qubit_codec_misc/
c_integer_literal_codec.rs

1// =============================================================================
2//    Copyright (c) 2026 Haixing Hu.
3//
4//    SPDX-License-Identifier: Apache-2.0
5//
6//    Licensed under the Apache License, Version 2.0.
7// =============================================================================
8//! C integer literal decoder.
9
10use qubit_codec::ValueDecoder;
11
12use crate::MiscCodecError;
13use crate::MiscCodecResult;
14use crate::internal::LiteralComponents;
15
16/// Decodes restricted non-negative C integer literal fragments.
17///
18/// This codec accepts decimal literals such as `123`, octal literals such as
19/// `0123`, and hexadecimal literals such as `0x123` or `0X123`. It trims
20/// surrounding whitespace and returns a `u64`. It intentionally accepts only
21/// a complete unsigned token: signs, integer suffixes, digit separators, and
22/// token-stream boundaries are outside this codec's contract.
23#[must_use]
24#[derive(Debug, Clone, Copy, Default, PartialEq, Eq)]
25pub struct CIntegerLiteralCodec;
26
27impl CIntegerLiteralCodec {
28    /// Creates a C integer literal codec.
29    ///
30    /// # Returns
31    /// A stateless C integer literal codec.
32    #[inline]
33    pub fn new() -> Self {
34        Self
35    }
36
37    /// Decodes a C integer literal into a `u64`.
38    ///
39    /// # Parameters
40    /// - `text`: C integer literal text.
41    ///
42    /// # Returns
43    /// Parsed integer value.
44    ///
45    /// # Errors
46    /// Returns [`MiscCodecError::InvalidInput`] when the input is empty, lacks
47    /// digits, or overflows `u64`; returns [`MiscCodecError::InvalidDigit`]
48    /// when a character is not valid for the detected radix.
49    #[inline]
50    pub fn decode(&self, text: &str) -> MiscCodecResult<u64> {
51        let (trimmed, trim_offset) = trim_with_offset(text);
52        if trimmed.is_empty() {
53            return Err(invalid_c_integer_input("expected at least one digit"));
54        }
55        let components = LiteralComponents::parse(trimmed, trim_offset)?;
56        validate_digits(components)?;
57        u64::from_str_radix(components.digits, components.radix).map_err(|error| {
58            invalid_c_integer_input(&format!("integer literal is out of range: {error}"))
59        })
60    }
61}
62
63impl ValueDecoder<str> for CIntegerLiteralCodec {
64    type Error = MiscCodecError;
65    type Output = u64;
66
67    /// Decodes a C integer literal into a `u64`.
68    #[inline]
69    fn decode(&mut self, input: &str) -> Result<Self::Output, Self::Error> {
70        CIntegerLiteralCodec::decode(self, input)
71    }
72}
73
74/// Trims surrounding whitespace while preserving the start byte offset.
75///
76/// # Parameters
77/// - `text`: Input text.
78///
79/// # Returns
80/// Trimmed text and the byte offset where it starts in `text`.
81#[inline]
82fn trim_with_offset(text: &str) -> (&str, usize) {
83    let trimmed_start = text.trim_start();
84    let start = text.len() - trimmed_start.len();
85    (trimmed_start.trim_end(), start)
86}
87
88/// Validates that every character is valid for the detected radix.
89///
90/// # Parameters
91/// - `components`: Parsed literal components.
92///
93/// # Errors
94/// Returns [`MiscCodecError::InvalidDigit`] with the original input byte index
95/// of the invalid character.
96fn validate_digits(components: LiteralComponents<'_>) -> MiscCodecResult<()> {
97    for (index, character) in components.digits.char_indices() {
98        if character.is_digit(components.radix) {
99            continue;
100        }
101        return Err(MiscCodecError::InvalidDigit {
102            radix: components.radix,
103            index: components.digits_offset + index,
104            character,
105        });
106    }
107    Ok(())
108}
109
110/// Builds an invalid C integer literal input error.
111///
112/// # Parameters
113/// - `reason`: Human-readable reason the input was rejected.
114///
115/// # Returns
116/// An invalid input error for the C integer literal codec.
117fn invalid_c_integer_input(reason: &str) -> MiscCodecError {
118    MiscCodecError::InvalidInput {
119        codec: "c-integer-literal",
120        reason: reason.to_owned(),
121    }
122}
qubit_codec_misc/c_integer_literal_codec.rs

qubit_codec_misc/
c_integer_literal_codec.rs
