1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
/*!
The globset crate provides cross platform single glob and glob set matching.

Glob set matching is the process of matching one or more glob patterns against
a single candidate path simultaneously, and returning all of the globs that
matched. For example, given this set of globs:

* `*.rs`
* `src/lib.rs`
* `src/**/foo.rs`

and a path `src/bar/baz/foo.rs`, then the set would report the first and third
globs as matching.

# Example: one glob

This example shows how to match a single glob against a single file path.

```
use globset::Glob;

let glob = Glob::new("*.rs")?.compile_matcher();

assert!(glob.is_match("foo.rs"));
assert!(glob.is_match("foo/bar.rs"));
assert!(!glob.is_match("Cargo.toml"));
# Ok::<(), Box<dyn std::error::Error>>(())
```

# Example: configuring a glob matcher

This example shows how to use a `GlobBuilder` to configure aspects of match
semantics. In this example, we prevent wildcards from matching path separators.

```
use globset::GlobBuilder;

let glob = GlobBuilder::new("*.rs")
    .literal_separator(true).build()?.compile_matcher();

assert!(glob.is_match("foo.rs"));
assert!(!glob.is_match("foo/bar.rs")); // no longer matches
assert!(!glob.is_match("Cargo.toml"));
# Ok::<(), Box<dyn std::error::Error>>(())
```

# Example: match multiple globs at once

This example shows how to match multiple glob patterns at once.

```
use globset::{Glob, GlobSetBuilder};

let mut builder = GlobSetBuilder::new();
// A GlobBuilder can be used to configure each glob's match semantics
// independently.
builder.add(Glob::new("*.rs")?);
builder.add(Glob::new("src/lib.rs")?);
builder.add(Glob::new("src/**/foo.rs")?);
let set = builder.build()?;

assert_eq!(set.matches("src/bar/baz/foo.rs"), vec![0, 2]);
# Ok::<(), Box<dyn std::error::Error>>(())
```

# Syntax

Standard Unix-style glob syntax is supported:

* `?` matches any single character. (If the `literal_separator` option is
  enabled, then `?` can never match a path separator.)
* `*` matches zero or more characters. (If the `literal_separator` option is
  enabled, then `*` can never match a path separator.)
* `**` recursively matches directories but are only legal in three situations.
  First, if the glob starts with <code>\*\*&#x2F;</code>, then it matches
  all directories. For example, <code>\*\*&#x2F;foo</code> matches `foo`
  and `bar/foo` but not `foo/bar`. Secondly, if the glob ends with
  <code>&#x2F;\*\*</code>, then it matches all sub-entries. For example,
  <code>foo&#x2F;\*\*</code> matches `foo/a` and `foo/a/b`, but not `foo`.
  Thirdly, if the glob contains <code>&#x2F;\*\*&#x2F;</code> anywhere within
  the pattern, then it matches zero or more directories. Using `**` anywhere
  else is illegal (N.B. the glob `**` is allowed and means "match everything").
* `{a,b}` matches `a` or `b` where `a` and `b` are arbitrary glob patterns.
  (N.B. Nesting `{...}` is not currently allowed.)
* `[ab]` matches `a` or `b` where `a` and `b` are characters. Use
  `[!ab]` to match any character except for `a` and `b`.
* Metacharacters such as `*` and `?` can be escaped with character class
  notation. e.g., `[*]` matches `*`.
* When backslash escapes are enabled, a backslash (`\`) will escape all meta
  characters in a glob. If it precedes a non-meta character, then the slash is
  ignored. A `\\` will match a literal `\\`. Note that this mode is only
  enabled on Unix platforms by default, but can be enabled on any platform
  via the `backslash_escape` setting on `Glob`.

