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
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
// Copyright 2015 The Rust Project Developers. See the COPYRIGHT
// file at the top-level directory of this distribution and at
// http://rust-lang.org/COPYRIGHT.
//
// Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
// http://www.apache.org/licenses/LICENSE-2.0> or the MIT license
// <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
// option. This file may not be copied, modified, or distributed
// except according to those terms.

//! The string Pattern API.
//!
//! For more details, see the traits `Pattern`, `Searcher`,
//! `ReverseSearcher` and `DoubleEndedSearcher`.

#![unstable(feature = "pattern",
            reason = "API not fully fleshed out and ready to be stabilized",
            issue = "27721")]

use cmp;
use fmt;
use slice::memchr;
use usize;

// Pattern

/// A string pattern.
///
/// A `Pattern<'a>` expresses that the implementing type
/// can be used as a string pattern for searching in a `&'a str`.
///
/// For example, both `'a'` and `"aa"` are patterns that
/// would match at index `1` in the string `"baaaab"`.
///
/// The trait itself acts as a builder for an associated
/// `Searcher` type, which does the actual work of finding
/// occurrences of the pattern in a string.
pub trait Pattern<'a>: Sized {
    /// Associated searcher for this pattern
    type Searcher: Searcher<'a>;

    /// Constructs the associated searcher from
    /// `self` and the `haystack` to search in.
    fn into_searcher(self, haystack: &'a str) -> Self::Searcher;

    /// Checks whether the pattern matches anywhere in the haystack
    #[inline]
    fn is_contained_in(self, haystack: &'a str) -> bool {
        self.into_searcher(haystack).next_match().is_some()
    }

    /// Checks whether the pattern matches at the front of the haystack
    #[inline]
    fn is_prefix_of(self, haystack: &'a str) -> bool {
        match self.into_searcher(haystack).next() {
            SearchStep::Match(0, _) => true,
            _ => false,
        }
    }

    /// Checks whether the pattern matches at the back of the haystack
    #[inline]
    fn is_suffix_of(self, haystack: &'a str) -> bool
        where Self::Searcher: ReverseSearcher<'a>
    {
        match self.into_searcher(haystack).next_back() {
            SearchStep::Match(_, j) if haystack.len() == j => true,
            _ => false,
        }
    }
}

// Searcher

/// Result of calling `Searcher::next()` or `ReverseSearcher::next_back()`.
#[derive(Copy, Clone, Eq, PartialEq, Debug)]
pub enum SearchStep {
    /// Expresses that a match of the pattern has been found at
    /// `haystack[a..b]`.
    Match(usize, usize),
    /// Expresses that `haystack[a..b]` has been rejected as a possible match
    /// of the pattern.
    ///
    /// Note that there might be more than one `Reject` between two `Match`es,
    /// there is no requirement for them to be combined into one.
    Reject(usize, usize),
    /// Expresses that every byte of the haystack has been visited, ending
    /// the iteration.
    Done
}

/// A searcher for a string pattern.
///
/// This trait provides methods for searching for non-overlapping
/// matches of a pattern starting from the front (left) of a string.
///
/// It will be implemented by associated `Searcher`
/// types of the `Pattern` trait.
///
/// The trait is marked unsafe because the indices returned by the
/// `next()` methods are required to lie on valid utf8 boundaries in
/// the haystack. This enables consumers of this trait to
/// slice the haystack without additional runtime checks.
pub unsafe trait Searcher<'a> {
    /// Getter for the underlying string to be searched in
    ///
    /// Will always return the same `&str`
    fn haystack(&self) -> &'a str;

    /// Performs the next search step starting from the front.
    ///
    /// - Returns `Match(a, b)` if `haystack[a..b]` matches the pattern.
    /// - Returns `Reject(a, b)` if `haystack[a..b]` can not match the
    ///   pattern, even partially.
    /// - Returns `Done` if every byte of the haystack has been visited
    ///
    /// The stream of `Match` and `Reject` values up to a `Done`
    /// will contain index ranges that are adjacent, non-overlapping,
    /// covering the whole haystack, and laying on utf8 boundaries.
    ///
    /// A `Match` result needs to contain the whole matched pattern,
    /// however `Reject` results may be split up into arbitrary
    /// many adjacent fragments. Both ranges may have zero length.
    ///
    /// As an example, the pattern `"aaa"` and the haystack `"cbaaaaab"`
    /// might produce the stream
    /// `[Reject(0, 1), Reject(1, 2), Match(2, 5), Reject(5, 8)]`
    fn next(&mut self) -> SearchStep;

    /// Find the next `Match` result. See `next()`
    ///
    /// Unlike next(), there is no guarantee that the returned ranges
    /// of this and next_reject will overlap. This will return (start_match, end_match),
    /// where start_match is the index of where the match begins, and end_match is
    /// the index after the end of the match.
    #[inline]
    fn next_match(&mut self) -> Option<(usize, usize)> {
        loop {
            match self.next() {
                SearchStep::Match(a, b) => return Some((a, b)),
                SearchStep::Done => return None,
                _ => continue,
            }
        }
    }

    /// Find the next `Reject` result. See `next()` and `next_match()`
    ///
    /// Unlike next(), there is no guarantee that the returned ranges
    /// of this and next_match will overlap.
    #[inline]
    fn next_reject(&mut self) -> Option<(usize, usize)> {
        loop {
            match self.next() {
                SearchStep::Reject(a, b) => return Some((a, b)),
                SearchStep::Done => return None,
                _ => continue,
            }
        }
    }
}

