Coverage Report

Created: 2025-12-31 06:16

next uncovered line (L), next uncovered region (R), next uncovered branch (B)
/rust/registry/src/index.crates.io-1949cf8c6b5b557f/uuid-1.8.0/src/parser.rs
Line
Count
Source
1
// Copyright 2013-2014 The Rust Project Developers.
2
// Copyright 2018 The Uuid Project Developers.
3
//
4
// See the COPYRIGHT file at the top-level directory of this distribution.
5
//
6
// Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
7
// http://www.apache.org/licenses/LICENSE-2.0> or the MIT license
8
// <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
9
// option. This file may not be copied, modified, or distributed
10
// except according to those terms.
11
12
//! [`Uuid`] parsing constructs and utilities.
13
//!
14
//! [`Uuid`]: ../struct.Uuid.html
15
16
use crate::{
17
    error::*,
18
    std::{convert::TryFrom, str},
19
    Uuid,
20
};
21
22
impl str::FromStr for Uuid {
23
    type Err = Error;
24
25
0
    fn from_str(uuid_str: &str) -> Result<Self, Self::Err> {
26
0
        Uuid::parse_str(uuid_str)
27
0
    }
28
}
29
30
impl TryFrom<&'_ str> for Uuid {
31
    type Error = Error;
32
33
0
    fn try_from(uuid_str: &'_ str) -> Result<Self, Self::Error> {
34
0
        Uuid::parse_str(uuid_str)
35
0
    }
36
}
37
38
impl Uuid {
39
    /// Parses a `Uuid` from a string of hexadecimal digits with optional
40
    /// hyphens.
41
    ///
42
    /// Any of the formats generated by this module (simple, hyphenated, urn,
43
    /// Microsoft GUID) are supported by this parsing function.
44
    ///
45
    /// Prefer [`try_parse`] unless you need detailed user-facing diagnostics.
46
    /// This method will be eventually deprecated in favor of `try_parse`.
47
    ///
48
    /// # Examples
49
    ///
50
    /// Parse a hyphenated UUID:
51
    ///
52
    /// ```
53
    /// # use uuid::{Uuid, Version, Variant};
54
    /// # fn main() -> Result<(), uuid::Error> {
55
    /// let uuid = Uuid::parse_str("550e8400-e29b-41d4-a716-446655440000")?;
56
    ///
57
    /// assert_eq!(Some(Version::Random), uuid.get_version());
58
    /// assert_eq!(Variant::RFC4122, uuid.get_variant());
59
    /// # Ok(())
60
    /// # }
61
    /// ```
62
    ///
63
    /// [`try_parse`]: #method.try_parse
64
0
    pub fn parse_str(input: &str) -> Result<Uuid, Error> {
65
0
        try_parse(input.as_bytes())
66
0
            .map(Uuid::from_bytes)
67
0
            .map_err(InvalidUuid::into_err)
68
0
    }
69
70
    /// Parses a `Uuid` from a string of hexadecimal digits with optional
71
    /// hyphens.
72
    ///
73
    /// This function is similar to [`parse_str`], in fact `parse_str` shares
74
    /// the same underlying parser. The difference is that if `try_parse`
75
    /// fails, it won't generate very useful error messages. The `parse_str`
76
    /// function will eventually be deprecated in favor of `try_parse`.
77
    ///
78
    /// To parse a UUID from a byte stream instead of a UTF8 string, see
79
    /// [`try_parse_ascii`].
80
    ///
81
    /// # Examples
82
    ///
83
    /// Parse a hyphenated UUID:
84
    ///
85
    /// ```
86
    /// # use uuid::{Uuid, Version, Variant};
87
    /// # fn main() -> Result<(), uuid::Error> {
88
    /// let uuid = Uuid::try_parse("550e8400-e29b-41d4-a716-446655440000")?;
89
    ///
90
    /// assert_eq!(Some(Version::Random), uuid.get_version());
91
    /// assert_eq!(Variant::RFC4122, uuid.get_variant());
92
    /// # Ok(())
93
    /// # }
94
    /// ```
95
    ///
96
    /// [`parse_str`]: #method.parse_str
97
    /// [`try_parse_ascii`]: #method.try_parse_ascii
98
0
    pub const fn try_parse(input: &str) -> Result<Uuid, Error> {
99
0
        Self::try_parse_ascii(input.as_bytes())
100
0
    }
101
102
    /// Parses a `Uuid` from a string of hexadecimal digits with optional
103
    /// hyphens.