A `GlobBuilder` can be used to prevent wildcards from matching path separators,
or to enable case insensitive matching.
*/

#![deny(missing_docs)]

use std::{
    borrow::Cow,
    panic::{RefUnwindSafe, UnwindSafe},
    path::Path,
    sync::Arc,
};

use {
    aho_corasick::AhoCorasick,
    bstr::{ByteSlice, ByteVec, B},
    regex_automata::{
        meta::Regex,
        util::pool::{Pool, PoolGuard},
        PatternSet,
    },
};

use crate::{
    glob::MatchStrategy,
    pathutil::{file_name, file_name_ext, normalize_path},
};

pub use crate::glob::{Glob, GlobBuilder, GlobMatcher};

mod fnv;
mod glob;
mod pathutil;

#[cfg(feature = "serde1")]
mod serde_impl;

#[cfg(feature = "log")]
macro_rules! debug {
    ($($token:tt)*) => (::log::debug!($($token)*);)
}

#[cfg(not(feature = "log"))]
macro_rules! debug {
    ($($token:tt)*) => {};
}

/// Represents an error that can occur when parsing a glob pattern.
#[derive(Clone, Debug, Eq, PartialEq)]
pub struct Error {
    /// The original glob provided by the caller.
    glob: Option<String>,
    /// The kind of error.
    kind: ErrorKind,
}

/// The kind of error that can occur when parsing a glob pattern.
#[derive(Clone, Debug, Eq, PartialEq)]
pub enum ErrorKind {
    /// **DEPRECATED**.
    ///
    /// This error used to occur for consistency with git's glob specification,
    /// but the specification now accepts all uses of `**`. When `**` does not
    /// appear adjacent to a path separator or at the beginning/end of a glob,
    /// it is now treated as two consecutive `*` patterns. As such, this error
    /// is no longer used.
    InvalidRecursive,
    /// Occurs when a character class (e.g., `[abc]`) is not closed.
    UnclosedClass,
    /// Occurs when a range in a character (e.g., `[a-z]`) is invalid. For
    /// example, if the range starts with a lexicographically larger character
    /// than it ends with.
    InvalidRange(char, char),
    /// Occurs when a `}` is found without a matching `{`.
    UnopenedAlternates,
    /// Occurs when a `{` is found without a matching `}`.
    UnclosedAlternates,
    /// Occurs when an alternating group is nested inside another alternating
    /// group, e.g., `{{a,b},{c,d}}`.
    NestedAlternates,
    /// Occurs when an unescaped '\' is found at the end of a glob.
    DanglingEscape,
    /// An error associated with parsing or compiling a regex.
    Regex(String),
    /// Hints that destructuring should not be exhaustive.
    ///
    /// This enum may grow additional variants, so this makes sure clients
    /// don't count on exhaustive matching. (Otherwise, adding a new variant
    /// could break existing code.)
    #[doc(hidden)]
    __Nonexhaustive,
}

impl std::error::Error for Error {
    fn description(&self) -> &str {
        self.kind.description()
    }
}

impl Error {
    /// Return the glob that caused this error, if one exists.
    pub fn glob(&self) -> Option<&str> {
        self.glob.as_ref().map(|s| &**s)
    }

    /// Return the kind of this error.
    pub fn kind(&self) -> &ErrorKind {
        &self.kind
    }
}

impl ErrorKind {
    fn description(&self) -> &str {
        match *self {
            ErrorKind::InvalidRecursive => {
                "invalid use of **; must be one path component"
            }
            ErrorKind::UnclosedClass => {
                "unclosed character class; missing ']'"
            }
            ErrorKind::InvalidRange(_, _) => "invalid character range",
            ErrorKind::UnopenedAlternates => {
                "unopened alternate group; missing '{' \
                (maybe escape '}' with '[}]'?)"
            }
            ErrorKind::UnclosedAlternates => {
                "unclosed alternate group; missing '}' \
                (maybe escape '{' with '[{]'?)"
            }
            ErrorKind::NestedAlternates => {
                "nested alternate groups are not allowed"
            }
            ErrorKind::DanglingEscape => "dangling '\\'",
            ErrorKind::Regex(ref err) => err,
            ErrorKind::__Nonexhaustive => unreachable!(),
        }
    }
}

impl std::fmt::Display for Error {
    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
        match self.glob {
            None => self.kind.fmt(f),
            Some(ref glob) => {
                write!(f, "error parsing glob '{}': {}", glob, self.kind)
            }
        }
    }
}