/// A reverse searcher for a string pattern.
///
/// This trait provides methods for searching for non-overlapping
/// matches of a pattern starting from the back (right) of a string.
///
/// It will be implemented by associated `Searcher`
/// types of the `Pattern` trait if the pattern supports searching
/// for it from the back.
///
/// The index ranges returned by this trait are not required
/// to exactly match those of the forward search in reverse.
///
/// For the reason why this trait is marked unsafe, see them
/// parent trait `Searcher`.
pub unsafe trait ReverseSearcher<'a>: Searcher<'a> {
    /// Performs the next search step starting from the back.
    ///
    /// - Returns `Match(a, b)` if `haystack[a..b]` matches the pattern.
    /// - Returns `Reject(a, b)` if `haystack[a..b]` can not match the
    ///   pattern, even partially.
    /// - Returns `Done` if every byte of the haystack has been visited
    ///
    /// The stream of `Match` and `Reject` values up to a `Done`
    /// will contain index ranges that are adjacent, non-overlapping,
    /// covering the whole haystack, and laying on utf8 boundaries.
    ///
    /// A `Match` result needs to contain the whole matched pattern,
    /// however `Reject` results may be split up into arbitrary
    /// many adjacent fragments. Both ranges may have zero length.
    ///
    /// As an example, the pattern `"aaa"` and the haystack `"cbaaaaab"`
    /// might produce the stream
    /// `[Reject(7, 8), Match(4, 7), Reject(1, 4), Reject(0, 1)]`
    fn next_back(&mut self) -> SearchStep;

    /// Find the next `Match` result. See `next_back()`
    #[inline]
    fn next_match_back(&mut self) -> Option<(usize, usize)>{
        loop {
            match self.next_back() {
                SearchStep::Match(a, b) => return Some((a, b)),
                SearchStep::Done => return None,
                _ => continue,
            }
        }
    }

    /// Find the next `Reject` result. See `next_back()`
    #[inline]
    fn next_reject_back(&mut self) -> Option<(usize, usize)>{
        loop {
            match self.next_back() {
                SearchStep::Reject(a, b) => return Some((a, b)),
                SearchStep::Done => return None,
                _ => continue,
            }
        }
    }
}

/// A marker trait to express that a `ReverseSearcher`
/// can be used for a `DoubleEndedIterator` implementation.
///
/// For this, the impl of `Searcher` and `ReverseSearcher` need
/// to follow these conditions:
///
/// - All results of `next()` need to be identical
///   to the results of `next_back()` in reverse order.
/// - `next()` and `next_back()` need to behave as
///   the two ends of a range of values, that is they
///   can not "walk past each other".
///
/// # Examples
///
/// `char::Searcher` is a `DoubleEndedSearcher` because searching for a
/// `char` only requires looking at one at a time, which behaves the same
/// from both ends.
///
/// `(&str)::Searcher` is not a `DoubleEndedSearcher` because
/// the pattern `"aa"` in the haystack `"aaa"` matches as either
/// `"[aa]a"` or `"a[aa]"`, depending from which side it is searched.
pub trait DoubleEndedSearcher<'a>: ReverseSearcher<'a> {}


/////////////////////////////////////////////////////////////////////////////
// Impl for char
/////////////////////////////////////////////////////////////////////////////

/// Associated type for `<char as Pattern<'a>>::Searcher`.
#[derive(Clone, Debug)]
pub struct CharSearcher<'a> {
    haystack: &'a str,
    // safety invariant: `finger`/`finger_back` must be a valid utf8 byte index of `haystack`
    // This invariant can be broken *within* next_match and next_match_back, however
    // they must exit with fingers on valid code point boundaries.

    /// `finger` is the current byte index of the forward search.
    /// Imagine that it exists before the byte at its index, i.e.
    /// `haystack[finger]` is the first byte of the slice we must inspect during
    /// forward searching
    finger: usize,
    /// `finger_back` is the current byte index of the reverse search.
    /// Imagine that it exists after the byte at its index, i.e.
    /// haystack[finger_back - 1] is the last byte of the slice we must inspect during
    /// forward searching (and thus the first byte to be inspected when calling next_back())
    finger_back: usize,
    /// The character being searched for
    needle: char,

    // safety invariant: `utf8_size` must be less than 5
    /// The number of bytes `needle` takes up when encoded in utf8
    utf8_size: usize,
    /// A utf8 encoded copy of the `needle`
    utf8_encoded: [u8; 4],
}

unsafe impl<'a> Searcher<'a> for CharSearcher<'a> {
    #[inline]
    fn haystack(&self) -> &'a str {
        self.haystack
    }
    #[inline]
    fn next(&mut self) -> SearchStep {
        let old_finger = self.finger;
        let slice = unsafe { self.haystack.get_unchecked(old_finger..self.finger_back) };
        let mut iter = slice.chars();
        let old_len = iter.iter.len();
        if let Some(ch) = iter.next() {
            // add byte offset of current character
            // without re-encoding as utf-8
            self.finger += old_len - iter.iter.len();
            if ch == self.needle {
                SearchStep::Match(old_finger, self.finger)
            } else {
                SearchStep::Reject(old_finger, self.finger)
            }
        } else {
            SearchStep::Done
        }
    }
    #[inline]
    fn next_match(&mut self) -> Option<(usize, usize)> {
        loop {
            // get the haystack after the last character found
            let bytes = if let Some(slice) = self.haystack.as_bytes()
                                                 .get(self.finger..self.finger_back) {
                slice
            } else {
                return None;
            };
            // the last byte of the utf8 encoded needle
            let last_byte = unsafe { *self.utf8_encoded.get_unchecked(self.utf8_size - 1) };
            if let Some(index) = memchr::memchr(last_byte, bytes) {
                // The new finger is the index of the byte we found,
                // plus one, since we memchr'd for the last byte of the character.
                //
                // Note that this doesn't always give us a finger on a UTF8 boundary.
                // If we *didn't* find our character
                // we may have indexed to the non-last byte of a 3-byte or 4-byte character.
                // We can't just skip to the next valid starting byte because a character like
                // ꁁ (U+A041 YI SYLLABLE PA), utf-8 `EA 81 81` will have us always find
                // the second byte when searching for the third.
                //
                // However, this is totally okay. While we have the invariant that
                // self.finger is on a UTF8 boundary, this invariant is not relied upon
                // within this method (it is relied upon in CharSearcher::next()).
                //
                // We only exit this method when we reach the end of the string, or if we
                // find something. When we find something the `finger` will be set
                // to a UTF8 boundary.
                self.finger += index + 1;
                if self.finger >= self.utf8_size {
                    let found_char = self.finger - self.utf8_size;
                    if let Some(slice) = self.haystack.as_bytes().get(found_char..self.finger) {
                        if slice == &self.utf8_encoded[0..self.utf8_size] {
                            return Some((found_char, self.finger));
                        }
                    }
                }
            } else {
                // found nothing, exit
                self.finger = self.finger_back;
                return None;
            }
        }
    }