104
    ///
105
    /// The input is expected to be a string of ASCII characters. This method
106
    /// can be more convenient than [`try_parse`] if the UUID is being
107
    /// parsed from a byte stream instead of from a UTF8 string.
108
    ///
109
    /// # Examples
110
    ///
111
    /// Parse a hyphenated UUID:
112
    ///
113
    /// ```
114
    /// # use uuid::{Uuid, Version, Variant};
115
    /// # fn main() -> Result<(), uuid::Error> {
116
    /// let uuid = Uuid::try_parse_ascii(b"550e8400-e29b-41d4-a716-446655440000")?;
117
    ///
118
    /// assert_eq!(Some(Version::Random), uuid.get_version());
119
    /// assert_eq!(Variant::RFC4122, uuid.get_variant());
120
    /// # Ok(())
121
    /// # }
122
    /// ```
123
    ///
124
    /// [`try_parse`]: #method.try_parse
125
0
    pub const fn try_parse_ascii(input: &[u8]) -> Result<Uuid, Error> {
126
0
        match try_parse(input) {
127
0
            Ok(bytes) => Ok(Uuid::from_bytes(bytes)),
128
            // If parsing fails then we don't know exactly what went wrong
129
            // In this case, we just return a generic error
130
0
            Err(_) => Err(Error(ErrorKind::Other)),
131
        }
132
0
    }
133
}
134
135
0
const fn try_parse(input: &[u8]) -> Result<[u8; 16], InvalidUuid> {
136
0
    match (input.len(), input) {
137
        // Inputs of 32 bytes must be a non-hyphenated UUID
138
0
        (32, s) => parse_simple(s),
139
        // Hyphenated UUIDs may be wrapped in various ways:
140
        // - `{UUID}` for braced UUIDs
141
        // - `urn:uuid:UUID` for URNs
142
        // - `UUID` for a regular hyphenated UUID
143
0
        (36, s)
144
0
        | (38, [b'{', s @ .., b'}'])
145
0
        | (45, [b'u', b'r', b'n', b':', b'u', b'u', b'i', b'd', b':', s @ ..]) => {
146
0
            parse_hyphenated(s)
147
        }
148
        // Any other shaped input is immediately invalid
149
0
        _ => Err(InvalidUuid(input)),
150
    }
151
0
}
152
153
#[inline]
154
0
pub(crate) const fn parse_braced(input: &[u8]) -> Result<[u8; 16], InvalidUuid> {
155
0
    if let (38, [b'{', s @ .., b'}']) = (input.len(), input) {
156
0
        parse_hyphenated(s)
157
    } else {
158
0
        Err(InvalidUuid(input))
159
    }
160
0
}
161
162
#[inline]
163
0
pub(crate) const fn parse_urn(input: &[u8]) -> Result<[u8; 16], InvalidUuid> {
164
0
    if let (45, [b'u', b'r', b'n', b':', b'u', b'u', b'i', b'd', b':', s @ ..]) =
165
0
        (input.len(), input)
166
    {
167
0
        parse_hyphenated(s)
168
    } else {
169
0
        Err(InvalidUuid(input))
170
    }
171
0
}
172
173
#[inline]
174
0
pub(crate) const fn parse_simple(s: &[u8]) -> Result<[u8; 16], InvalidUuid> {
175
    // This length check here removes all other bounds
176
    // checks in this function
177
0
    if s.len() != 32 {
178
0
        return Err(InvalidUuid(s));
179
0
    }
180
181
0
    let mut buf: [u8; 16] = [0; 16];
182
0
    let mut i = 0;
183
184
0
    while i < 16 {
185
        // Convert a two-char hex value (like `A8`)
186
        // into a byte (like `10101000`)
187
0
        let h1 = HEX_TABLE[s[i * 2] as usize];
188
0
        let h2 = HEX_TABLE[s[i * 2 + 1] as usize];
189
190
        // We use `0xff` as a sentinel value to indicate
191
        // an invalid hex character sequence (like the letter `G`)
192
0
        if h1 | h2 == 0xff {
193
0
            return Err(InvalidUuid(s));
194
0
        }
195
196
        // The upper nibble needs to be shifted into position
197
        // to produce the final byte value
198
0
        buf[i] = SHL4_TABLE[h1 as usize] | h2;
199
0
        i += 1;
200
    }
201
202
0
    Ok(buf)
203
0
}
204
205
#[inline]
206
0
const fn parse_hyphenated(s: &[u8]) -> Result<[u8; 16], InvalidUuid> {
207
    // This length check here removes all other bounds
208
    // checks in this function
209
0
    if s.len() != 36 {
210
0
        return Err(InvalidUuid(s));
211
0
    }
212
213
    // We look at two hex-encoded values (4 chars) at a time because
214
    // that's the size of the smallest group in a hyphenated UUID.
