/rust/registry/src/index.crates.io-1949cf8c6b5b557f/tinystr-0.8.1/src/int_ops.rs
Line | Count | Source |
1 | | // This file is part of ICU4X. For terms of use, please see the file |
2 | | // called LICENSE at the top level of the ICU4X source tree |
3 | | // (online at: https://github.com/unicode-org/icu4x/blob/main/LICENSE ). |
4 | | |
5 | | use crate::asciibyte::AsciiByte; |
6 | | |
7 | | /// Internal helper struct that performs operations on aligned integers. |
8 | | /// Supports strings up to 4 bytes long. |
9 | | #[repr(transparent)] |
10 | | pub struct Aligned4(u32); |
11 | | |
12 | | impl Aligned4 { |
13 | | /// # Panics |
14 | | /// Panics if N is greater than 4 |
15 | | #[inline] |
16 | 0 | pub const fn from_utf8<const N: usize>(src: &[u8; N]) -> Self { |
17 | 0 | let mut bytes = [0; 4]; |
18 | 0 | let mut i = 0; |
19 | | // The function documentation defines when panics may occur |
20 | | #[allow(clippy::indexing_slicing)] |
21 | 0 | while i < N { |
22 | 0 | bytes[i] = src[i]; |
23 | 0 | i += 1; |
24 | 0 | } |
25 | 0 | Self(u32::from_ne_bytes(bytes)) |
26 | 0 | } Unexecuted instantiation: <tinystr::int_ops::Aligned4>::from_utf8::<2> Unexecuted instantiation: <tinystr::int_ops::Aligned4>::from_utf8::<3> Unexecuted instantiation: <tinystr::int_ops::Aligned4>::from_utf8::<4> Unexecuted instantiation: <tinystr::int_ops::Aligned4>::from_utf8::<8> Unexecuted instantiation: <tinystr::int_ops::Aligned4>::from_utf8::<_> |
27 | | |
28 | | #[inline] |
29 | 0 | pub const fn from_ascii_bytes<const N: usize>(src: &[AsciiByte; N]) -> Self { |
30 | 0 | Self::from_utf8::<N>(unsafe { core::mem::transmute::<&[AsciiByte; N], &[u8; N]>(src) }) |
31 | 0 | } Unexecuted instantiation: <tinystr::int_ops::Aligned4>::from_ascii_bytes::<2> Unexecuted instantiation: <tinystr::int_ops::Aligned4>::from_ascii_bytes::<3> Unexecuted instantiation: <tinystr::int_ops::Aligned4>::from_ascii_bytes::<4> Unexecuted instantiation: <tinystr::int_ops::Aligned4>::from_ascii_bytes::<8> Unexecuted instantiation: <tinystr::int_ops::Aligned4>::from_ascii_bytes::<_> |
32 | | |
33 | | #[inline] |
34 | 0 | pub const fn to_bytes(&self) -> [u8; 4] { |
35 | 0 | self.0.to_ne_bytes() |
36 | 0 | } Unexecuted instantiation: <tinystr::int_ops::Aligned4>::to_bytes Unexecuted instantiation: <tinystr::int_ops::Aligned4>::to_bytes |
37 | | |
38 | | #[inline] |
39 | 0 | pub const fn to_ascii_bytes(&self) -> [AsciiByte; 4] { |
40 | 0 | unsafe { core::mem::transmute(self.to_bytes()) } |
41 | 0 | } Unexecuted instantiation: <tinystr::int_ops::Aligned4>::to_ascii_bytes Unexecuted instantiation: <tinystr::int_ops::Aligned4>::to_ascii_bytes |
42 | | |
43 | 0 | pub const fn len(&self) -> usize { |
44 | 0 | let word = self.0; |
45 | | #[cfg(target_endian = "little")] |
46 | 0 | let len = (4 - word.leading_zeros() / 8) as usize; |
47 | | #[cfg(target_endian = "big")] |
48 | | let len = (4 - word.trailing_zeros() / 8) as usize; |
49 | 0 | len |
50 | 0 | } |
51 | | |
52 | 0 | pub const fn is_ascii_alphabetic(&self) -> bool { |
53 | 0 | let word = self.0; |
54 | | // Each of the following bitmasks set *the high bit* (0x8) to 0 for valid and 1 for invalid. |
