qubit_text_codec/codec/
utf32_u32_codec.rs

1/*******************************************************************************
2 *
3 *    Copyright (c) 2026 Haixing Hu.
4 *
5 *    SPDX-License-Identifier: Apache-2.0
6 *
7 *    Licensed under the Apache License, Version 2.0.
8 *
9 ******************************************************************************/
10use super::inner::utf32;
11use crate::{
12    Charset,
13    CharsetCodec,
14    CharsetDecodeResult,
15    CharsetEncodeResult,
16    DecodeStatus,
17    Utf32,
18};
19
20/// Combined UTF-32 `u32` code-unit codec.
21///
22/// `Utf32U32Codec` works with raw UTF-32 scalar-value units rather than
23/// serialized bytes. Use [`crate::Utf32ByteCodec`] for byte streams with an
24/// explicit byte order.
25///
26/// # Examples
27///
28/// ```rust
29/// use qubit_text_codec::{
30///     CharsetCodec,
31///     DecodeStatus,
32///     Charset,
33///     Utf32,
34///     Utf32U32Codec,
35/// };
36///
37/// let codec = Utf32U32Codec;
38/// assert_eq!(Charset::UTF_32, codec.charset());
39/// assert_eq!(Utf32::MAX_UNITS_PER_CHAR, codec.max_units_per_char());
40///
41/// let mut output = [0_u32; Utf32::MAX_UNITS_PER_CHAR];
42/// let written = codec.encode_one('中', &mut output, 0).expect("buffer fits");
43/// assert_eq!(
44///     DecodeStatus::Complete {
45///         value: '中',
46///         consumed: written,
47///     },
48///     codec.decode_one(&output[..written], 0).expect("valid UTF-32"),
49/// );
50/// ```
51#[derive(Clone, Copy, Debug, Default, Eq, Hash, PartialEq)]
52pub struct Utf32U32Codec;
53
54impl Utf32U32Codec {
55    /// Returns the UTF-32 encoding descriptor.
56    ///
57    /// # Returns
58    ///
59    /// Returns [`Charset::UTF_32`].
60    #[must_use]
61    #[inline]
62    pub const fn charset(self) -> Charset {
63        Charset::UTF_32
64    }
65
66    /// Returns the maximum number of UTF-32 code units needed for one character.
67    ///
68    /// # Returns
69    ///
70    /// Returns [`Utf32::MAX_UNITS_PER_CHAR`].
71    #[must_use]
72    #[inline]
73    pub const fn max_units_per_char(self) -> usize {
74        Utf32::MAX_UNITS_PER_CHAR
75    }
76}
77
78impl CharsetCodec for Utf32U32Codec {
79    type Unit = u32;
80    /// Returns UTF-32 charset descriptor.
81    ///
82    /// # Returns
83    ///
84    /// Returns [`Charset::UTF_32`].
85    #[inline]
86    fn charset(&self) -> Charset {
87        Charset::UTF_32
88    }
89
90    /// Returns the fixed size (1 unit) for one UTF-32 scalar value.
91    ///
92    /// # Returns
93    ///
94    /// Returns [`Utf32::MAX_UNITS_PER_CHAR`].
95    #[inline]
96    fn max_units_per_char(&self) -> usize {
97        Utf32::MAX_UNITS_PER_CHAR
98    }
99
100    /// Decodes one UTF-32 scalar value from a `u32` prefix.
101    ///
102    /// # Arguments
103    ///
104    /// * `input` - UTF-32 unit slice.
105    /// * `index` - Start offset for parsing; must satisfy `index <= input.len()`.
106    ///
107    /// # Returns
108    ///
109    /// * `Ok(DecodeStatus::NeedMore { required, available })` when no unit is available.
110    /// * `Ok(DecodeStatus::Complete { value, consumed })` with `consumed == 1`.
111    ///
112    /// # Errors
113    ///
114    /// * [`crate::CharsetDecodeErrorKind::MalformedSequence`] when index is out
115    ///   of bounds.
116    /// * [`crate::CharsetDecodeErrorKind::InvalidCodePoint`] when unit is not a
117    ///   valid scalar.
118    fn decode_one(&self, input: &[u32], index: usize) -> CharsetDecodeResult<DecodeStatus> {
119        utf32::decode_units_prefix(input, index)
120    }
121
122    /// Encodes one Unicode scalar value into a `u32` unit at `index`.
123    ///
124    /// # Arguments
125    ///
126    /// * `ch` - The Unicode scalar value to encode.
127    /// * `output` - Destination `u32` buffer.
128    /// * `index` - Start offset where one unit is written; must satisfy
129    ///   `index < output.len()`.
130    ///
131    /// # Returns
132    ///
133    /// Always returns `Ok(1)` on success.
134    ///
135    /// # Errors
136    ///
137    /// * [`crate::CharsetEncodeErrorKind::BufferTooSmall`] if `output` has no
138    ///   room at `index`.
139    fn encode_one(&self, ch: char, output: &mut [u32], index: usize) -> CharsetEncodeResult<usize> {
140        utf32::encode_units_char(ch, output, index)
141    }
142}
qubit_text_codec/codec/utf32_u32_codec.rs

qubit_text_codec/codec/
utf32_u32_codec.rs