impl std::fmt::Display for ErrorKind {
    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
        match *self {
            ErrorKind::InvalidRecursive
            | ErrorKind::UnclosedClass
            | ErrorKind::UnopenedAlternates
            | ErrorKind::UnclosedAlternates
            | ErrorKind::NestedAlternates
            | ErrorKind::DanglingEscape
            | ErrorKind::Regex(_) => write!(f, "{}", self.description()),
            ErrorKind::InvalidRange(s, e) => {
                write!(f, "invalid range; '{}' > '{}'", s, e)
            }
            ErrorKind::__Nonexhaustive => unreachable!(),
        }
    }
}

fn new_regex(pat: &str) -> Result<Regex, Error> {
    let syntax = regex_automata::util::syntax::Config::new()
        .utf8(false)
        .dot_matches_new_line(true);
    let config = Regex::config()
        .utf8_empty(false)
        .nfa_size_limit(Some(10 * (1 << 20)))
        .hybrid_cache_capacity(10 * (1 << 20));
    Regex::builder().syntax(syntax).configure(config).build(pat).map_err(
        |err| Error {
            glob: Some(pat.to_string()),
            kind: ErrorKind::Regex(err.to_string()),
        },
    )
}

fn new_regex_set(pats: Vec<String>) -> Result<Regex, Error> {
    let syntax = regex_automata::util::syntax::Config::new()
        .utf8(false)
        .dot_matches_new_line(true);
    let config = Regex::config()
        .match_kind(regex_automata::MatchKind::All)
        .utf8_empty(false)
        .nfa_size_limit(Some(10 * (1 << 20)))
        .hybrid_cache_capacity(10 * (1 << 20));
    Regex::builder()
        .syntax(syntax)
        .configure(config)
        .build_many(&pats)
        .map_err(|err| Error {
            glob: None,
            kind: ErrorKind::Regex(err.to_string()),
        })
}

/// GlobSet represents a group of globs that can be matched together in a
/// single pass.
#[derive(Clone, Debug)]
pub struct GlobSet {
    len: usize,
    strats: Vec<GlobSetMatchStrategy>,
}

impl GlobSet {
    /// Create a new [`GlobSetBuilder`]. A `GlobSetBuilder` can be used to add
    /// new patterns. Once all patterns have been added, `build` should be
    /// called to produce a `GlobSet`, which can then be used for matching.
    #[inline]
    pub fn builder() -> GlobSetBuilder {
        GlobSetBuilder::new()
    }

    /// Create an empty `GlobSet`. An empty set matches nothing.
    #[inline]
    pub fn empty() -> GlobSet {
        GlobSet { len: 0, strats: vec![] }
    }

    /// Returns true if this set is empty, and therefore matches nothing.
    #[inline]
    pub fn is_empty(&self) -> bool {
        self.len == 0
    }

    /// Returns the number of globs in this set.
    #[inline]
    pub fn len(&self) -> usize {
        self.len
    }

    /// Returns true if any glob in this set matches the path given.
    pub fn is_match<P: AsRef<Path>>(&self, path: P) -> bool {
        self.is_match_candidate(&Candidate::new(path.as_ref()))
    }

    /// Returns true if any glob in this set matches the path given.
    ///
    /// This takes a Candidate as input, which can be used to amortize the
    /// cost of preparing a path for matching.
    pub fn is_match_candidate(&self, path: &Candidate<'_>) -> bool {
        if self.is_empty() {
            return false;
        }
        for strat in &self.strats {
            if strat.is_match(path) {
                return true;
            }
        }
        false
    }

    /// Returns the sequence number of every glob pattern that matches the
    /// given path.
    pub fn matches<P: AsRef<Path>>(&self, path: P) -> Vec<usize> {
        self.matches_candidate(&Candidate::new(path.as_ref()))
    }

    /// Returns the sequence number of every glob pattern that matches the
    /// given path.
    ///
    /// This takes a Candidate as input, which can be used to amortize the
    /// cost of preparing a path for matching.
    pub fn matches_candidate(&self, path: &Candidate<'_>) -> Vec<usize> {
        let mut into = vec![];
        if self.is_empty() {
            return into;
        }
        self.matches_candidate_into(path, &mut into);
        into
    }

