Files
addr2line
adler
ahash
aho_corasick
ansi_term
anyhow
arc_swap
arrayref
arrayvec
ascii
assert_matches
async_stream
async_stream_impl
async_trait
atty
auto_enums
auto_enums_core
auto_enums_derive
backoff
backtrace
base32
base64
bincode
bip39
bitflags
bitvec
blake3
block_buffer
block_padding
borsh
borsh_derive
borsh_derive_internal
borsh_schema_derive_internal
bs58
bstr
bv
byte_slice_cast
byte_unit
bytecount
byteorder
bytes
bzip2
bzip2_sys
cargo_build_bpf
cargo_metadata
cargo_platform
cargo_test_bpf
cast
cc
cfg_if
chrono
chrono_humanize
clap
colored
combine
console
const_fn
constant_time_eq
core_affinity
cpufeatures
crc32fast
criterion_stats
crossbeam_channel
crossbeam_deque
crossbeam_epoch
crossbeam_queue
crossbeam_utils
crunchy
crypto_mac
csv
csv_core
ctrlc
curve25519_dalek
dashmap
derivative
derive_more
derive_utils
dialoguer
digest
dir_diff
dirs_next
dirs_sys_next
dlopen
dlopen_derive
doc_comment
dtoa
ed25519
ed25519_dalek
either
encoding_rs
enum_iterator
enum_iterator_derive
env_logger
ethabi
ethbloom
ethereum
ethereum_types
evm
evm_bridge
evm_core
evm_gasometer
evm_rpc
evm_runtime
evm_state
evm_utils
failure
failure_derive
fake_simd
fast_math
fd_lock
filetime
fixed_hash
flate2
fnv
foreign_types
foreign_types_shared
form_urlencoded
fs_extra
futures
futures_channel
futures_core
futures_executor
futures_io
futures_macro
futures_sink
futures_task
futures_util
async_await
future
io
lock
sink
stream
task
gag
generic_array
gethostname
getrandom
gimli
globset
goauth
goblin
h2
half
hash256_std_hasher
hash32
hash_db
hashbrown
heck
hex
hidapi
histogram
hmac
hmac_drbg
http
http_body
httparse
httpdate
humantime
hyper
hyper_rustls
hyper_tls
idna
ieee754
impl_codec
impl_rlp
impl_serde
indexed
indexmap
indicatif
inflector
cases
camelcase
case
classcase
kebabcase
pascalcase
screamingsnakecase
sentencecase
snakecase
tablecase
titlecase
traincase
numbers
deordinalize
ordinalize
string
constants
deconstantize
demodulize
pluralize
singularize
suffix
foreignkey
input_buffer
instant
iovec
ipnet
itertools
itoa
jemalloc_ctl
jemalloc_sys
jemallocator
jobserver
jsonrpc_client_transports
jsonrpc_core
jsonrpc_core_client
jsonrpc_derive
jsonrpc_http_server
jsonrpc_pubsub
jsonrpc_server_utils
jsonrpc_ws_server
keccak
keccak_hash
keccak_hasher
kernel32
lazy_static
lazycell
libc
libloading
librocksdb_sys
linked_hash_map
lock_api
log
lru
matches
maybe_uninit
memchr
memmap2
memoffset
mime
mime_guess
miniz_oxide
mio
mio_extras
miow
native_tls
net2
nix
num_cpus
num_derive
num_enum
num_enum_derive
num_integer
num_traits
number_prefix
object
once_cell
opaque_debug
openssl
openssl_probe
openssl_sys
ouroboros
ouroboros_macro
parity_scale_codec
parity_scale_codec_derive
parity_ws
parking_lot
parking_lot_core
paste
paste_impl
paw
paw_attributes
paw_raw
pbkdf2
percent_encoding
pest
pickledb
pin_project
pin_project_lite
pin_utils
plain
ppv_lite86
pretty_hex
primitive_types
proc_macro2
proc_macro_crate
proc_macro_error
proc_macro_error_attr
proc_macro_hack
proc_macro_nested
prost
prost_derive
prost_types
quote
radium
rand
rand_chacha
rand_core
rand_isaac
raptorq
rayon
rayon_core
reed_solomon_erasure
regex
regex_automata
regex_syntax
remove_dir_all
reqwest
retain_mut
ring
ripemd160
rlp
rlp_derive
rocksdb
rpassword
rustc_demangle
rustc_hash
rustc_hex
rustls
rustversion
ryu
same_file
scopeguard
scroll
scroll_derive
sct
secp256k1
secp256k1_sys
semver
semver_parser
serde
serde_bytes
serde_cbor
serde_derive
serde_json
serde_urlencoded
serde_yaml
sha1
sha2
sha3
signal_hook
signal_hook_registry
signature
simpl
simple_logger
slab
smallvec
smpl_jwt
snafu
snafu_derive
socket2
solana_account_decoder
solana_accounts_bench
solana_banking_bench
solana_banks_client
solana_banks_interface
solana_banks_server
solana_bench_exchange
solana_bench_streamer
solana_bench_tps
solana_bench_tps_evm
solana_bpf_loader_program
solana_budget_program
solana_clap_utils
solana_cli
solana_cli_config
solana_cli_output
solana_client
solana_config_program
solana_core
solana_crate_features
solana_csv_to_validator_infos
solana_dos
solana_download_utils
solana_evm_loader_program
solana_exchange_program
solana_failure_program
solana_faucet
solana_frozen_abi
solana_frozen_abi_macro
solana_genesis
solana_ip_address
solana_ip_address_server
solana_ledger
solana_ledger_tool
solana_ledger_udev
solana_local_cluster
solana_log_analyzer
solana_logger
solana_measure
solana_merkle_root_bench
solana_merkle_tree
solana_metrics
solana_net_shaper
solana_net_utils
solana_noop_program
solana_notifier
solana_ownable
solana_perf
solana_poh_bench
solana_program
solana_program_test
solana_ramp_tps
solana_rayon_threadlimit
solana_rbpf
solana_remote_wallet
solana_runtime
solana_sdk
solana_sdk_macro
solana_secp256k1_program
solana_sleep_program
solana_stake_accounts
solana_stake_monitor
solana_stake_o_matic
solana_stake_program
solana_storage_bigtable
solana_storage_proto
solana_store_tool
solana_streamer
solana_sys_tuner
solana_tokens
solana_transaction_status
solana_upload_perf
solana_version
solana_vest_program
solana_vote_program
solana_watchtower
spin
spl_associated_token_account
spl_memo
spl_token
stable_deref_trait
standback
static_assertions
strsim
structopt
structopt_derive
subtle
symlink
syn
synstructure
sysctl
tar
tarpc
tarpc_plugins
tempfile
termcolor
terminal_size
textwrap
thiserror
thiserror_impl
thread_scoped
time
time_macros
time_macros_impl
tiny_keccak
tinyvec
tinyvec_macros
tokio
fs
future
io
loom
macros
net
park
process
runtime
signal
stream
sync
task
time
util
tokio_codec
tokio_executor
tokio_fs
tokio_io
tokio_reactor
tokio_rustls
tokio_serde
tokio_sync
tokio_tcp
tokio_threadpool
tokio_tls
tokio_util
toml
tonic
tower
tower_balance
tower_buffer
tower_discover
tower_layer
tower_limit
tower_load
tower_load_shed
tower_make
tower_ready_cache
tower_retry
tower_service
tower_timeout
tower_util
tracing
tracing_attributes
tracing_core
tracing_futures
trees
triedb
triehash
try_lock
tungstenite
typenum
ucd_trie
uint
unicase
unicode_bidi
unicode_normalization
unicode_segmentation
unicode_width
unicode_xid
unix_socket
unreachable
untrusted
url
users
utf8
utf8_width
vec_map
velas
velas_account_program
velas_faucet
velas_genesis
velas_gossip
velas_install
velas_install_init
velas_keygen
velas_test_validator
velas_validator
void
walkdir
want
webpki
webpki_roots
websocket
websocket_base
winapi
ws2_32
xattr
yaml_rust
zeroize
zeroize_derive
zstd
zstd_safe
zstd_sys
   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
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
use std::cell::RefCell;
use std::collections::HashMap;
use std::panic::AssertUnwindSafe;
use std::sync::Arc;