215
    // The indexes we're interested in are:
216
    //
217
    // uuid     : 936da01f-9abd-4d9d-80c7-02af85c822a8
218
    //            |   |   ||   ||   ||   ||   |   |
219
    // hyphens  : |   |   8|  13|  18|  23|   |   |
220
    // positions: 0   4    9   14   19   24  28  32
221
222
    // First, ensure the hyphens appear in the right places
223
0
    match [s[8], s[13], s[18], s[23]] {
224
0
        [b'-', b'-', b'-', b'-'] => {}
225
0
        _ => return Err(InvalidUuid(s)),
226
    }
227
228
0
    let positions: [u8; 8] = [0, 4, 9, 14, 19, 24, 28, 32];
229
0
    let mut buf: [u8; 16] = [0; 16];
230
0
    let mut j = 0;
231
232
0
    while j < 8 {
233
0
        let i = positions[j];
234
235
        // The decoding here is the same as the simple case
236
        // We're just dealing with two values instead of one
237
0
        let h1 = HEX_TABLE[s[i as usize] as usize];
238
0
        let h2 = HEX_TABLE[s[(i + 1) as usize] as usize];
239
0
        let h3 = HEX_TABLE[s[(i + 2) as usize] as usize];
240
0
        let h4 = HEX_TABLE[s[(i + 3) as usize] as usize];
241
242
0
        if h1 | h2 | h3 | h4 == 0xff {
243
0
            return Err(InvalidUuid(s));
244
0
        }
245
246
0
        buf[j * 2] = SHL4_TABLE[h1 as usize] | h2;
247
0
        buf[j * 2 + 1] = SHL4_TABLE[h3 as usize] | h4;
248
0
        j += 1;
249
    }
250
251
0
    Ok(buf)
252
0
}
253
254
const HEX_TABLE: &[u8; 256] = &{
255
    let mut buf = [0; 256];
256
    let mut i: u8 = 0;
257
258
    loop {
259
        buf[i as usize] = match i {
260
            b'0'..=b'9' => i - b'0',
261
            b'a'..=b'f' => i - b'a' + 10,
262
            b'A'..=b'F' => i - b'A' + 10,
263
            _ => 0xff,
264
        };
265
266
        if i == 255 {
267
            break buf;
268
        }
269
270
        i += 1
271
    }
272
};
273
274
const SHL4_TABLE: &[u8; 256] = &{
275
    let mut buf = [0; 256];
276
    let mut i: u8 = 0;
277
278
    loop {
279
        buf[i as usize] = i.wrapping_shl(4);
280
281
        if i == 255 {
282
            break buf;
283
        }
284
285
        i += 1;
286
    }
287
};
288
289
#[cfg(test)]
290
mod tests {
291
    use super::*;
292
    use crate::{std::string::ToString, tests::new};
293
294
    #[test]
295
    fn test_parse_uuid_v4_valid() {
296
        let from_hyphenated = Uuid::parse_str("67e55044-10b1-426f-9247-bb680e5fe0c8").unwrap();
297
        let from_simple = Uuid::parse_str("67e5504410b1426f9247bb680e5fe0c8").unwrap();
298
        let from_urn = Uuid::parse_str("urn:uuid:67e55044-10b1-426f-9247-bb680e5fe0c8").unwrap();
299
        let from_guid = Uuid::parse_str("{67e55044-10b1-426f-9247-bb680e5fe0c8}").unwrap();
300
301
        assert_eq!(from_hyphenated, from_simple);
302
        assert_eq!(from_hyphenated, from_urn);
303
        assert_eq!(from_hyphenated, from_guid);
304
305
        assert!(Uuid::parse_str("00000000000000000000000000000000").is_ok());
306
        assert!(Uuid::parse_str("67e55044-10b1-426f-9247-bb680e5fe0c8").is_ok());
307
        assert!(Uuid::parse_str("F9168C5E-CEB2-4faa-B6BF-329BF39FA1E4").is_ok());
308
        assert!(Uuid::parse_str("67e5504410b1426f9247bb680e5fe0c8").is_ok());
309
        assert!(Uuid::parse_str("01020304-1112-2122-3132-414243444546").is_ok());
310
        assert!(Uuid::parse_str("urn:uuid:67e55044-10b1-426f-9247-bb680e5fe0c8").is_ok());
311
        assert!(Uuid::parse_str("{6d93bade-bd9f-4e13-8914-9474e1e3567b}").is_ok());
312
313
        // Nil
314
        let nil = Uuid::nil();
315
        assert_eq!(
316
            Uuid::parse_str("00000000000000000000000000000000").unwrap(),
317
            nil
318
        );
319
        assert_eq!(
320
            Uuid::parse_str("00000000-0000-0000-0000-000000000000").unwrap(),
321
            nil
322
        );
323
    }
324
325
    #[test]