    /// Adds the sequence number of every glob pattern that matches the given
    /// path to the vec given.
    ///
    /// `into` is cleared before matching begins, and contains the set of
    /// sequence numbers (in ascending order) after matching ends. If no globs
    /// were matched, then `into` will be empty.
    pub fn matches_into<P: AsRef<Path>>(
        &self,
        path: P,
        into: &mut Vec<usize>,
    ) {
        self.matches_candidate_into(&Candidate::new(path.as_ref()), into);
    }

    /// Adds the sequence number of every glob pattern that matches the given
    /// path to the vec given.
    ///
    /// `into` is cleared before matching begins, and contains the set of
    /// sequence numbers (in ascending order) after matching ends. If no globs
    /// were matched, then `into` will be empty.
    ///
    /// This takes a Candidate as input, which can be used to amortize the
    /// cost of preparing a path for matching.
    pub fn matches_candidate_into(
        &self,
        path: &Candidate<'_>,
        into: &mut Vec<usize>,
    ) {
        into.clear();
        if self.is_empty() {
            return;
        }
        for strat in &self.strats {
            strat.matches_into(path, into);
        }
        into.sort();
        into.dedup();
    }

    fn new(pats: &[Glob]) -> Result<GlobSet, Error> {
        if pats.is_empty() {
            return Ok(GlobSet { len: 0, strats: vec![] });
        }
        let mut lits = LiteralStrategy::new();
        let mut base_lits = BasenameLiteralStrategy::new();
        let mut exts = ExtensionStrategy::new();
        let mut prefixes = MultiStrategyBuilder::new();
        let mut suffixes = MultiStrategyBuilder::new();
        let mut required_exts = RequiredExtensionStrategyBuilder::new();
        let mut regexes = MultiStrategyBuilder::new();
        for (i, p) in pats.iter().enumerate() {
            match MatchStrategy::new(p) {
                MatchStrategy::Literal(lit) => {
                    lits.add(i, lit);
                }
                MatchStrategy::BasenameLiteral(lit) => {
                    base_lits.add(i, lit);
                }
                MatchStrategy::Extension(ext) => {
                    exts.add(i, ext);
                }
                MatchStrategy::Prefix(prefix) => {
                    prefixes.add(i, prefix);
                }
                MatchStrategy::Suffix { suffix, component } => {
                    if component {
                        lits.add(i, suffix[1..].to_string());
                    }
                    suffixes.add(i, suffix);
                }
                MatchStrategy::RequiredExtension(ext) => {
                    required_exts.add(i, ext, p.regex().to_owned());
                }
                MatchStrategy::Regex => {
                    debug!("glob converted to regex: {:?}", p);
                    regexes.add(i, p.regex().to_owned());
                }
            }
        }
        debug!(
            "built glob set; {} literals, {} basenames, {} extensions, \
                {} prefixes, {} suffixes, {} required extensions, {} regexes",
            lits.0.len(),
            base_lits.0.len(),
            exts.0.len(),
            prefixes.literals.len(),
            suffixes.literals.len(),
            required_exts.0.len(),
            regexes.literals.len()
        );
        Ok(GlobSet {
            len: pats.len(),
            strats: vec![
                GlobSetMatchStrategy::Extension(exts),
                GlobSetMatchStrategy::BasenameLiteral(base_lits),
                GlobSetMatchStrategy::Literal(lits),
                GlobSetMatchStrategy::Suffix(suffixes.suffix()),
                GlobSetMatchStrategy::Prefix(prefixes.prefix()),
                GlobSetMatchStrategy::RequiredExtension(
                    required_exts.build()?,
                ),
                GlobSetMatchStrategy::Regex(regexes.regex_set()?),
            ],
        })
    }
}

impl Default for GlobSet {
    /// Create a default empty GlobSet.
    fn default() -> Self {
        GlobSet::empty()
    }
}

