From 013afa007e8d5345f7f06e3f4f32bda858944ec9 Mon Sep 17 00:00:00 2001
From: Wilson Lin <code@wilsonl.in>
Date: Thu, 5 Jan 2023 14:55:45 +1100
Subject: [PATCH] Improve handling of RCDATA text content in edge cases

---
 CHANGELOG.md                    |  4 ++++
 rust/main/src/ast/c14n.rs       |  4 ++--
 rust/main/src/ast/mod.rs        | 14 +++++++++++---
 rust/main/src/minify/content.rs |  2 +-
 rust/main/src/minify/rcdata.rs  | 18 +++++++++++++++---
 rust/main/src/parse/textarea.rs |  5 +++--
 rust/main/src/parse/title.rs    |  5 +++--
 7 files changed, 39 insertions(+), 13 deletions(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index ea5a337..16d6208 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -1,5 +1,9 @@
 # minify-html changelog
 
+## Pending
+
+- Improve handling of RCDATA text content in edge cases.
+
 ## 0.10.5
 
 - Do not encode entities in RCDATA text content (e.g. contents of `<textarea>` and `<title>`).
diff --git a/rust/main/src/ast/c14n.rs b/rust/main/src/ast/c14n.rs
index 4dc6d7c..724e745 100644
--- a/rust/main/src/ast/c14n.rs
+++ b/rust/main/src/ast/c14n.rs
@@ -129,8 +129,8 @@ 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::RcdataContent { typ, text } => {
+            out.write_all(&TEXT_REPLACER.replace_all(text))?;
         }
         NodeData::ScriptOrStyleContent { code, .. } => {
             out.write_all(code)?;
diff --git a/rust/main/src/ast/mod.rs b/rust/main/src/ast/mod.rs
index b988ef1..5579100 100644
--- a/rust/main/src/ast/mod.rs
+++ b/rust/main/src/ast/mod.rs
@@ -48,6 +48,12 @@ impl PartialEq for AttrVal {
 
 impl Eq for AttrVal {}
 
+#[derive(Eq, PartialEq, Debug)]
+pub enum RcdataContentType {
+  Textarea,
+  Title,
+}
+
 // Derive Eq for testing.
 #[derive(Eq, PartialEq)]
 pub enum NodeData {
@@ -85,7 +91,8 @@ pub enum NodeData {
     },
     // <title> or <textarea> element contents.
     RcdataContent {
-        content: Vec<u8>,
+        typ: RcdataContentType,
+        text: Vec<u8>,
     },
     // Entities should not be decoded in ScriptOrStyleContent.
     ScriptOrStyleContent {
@@ -143,9 +150,10 @@ impl Debug for NodeData {
                 .field("code", &from_utf8(code).unwrap().to_string())
                 .field("ended", ended)
                 .finish(),
-            NodeData::RcdataContent { content } => f
+            NodeData::RcdataContent { typ, text } => f
                 .debug_struct("RcdataContent")
-                .field("content", &from_utf8(content).unwrap().to_string())
+                .field("typ", typ)
+                .field("text", &from_utf8(text).unwrap().to_string())
                 .finish(),
             NodeData::ScriptOrStyleContent { code, lang } => f
                 .debug_struct("ScriptOrStyleContent")
diff --git a/rust/main/src/minify/content.rs b/rust/main/src/minify/content.rs
index 15cf6a1..45db670 100644
--- a/rust/main/src/minify/content.rs
+++ b/rust/main/src/minify/content.rs
@@ -144,7 +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::RcdataContent { typ, text } => minify_rcdata(out, typ, &text),
             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/rcdata.rs b/rust/main/src/minify/rcdata.rs
index efca760..e419252 100644
--- a/rust/main/src/minify/rcdata.rs
+++ b/rust/main/src/minify/rcdata.rs
@@ -1,5 +1,17 @@
-use crate::cfg::Cfg;
+use crate::{ast::RcdataContentType, tag::{TAG_TEXTAREA_END, TAG_TITLE_END}, entity::encode::encode_entities};
 
-pub fn minify_rcdata(cfg: &Cfg, out: &mut Vec<u8>, content: &[u8]) {
-    out.extend_from_slice(content);
+pub fn minify_rcdata(out: &mut Vec<u8>, typ: RcdataContentType, text: &[u8]) {
+    // Encode entities, since they're still decoded by the browser.
+    let html = encode_entities(text, false);
+
+    // Since the text has been decoded, there may be unintentional matches to end tags that we must escape.
+    let html = match typ {
+      RcdataContentType::Textarea => &*TAG_TEXTAREA_END,
+      RcdataContentType::Title => &*TAG_TITLE_END,
+    }.replace_all_bytes(&html, &[match typ {
+        RcdataContentType::Textarea => b"&LT/textarea".as_slice(),
+        RcdataContentType::Title => b"&LT/title".as_slice(),
+    }]);
+
+    out.extend_from_slice(&html);
 }
diff --git a/rust/main/src/parse/textarea.rs b/rust/main/src/parse/textarea.rs
index 51604df..fa414cc 100644
--- a/rust/main/src/parse/textarea.rs
+++ b/rust/main/src/parse/textarea.rs
@@ -1,4 +1,4 @@
-use crate::ast::NodeData;
+use crate::ast::{NodeData, RcdataContentType};
 use crate::entity::decode::decode_entities;
 use crate::parse::content::ParsedContent;
 use crate::parse::Code;
@@ -12,7 +12,8 @@ pub fn parse_textarea_content(code: &mut Code) -> ParsedContent {
     ParsedContent {
         closing_tag_omitted,
         children: vec![NodeData::RcdataContent {
-            content: decode_entities(code.slice_and_shift(len), false),
+            typ: RcdataContentType::Textarea,
+            text: 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 00998bb..835f9b0 100644
--- a/rust/main/src/parse/title.rs
+++ b/rust/main/src/parse/title.rs
@@ -1,4 +1,4 @@
-use crate::ast::NodeData;
+use crate::ast::{NodeData, RcdataContentType};
 use crate::entity::decode::decode_entities;
 use crate::parse::content::ParsedContent;
 use crate::parse::Code;
@@ -12,7 +12,8 @@ pub fn parse_title_content(code: &mut Code) -> ParsedContent {
     ParsedContent {
         closing_tag_omitted,
         children: vec![NodeData::RcdataContent {
-            content: decode_entities(code.slice_and_shift(len), false),
+            typ: RcdataContentType::Title,
+            text: decode_entities(code.slice_and_shift(len), false),
         }],
     }
 }