326
    fn test_parse_uuid_v4_invalid() {
327
        // Invalid
328
        assert_eq!(
329
            Uuid::parse_str(""),
330
            Err(Error(ErrorKind::SimpleLength { len: 0 }))
331
        );
332
333
        assert_eq!(
334
            Uuid::parse_str("!"),
335
            Err(Error(ErrorKind::Char {
336
                character: '!',
337
                index: 1,
338
            }))
339
        );
340
341
        assert_eq!(
342
            Uuid::parse_str("F9168C5E-CEB2-4faa-B6BF-329BF39FA1E45"),
343
            Err(Error(ErrorKind::GroupLength {
344
                group: 4,
345
                len: 13,
346
                index: 25,
347
            }))
348
        );
349
350
        assert_eq!(
351
            Uuid::parse_str("F9168C5E-CEB2-4faa-BBF-329BF39FA1E4"),
352
            Err(Error(ErrorKind::GroupLength {
353
                group: 3,
354
                len: 3,
355
                index: 20,
356
            }))
357
        );
358
359
        assert_eq!(
360
            Uuid::parse_str("F9168C5E-CEB2-4faa-BGBF-329BF39FA1E4"),
361
            Err(Error(ErrorKind::Char {
362
                character: 'G',
363
                index: 21,
364
            }))
365
        );
366
367
        assert_eq!(
368
            Uuid::parse_str("F9168C5E-CEB2F4faaFB6BFF329BF39FA1E4"),
369
            Err(Error(ErrorKind::GroupCount { count: 2 }))
370
        );
371
372
        assert_eq!(
373
            Uuid::parse_str("F9168C5E-CEB2-4faaFB6BFF329BF39FA1E4"),
374
            Err(Error(ErrorKind::GroupCount { count: 3 }))
375
        );
376
377
        assert_eq!(
378
            Uuid::parse_str("F9168C5E-CEB2-4faa-B6BFF329BF39FA1E4"),
379
            Err(Error(ErrorKind::GroupCount { count: 4 }))
380
        );
381
382
        assert_eq!(
383
            Uuid::parse_str("F9168C5E-CEB2-4faa"),
384
            Err(Error(ErrorKind::GroupCount { count: 3 }))
385
        );
386
387
        assert_eq!(
388
            Uuid::parse_str("F9168C5E-CEB2-4faaXB6BFF329BF39FA1E4"),
389
            Err(Error(ErrorKind::Char {
390
                character: 'X',
391
                index: 19,
392
            }))
393
        );
394
395
        assert_eq!(
396
            Uuid::parse_str("{F9168C5E-CEB2-4faa9B6BFF329BF39FA1E41"),
397
            Err(Error(ErrorKind::Char {
398
                character: '{',
399
                index: 1,
400
            }))
401
        );
402
403
        assert_eq!(
404
            Uuid::parse_str("{F9168C5E-CEB2-4faa9B6BFF329BF39FA1E41}"),
405
            Err(Error(ErrorKind::GroupCount { count: 3 }))
406
        );
407
408
        assert_eq!(
409
            Uuid::parse_str("F9168C5E-CEB-24fa-eB6BFF32-BF39FA1E4"),
410
            Err(Error(ErrorKind::GroupLength {
411
                group: 1,
412
                len: 3,
413
                index: 10,
414
            }))
415
        );
416
417
        // // (group, found, expecting)
418
        // //
419
        assert_eq!(
420
            Uuid::parse_str("01020304-1112-2122-3132-41424344"),
421
            Err(Error(ErrorKind::GroupLength {
422
                group: 4,
423
                len: 8,
424
                index: 25,
425
            }))
426
        );
427
428
        assert_eq!(
429
            Uuid::parse_str("67e5504410b1426f9247bb680e5fe0c"),
430
            Err(Error(ErrorKind::SimpleLength { len: 31 }))
431
        );
432
433
        assert_eq!(
434
            Uuid::parse_str("67e5504410b1426f9247bb680e5fe0c88"),
435
            Err(Error(ErrorKind::SimpleLength { len: 33 }))
436
        );
437
438
        assert_eq!(
439
            Uuid::parse_str("67e5504410b1426f9247bb680e5fe0cg8"),
440
            Err(Error(ErrorKind::Char {
441
                character: 'g',
442
                index: 32,
443
            }))
444
        );
445
446