/// GlobSetBuilder builds a group of patterns that can be used to
/// simultaneously match a file path.
#[derive(Clone, Debug)]
pub struct GlobSetBuilder {
    pats: Vec<Glob>,
}

impl GlobSetBuilder {
    /// Create a new `GlobSetBuilder`. A `GlobSetBuilder` can be used to add new
    /// patterns. Once all patterns have been added, `build` should be called
    /// to produce a [`GlobSet`], which can then be used for matching.
    pub fn new() -> GlobSetBuilder {
        GlobSetBuilder { pats: vec![] }
    }

    /// Builds a new matcher from all of the glob patterns added so far.
    ///
    /// Once a matcher is built, no new patterns can be added to it.
    pub fn build(&self) -> Result<GlobSet, Error> {
        GlobSet::new(&self.pats)
    }

    /// Add a new pattern to this set.
    pub fn add(&mut self, pat: Glob) -> &mut GlobSetBuilder {
        self.pats.push(pat);
        self
    }
}

/// A candidate path for matching.
///
/// All glob matching in this crate operates on `Candidate` values.
/// Constructing candidates has a very small cost associated with it, so
/// callers may find it beneficial to amortize that cost when matching a single
/// path against multiple globs or sets of globs.
#[derive(Clone)]
pub struct Candidate<'a> {
    path: Cow<'a, [u8]>,
    basename: Cow<'a, [u8]>,
    ext: Cow<'a, [u8]>,
}

impl<'a> std::fmt::Debug for Candidate<'a> {
    fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
        f.debug_struct("Candidate")
            .field("path", &self.path.as_bstr())
            .field("basename", &self.basename.as_bstr())
            .field("ext", &self.ext.as_bstr())
            .finish()
    }
}

impl<'a> Candidate<'a> {
    /// Create a new candidate for matching from the given path.
    pub fn new<P: AsRef<Path> + ?Sized>(path: &'a P) -> Candidate<'a> {
        let path = normalize_path(Vec::from_path_lossy(path.as_ref()));
        let basename = file_name(&path).unwrap_or(Cow::Borrowed(B("")));
        let ext = file_name_ext(&basename).unwrap_or(Cow::Borrowed(B("")));
        Candidate { path, basename, ext }
    }

    fn path_prefix(&self, max: usize) -> &[u8] {
        if self.path.len() <= max {
            &*self.path
        } else {
            &self.path[..max]
        }
    }

    fn path_suffix(&self, max: usize) -> &[u8] {
        if self.path.len() <= max {
            &*self.path
        } else {
            &self.path[self.path.len() - max..]
        }
    }
}

#[derive(Clone, Debug)]
enum GlobSetMatchStrategy {
    Literal(LiteralStrategy),
    BasenameLiteral(BasenameLiteralStrategy),
    Extension(ExtensionStrategy),
    Prefix(PrefixStrategy),
    Suffix(SuffixStrategy),
    RequiredExtension(RequiredExtensionStrategy),
    Regex(RegexSetStrategy),
}

impl GlobSetMatchStrategy {
    fn is_match(&self, candidate: &Candidate<'_>) -> bool {
        use self::GlobSetMatchStrategy::*;
        match *self {
            Literal(ref s) => s.is_match(candidate),
            BasenameLiteral(ref s) => s.is_match(candidate),
            Extension(ref s) => s.is_match(candidate),
            Prefix(ref s) => s.is_match(candidate),
            Suffix(ref s) => s.is_match(candidate),
            RequiredExtension(ref s) => s.is_match(candidate),
            Regex(ref s) => s.is_match(candidate),
        }
    }

    fn matches_into(
        &self,
        candidate: &Candidate<'_>,
        matches: &mut Vec<usize>,
    ) {
        use self::GlobSetMatchStrategy::*;
        match *self {
            Literal(ref s) => s.matches_into(candidate, matches),
            BasenameLiteral(ref s) => s.matches_into(candidate, matches),
            Extension(ref s) => s.matches_into(candidate, matches),
            Prefix(ref s) => s.matches_into(candidate, matches),
            Suffix(ref s) => s.matches_into(candidate, matches),
            RequiredExtension(ref s) => s.matches_into(candidate, matches),
            Regex(ref s) => s.matches_into(candidate, matches),
        }
    }
}