    // let next_reject use the default implementation from the Searcher trait
}

unsafe impl<'a> ReverseSearcher<'a> for CharSearcher<'a> {
    #[inline]
    fn next_back(&mut self) -> SearchStep {
        let old_finger = self.finger_back;
        let slice = unsafe { self.haystack.get_unchecked(self.finger..old_finger) };
        let mut iter = slice.chars();
        let old_len = iter.iter.len();
        if let Some(ch) = iter.next_back() {
            // subtract byte offset of current character
            // without re-encoding as utf-8
            self.finger_back -= old_len - iter.iter.len();
            if ch == self.needle {
                SearchStep::Match(self.finger_back, old_finger)
            } else {
                SearchStep::Reject(self.finger_back, old_finger)
            }
        } else {
            SearchStep::Done
        }
    }
    #[inline]
    fn next_match_back(&mut self) -> Option<(usize, usize)> {
        let haystack = self.haystack.as_bytes();
        loop {
            // get the haystack up to but not including the last character searched
            let bytes = if let Some(slice) = haystack.get(self.finger..self.finger_back) {
                slice
            } else {
                return None;
            };
            // the last byte of the utf8 encoded needle
            let last_byte = unsafe { *self.utf8_encoded.get_unchecked(self.utf8_size - 1) };
            if let Some(index) = memchr::memrchr(last_byte, bytes) {
                // we searched a slice that was offset by self.finger,
                // add self.finger to recoup the original index
                let index = self.finger + index;
                // memrchr will return the index of the byte we wish to
                // find. In case of an ASCII character, this is indeed
                // were we wish our new finger to be ("after" the found
                // char in the paradigm of reverse iteration). For
                // multibyte chars we need to skip down by the number of more
                // bytes they have than ASCII
                let shift = self.utf8_size - 1;
                if index >= shift {
                    let found_char = index - shift;
                    if let Some(slice) = haystack.get(found_char..(found_char + self.utf8_size)) {
                        if slice == &self.utf8_encoded[0..self.utf8_size] {
                            // move finger to before the character found (i.e. at its start index)
                            self.finger_back = found_char;
                            return Some((self.finger_back, self.finger_back + self.utf8_size));
                        }
                    }
                }
                // We can't use finger_back = index - size + 1 here. If we found the last char
                // of a different-sized character (or the middle byte of a different character)
                // we need to bump the finger_back down to `index`. This similarly makes
                // `finger_back` have the potential to no longer be on a boundary,
                // but this is OK since we only exit this function on a boundary
                // or when the haystack has been searched completely.
                //
                // Unlike next_match this does not
                // have the problem of repeated bytes in utf-8 because
                // we're searching for the last byte, and we can only have
                // found the last byte when searching in reverse.
                self.finger_back = index;
            } else {
                self.finger_back = self.finger;
                // found nothing, exit
                return None;
            }
        }
    }

    // let next_reject_back use the default implementation from the Searcher trait
}

impl<'a> DoubleEndedSearcher<'a> for CharSearcher<'a> {}

/// Searches for chars that are equal to a given char
impl<'a> Pattern<'a> for char {
    type Searcher = CharSearcher<'a>;

    #[inline]
    fn into_searcher(self, haystack: &'a str) -> Self::Searcher {
        let mut utf8_encoded = [0; 4];
        self.encode_utf8(&mut utf8_encoded);
        let utf8_size = self.len_utf8();
        CharSearcher {
            haystack,
            finger: 0,
            finger_back: haystack.len(),
            needle: self,
            utf8_size,
            utf8_encoded
        }
    }

    #[inline]
    fn is_contained_in(self, haystack: &'a str) -> bool {
        if (self as u32) < 128 {
            haystack.as_bytes().contains(&(self as u8))
        } else {
            let mut buffer = [0u8; 4];
            self.encode_utf8(&mut buffer).is_contained_in(haystack)
        }
    }

    #[inline]
    fn is_prefix_of(self, haystack: &'a str) -> bool {
        if let Some(ch) = haystack.chars().next() {
            self == ch
        } else {
            false
        }
    }

    #[inline]
    fn is_suffix_of(self, haystack: &'a str) -> bool where Self::Searcher: ReverseSearcher<'a>
    {
        if let Some(ch) = haystack.chars().next_back() {
            self == ch
        } else {
            false
        }
    }
}

/////////////////////////////////////////////////////////////////////////////
// Impl for a MultiCharEq wrapper
/////////////////////////////////////////////////////////////////////////////

#[doc(hidden)]
trait MultiCharEq {
    fn matches(&mut self, c: char) -> bool;
}

impl<F> MultiCharEq for F where F: FnMut(char) -> bool {
    #[inline]
    fn matches(&mut self, c: char) -> bool { (*self)(c) }
}

impl MultiCharEq for &[char] {
    #[inline]
    fn matches(&mut self, c: char) -> bool {
        self.iter().any(|&m| { m == c })
    }
}

struct MultiCharEqPattern<C: MultiCharEq>(C);

#[derive(Clone, Debug)]
struct MultiCharEqSearcher<'a, C: MultiCharEq> {
    char_eq: C,
    haystack: &'a str,
    char_indices: super::CharIndices<'a>,
}

impl<'a, C: MultiCharEq> Pattern<'a> for MultiCharEqPattern<C> {
    type Searcher = MultiCharEqSearcher<'a, C>;

