributes.get(b"type".as_ref()) {
             Some(mime) if !JAVASCRIPT_MIME_TYPES.contains(mime.as_slice()) => {
                 parse_script_content(code, ScriptOrStyleLang::Data)
             }
@@ -174,9 +174,9 @@ pub fn parse_element(code: &mut Code, ns: Namespace, parent: &[u8]) -> NodeData
             }
             _ => parse_script_content(code, ScriptOrStyleLang::JS),
         },
-        b"style" => parse_style_content(code),
-        b"textarea" => parse_textarea_content(code),
-        b"title" => parse_title_content(code),
+        (_, b"style") => parse_style_content(code),
+        (Namespace::Html, b"textarea") => parse_textarea_content(code),
+        (Namespace::Html, b"title") => parse_title_content(code),
         _ => parse_content(code, ns, parent, &elem_name),
     };
 
diff --git a/rust/main/src/parse/textarea.rs b/rust/main/src/parse/textarea.rs
index 629a5ca..51604df 100644
--- a/rust/main/src/parse/textarea.rs
+++ b/rust/main/src/parse/textarea.rs
@@ -1,27 +1,18 @@
-use aho_corasick::AhoCorasick;
-use aho_corasick::AhoCorasickBuilder;
-use lazy_static::lazy_static;
-
 use crate::ast::NodeData;
 use crate::entity::decode::decode_entities;
 use crate::parse::content::ParsedContent;
 use crate::parse::Code;
-
-lazy_static! {
-    static ref END: AhoCorasick = AhoCorasickBuilder::new()
-        .ascii_case_insensitive(true)
-        .build(&["</textarea"]);
-}
+use crate::tag::TAG_TEXTAREA_END;
 
 pub fn parse_textarea_content(code: &mut Code) -> ParsedContent {
-    let (len, closing_tag_omitted) = match END.find(code.as_slice()) {
+    let (len, closing_tag_omitted) = match TAG_TEXTAREA_END.find(code.as_slice()) {
         Some(m) => (m.start(), false),
         None => (code.rem(), true),
     };
     ParsedContent {
         closing_tag_omitted,
-        children: vec![NodeData::Text {
-            value: decode_entities(code.slice_and_shift(len), false),
+        children: vec![NodeData::RcdataContent {
+            content: decode_entities(code.slice_and_shift(len), false),
         }],
     }
 }
diff --git a/rust/main/src/parse/title.rs b/rust/main/src/parse/title.rs
index 2c55014..00998bb 100644
--- a/rust/main/src/parse/title.rs
+++ b/rust/main/src/parse/title.rs
@@ -1,27 +1,18 @@
-use aho_corasick::AhoCorasick;
-use aho_corasick::AhoCorasickBuilder;
-use lazy_static::lazy_static;
-
 use crate::ast::NodeData;
 use crate::entity::decode::decode_entities;
 use crate::parse::content::ParsedContent;
 use crate::parse::Code;
-
-lazy_static! {
-    static ref END: AhoCorasick = AhoCorasickBuilder::new()
-        .ascii_case_insensitive(true)
-        .build(&["</title"]);
-}
+use crate::tag::TAG_TITLE_END;
 
 pub fn parse_title_content(code: &mut Code) -> ParsedContent {
-    let (len, closing_tag_omitted) = match END.find(code.as_slice()) {
+    let (len, closing_tag_omitted) = match TAG_TITLE_END.find(code.as_slice()) {
         Some(m) => (m.start(), false),
         None => (code.rem(), true),
     };
     ParsedContent {
         closing_tag_omitted,
-        children: vec![NodeData::Text {
-            value: decode_entities(code.slice_and_shift(len), false),
+        children: vec![NodeData::RcdataContent {
+            content: decode_entities(code.slice_and_shift(len), false),
         }],
     }
 }
diff --git a/rust/main/src/tag/mod.rs b/rust/main/src/tag/mod.rs
new file mode 100644
index 0000000..eec14f5
--- /dev/null
+++ b/rust/main/src/tag/mod.rs
@@ -0,0 +1,12 @@
+use aho_corasick::AhoCorasick;
+use aho_corasick::AhoCorasickBuilder;
+use lazy_static::lazy_static;
+
+lazy_static! {
+    pub static ref TAG_TEXTAREA_END: AhoCorasick = AhoCorasickBuilder::new()
+        .ascii_case_insensitive(true)
+        .build(&["</textarea"]);
+    pub static ref TAG_TITLE_END: AhoCorasick = AhoCorasickBuilder::new()
+        .ascii_case_insensitive(true)
+        .build(&["</title"]);
+}