Coverage Report

Created: 2026-01-13 06:57

next uncovered line (L), next uncovered region (R), next uncovered branch (B)
/src/html5ever/tendril/src/utf8_decode.rs
Line
Count
Source
1
// Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
2
// https://www.apache.org/licenses/LICENSE-2.0> or the MIT license
3
// <LICENSE-MIT or https://opensource.org/licenses/MIT>, at your
4
// option. This file may not be copied, modified, or distributed
5
// except according to those terms.
6
7
use crate::fmt;
8
use crate::{Atomicity, Tendril};
9
10
pub struct IncompleteUtf8(utf8::Incomplete);
11
12
impl<A> Tendril<fmt::Bytes, A>
13
where
14
    A: Atomicity,
15
{
16
0
    pub fn decode_utf8_lossy<F>(mut self, mut push_utf8: F) -> Option<IncompleteUtf8>
17
0
    where
18
0
        F: FnMut(Tendril<fmt::UTF8, A>),
19
    {
20
        loop {
21
0
            if self.is_empty() {
22
0
                return None;
23
0
            }
24
0
            let unborrowed_result = match utf8::decode(&self) {
25
0
                Ok(s) => {
26
0
                    debug_assert!(s.as_ptr() == self.as_ptr());
27
0
                    debug_assert!(s.len() == self.len());
28
0
                    Ok(())
29
                },
30
                Err(utf8::DecodeError::Invalid {
31
0
                    valid_prefix,
32
0
                    invalid_sequence,
33
                    ..
34
                }) => {
35
0
                    debug_assert!(valid_prefix.as_ptr() == self.as_ptr());
36
0
                    debug_assert!(valid_prefix.len() <= self.len());
37
0
                    Err((
38
0
                        valid_prefix.len(),
39
0
                        Err(valid_prefix.len() + invalid_sequence.len()),
40
0
                    ))
41
                },
42
                Err(utf8::DecodeError::Incomplete {
43
0
                    valid_prefix,
44
0
                    incomplete_suffix,
45
                }) => {
46
0
                    debug_assert!(valid_prefix.as_ptr() == self.as_ptr());
47
0
                    debug_assert!(valid_prefix.len() <= self.len());
48
0
                    Err((valid_prefix.len(), Ok(incomplete_suffix)))
49
                },
50
            };
51
0
            match unborrowed_result {
52
                Ok(()) => {
53
0
                    unsafe { push_utf8(self.reinterpret_without_validating()) }
54
0
                    return None;
55
                },
56
0
                Err((valid_len, and_then)) => {
57
0
                    if valid_len > 0 {
58
0
                        let subtendril = self.subtendril(0, valid_len as u32);
59
0
                        unsafe { push_utf8(subtendril.reinterpret_without_validating()) }
60
0
                    }
61
0
                    match and_then {
62
0
                        Ok(incomplete) => return Some(IncompleteUtf8(incomplete)),
63
0
                        Err(offset) => {
64
0
                            push_utf8(Tendril::from_slice(utf8::REPLACEMENT_CHARACTER));
65
0
                            self.pop_front(offset as u32)
66
                        },
67
                    }
68
                },
69
            }
70
        }
71
0
    }
72
}
73
74
impl IncompleteUtf8 {
75
0
    pub fn try_complete<A, F>(
76
0
        &mut self,
77
0
        mut input: Tendril<fmt::Bytes, A>,
78
0
        mut push_utf8: F,
79
0
    ) -> Result<Tendril<fmt::Bytes, A>, ()>
80
0
    where
81
0
        A: Atomicity,
82
0
        F: FnMut(Tendril<fmt::UTF8, A>),
83
    {
84
0
        let resume_at = match self.0.try_complete(&input) {
85
0
            None => return Err(()),
86
0
            Some((result, rest)) => {
87
0
                push_utf8(Tendril::from_slice(
88
0
                    result.unwrap_or(utf8::REPLACEMENT_CHARACTER),
89
0
                ));
90
0
                input.len() - rest.len()
91
            },
92
        };
93
0
        input.pop_front(resume_at as u32);
94
0
        Ok(input)
95
0
    }
96
}