    #[inline]
    fn into_searcher(self, haystack: &'a str) -> MultiCharEqSearcher<'a, C> {
        MultiCharEqSearcher {
            haystack,
            char_eq: self.0,
            char_indices: haystack.char_indices(),
        }
    }
}

unsafe impl<'a, C: MultiCharEq> Searcher<'a> for MultiCharEqSearcher<'a, C> {
    #[inline]
    fn haystack(&self) -> &'a str {
        self.haystack
    }

    #[inline]
    fn next(&mut self) -> SearchStep {
        let s = &mut self.char_indices;
        // Compare lengths of the internal byte slice iterator
        // to find length of current char
        let pre_len = s.iter.iter.len();
        if let Some((i, c)) = s.next() {
            let len = s.iter.iter.len();
            let char_len = pre_len - len;
            if self.char_eq.matches(c) {
                return SearchStep::Match(i, i + char_len);
            } else {
                return SearchStep::Reject(i, i + char_len);
            }
        }
        SearchStep::Done
    }
}

unsafe impl<'a, C: MultiCharEq> ReverseSearcher<'a> for MultiCharEqSearcher<'a, C> {
    #[inline]
    fn next_back(&mut self) -> SearchStep {
        let s = &mut self.char_indices;
        // Compare lengths of the internal byte slice iterator
        // to find length of current char
        let pre_len = s.iter.iter.len();
        if let Some((i, c)) = s.next_back() {
            let len = s.iter.iter.len();
            let char_len = pre_len - len;
            if self.char_eq.matches(c) {
                return SearchStep::Match(i, i + char_len);
            } else {
                return SearchStep::Reject(i, i + char_len);
            }
        }
        SearchStep::Done
    }
}

impl<'a, C: MultiCharEq> DoubleEndedSearcher<'a> for MultiCharEqSearcher<'a, C> {}

/////////////////////////////////////////////////////////////////////////////

macro_rules! pattern_methods {
    ($t:ty, $pmap:expr, $smap:expr) => {
        type Searcher = $t;

        #[inline]
        fn into_searcher(self, haystack: &'a str) -> $t {
            ($smap)(($pmap)(self).into_searcher(haystack))
        }

        #[inline]
        fn is_contained_in(self, haystack: &'a str) -> bool {
            ($pmap)(self).is_contained_in(haystack)
        }

        #[inline]
        fn is_prefix_of(self, haystack: &'a str) -> bool {
            ($pmap)(self).is_prefix_of(haystack)
        }

        #[inline]
        fn is_suffix_of(self, haystack: &'a str) -> bool
            where $t: ReverseSearcher<'a>
        {
            ($pmap)(self).is_suffix_of(haystack)
        }
    }
}

macro_rules! searcher_methods {
    (forward) => {
        #[inline]
        fn haystack(&self) -> &'a str {
            self.0.haystack()
        }
        #[inline]
        fn next(&mut self) -> SearchStep {
            self.0.next()
        }
        #[inline]
        fn next_match(&mut self) -> Option<(usize, usize)> {
            self.0.next_match()
        }
        #[inline]
        fn next_reject(&mut self) -> Option<(usize, usize)> {
            self.0.next_reject()
        }
    };
    (reverse) => {
        #[inline]
        fn next_back(&mut self) -> SearchStep {
            self.0.next_back()
        }
        #[inline]
        fn next_match_back(&mut self) -> Option<(usize, usize)> {
            self.0.next_match_back()
        }
        #[inline]
        fn next_reject_back(&mut self) -> Option<(usize, usize)> {
            self.0.next_reject_back()
        }
    }
}

/////////////////////////////////////////////////////////////////////////////
// Impl for &[char]
/////////////////////////////////////////////////////////////////////////////

// Todo: Change / Remove due to ambiguity in meaning.

/// Associated type for `<&[char] as Pattern<'a>>::Searcher`.
#[derive(Clone, Debug)]
pub struct CharSliceSearcher<'a, 'b>(<MultiCharEqPattern<&'b [char]> as Pattern<'a>>::Searcher);

unsafe impl<'a, 'b> Searcher<'a> for CharSliceSearcher<'a, 'b> {
    searcher_methods!(forward);
}

unsafe impl<'a, 'b> ReverseSearcher<'a> for CharSliceSearcher<'a, 'b> {
    searcher_methods!(reverse);
}

impl<'a, 'b> DoubleEndedSearcher<'a> for CharSliceSearcher<'a, 'b> {}

/// Searches for chars that are equal to any of the chars in the array
impl<'a, 'b> Pattern<'a> for &'b [char] {
    pattern_methods!(CharSliceSearcher<'a, 'b>, MultiCharEqPattern, CharSliceSearcher);
}

/////////////////////////////////////////////////////////////////////////////
// Impl for F: FnMut(char) -> bool
/////////////////////////////////////////////////////////////////////////////

/// Associated type for `<F as Pattern<'a>>::Searcher`.
#[derive(Clone)]
pub struct CharPredicateSearcher<'a, F>(<MultiCharEqPattern<F> as Pattern<'a>>::Searcher)
    where F: FnMut(char) -> bool;

impl<F> fmt::Debug for CharPredicateSearcher<'_, F>
    where F: FnMut(char) -> bool
{
    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
        f.debug_struct("CharPredicateSearcher")
            .field("haystack", &self.0.haystack)
            .field("char_indices", &self.0.char_indices)
            .finish()
    }
}
unsafe impl<'a, F> Searcher<'a> for CharPredicateSearcher<'a, F>
    where F: FnMut(char) -> bool
{
    searcher_methods!(forward);
}

unsafe impl<'a, F> ReverseSearcher<'a> for CharPredicateSearcher<'a, F>
    where F: FnMut(char) -> bool
{
    searcher_methods!(reverse);
}

impl<'a, F> DoubleEndedSearcher<'a> for CharPredicateSearcher<'a, F>
    where F: FnMut(char) -> bool {}

/// Searches for chars that match the given predicate
impl<'a, F> Pattern<'a> for F where F: FnMut(char) -> bool {
    pattern_methods!(CharPredicateSearcher<'a, F>, MultiCharEqPattern, CharPredicateSearcher);
}