55 | | // `mask` sets all NUL bytes to 0. |
56 | 0 | let mask = (word + 0x7f7f_7f7f) & 0x8080_8080; |
57 | | // `lower` converts the string to lowercase. It may also change the value of non-alpha |
58 | | // characters, but this does not matter for the alphabetic test that follows. |
59 | 0 | let lower = word | 0x2020_2020; |
60 | | // `alpha` sets all alphabetic bytes to 0. We only need check for lowercase characters. |
61 | 0 | let alpha = !(lower + 0x1f1f_1f1f) | (lower + 0x0505_0505); |
62 | | // The overall string is valid if every character passes at least one test. |
63 | | // We performed two tests here: non-NUL (`mask`) and alphabetic (`alpha`). |
64 | 0 | (alpha & mask) == 0 |
65 | 0 | } |
66 | | |
67 | 0 | pub const fn is_ascii_alphanumeric(&self) -> bool { |
68 | 0 | let word = self.0; |
69 | | // See explanatory comments in is_ascii_alphabetic |
70 | 0 | let mask = (word + 0x7f7f_7f7f) & 0x8080_8080; |
71 | 0 | let numeric = !(word + 0x5050_5050) | (word + 0x4646_4646); |
72 | 0 | let lower = word | 0x2020_2020; |
73 | 0 | let alpha = !(lower + 0x1f1f_1f1f) | (lower + 0x0505_0505); |
74 | 0 | (alpha & numeric & mask) == 0 |
75 | 0 | } |
76 | | |
77 | 0 | pub const fn is_ascii_numeric(&self) -> bool { |
78 | 0 | let word = self.0; |
79 | | // See explanatory comments in is_ascii_alphabetic |
80 | 0 | let mask = (word + 0x7f7f_7f7f) & 0x8080_8080; |
81 | 0 | let numeric = !(word + 0x5050_5050) | (word + 0x4646_4646); |
82 | 0 | (numeric & mask) == 0 |
83 | 0 | } |
84 | | |
85 | 0 | pub const fn is_ascii_lowercase(&self) -> bool { |
86 | 0 | let word = self.0; |
87 | | // For efficiency, this function tests for an invalid string rather than a valid string. |
88 | | // A string is ASCII lowercase iff it contains no uppercase ASCII characters. |
89 | | // `invalid_case` sets all uppercase ASCII characters to 0 and all others to 1. |
90 | 0 | let invalid_case = !(word + 0x3f3f_3f3f) | (word + 0x2525_2525); |
91 | | // The string is valid if it contains no invalid characters (if all high bits are 1). |
92 | 0 | (invalid_case & 0x8080_8080) == 0x8080_8080 |
93 | 0 | } |
94 | | |
95 | 0 | pub const fn is_ascii_titlecase(&self) -> bool { |
96 | 0 | let word = self.0; |
97 | | // See explanatory comments in is_ascii_lowercase |
98 | 0 | let invalid_case = if cfg!(target_endian = "little") { |
99 | 0 | !(word + 0x3f3f_3f1f) | (word + 0x2525_2505) |
100 | | } else { |
101 | 0 | !(word + 0x1f3f_3f3f) | (word + 0x0525_2525) |
102 | | }; |
103 | 0 | (invalid_case & 0x8080_8080) == 0x8080_8080 |
104 | 0 | } |
105 | | |
106 | 0 | pub const fn is_ascii_uppercase(&self) -> bool { |
107 | 0 | let word = self.0; |
108 | | // See explanatory comments in is_ascii_lowercase |
109 | 0 | let invalid_case = !(word + 0x1f1f_1f1f) | (word + 0x0505_0505); |
110 | 0 | (invalid_case & 0x8080_8080) == 0x8080_8080 |
111 | 0 | } |
112 | | |
113 | 0 | pub const fn is_ascii_alphabetic_lowercase(&self) -> bool { |
114 | 0 | let word = self.0; |
115 | | // `mask` sets all NUL bytes to 0. |
116 | 0 | let mask = (word + 0x7f7f_7f7f) & 0x8080_8080; |
117 | | // `lower_alpha` sets all lowercase ASCII characters to 0 and all others to 1. |
