Support dynamic charset change on meta tags in HtmlRewriter.

Cargo.toml

@@ -45,6 +45,7 @@ safemem = "0.3.3"
 selectors = "0.22.0"
 thiserror = "1.0.2"
 hashbrown = "0.13.1"
+mime = "0.3.16"
 
 [dev-dependencies]
 criterion = "0.4.0"

c-api/src/rewriter.rs

@@ -57,6 +57,7 @@ pub extern "C" fn lol_html_rewriter_build(
         memory_settings,
         strict,
         enable_esi_tags: false,
+        adjust_charset_on_meta_tag: false,
     };
 
     let output_sink = ExternOutputSink::new(output_sink, output_sink_user_data);
@@ -88,6 +89,7 @@ pub extern "C" fn unstable_lol_html_rewriter_build_with_esi_tags(
         memory_settings,
         strict,
         enable_esi_tags: true,
+        adjust_charset_on_meta_tag: false,
     };
 
     let output_sink = ExternOutputSink::new(output_sink, output_sink_user_data);

fuzz/test_case/src/lib.rs

@@ -180,6 +180,7 @@ fn run_rewriter_iter(data: &[u8], selector: &str, encoding: &'static Encoding) -
             encoding: encoding.try_into().unwrap(),
             memory_settings: MemorySettings::default(),
             strict: false,
+            adjust_charset_on_meta_tag: false,
         },
         |_: &[u8]| {},
     );

src/base/encoding.rs

@@ -0,0 +1,34 @@
+use crate::rewriter::AsciiCompatibleEncoding;
+use encoding_rs::Encoding;
+use std::cell::Cell;
+use std::ops::Deref;
+use std::rc::Rc;
+
+#[derive(Clone)]
+pub struct SharedEncoding {
+    encoding: Rc<Cell<AsciiCompatibleEncoding>>,
+}
+
+impl SharedEncoding {
+    pub fn new(encoding: AsciiCompatibleEncoding) -> SharedEncoding {
+        SharedEncoding {
+            encoding: Rc::new(Cell::new(encoding)),
+        }
+    }
+
+    pub fn get(&self) -> &'static Encoding {
+        self.encoding.get().into()
+    }
+
+    pub fn set(&self, encoding: AsciiCompatibleEncoding) {
+        self.encoding.set(encoding);
+    }
+}
+
+impl Deref for SharedEncoding {
+    type Target = Encoding;
+
+    fn deref(&self) -> &'static Encoding {
+        self.get()
+    }
+}

src/base/mod.rs

@@ -3,8 +3,10 @@ mod debug_trace;
 
 mod align;
 mod bytes;
+mod encoding;
 mod range;
 
 pub use self::align::Align;
 pub use self::bytes::{Bytes, HasReplacementsError};
+pub use self::encoding::SharedEncoding;
 pub use self::range::Range;

src/lib.rs

@@ -154,6 +154,8 @@ cfg_if! {
     if #[cfg(feature = "integration_test")] {
         pub mod selectors_vm;
 
+        pub use self::base::SharedEncoding;
+
         pub use self::transform_stream::{
             StartTagHandlingResult, TransformController, TransformStream,
             TransformStreamSettings

src/rewritable_units/document_end.rs

@@ -1,6 +1,5 @@
 use super::mutations::content_to_bytes;
 use super::ContentType;
-
 use encoding_rs::Encoding;
 
 use crate::transform_stream::OutputSink;

src/rewritable_units/mod.rs

@@ -87,6 +87,7 @@ mod tokens;
 
 #[cfg(test)]
 mod test_utils {
+    use crate::rewriter::AsciiCompatibleEncoding;
     use crate::test_utils::{Output, ASCII_COMPATIBLE_ENCODINGS};
     use crate::*;
     use encoding_rs::Encoding;

src/rewritable_units/tokens/capturer/mod.rs

@@ -3,10 +3,10 @@ mod to_token;
 
 use self::text_decoder::TextDecoder;
 use super::*;
+use crate::base::SharedEncoding;
 use crate::parser::Lexeme;
 use crate::rewriter::RewritingError;
 use bitflags::bitflags;
-use encoding_rs::Encoding;
 
 pub use self::to_token::{ToToken, ToTokenResult};
 
@@ -30,15 +30,15 @@ pub enum TokenCapturerEvent<'i> {
 type CapturerEventHandler<'h> = &'h mut dyn FnMut(TokenCapturerEvent) -> Result<(), RewritingError>;
 
 pub struct TokenCapturer {
-    encoding: &'static Encoding,
+    encoding: SharedEncoding,
     text_decoder: TextDecoder,
     capture_flags: TokenCaptureFlags,
 }
 
 impl TokenCapturer {
-    pub fn new(capture_flags: TokenCaptureFlags, encoding: &'static Encoding) -> Self {
+    pub fn new(capture_flags: TokenCaptureFlags, encoding: SharedEncoding) -> Self {
         TokenCapturer {
-            encoding,
+            encoding: SharedEncoding::clone(&encoding),
             text_decoder: TextDecoder::new(encoding),
             capture_flags,
         }
@@ -70,7 +70,7 @@ impl TokenCapturer {
     where
         Lexeme<'i, T>: ToToken,
     {
-        match lexeme.to_token(&mut self.capture_flags, self.encoding) {
+        match lexeme.to_token(&mut self.capture_flags, self.encoding.get()) {
             ToTokenResult::Token(token) => {
                 self.flush_pending_text(&mut event_handler)?;
                 event_handler(TokenCapturerEvent::LexemeConsumed)?;

src/rewritable_units/tokens/capturer/text_decoder.rs

@@ -1,27 +1,28 @@
 use super::*;
+use crate::base::SharedEncoding;
 use crate::html::TextType;
 use crate::rewriter::RewritingError;
-use encoding_rs::{CoderResult, Decoder, Encoding};
+use encoding_rs::{CoderResult, Decoder};
 
 // NOTE: this can't be refactored into method, because we hold a mutable reference for `self`
 // during the decoding loop in `feed_text`.
 macro_rules! emit {
     ($self:tt, $text:expr, $last:ident, $event_handler:ident) => {{
-        let token = TextChunk::new_token($text, $self.last_text_type, $last, $self.encoding);
+        let token = TextChunk::new_token($text, $self.last_text_type, $last, $self.encoding.get());
 
         $event_handler(TokenCapturerEvent::TokenProduced(Box::new(token)))
     }};
 }
 
 pub struct TextDecoder {
-    encoding: &'static Encoding,
+    encoding: SharedEncoding,
     pending_text_streaming_decoder: Option<Decoder>,
     text_buffer: String,
     last_text_type: TextType,
 }
 
 impl TextDecoder {
-    pub fn new(encoding: &'static Encoding) -> Self {
+    pub fn new(encoding: SharedEncoding) -> Self {
         TextDecoder {
             encoding,
             pending_text_streaming_decoder: None,
@@ -49,7 +50,7 @@ impl TextDecoder {
         last: bool,
         event_handler: CapturerEventHandler,
     ) -> Result<(), RewritingError> {
-        let encoding = self.encoding;
+        let encoding = self.encoding.get();
         let buffer = self.text_buffer.as_mut_str();
 
         let decoder = self