#[cfg(feature = "perf-literal")]
use aho_corasick::{AhoCorasick, AhoCorasickBuilder, MatchKind};
use syntax::hir::literal::Literals;
use syntax::hir::Hir;
use syntax::ParserBuilder;

use backtrack;
use compile::Compiler;
#[cfg(feature = "perf-dfa")]
use dfa;
use error::Error;
use input::{ByteInput, CharInput};
use literal::LiteralSearcher;
use pikevm;
use pool::{Pool, PoolGuard};
use prog::Program;
use re_builder::RegexOptions;
use re_bytes;
use re_set;
use re_trait::{Locations, RegularExpression, Slot};
use re_unicode;
use utf8::next_utf8;

/// `Exec` manages the execution of a regular expression.
///
/// In particular, this manages the various compiled forms of a single regular
/// expression and the choice of which matching engine to use to execute a
/// regular expression.
#[derive(Debug)]
pub struct Exec {
    /// All read only state.
    ro: Arc<ExecReadOnly>,
    /// A pool of reusable values for the various matching engines.
    ///
    /// Note that boxing this value is not strictly necessary, but it is an
    /// easy way to ensure that T does not bloat the stack sized used by a pool
    /// in the case where T is big. And this turns out to be the case at the
    /// time of writing for regex's use of this pool. At the time of writing,
    /// the size of a Regex on the stack is 856 bytes. Boxing this value
    /// reduces that size to 16 bytes.
    pool: Box<Pool<ProgramCache>>,
}