        assert_eq!(
447
            Uuid::parse_str("67e5504410b1426%9247bb680e5fe0c8"),
448
            Err(Error(ErrorKind::Char {
449
                character: '%',
450
                index: 16,
451
            }))
452
        );
453
454
        assert_eq!(
455
            Uuid::parse_str("231231212212423424324323477343246663"),
456
            Err(Error(ErrorKind::SimpleLength { len: 36 }))
457
        );
458
459
        assert_eq!(
460
            Uuid::parse_str("{00000000000000000000000000000000}"),
461
            Err(Error(ErrorKind::GroupCount { count: 1 }))
462
        );
463
464
        assert_eq!(
465
            Uuid::parse_str("67e5504410b1426f9247bb680e5fe0c"),
466
            Err(Error(ErrorKind::SimpleLength { len: 31 }))
467
        );
468
469
        assert_eq!(
470
            Uuid::parse_str("67e550X410b1426f9247bb680e5fe0cd"),
471
            Err(Error(ErrorKind::Char {
472
                character: 'X',
473
                index: 7,
474
            }))
475
        );
476
477
        assert_eq!(
478
            Uuid::parse_str("67e550-4105b1426f9247bb680e5fe0c"),
479
            Err(Error(ErrorKind::GroupCount { count: 2 }))
480
        );
481
482
        assert_eq!(
483
            Uuid::parse_str("F9168C5E-CEB2-4faa-B6BF1-02BF39FA1E4"),
484
            Err(Error(ErrorKind::GroupLength {
485
                group: 3,
486
                len: 5,
487
                index: 20,
488
            }))
489
        );
490
491
        assert_eq!(
492
            Uuid::parse_str("\u{bcf3c}"),
493
            Err(Error(ErrorKind::Char {
494
                character: '\u{bcf3c}',
495
                index: 1
496
            }))
497
        );
498
    }
499
500
    #[test]
501
    fn test_roundtrip_default() {
502
        let uuid_orig = new();
503
        let orig_str = uuid_orig.to_string();
504
        let uuid_out = Uuid::parse_str(&orig_str).unwrap();
505
        assert_eq!(uuid_orig, uuid_out);
506
    }
507
508
    #[test]
509
    fn test_roundtrip_hyphenated() {
510
        let uuid_orig = new();
511
        let orig_str = uuid_orig.hyphenated().to_string();
512
        let uuid_out = Uuid::parse_str(&orig_str).unwrap();
513
        assert_eq!(uuid_orig, uuid_out);
514
    }
515
516
    #[test]
517
    fn test_roundtrip_simple() {
518
        let uuid_orig = new();
519
        let orig_str = uuid_orig.simple().to_string();
520
        let uuid_out = Uuid::parse_str(&orig_str).unwrap();
521
        assert_eq!(uuid_orig, uuid_out);
522
    }
523
524
    #[test]
525
    fn test_roundtrip_urn() {
526
        let uuid_orig = new();
527
        let orig_str = uuid_orig.urn().to_string();
528
        let uuid_out = Uuid::parse_str(&orig_str).unwrap();
529
        assert_eq!(uuid_orig, uuid_out);
530
    }
531
532
    #[test]
533
    fn test_roundtrip_braced() {
534
        let uuid_orig = new();
535
        let orig_str = uuid_orig.braced().to_string();
536
        let uuid_out = Uuid::parse_str(&orig_str).unwrap();
537
        assert_eq!(uuid_orig, uuid_out);
538
    }
539
540
    #[test]
541
    fn test_roundtrip_parse_urn() {
542
        let uuid_orig = new();
543
        let orig_str = uuid_orig.urn().to_string();
544
        let uuid_out = Uuid::from_bytes(parse_urn(orig_str.as_bytes()).unwrap());
545
        assert_eq!(uuid_orig, uuid_out);
546
    }
547
548
    #[test]
549
    fn test_roundtrip_parse_braced() {
550
        let uuid_orig = new();
551
        let orig_str = uuid_orig.braced().to_string();
552
        let uuid_out = Uuid::from_bytes(parse_braced(orig_str.as_bytes()).unwrap());
553
        assert_eq!(uuid_orig, uuid_out);
554
    }
555
556
    #[test]
557
    fn test_try_parse_ascii_non_utf8() {
558
        assert!(Uuid::try_parse_ascii(b"67e55044-10b1-426f-9247-bb680e5\0e0c8").is_err());
559
    }
560
}