@@ -22,16 +22,18 @@ use self::states::{Rawtext, Rcdata, ScriptData, ScriptDataEscaped};
22
22
use self :: char_ref:: { CharRef , CharRefTokenizer } ;
23
23
24
24
use crate :: util:: str:: lower_ascii_letter;
25
-
26
25
use log:: { debug, trace} ;
27
26
use mac:: format_if;
28
- use markup5ever:: { ns, small_char_set, TokenizerResult } ;
27
+ use markup5ever:: {
28
+ buffer_queue:: BufferQueue , namespace_url, ns, small_char_set, InputSink , InputSinkResult ,
29
+ TokenizerResult ,
30
+ } ;
29
31
use std:: borrow:: Cow :: { self , Borrowed } ;
30
32
use std:: cell:: { Cell , RefCell , RefMut } ;
31
33
use std:: collections:: BTreeMap ;
32
- use std:: mem;
34
+ use std:: { iter , mem} ;
33
35
34
- pub use crate :: buffer_queue:: { BufferQueue , FromSet , NotFromSet , SetResult } ;
36
+ pub use crate :: buffer_queue:: { FromSet , NotFromSet , SetResult } ;
35
37
use crate :: tendril:: StrTendril ;
36
38
use crate :: { Attribute , LocalName , QualName , SmallCharSet } ;
37
39
@@ -43,6 +45,8 @@ pub enum ProcessResult<Handle> {
43
45
Continue ,
44
46
Suspend ,
45
47
Script ( Handle ) ,
48
+ #[ cfg( feature = "encoding" ) ]
49
+ MaybeChangeEncodingAndStartOver ( & ' static encoding_rs:: Encoding ) ,
46
50
}
47
51
48
52
fn option_push ( opt_str : & mut Option < StrTendril > , c : char ) {
@@ -357,6 +361,10 @@ impl<Sink: TokenSink> Tokenizer<Sink> {
357
361
ProcessResult :: Continue => ( ) ,
358
362
ProcessResult :: Suspend => break ,
359
363
ProcessResult :: Script ( node) => return TokenizerResult :: Script ( node) ,
364
+ #[ cfg( feature = "encoding" ) ]
365
+ ProcessResult :: MaybeChangeEncodingAndStartOver ( encoding) => {
366
+ return TokenizerResult :: MaybeChangeEncodingAndStartOver ( encoding)
367
+ } ,
360
368
}
361
369
}
362
370
} else {
@@ -365,6 +373,10 @@ impl<Sink: TokenSink> Tokenizer<Sink> {
365
373
ProcessResult :: Continue => ( ) ,
366
374
ProcessResult :: Suspend => break ,
367
375
ProcessResult :: Script ( node) => return TokenizerResult :: Script ( node) ,
376
+ #[ cfg( feature = "encoding" ) ]
377
+ ProcessResult :: MaybeChangeEncodingAndStartOver ( encoding) => {
378
+ return TokenizerResult :: MaybeChangeEncodingAndStartOver ( encoding)
379
+ } ,
368
380
}
369
381
}
370
382
}
@@ -456,6 +468,10 @@ impl<Sink: TokenSink> Tokenizer<Sink> {
456
468
self . state . set ( states:: RawData ( kind) ) ;
457
469
ProcessResult :: Continue
458
470
} ,
471
+ #[ cfg( feature = "encoding" ) ]
472
+ TokenSinkResult :: MaybeChangeEncodingAndStartOver ( encoding) => {
473
+ ProcessResult :: MaybeChangeEncodingAndStartOver ( encoding)
474
+ } ,
459
475
}
460
476
}
461
477
@@ -1725,6 +1741,8 @@ impl<Sink: TokenSink> Tokenizer<Sink> {
1725
1741
ProcessResult :: Continue => ( ) ,
1726
1742
ProcessResult :: Suspend => break ,
1727
1743
ProcessResult :: Script ( _) => unreachable ! ( ) ,
1744
+ #[ cfg( feature = "encoding" ) ]
1745
+ ProcessResult :: MaybeChangeEncodingAndStartOver ( _) => unreachable ! ( ) ,
1728
1746
}
1729
1747
}
1730
1748
@@ -2001,13 +2019,27 @@ impl<Sink: TokenSink> Tokenizer<Sink> {
2001
2019
}
2002
2020
}
2003
2021
2022
+ impl < Sink > InputSink for Tokenizer < Sink >
2023
+ where
2024
+ Sink : TokenSink ,
2025
+ {
2026
+ type Handle = Sink :: Handle ;
2027
+
2028
+ fn feed < ' a > (
2029
+ & ' a self ,
2030
+ input : & ' a BufferQueue ,
2031
+ ) -> impl Iterator < Item = InputSinkResult < Self :: Handle > > + ' a {
2032
+ iter:: from_fn ( || self . feed ( input) . into ( ) )
2033
+ }
2034
+ }
2035
+
2004
2036
#[ cfg( test) ]
2005
2037
#[ allow( non_snake_case) ]
2006
2038
mod test {
2007
2039
use super :: option_push; // private items
2008
- use crate :: tendril:: { SliceExt , StrTendril } ;
2009
-
2010
2040
use super :: { TokenSink , TokenSinkResult , Tokenizer , TokenizerOpts } ;
2041
+ use crate :: tendril:: { SliceExt , StrTendril } ;
2042
+ use crate :: LocalName ;
2011
2043
2012
2044
use super :: interface:: { CharacterTokens , EOFToken , NullCharacterToken , ParseError } ;
2013
2045
use super :: interface:: { EndTag , StartTag , Tag , TagKind } ;
@@ -2016,8 +2048,6 @@ mod test {
2016
2048
use markup5ever:: buffer_queue:: BufferQueue ;
2017
2049
use std:: cell:: RefCell ;
2018
2050
2019
- use crate :: LocalName ;
2020
-
2021
2051
// LinesMatch implements the TokenSink trait. It is used for testing to see
2022
2052
// if current_line is being updated when process_token is called. The lines
2023
2053
// vector is a collection of the line numbers that each token is on.
0 commit comments