118 | 0 | let lower_alpha = !(word + 0x1f1f_1f1f) | (word + 0x0505_0505); |
119 | | // The overall string is valid if every character passes at least one test. |
120 | | // We performed two tests here: non-NUL (`mask`) and lowercase ASCII character (`alpha`). |
121 | 0 | (lower_alpha & mask) == 0 |
122 | 0 | } |
123 | | |
124 | 0 | pub const fn is_ascii_alphabetic_titlecase(&self) -> bool { |
125 | 0 | let word = self.0; |
126 | | // See explanatory comments in is_ascii_alphabetic_lowercase |
127 | 0 | let mask = (word + 0x7f7f_7f7f) & 0x8080_8080; |
128 | 0 | let title_case = if cfg!(target_endian = "little") { |
129 | 0 | !(word + 0x1f1f_1f3f) | (word + 0x0505_0525) |
130 | | } else { |
131 | 0 | !(word + 0x3f1f_1f1f) | (word + 0x2505_0505) |
132 | | }; |
133 | 0 | (title_case & mask) == 0 |
134 | 0 | } |
135 | | |
136 | 0 | pub const fn is_ascii_alphabetic_uppercase(&self) -> bool { |
137 | 0 | let word = self.0; |
138 | | // See explanatory comments in is_ascii_alphabetic_lowercase |
139 | 0 | let mask = (word + 0x7f7f_7f7f) & 0x8080_8080; |
140 | 0 | let upper_alpha = !(word + 0x3f3f_3f3f) | (word + 0x2525_2525); |
141 | 0 | (upper_alpha & mask) == 0 |
142 | 0 | } |
143 | | |
144 | 0 | pub const fn to_ascii_lowercase(&self) -> Self { |
145 | 0 | let word = self.0; |
146 | 0 | let result = word | (((word + 0x3f3f_3f3f) & !(word + 0x2525_2525) & 0x8080_8080) >> 2); |
147 | 0 | Self(result) |
148 | 0 | } |
149 | | |
150 | 0 | pub const fn to_ascii_titlecase(&self) -> Self { |
151 | 0 | let word = self.0.to_le(); |
152 | 0 | let mask = ((word + 0x3f3f_3f1f) & !(word + 0x2525_2505) & 0x8080_8080) >> 2; |
153 | 0 | let result = (word | mask) & !(0x20 & mask); |
154 | 0 | Self(u32::from_le(result)) |
155 | 0 | } |
156 | | |
157 | 0 | pub const fn to_ascii_uppercase(&self) -> Self { |
158 | 0 | let word = self.0; |
159 | 0 | let result = word & !(((word + 0x1f1f_1f1f) & !(word + 0x0505_0505) & 0x8080_8080) >> 2); |
160 | 0 | Self(result) |
161 | 0 | } |
162 | | } |
163 | | |
164 | | /// Internal helper struct that performs operations on aligned integers. |
165 | | /// Supports strings up to 8 bytes long. |
166 | | #[repr(transparent)] |
167 | | pub struct Aligned8(u64); |
168 | | |
169 | | impl Aligned8 { |
170 | | /// # Panics |
171 | | /// Panics if N is greater than 8 |
172 | | #[inline] |
173 | 0 | pub const fn from_utf8<const N: usize>(src: &[u8; N]) -> Self { |
174 | 0 | let mut bytes = [0; 8]; |
175 | 0 | let mut i = 0; |
176 | | // The function documentation defines when panics may occur |
177 | | #[allow(clippy::indexing_slicing)] |
178 | 0 | while i < N { |
179 | 0 | bytes[i] = src[i]; |
180 | 0 | i += 1; |
181 | 0 | } |
182 | 0 | Self(u64::from_ne_bytes(bytes)) |
183 | 0 | } Unexecuted instantiation: <tinystr::int_ops::Aligned8>::from_utf8::<2> Unexecuted instantiation: <tinystr::int_ops::Aligned8>::from_utf8::<3> Unexecuted instantiation: <tinystr::int_ops::Aligned8>::from_utf8::<4> Unexecuted instantiation: <tinystr::int_ops::Aligned8>::from_utf8::<8> Unexecuted instantiation: <tinystr::int_ops::Aligned8>::from_utf8::<_> |
184 | | |
185 | | #[inline] |