/////////////////////////////////////////////////////////////////////////////
// Impl for &&str
/////////////////////////////////////////////////////////////////////////////

/// Delegates to the `&str` impl.
impl<'a, 'b, 'c> Pattern<'a> for &'c &'b str {
    pattern_methods!(StrSearcher<'a, 'b>, |&s| s, |s| s);
}

/////////////////////////////////////////////////////////////////////////////
// Impl for &str
/////////////////////////////////////////////////////////////////////////////

/// Non-allocating substring search.
///
/// Will handle the pattern `""` as returning empty matches at each character
/// boundary.
impl<'a, 'b> Pattern<'a> for &'b str {
    type Searcher = StrSearcher<'a, 'b>;

    #[inline]
    fn into_searcher(self, haystack: &'a str) -> StrSearcher<'a, 'b> {
        StrSearcher::new(haystack, self)
    }

    /// Checks whether the pattern matches at the front of the haystack
    #[inline]
    fn is_prefix_of(self, haystack: &'a str) -> bool {
        haystack.is_char_boundary(self.len()) &&
            self == &haystack[..self.len()]
    }

    /// Checks whether the pattern matches at the back of the haystack
    #[inline]
    fn is_suffix_of(self, haystack: &'a str) -> bool {
        self.len() <= haystack.len() &&
            haystack.is_char_boundary(haystack.len() - self.len()) &&
            self == &haystack[haystack.len() - self.len()..]
    }
}


/////////////////////////////////////////////////////////////////////////////
// Two Way substring searcher
/////////////////////////////////////////////////////////////////////////////

#[derive(Clone, Debug)]
/// Associated type for `<&str as Pattern<'a>>::Searcher`.
pub struct StrSearcher<'a, 'b> {
    haystack: &'a str,
    needle: &'b str,

    searcher: StrSearcherImpl,
}

#[derive(Clone, Debug)]
enum StrSearcherImpl {
    Empty(EmptyNeedle),
    TwoWay(TwoWaySearcher),
}

#[derive(Clone, Debug)]
struct EmptyNeedle {
    position: usize,
    end: usize,
    is_match_fw: bool,
    is_match_bw: bool,
}

impl<'a, 'b> StrSearcher<'a, 'b> {
    fn new(haystack: &'a str, needle: &'b str) -> StrSearcher<'a, 'b> {
        if needle.is_empty() {
            StrSearcher {
                haystack,
                needle,
                searcher: StrSearcherImpl::Empty(EmptyNeedle {
                    position: 0,
                    end: haystack.len(),
                    is_match_fw: true,
                    is_match_bw: true,
                }),
            }
        } else {
            StrSearcher {
                haystack,
                needle,
                searcher: StrSearcherImpl::TwoWay(
                    TwoWaySearcher::new(needle.as_bytes(), haystack.len())
                ),
            }
        }
    }
}

unsafe impl<'a, 'b> Searcher<'a> for StrSearcher<'a, 'b> {
    #[inline]
    fn haystack(&self) -> &'a str {
        self.haystack
    }

    #[inline]
    fn next(&mut self) -> SearchStep {
        match self.searcher {
            StrSearcherImpl::Empty(ref mut searcher) => {
                // empty needle rejects every char and matches every empty string between them
                let is_match = searcher.is_match_fw;
                searcher.is_match_fw = !searcher.is_match_fw;
                let pos = searcher.position;
                match self.haystack[pos..].chars().next() {
                    _ if is_match => SearchStep::Match(pos, pos),
                    None => SearchStep::Done,
                    Some(ch) => {
                        searcher.position += ch.len_utf8();
                        SearchStep::Reject(pos, searcher.position)
                    }
                }
            }
            StrSearcherImpl::TwoWay(ref mut searcher) => {
                // TwoWaySearcher produces valid *Match* indices that split at char boundaries
                // as long as it does correct matching and that haystack and needle are
                // valid UTF-8
                // *Rejects* from the algorithm can fall on any indices, but we will walk them
                // manually to the next character boundary, so that they are utf-8 safe.
                if searcher.position == self.haystack.len() {
                    return SearchStep::Done;
                }
                let is_long = searcher.memory == usize::MAX;
                match searcher.next::<RejectAndMatch>(self.haystack.as_bytes(),
                                                      self.needle.as_bytes(),
                                                      is_long)
                {
                    SearchStep::Reject(a, mut b) => {
                        // skip to next char boundary
                        while !self.haystack.is_char_boundary(b) {
                            b += 1;
                        }
                        searcher.position = cmp::max(b, searcher.position);
                        SearchStep::Reject(a, b)
                    }
                    otherwise => otherwise,
                }
            }
        }
    }

    #[inline]
    fn next_match(&mut self) -> Option<(usize, usize)> {
        match self.searcher {
            StrSearcherImpl::Empty(..) => {
                loop {
                    match self.next() {
                        SearchStep::Match(a, b) => return Some((a, b)),
                        SearchStep::Done => return None,
                        SearchStep::Reject(..) => { }
                    }
                }
            }
            StrSearcherImpl::TwoWay(ref mut searcher) => {
                let is_long = searcher.memory == usize::MAX;
                // write out `true` and `false` cases to encourage the compiler
                // to specialize the two cases separately.
                if is_long {
                    searcher.next::<MatchOnly>(self.haystack.as_bytes(),
                                               self.needle.as_bytes(),
                                               true)
                } else {
                    searcher.next::<MatchOnly>(self.haystack.as_bytes(),
                                               self.needle.as_bytes(),
                                               false)
                }
            }
        }
    }
}