#[derive(Clone, Debug)]
struct LiteralStrategy(fnv::HashMap<Vec<u8>, Vec<usize>>);

impl LiteralStrategy {
    fn new() -> LiteralStrategy {
        LiteralStrategy(fnv::HashMap::default())
    }

    fn add(&mut self, global_index: usize, lit: String) {
        self.0.entry(lit.into_bytes()).or_insert(vec![]).push(global_index);
    }

    fn is_match(&self, candidate: &Candidate<'_>) -> bool {
        self.0.contains_key(candidate.path.as_bytes())
    }

    #[inline(never)]
    fn matches_into(
        &self,
        candidate: &Candidate<'_>,
        matches: &mut Vec<usize>,
    ) {
        if let Some(hits) = self.0.get(candidate.path.as_bytes()) {
            matches.extend(hits);
        }
    }
}

#[derive(Clone, Debug)]
struct BasenameLiteralStrategy(fnv::HashMap<Vec<u8>, Vec<usize>>);

impl BasenameLiteralStrategy {
    fn new() -> BasenameLiteralStrategy {
        BasenameLiteralStrategy(fnv::HashMap::default())
    }

    fn add(&mut self, global_index: usize, lit: String) {
        self.0.entry(lit.into_bytes()).or_insert(vec![]).push(global_index);
    }

    fn is_match(&self, candidate: &Candidate<'_>) -> bool {
        if candidate.basename.is_empty() {
            return false;
        }
        self.0.contains_key(candidate.basename.as_bytes())
    }

    #[inline(never)]
    fn matches_into(
        &self,
        candidate: &Candidate<'_>,
        matches: &mut Vec<usize>,
    ) {
        if candidate.basename.is_empty() {
            return;
        }
        if let Some(hits) = self.0.get(candidate.basename.as_bytes()) {
            matches.extend(hits);
        }
    }
}

#[derive(Clone, Debug)]
struct ExtensionStrategy(fnv::HashMap<Vec<u8>, Vec<usize>>);

impl ExtensionStrategy {
    fn new() -> ExtensionStrategy {
        ExtensionStrategy(fnv::HashMap::default())
    }

    fn add(&mut self, global_index: usize, ext: String) {
        self.0.entry(ext.into_bytes()).or_insert(vec![]).push(global_index);
    }

    fn is_match(&self, candidate: &Candidate<'_>) -> bool {
        if candidate.ext.is_empty() {
            return false;
        }
        self.0.contains_key(candidate.ext.as_bytes())
    }

    #[inline(never)]
    fn matches_into(
        &self,
        candidate: &Candidate<'_>,
        matches: &mut Vec<usize>,
    ) {
        if candidate.ext.is_empty() {
            return;
        }
        if let Some(hits) = self.0.get(candidate.ext.as_bytes()) {
            matches.extend(hits);
        }
    }
}

#[derive(Clone, Debug)]
struct PrefixStrategy {
    matcher: AhoCorasick,
    map: Vec<usize>,
    longest: usize,
}

impl PrefixStrategy {
    fn is_match(&self, candidate: &Candidate<'_>) -> bool {
        let path = candidate.path_prefix(self.longest);
        for m in self.matcher.find_overlapping_iter(path) {
            if m.start() == 0 {
                return true;
            }
        }
        false
    }

    fn matches_into(
        &self,
        candidate: &Candidate<'_>,
        matches: &mut Vec<usize>,
    ) {
        let path = candidate.path_prefix(self.longest);
        for m in self.matcher.find_overlapping_iter(path) {
            if m.start() == 0 {
                matches.push(self.map[m.pattern()]);
            }
        }
    }
}

#[derive(Clone, Debug)]
struct SuffixStrategy {
    matcher: AhoCorasick,
    map: Vec<usize>,
    longest: usize,
}

impl SuffixStrategy {
    fn is_match(&self, candidate: &Candidate<'_>) -> bool {
        let path = candidate.path_suffix(self.longest);
        for m in self.matcher.find_overlapping_iter(path) {
            if m.end() == path.len() {
                return true;
            }
        }
        false
    }

    fn matches_into(
        &self,
        candidate: &Candidate<'_>,
        matches: &mut Vec<usize>,
    ) {
        let path = candidate.path_suffix(self.longest);
        for m in self.matcher.find_overlapping_iter(path) {
            if m.end() == path.len() {
                matches.push(self.map[m.pattern()]);
            }
        }
    }
}

