diff options
Diffstat (limited to 'src/naive_parser.rs')
| -rw-r--r-- | src/naive_parser.rs | 21 |
1 files changed, 17 insertions, 4 deletions
diff --git a/src/naive_parser.rs b/src/naive_parser.rs index 2626209..f126dfd 100644 --- a/src/naive_parser.rs +++ b/src/naive_parser.rs @@ -1,7 +1,8 @@ use crate::emitter::DefaultEmitter; use crate::offset::{Offset, Position}; use crate::reader::{IntoReader, Reader}; -use crate::{Emitter, State, Tokenizer}; +use crate::tokenizer::CdataAction; +use crate::{Emitter, Event, State, Tokenizer}; /// A naive HTML parser (**not** spec-compliant since it doesn't do tree construction). /// @@ -53,9 +54,21 @@ impl<R: Reader + Position<O>, O: Offset, E: Emitter<O>> Iterator for NaiveParser type Item = Result<E::Token, R::Error>; fn next(&mut self) -> Option<Self::Item> { - // A proper parser would follow the steps described under section '13.2.6 Tree construction' - // of the spec. Since this parser is naive, we directly return the token instead. - self.tokenizer.next() + loop { + let event = self.tokenizer.next()?; + match event { + Err(e) => return Some(Err(e)), + Ok(Event::Token(t)) => { + // A proper parser would follow the steps described under section '13.2.6 Tree construction' + // of the spec. Since this parser is naive, we directly return the token instead. + return Some(Ok(t)); + } + Ok(Event::CdataOpen) => { + // Naively parse any CDATA sections as bogus comments. + self.tokenizer.handle_cdata_open(CdataAction::BogusComment) + } + } + } } } |