unsafe impl<'a, 'b> ReverseSearcher<'a> for StrSearcher<'a, 'b> {
    #[inline]
    fn next_back(&mut self) -> SearchStep {
        match self.searcher {
            StrSearcherImpl::Empty(ref mut searcher) => {
                let is_match = searcher.is_match_bw;
                searcher.is_match_bw = !searcher.is_match_bw;
                let end = searcher.end;
                match self.haystack[..end].chars().next_back() {
                    _ if is_match => SearchStep::Match(end, end),
                    None => SearchStep::Done,
                    Some(ch) => {
                        searcher.end -= ch.len_utf8();
                        SearchStep::Reject(searcher.end, end)
                    }
                }
            }
            StrSearcherImpl::TwoWay(ref mut searcher) => {
                if searcher.end == 0 {
                    return SearchStep::Done;
                }
                let is_long = searcher.memory == usize::MAX;
                match searcher.next_back::<RejectAndMatch>(self.haystack.as_bytes(),
                                                           self.needle.as_bytes(),
                                                           is_long)
                {
                    SearchStep::Reject(mut a, b) => {
                        // skip to next char boundary
                        while !self.haystack.is_char_boundary(a) {
                            a -= 1;
                        }
                        searcher.end = cmp::min(a, searcher.end);
                        SearchStep::Reject(a, b)
                    }
                    otherwise => otherwise,
                }
            }
        }
    }

    #[inline]
    fn next_match_back(&mut self) -> Option<(usize, usize)> {
        match self.searcher {
            StrSearcherImpl::Empty(..) => {
                loop {
                    match self.next_back() {
                        SearchStep::Match(a, b) => return Some((a, b)),
                        SearchStep::Done => return None,
                        SearchStep::Reject(..) => { }
                    }
                }
            }
            StrSearcherImpl::TwoWay(ref mut searcher) => {
                let is_long = searcher.memory == usize::MAX;
                // write out `true` and `false`, like `next_match`
                if is_long {
                    searcher.next_back::<MatchOnly>(self.haystack.as_bytes(),
                                                    self.needle.as_bytes(),
                                                    true)
                } else {
                    searcher.next_back::<MatchOnly>(self.haystack.as_bytes(),
                                                    self.needle.as_bytes(),
                                                    false)
                }
            }
        }
    }
}

/// The internal state of the two-way substring search algorithm.
#[derive(Clone, Debug)]
struct TwoWaySearcher {
    // constants
    /// critical factorization index
    crit_pos: usize,
    /// critical factorization index for reversed needle
    crit_pos_back: usize,
    period: usize,
    /// `byteset` is an extension (not part of the two way algorithm);
    /// it's a 64-bit "fingerprint" where each set bit `j` corresponds
    /// to a (byte & 63) == j present in the needle.
    byteset: u64,

    // variables
    position: usize,
    end: usize,
    /// index into needle before which we have already matched
    memory: usize,
    /// index into needle after which we have already matched
    memory_back: usize,
}

/*
    This is the Two-Way search algorithm, which was introduced in the paper:
    Crochemore, M., Perrin, D., 1991, Two-way string-matching, Journal of the ACM 38(3):651-675.

    Here's some background information.

    A *word* is a string of symbols. The *length* of a word should be a familiar
    notion, and here we denote it for any word x by |x|.
    (We also allow for the possibility of the *empty word*, a word of length zero).

    If x is any non-empty word, then an integer p with 0 < p <= |x| is said to be a
    *period* for x iff for all i with 0 <= i <= |x| - p - 1, we have x[i] == x[i+p].
    For example, both 1 and 2 are periods for the string "aa". As another example,
    the only period of the string "abcd" is 4.

    We denote by period(x) the *smallest* period of x (provided that x is non-empty).
    This is always well-defined since every non-empty word x has at least one period,
    |x|. We sometimes call this *the period* of x.

    If u, v and x are words such that x = uv, where uv is the concatenation of u and
    v, then we say that (u, v) is a *factorization* of x.

    Let (u, v) be a factorization for a word x. Then if w is a non-empty word such
    that both of the following hold

      - either w is a suffix of u or u is a suffix of w
      - either w is a prefix of v or v is a prefix of w

    then w is said to be a *repetition* for the factorization (u, v).

    Just to unpack this, there are four possibilities here. Let w = "abc". Then we
    might have:

      - w is a suffix of u and w is a prefix of v. ex: ("lolabc", "abcde")
      - w is a suffix of u and v is a prefix of w. ex: ("lolabc", "ab")
      - u is a suffix of w and w is a prefix of v. ex: ("bc", "abchi")
      - u is a suffix of w and v is a prefix of w. ex: ("bc", "a")

    Note that the word vu is a repetition for any factorization (u,v) of x = uv,
    so every factorization has at least one repetition.

    If x is a string and (u, v) is a factorization for x, then a *local period* for
    (u, v) is an integer r such that there is some word w such that |w| = r and w is
    a repetition for (u, v).

    We denote by local_period(u, v) the smallest local period of (u, v). We sometimes
    call this *the local period* of (u, v). Provided that x = uv is non-empty, this
    is well-defined (because each non-empty word has at least one factorization, as
    noted above).

    It can be proven that the following is an equivalent definition of a local period
    for a factorization (u, v): any positive integer r such that x[i] == x[i+r] for
    all i such that |u| - r <= i <= |u| - 1 and such that both x[i] and x[i+r] are
    defined. (i.e. i > 0 and i + r < |x|).

    Using the above reformulation, it is easy to prove that

        1 <= local_period(u, v) <= period(uv)

    A factorization (u, v) of x such that local_period(u,v) = period(x) is called a
    *critical factorization*.

    The algorithm hinges on the following theorem, which is stated without proof:

    **Critical Factorization Theorem** Any word x has at least one critical
    factorization (u, v) such that |u| < period(x).

    The purpose of maximal_suffix is to find such a critical factorization.

    If the period is short, compute another factorization x = u' v' to use
    for reverse search, chosen instead so that |v'| < period(x).