#[derive(Clone, Debug)]
struct RequiredExtensionStrategy(fnv::HashMap<Vec<u8>, Vec<(usize, Regex)>>);

impl RequiredExtensionStrategy {
    fn is_match(&self, candidate: &Candidate<'_>) -> bool {
        if candidate.ext.is_empty() {
            return false;
        }
        match self.0.get(candidate.ext.as_bytes()) {
            None => false,
            Some(regexes) => {
                for &(_, ref re) in regexes {
                    if re.is_match(candidate.path.as_bytes()) {
                        return true;
                    }
                }
                false
            }
        }
    }

    #[inline(never)]
    fn matches_into(
        &self,
        candidate: &Candidate<'_>,
        matches: &mut Vec<usize>,
    ) {
        if candidate.ext.is_empty() {
            return;
        }
        if let Some(regexes) = self.0.get(candidate.ext.as_bytes()) {
            for &(global_index, ref re) in regexes {
                if re.is_match(candidate.path.as_bytes()) {
                    matches.push(global_index);
                }
            }
        }
    }
}

#[derive(Clone, Debug)]
struct RegexSetStrategy {
    matcher: Regex,
    map: Vec<usize>,
    // We use a pool of PatternSets to hopefully allocating a fresh one on each
    // call.
    //
    // TODO: In the next semver breaking release, we should drop this pool and
    // expose an opaque type that wraps PatternSet. Then callers can provide
    // it to `matches_into` directly. Callers might still want to use a pool
    // or similar to amortize allocation, but that matches the status quo and
    // absolves us of needing to do it here.
    patset: Arc<Pool<PatternSet, PatternSetPoolFn>>,
}

type PatternSetPoolFn =
    Box<dyn Fn() -> PatternSet + Send + Sync + UnwindSafe + RefUnwindSafe>;

impl RegexSetStrategy {
    fn is_match(&self, candidate: &Candidate<'_>) -> bool {
        self.matcher.is_match(candidate.path.as_bytes())
    }

    fn matches_into(
        &self,
        candidate: &Candidate<'_>,
        matches: &mut Vec<usize>,
    ) {
        let input = regex_automata::Input::new(candidate.path.as_bytes());
        let mut patset = self.patset.get();
        patset.clear();
        self.matcher.which_overlapping_matches(&input, &mut patset);
        for i in patset.iter() {
            matches.push(self.map[i]);
        }
        PoolGuard::put(patset);
    }
}

#[derive(Clone, Debug)]
struct MultiStrategyBuilder {
    literals: Vec<String>,
    map: Vec<usize>,
    longest: usize,
}

impl MultiStrategyBuilder {
    fn new() -> MultiStrategyBuilder {
        MultiStrategyBuilder { literals: vec![], map: vec![], longest: 0 }
    }

    fn add(&mut self, global_index: usize, literal: String) {
        if literal.len() > self.longest {
            self.longest = literal.len();
        }
        self.map.push(global_index);
        self.literals.push(literal);
    }

    fn prefix(self) -> PrefixStrategy {
        PrefixStrategy {
            matcher: AhoCorasick::new(&self.literals).unwrap(),
            map: self.map,
            longest: self.longest,
        }
    }

    fn suffix(self) -> SuffixStrategy {
        SuffixStrategy {
            matcher: AhoCorasick::new(&self.literals).unwrap(),
            map: self.map,
            longest: self.longest,
        }
    }

    fn regex_set(self) -> Result<RegexSetStrategy, Error> {
        let matcher = new_regex_set(self.literals)?;
        let pattern_len = matcher.pattern_len();
        let create: PatternSetPoolFn =
            Box::new(move || PatternSet::new(pattern_len));
        Ok(RegexSetStrategy {
            matcher,
            map: self.map,
            patset: Arc::new(Pool::new(create)),
        })
    }
}

