diff options
author | Markus Unterwaditzer <markus-honeypot@unterwaditzer.net> | 2021-11-27 23:33:26 +0100 |
---|---|---|
committer | Markus Unterwaditzer <markus-honeypot@unterwaditzer.net> | 2021-11-27 23:43:54 +0100 |
commit | 95afc5359e940398498310d46e81352f04b43a49 (patch) | |
tree | b89915478a19f7f8673bb4663e9e7fa112abf09c /src | |
parent | 96808d0d940e1580cf86e433d0c844e943157e0d (diff) |
fix crash in try_read_string
Diffstat (limited to 'src')
-rw-r--r-- | src/machine.rs | 10 | ||||
-rw-r--r-- | src/tokenizer.rs | 25 | ||||
-rw-r--r-- | src/utils.rs | 8 |
3 files changed, 27 insertions, 16 deletions
diff --git a/src/machine.rs b/src/machine.rs index 5991912..9f728dd 100644 --- a/src/machine.rs +++ b/src/machine.rs @@ -1,16 +1,10 @@ use crate::entities::try_read_character_reference; use crate::utils::{ - ascii_digit_pat, control_pat, noncharacter_pat, surrogate_pat, whitespace_pat, ControlToken, - State, + ascii_digit_pat, control_pat, ctostr, noncharacter_pat, surrogate_pat, whitespace_pat, + ControlToken, State, }; use crate::{Emitter, Error, Reader, Tokenizer}; -macro_rules! ctostr { - ($c:expr) => { - &*$c.encode_utf8(&mut [0; 4]) - }; -} - // Note: This is not implemented as a method on Tokenizer because there's fields on Tokenizer that // should not be available in this method, such as Tokenizer.to_reconsume or the Reader instance #[inline] diff --git a/src/tokenizer.rs b/src/tokenizer.rs index d7e60ac..7430cbc 100644 --- a/src/tokenizer.rs +++ b/src/tokenizer.rs @@ -3,7 +3,7 @@ use crate::utils::{control_pat, noncharacter_pat, surrogate_pat, ControlToken, S use crate::{DefaultEmitter, Emitter, Error, Never, Readable, Reader}; // this is a stack that can hold 0 to 2 Ts -#[derive(Debug, Default)] +#[derive(Debug, Default, Clone, Copy)] struct Stack2<T: Copy>(Option<(T, Option<T>)>); impl<T: Copy> Stack2<T> { @@ -26,11 +26,6 @@ impl<T: Copy> Stack2<T> { self.0 = new_self; rv } - - #[inline] - fn is_empty(&self) -> bool { - matches!(self.0, None) - } } /// A HTML tokenizer. See crate-level docs for basic usage. @@ -154,11 +149,25 @@ impl<R: Reader, E: Emitter> Tokenizer<R, E> { #[inline] pub(crate) fn try_read_string( &mut self, - s: &str, + mut s: &str, case_sensitive: bool, ) -> Result<bool, R::Error> { debug_assert!(!s.is_empty()); - debug_assert!(self.to_reconsume.is_empty()); + + let to_reconsume_bak = self.to_reconsume; + let mut chars = s.chars(); + while let Some(c) = self.to_reconsume.pop() { + if let (Some(x), Some(x2)) = (c, chars.next()) { + if x == x2 || (!case_sensitive && x.to_ascii_lowercase() == x2.to_ascii_lowercase()) { + s = &s[x.len_utf8()..]; + continue; + } + } + + self.to_reconsume = to_reconsume_bak; + return Ok(false); + } + self.reader.try_read_string(s, case_sensitive) } diff --git a/src/utils.rs b/src/utils.rs index 67db1b9..c3d2f1a 100644 --- a/src/utils.rs +++ b/src/utils.rs @@ -162,3 +162,11 @@ pub enum ControlToken { Eof, Continue, } + +macro_rules! ctostr { + ($c:expr) => { + &*$c.encode_utf8(&mut [0; 4]) + }; +} + +pub(crate) use ctostr; |