*/
impl TwoWaySearcher {
    fn new(needle: &[u8], end: usize) -> TwoWaySearcher {
        let (crit_pos_false, period_false) = TwoWaySearcher::maximal_suffix(needle, false);
        let (crit_pos_true, period_true) = TwoWaySearcher::maximal_suffix(needle, true);

        let (crit_pos, period) =
            if crit_pos_false > crit_pos_true {
                (crit_pos_false, period_false)
            } else {
                (crit_pos_true, period_true)
            };

        // A particularly readable explanation of what's going on here can be found
        // in Crochemore and Rytter's book "Text Algorithms", ch 13. Specifically
        // see the code for "Algorithm CP" on p. 323.
        //
        // What's going on is we have some critical factorization (u, v) of the
        // needle, and we want to determine whether u is a suffix of
        // &v[..period]. If it is, we use "Algorithm CP1". Otherwise we use
        // "Algorithm CP2", which is optimized for when the period of the needle
        // is large.
        if &needle[..crit_pos] == &needle[period.. period + crit_pos] {
            // short period case -- the period is exact
            // compute a separate critical factorization for the reversed needle
            // x = u' v' where |v'| < period(x).
            //
            // This is sped up by the period being known already.
            // Note that a case like x = "acba" may be factored exactly forwards
            // (crit_pos = 1, period = 3) while being factored with approximate
            // period in reverse (crit_pos = 2, period = 2). We use the given
            // reverse factorization but keep the exact period.
            let crit_pos_back = needle.len() - cmp::max(
                TwoWaySearcher::reverse_maximal_suffix(needle, period, false),
                TwoWaySearcher::reverse_maximal_suffix(needle, period, true));

            TwoWaySearcher {
                crit_pos,
                crit_pos_back,
                period,
                byteset: Self::byteset_create(&needle[..period]),

                position: 0,
                end,
                memory: 0,
                memory_back: needle.len(),
            }
        } else {
            // long period case -- we have an approximation to the actual period,
            // and don't use memorization.
            //
            // Approximate the period by lower bound max(|u|, |v|) + 1.
            // The critical factorization is efficient to use for both forward and
            // reverse search.

            TwoWaySearcher {
                crit_pos,
                crit_pos_back: crit_pos,
                period: cmp::max(crit_pos, needle.len() - crit_pos) + 1,
                byteset: Self::byteset_create(needle),

                position: 0,
                end,
                memory: usize::MAX, // Dummy value to signify that the period is long
                memory_back: usize::MAX,
            }
        }
    }

    #[inline]
    fn byteset_create(bytes: &[u8]) -> u64 {
        bytes.iter().fold(0, |a, &b| (1 << (b & 0x3f)) | a)
    }

    #[inline]
    fn byteset_contains(&self, byte: u8) -> bool {
        (self.byteset >> ((byte & 0x3f) as usize)) & 1 != 0
    }

    // One of the main ideas of Two-Way is that we factorize the needle into
    // two halves, (u, v), and begin trying to find v in the haystack by scanning
    // left to right. If v matches, we try to match u by scanning right to left.
    // How far we can jump when we encounter a mismatch is all based on the fact
    // that (u, v) is a critical factorization for the needle.
    #[inline]
    fn next<S>(&mut self, haystack: &[u8], needle: &[u8], long_period: bool)
        -> S::Output
        where S: TwoWayStrategy
    {
        // `next()` uses `self.position` as its cursor
        let old_pos = self.position;
        let needle_last = needle.len() - 1;
        'search: loop {
            // Check that we have room to search in
            // position + needle_last can not overflow if we assume slices
            // are bounded by isize's range.
            let tail_byte = match haystack.get(self.position + needle_last) {
                Some(&b) => b,
                None => {
                    self.position = haystack.len();
                    return S::rejecting(old_pos, self.position);
                }
            };

            if S::use_early_reject() && old_pos != self.position {
                return S::rejecting(old_pos, self.position);
            }

            // Quickly skip by large portions unrelated to our substring
            if !self.byteset_contains(tail_byte) {
                self.position += needle.len();
                if !long_period {
                    self.memory = 0;
                }
                continue 'search;
            }

            // See if the right part of the needle matches
            let start = if long_period { self.crit_pos }
                        else { cmp::max(self.crit_pos, self.memory) };
            for i in start..needle.len() {
                if needle[i] != haystack[self.position + i] {
                    self.position += i - self.crit_pos + 1;
                    if !long_period {
                        self.memory = 0;
                    }
                    continue 'search;
                }
            }

            // See if the left part of the needle matches
            let start = if long_period { 0 } else { self.memory };
            for i in (start..self.crit_pos).rev() {
                if needle[i] != haystack[self.position + i] {
                    self.position += self.period;
                    if !long_period {
                        self.memory = needle.len() - self.period;
                    }
                    continue 'search;
                }
            }

            // We have found a match!
            let match_pos = self.position;

            // Note: add self.period instead of needle.len() to have overlapping matches
            self.position += needle.len();
            if !long_period {
                self.memory = 0; // set to needle.len() - self.period for overlapping matches
            }

            return S::matching(match_pos, match_pos + needle.len());
        }
    }

    // Follows the ideas in `next()`.
    //
    // The definitions are symmetrical, with period(x) = period(reverse(x))
    // and local_period(u, v) = local_period(reverse(v), reverse(u)), so if (u, v)
    // is a critical factorization, so is (reverse(v), reverse(u)).
    //
    // For the reverse case we have computed a critical factorization x = u' v'
    // (field `crit_pos_back`). We need |u| < period(x) for the forward case and
    // thus |v'| < period(x) for the reverse.
    //
    // To search in reverse through the haystack, we search forward through
    // a reversed haystack with a reversed needle, matching first u' and then v'.
    #[inline]
    fn next_back<S>(&mut self, haystack: &[u8], needle: &[u8], long_period: bool)
        -> S::Output
        where S: TwoWayStrategy
    {
        // `next_back()` uses `self.end` as its cursor -- so that `next()` and `next_back()`
        // are independent.
        let old_end = self.end;
        'search: loop {
            // Check that we have room to search in
            // end - needle.len() will wrap around when there is no more room,
            // but due to slice length limits it can never wrap all the way back
            // into the length of haystack.
            let front_byte = match haystack.get(self.end.wrapping_sub(needle.len())) {
                Some(&b) => b,
                None => {
                    self.end = 0;
                    return S::rejecting(0, old_end);
                }
            };

            if S::use_early_reject() && old_end != self.end {
                return S::rejecting(self.end, old_end);
            }