#[derive(Clone, Debug)]
struct RequiredExtensionStrategyBuilder(
    fnv::HashMap<Vec<u8>, Vec<(usize, String)>>,
);

impl RequiredExtensionStrategyBuilder {
    fn new() -> RequiredExtensionStrategyBuilder {
        RequiredExtensionStrategyBuilder(fnv::HashMap::default())
    }

    fn add(&mut self, global_index: usize, ext: String, regex: String) {
        self.0
            .entry(ext.into_bytes())
            .or_insert(vec![])
            .push((global_index, regex));
    }

    fn build(self) -> Result<RequiredExtensionStrategy, Error> {
        let mut exts = fnv::HashMap::default();
        for (ext, regexes) in self.0.into_iter() {
            exts.insert(ext.clone(), vec![]);
            for (global_index, regex) in regexes {
                let compiled = new_regex(&regex)?;
                exts.get_mut(&ext).unwrap().push((global_index, compiled));
            }
        }
        Ok(RequiredExtensionStrategy(exts))
    }
}

/// Escape meta-characters within the given glob pattern.
///
/// The escaping works by surrounding meta-characters with brackets. For
/// example, `*` becomes `[*]`.
pub fn escape(s: &str) -> String {
    let mut escaped = String::with_capacity(s.len());
    for c in s.chars() {
        match c {
            // note that ! does not need escaping because it is only special
            // inside brackets
            '?' | '*' | '[' | ']' => {
                escaped.push('[');
                escaped.push(c);
                escaped.push(']');
            }
            c => {
                escaped.push(c);
            }
        }
    }
    escaped
}

#[cfg(test)]
mod tests {
    use crate::glob::Glob;

    use super::{GlobSet, GlobSetBuilder};

    #[test]
    fn set_works() {
        let mut builder = GlobSetBuilder::new();
        builder.add(Glob::new("src/**/*.rs").unwrap());
        builder.add(Glob::new("*.c").unwrap());
        builder.add(Glob::new("src/lib.rs").unwrap());
        let set = builder.build().unwrap();

        assert!(set.is_match("foo.c"));
        assert!(set.is_match("src/foo.c"));
        assert!(!set.is_match("foo.rs"));
        assert!(!set.is_match("tests/foo.rs"));
        assert!(set.is_match("src/foo.rs"));
        assert!(set.is_match("src/grep/src/main.rs"));

        let matches = set.matches("src/lib.rs");
        assert_eq!(2, matches.len());
        assert_eq!(0, matches[0]);
        assert_eq!(2, matches[1]);
    }

    #[test]
    fn empty_set_works() {
        let set = GlobSetBuilder::new().build().unwrap();
        assert!(!set.is_match(""));
        assert!(!set.is_match("a"));
    }

    #[test]
    fn default_set_is_empty_works() {
        let set: GlobSet = Default::default();
        assert!(!set.is_match(""));
        assert!(!set.is_match("a"));
    }

    #[test]
    fn escape() {
        use super::escape;
        assert_eq!("foo", escape("foo"));
        assert_eq!("foo[*]", escape("foo*"));
        assert_eq!("[[][]]", escape("[]"));
        assert_eq!("[*][?]", escape("*?"));
        assert_eq!("src/[*][*]/[*].rs", escape("src/**/*.rs"));
        assert_eq!("bar[[]ab[]]baz", escape("bar[ab]baz"));
        assert_eq!("bar[[]!![]]!baz", escape("bar[!!]!baz"));
    }

    // This tests that regex matching doesn't "remember" the results of
    // previous searches. That is, if any memory is reused from a previous
    // search, then it should be cleared first.
    #[test]
    fn set_does_not_remember() {
        let mut builder = GlobSetBuilder::new();
        builder.add(Glob::new("*foo*").unwrap());
        builder.add(Glob::new("*bar*").unwrap());
        builder.add(Glob::new("*quux*").unwrap());
        let set = builder.build().unwrap();

        let matches = set.matches("ZfooZquuxZ");
        assert_eq!(2, matches.len());
        assert_eq!(0, matches[0]);
        assert_eq!(2, matches[1]);

        let matches = set.matches("nada");
        assert_eq!(0, matches.len());
    }
}