186 | 0 | pub const fn from_ascii_bytes<const N: usize>(src: &[AsciiByte; N]) -> Self { |
187 | 0 | Self::from_utf8::<N>(unsafe { core::mem::transmute::<&[AsciiByte; N], &[u8; N]>(src) }) |
188 | 0 | } Unexecuted instantiation: <tinystr::int_ops::Aligned8>::from_ascii_bytes::<2> Unexecuted instantiation: <tinystr::int_ops::Aligned8>::from_ascii_bytes::<3> Unexecuted instantiation: <tinystr::int_ops::Aligned8>::from_ascii_bytes::<4> Unexecuted instantiation: <tinystr::int_ops::Aligned8>::from_ascii_bytes::<8> Unexecuted instantiation: <tinystr::int_ops::Aligned8>::from_ascii_bytes::<_> |
189 | | |
190 | | #[inline] |
191 | 0 | pub const fn to_bytes(&self) -> [u8; 8] { |
192 | 0 | self.0.to_ne_bytes() |
193 | 0 | } Unexecuted instantiation: <tinystr::int_ops::Aligned8>::to_bytes Unexecuted instantiation: <tinystr::int_ops::Aligned8>::to_bytes |
194 | | |
195 | | #[inline] |
196 | 0 | pub const fn to_ascii_bytes(&self) -> [AsciiByte; 8] { |
197 | 0 | unsafe { core::mem::transmute(self.to_bytes()) } |
198 | 0 | } Unexecuted instantiation: <tinystr::int_ops::Aligned8>::to_ascii_bytes Unexecuted instantiation: <tinystr::int_ops::Aligned8>::to_ascii_bytes |
199 | | |
200 | 0 | pub const fn len(&self) -> usize { |
201 | 0 | let word = self.0; |
202 | | #[cfg(target_endian = "little")] |
203 | 0 | let len = (8 - word.leading_zeros() / 8) as usize; |
204 | | #[cfg(target_endian = "big")] |
205 | | let len = (8 - word.trailing_zeros() / 8) as usize; |
206 | 0 | len |
207 | 0 | } |
208 | | |
209 | 0 | pub const fn is_ascii_alphabetic(&self) -> bool { |
210 | 0 | let word = self.0; |
211 | 0 | let mask = (word + 0x7f7f_7f7f_7f7f_7f7f) & 0x8080_8080_8080_8080; |
212 | 0 | let lower = word | 0x2020_2020_2020_2020; |
213 | 0 | let alpha = !(lower + 0x1f1f_1f1f_1f1f_1f1f) | (lower + 0x0505_0505_0505_0505); |
214 | 0 | (alpha & mask) == 0 |
215 | 0 | } |
216 | | |
217 | 0 | pub const fn is_ascii_alphanumeric(&self) -> bool { |
218 | 0 | let word = self.0; |
219 | 0 | let mask = (word + 0x7f7f_7f7f_7f7f_7f7f) & 0x8080_8080_8080_8080; |
220 | 0 | let numeric = !(word + 0x5050_5050_5050_5050) | (word + 0x4646_4646_4646_4646); |
221 | 0 | let lower = word | 0x2020_2020_2020_2020; |
222 | 0 | let alpha = !(lower + 0x1f1f_1f1f_1f1f_1f1f) | (lower + 0x0505_0505_0505_0505); |
223 | 0 | (alpha & numeric & mask) == 0 |
224 | 0 | } |
225 | | |
226 | 0 | pub const fn is_ascii_numeric(&self) -> bool { |
227 | 0 | let word = self.0; |
228 | 0 | let mask = (word + 0x7f7f_7f7f_7f7f_7f7f) & 0x8080_8080_8080_8080; |
229 | 0 | let numeric = !(word + 0x5050_5050_5050_5050) | (word + 0x4646_4646_4646_4646); |
230 | 0 | (numeric & mask) == 0 |
231 | 0 | } |
232 | | |
233 | 0 | pub const fn is_ascii_lowercase(&self) -> bool { |
234 | 0 | let word = self.0; |
235 | 0 | let invalid_case = !(word + 0x3f3f_3f3f_3f3f_3f3f) | (word + 0x2525_2525_2525_2525); |
236 | 0 | (invalid_case & 0x8080_8080_8080_8080) == 0x8080_8080_8080_8080 |
237 | 0 | } |
238 | | |
239 | 0 | pub const fn is_ascii_titlecase(&self) -> bool { |
240 | 0 | let word = self.0; |
241 | 0 | let invalid_case = if cfg!(target_endian = "little") { |
242 | 0 | !(word + 0x3f3f_3f3f_3f3f_3f1f) | (word + 0x2525_2525_2525_2505) |