            // Quickly skip by large portions unrelated to our substring
            if !self.byteset_contains(front_byte) {
                self.end -= needle.len();
                if !long_period {
                    self.memory_back = needle.len();
                }
                continue 'search;
            }

            // See if the left part of the needle matches
            let crit = if long_period { self.crit_pos_back }
                       else { cmp::min(self.crit_pos_back, self.memory_back) };
            for i in (0..crit).rev() {
                if needle[i] != haystack[self.end - needle.len() + i] {
                    self.end -= self.crit_pos_back - i;
                    if !long_period {
                        self.memory_back = needle.len();
                    }
                    continue 'search;
                }
            }

            // See if the right part of the needle matches
            let needle_end = if long_period { needle.len() }
                             else { self.memory_back };
            for i in self.crit_pos_back..needle_end {
                if needle[i] != haystack[self.end - needle.len() + i] {
                    self.end -= self.period;
                    if !long_period {
                        self.memory_back = self.period;
                    }
                    continue 'search;
                }
            }

            // We have found a match!
            let match_pos = self.end - needle.len();
            // Note: sub self.period instead of needle.len() to have overlapping matches
            self.end -= needle.len();
            if !long_period {
                self.memory_back = needle.len();
            }

            return S::matching(match_pos, match_pos + needle.len());
        }
    }

    // Compute the maximal suffix of `arr`.
    //
    // The maximal suffix is a possible critical factorization (u, v) of `arr`.
    //
    // Returns (`i`, `p`) where `i` is the starting index of v and `p` is the
    // period of v.
    //
    // `order_greater` determines if lexical order is `<` or `>`. Both
    // orders must be computed -- the ordering with the largest `i` gives
    // a critical factorization.
    //
    // For long period cases, the resulting period is not exact (it is too short).
    #[inline]
    fn maximal_suffix(arr: &[u8], order_greater: bool) -> (usize, usize) {
        let mut left = 0; // Corresponds to i in the paper
        let mut right = 1; // Corresponds to j in the paper
        let mut offset = 0; // Corresponds to k in the paper, but starting at 0
                            // to match 0-based indexing.
        let mut period = 1; // Corresponds to p in the paper

        while let Some(&a) = arr.get(right + offset) {
            // `left` will be inbounds when `right` is.
            let b = arr[left + offset];
            if (a < b && !order_greater) || (a > b && order_greater) {
                // Suffix is smaller, period is entire prefix so far.
                right += offset + 1;
                offset = 0;
                period = right - left;
            } else if a == b {
                // Advance through repetition of the current period.
                if offset + 1 == period {
                    right += offset + 1;
                    offset = 0;
                } else {
                    offset += 1;
                }
            } else {
                // Suffix is larger, start over from current location.
                left = right;
                right += 1;
                offset = 0;
                period = 1;
            }
        }
        (left, period)
    }

    // Compute the maximal suffix of the reverse of `arr`.
    //
    // The maximal suffix is a possible critical factorization (u', v') of `arr`.
    //
    // Returns `i` where `i` is the starting index of v', from the back;
    // returns immediately when a period of `known_period` is reached.
    //
    // `order_greater` determines if lexical order is `<` or `>`. Both
    // orders must be computed -- the ordering with the largest `i` gives
    // a critical factorization.
    //
    // For long period cases, the resulting period is not exact (it is too short).
    fn reverse_maximal_suffix(arr: &[u8], known_period: usize,
                              order_greater: bool) -> usize
    {
        let mut left = 0; // Corresponds to i in the paper
        let mut right = 1; // Corresponds to j in the paper
        let mut offset = 0; // Corresponds to k in the paper, but starting at 0
                            // to match 0-based indexing.
        let mut period = 1; // Corresponds to p in the paper
        let n = arr.len();

        while right + offset < n {
            let a = arr[n - (1 + right + offset)];
            let b = arr[n - (1 + left + offset)];
            if (a < b && !order_greater) || (a > b && order_greater) {
                // Suffix is smaller, period is entire prefix so far.
                right += offset + 1;
                offset = 0;
                period = right - left;
            } else if a == b {
                // Advance through repetition of the current period.
                if offset + 1 == period {
                    right += offset + 1;
                    offset = 0;
                } else {
                    offset += 1;
                }
            } else {
                // Suffix is larger, start over from current location.
                left = right;
                right += 1;
                offset = 0;
                period = 1;
            }
            if period == known_period {
                break;
            }
        }
        debug_assert!(period <= known_period);
        left
    }
}

// TwoWayStrategy allows the algorithm to either skip non-matches as quickly
// as possible, or to work in a mode where it emits Rejects relatively quickly.
trait TwoWayStrategy {
    type Output;
    fn use_early_reject() -> bool;
    fn rejecting(a: usize, b: usize) -> Self::Output;
    fn matching(a: usize, b: usize) -> Self::Output;
}

/// Skip to match intervals as quickly as possible
enum MatchOnly { }

impl TwoWayStrategy for MatchOnly {
    type Output = Option<(usize, usize)>;

    #[inline]
    fn use_early_reject() -> bool { false }
    #[inline]
    fn rejecting(_a: usize, _b: usize) -> Self::Output { None }
    #[inline]
    fn matching(a: usize, b: usize) -> Self::Output { Some((a, b)) }
}

/// Emit Rejects regularly
enum RejectAndMatch { }

impl TwoWayStrategy for RejectAndMatch {
    type Output = SearchStep;

    #[inline]
    fn use_early_reject() -> bool { true }
    #[inline]
    fn rejecting(a: usize, b: usize) -> Self::Output { SearchStep::Reject(a, b) }
    #[inline]
    fn matching(a: usize, b: usize) -> Self::Output { SearchStep::Match(a, b) }
}