litrs/bytestr/
mod.rs

1use std::{fmt, ops::Range};
2
3use crate::{
4    err::{perr, ParseErrorKind::*},
5    escape::{scan_raw_string, unescape_string},
6    Buffer, ParseError,
7};
8
9
10/// A byte string or raw byte string literal, e.g. `b"hello"` or `br#"abc"def"#`.
11///
12/// See [the reference][ref] for more information.
13///
14/// [ref]: https://doc.rust-lang.org/reference/tokens.html#byte-string-literals
15#[derive(Debug, Clone, PartialEq, Eq)]
16pub struct ByteStringLit<B: Buffer> {
17    /// The raw input.
18    raw: B,
19
20    /// The string value (with all escaped unescaped), or `None` if there were
21    /// no escapes. In the latter case, `input` is the string value.
22    value: Option<Vec<u8>>,
23
24    /// The number of hash signs in case of a raw string literal, or `None` if
25    /// it's not a raw string literal.
26    num_hashes: Option<u8>,
27
28    /// Start index of the suffix or `raw.len()` if there is no suffix.
29    start_suffix: usize,
30}
31
32impl<B: Buffer> ByteStringLit<B> {
33    /// Parses the input as a (raw) byte string literal. Returns an error if the
34    /// input is invalid or represents a different kind of literal.
35    pub fn parse(input: B) -> Result<Self, ParseError> {
36        if input.is_empty() {
37            return Err(perr(None, Empty));
38        }
39        if !input.starts_with(r#"b""#) && !input.starts_with("br") {
40            return Err(perr(None, InvalidByteStringLiteralStart));
41        }
42
43        let (value, num_hashes, start_suffix) = parse_impl(&input)?;
44        Ok(Self { raw: input, value, num_hashes, start_suffix })
45    }
46
47    /// Returns the string value this literal represents (where all escapes have
48    /// been turned into their respective values).
49    pub fn value(&self) -> &[u8] {
50        self.value
51            .as_deref()
52            .unwrap_or(&self.raw.as_bytes()[self.inner_range()])
53    }
54
55    /// Like `value` but returns a potentially owned version of the value.
56    ///
57    /// The return value is either `Vec<u8>` if `B = String`, or
58    /// `Cow<'a, [u8]>` if `B = &'a str`.
59    pub fn into_value(self) -> B::ByteCow {
60        let inner_range = self.inner_range();
61        let Self { raw, value, .. } = self;
62        value
63            .map(B::ByteCow::from)
64            .unwrap_or_else(|| raw.cut(inner_range).into_byte_cow())
65    }
66
67    /// The optional suffix. Returns `""` if the suffix is empty/does not exist.
68    pub fn suffix(&self) -> &str {
69        &(*self.raw)[self.start_suffix..]
70    }
71
72    /// Returns whether this literal is a raw string literal (starting with
73    /// `r`).
74    pub fn is_raw_byte_string(&self) -> bool {
75        self.num_hashes.is_some()
76    }
77
78    /// Returns the raw input that was passed to `parse`.
79    pub fn raw_input(&self) -> &str {
80        &self.raw
81    }
82
83    /// Returns the raw input that was passed to `parse`, potentially owned.
84    pub fn into_raw_input(self) -> B {
85        self.raw
86    }
87
88    /// The range within `self.raw` that excludes the quotes and potential `r#`.
89    fn inner_range(&self) -> Range<usize> {
90        match self.num_hashes {
91            None => 2..self.start_suffix - 1,
92            Some(n) => 2 + n as usize + 1..self.start_suffix - n as usize - 1,
93        }
94    }
95}
96
97impl ByteStringLit<&str> {
98    /// Makes a copy of the underlying buffer and returns the owned version of
99    /// `Self`.
100    pub fn into_owned(self) -> ByteStringLit<String> {
101        ByteStringLit {
102            raw: self.raw.to_owned(),
103            value: self.value,
104            num_hashes: self.num_hashes,
105            start_suffix: self.start_suffix,
106        }
107    }
108}
109
110impl<B: Buffer> fmt::Display for ByteStringLit<B> {
111    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
112        f.pad(&self.raw)
113    }
114}
115
116
117/// Precondition: input has to start with either `b"` or `br`.
118#[inline(never)]
119fn parse_impl(input: &str) -> Result<(Option<Vec<u8>>, Option<u8>, usize), ParseError> {
120    if input.starts_with("br") {
121        scan_raw_string(input, 2, false, true)
122            .map(|(num, start_suffix)| (None, Some(num), start_suffix))
123    } else {
124        unescape_string::<Vec<u8>>(input, 2, false, true, true)
125            .map(|(v, start_suffix)| (v, None, start_suffix))
126    }
127}
128
129#[cfg(test)]
130mod tests;
litrs/bytestr/mod.rs

litrs/bytestr/
mod.rs