243 | | } else { |
244 | 0 | !(word + 0x1f3f_3f3f_3f3f_3f3f) | (word + 0x0525_2525_2525_2525) |
245 | | }; |
246 | 0 | (invalid_case & 0x8080_8080_8080_8080) == 0x8080_8080_8080_8080 |
247 | 0 | } |
248 | | |
249 | 0 | pub const fn is_ascii_uppercase(&self) -> bool { |
250 | 0 | let word = self.0; |
251 | 0 | let invalid_case = !(word + 0x1f1f_1f1f_1f1f_1f1f) | (word + 0x0505_0505_0505_0505); |
252 | 0 | (invalid_case & 0x8080_8080_8080_8080) == 0x8080_8080_8080_8080 |
253 | 0 | } |
254 | | |
255 | 0 | pub const fn is_ascii_alphabetic_lowercase(&self) -> bool { |
256 | 0 | let word = self.0; |
257 | | // `mask` sets all NUL bytes to 0. |
258 | 0 | let mask = (word + 0x7f7f_7f7f_7f7f_7f7f) & 0x8080_8080_8080_8080; |
259 | | // `lower_alpha` sets all lowercase ASCII characters to 0 and all others to 1. |
260 | 0 | let lower_alpha = !(word + 0x1f1f_1f1f_1f1f_1f1f) | (word + 0x0505_0505_0505_0505); |
261 | | // The overall string is valid if every character passes at least one test. |
262 | | // We performed two tests here: non-NUL (`mask`) and lowercase ASCII character (`alpha`). |
263 | 0 | (lower_alpha & mask) == 0 |
264 | 0 | } |
265 | | |
266 | 0 | pub const fn is_ascii_alphabetic_titlecase(&self) -> bool { |
267 | 0 | let word = self.0; |
268 | | // See explanatory comments in is_ascii_alphabetic_lowercase |
269 | 0 | let mask = (word + 0x7f7f_7f7f_7f7f_7f7f) & 0x8080_8080_8080_8080; |
270 | 0 | let title_case = if cfg!(target_endian = "little") { |
271 | 0 | !(word + 0x1f1f_1f1f_1f1f_1f3f) | (word + 0x0505_0505_0505_0525) |
272 | | } else { |
273 | 0 | !(word + 0x3f1f_1f1f_1f1f_1f1f) | (word + 0x2505_0505_0505_0505) |
274 | | }; |
275 | 0 | (title_case & mask) == 0 |
276 | 0 | } |
277 | | |
278 | 0 | pub const fn is_ascii_alphabetic_uppercase(&self) -> bool { |
279 | 0 | let word = self.0; |
280 | | // See explanatory comments in is_ascii_alphabetic_lowercase |
281 | 0 | let mask = (word + 0x7f7f_7f7f_7f7f_7f7f) & 0x8080_8080_8080_8080; |
282 | 0 | let upper_alpha = !(word + 0x3f3f_3f3f_3f3f_3f3f) | (word + 0x2525_2525_2525_2525); |
283 | 0 | (upper_alpha & mask) == 0 |
284 | 0 | } |
285 | | |
286 | 0 | pub const fn to_ascii_lowercase(&self) -> Self { |
287 | 0 | let word = self.0; |
288 | 0 | let result = word |
289 | 0 | | (((word + 0x3f3f_3f3f_3f3f_3f3f) |
290 | 0 | & !(word + 0x2525_2525_2525_2525) |
291 | 0 | & 0x8080_8080_8080_8080) |
292 | 0 | >> 2); |
293 | 0 | Self(result) |
294 | 0 | } |
295 | | |
296 | 0 | pub const fn to_ascii_titlecase(&self) -> Self { |
297 | 0 | let word = self.0.to_le(); |
298 | 0 | let mask = ((word + 0x3f3f_3f3f_3f3f_3f1f) |
299 | 0 | & !(word + 0x2525_2525_2525_2505) |
300 | 0 | & 0x8080_8080_8080_8080) |
301 | 0 | >> 2; |
302 | 0 | let result = (word | mask) & !(0x20 & mask); |
303 | 0 | Self(u64::from_le(result)) |
304 | 0 | } |
305 | | |
306 | 0 | pub const fn to_ascii_uppercase(&self) -> Self { |
307 | 0 | let word = self.0; |
308 | 0 | let result = word |
309 | 0 | & !(((word + 0x1f1f_1f1f_1f1f_1f1f) |
310 | 0 | & !(word + 0x0505_0505_0505_0505) |
311 | 0 | & 0x8080_8080_8080_8080) |
312 | 0 | >> 2); |
313 | 0 | Self(result) |
314 | 0 | } |
315 | | } |