/// `ExecNoSync` is like `Exec`, except it embeds a reference to a cache. This
/// means it is no longer Sync, but we can now avoid the overhead of
/// synchronization to fetch the cache.
#[derive(Debug)]
pub struct ExecNoSync<'c> {
    /// All read only state.
    ro: &'c Arc<ExecReadOnly>,
    /// Caches for the various matching engines.
    cache: PoolGuard<'c, ProgramCache>,
}

/// `ExecNoSyncStr` is like `ExecNoSync`, but matches on &str instead of &[u8].
#[derive(Debug)]
pub struct ExecNoSyncStr<'c>(ExecNoSync<'c>);

/// `ExecReadOnly` comprises all read only state for a regex. Namely, all such
/// state is determined at compile time and never changes during search.
#[derive(Debug)]
struct ExecReadOnly {
    /// The original regular expressions given by the caller to compile.
    res: Vec<String>,
    /// A compiled program that is used in the NFA simulation and backtracking.
    /// It can be byte-based or Unicode codepoint based.
    ///
    /// N.B. It is not possibly to make this byte-based from the public API.
    /// It is only used for testing byte based programs in the NFA simulations.
    nfa: Program,
    /// A compiled byte based program for DFA execution. This is only used
    /// if a DFA can be executed. (Currently, only word boundary assertions are
    /// not supported.) Note that this program contains an embedded `.*?`
    /// preceding the first capture group, unless the regex is anchored at the
    /// beginning.
    dfa: Program,
    /// The same as above, except the program is reversed (and there is no
    /// preceding `.*?`). This is used by the DFA to find the starting location
    /// of matches.
    dfa_reverse: Program,
    /// A set of suffix literals extracted from the regex.
    ///
    /// Prefix literals are stored on the `Program`, since they are used inside
    /// the matching engines.
    suffixes: LiteralSearcher,
    /// An Aho-Corasick automaton with leftmost-first match semantics.
    ///
    /// This is only set when the entire regex is a simple unanchored
    /// alternation of literals. We could probably use it more circumstances,
    /// but this is already hacky enough in this architecture.
    ///
    /// N.B. We use u32 as a state ID representation under the assumption that
    /// if we were to exhaust the ID space, we probably would have long
    /// surpassed the compilation size limit.
    #[cfg(feature = "perf-literal")]
    ac: Option<AhoCorasick<u32>>,
    /// match_type encodes as much upfront knowledge about how we're going to
    /// execute a search as possible.
    match_type: MatchType,
}

/// Facilitates the construction of an executor by exposing various knobs
/// to control how a regex is executed and what kinds of resources it's
/// permitted to use.
// `ExecBuilder` is only public via the `internal` module, so avoid deriving
// `Debug`.
#[allow(missing_debug_implementations)]
pub struct ExecBuilder {
    options: RegexOptions,
    match_type: Option<MatchType>,
    bytes: bool,
    only_utf8: bool,
}

/// Parsed represents a set of parsed regular expressions and their detected
/// literals.
struct Parsed {
    exprs: Vec<Hir>,
    prefixes: Literals,
    suffixes: Literals,
    bytes: bool,
}

impl ExecBuilder {
    /// Create a regex execution builder.
    ///
    /// This uses default settings for everything except the regex itself,
    /// which must be provided. Further knobs can be set by calling methods,
    /// and then finally, `build` to actually create the executor.
    pub fn new(re: &str) -> Self {
        Self::new_many(&[re])
    }

    /// Like new, but compiles the union of the given regular expressions.
    ///
    /// Note that when compiling 2 or more regular expressions, capture groups
    /// are completely unsupported. (This means both `find` and `captures`
    /// wont work.)
    pub fn new_many<I, S>(res: I) -> Self
    where
        S: AsRef<str>,
        I: IntoIterator<Item = S>,
    {
        let mut opts = RegexOptions::default();
        opts.pats = res.into_iter().map(|s| s.as_ref().to_owned()).collect();
        Self::new_options(opts)
    }

    /// Create a regex execution builder.
    pub fn new_options(opts: RegexOptions) -> Self {
        ExecBuilder {
            options: opts,
            match_type: None,
            bytes: false,
            only_utf8: true,
        }
    }

    /// Set the matching engine to be automatically determined.
    ///
    /// This is the default state and will apply whatever optimizations are
    /// possible, such as running a DFA.
    ///
    /// This overrides whatever was previously set via the `nfa` or
    /// `bounded_backtracking` methods.
    pub fn automatic(mut self) -> Self {
        self.match_type = None;
        self
    }

    /// Sets the matching engine to use the NFA algorithm no matter what
    /// optimizations are possible.
    ///
    /// This overrides whatever was previously set via the `automatic` or
    /// `bounded_backtracking` methods.
    pub fn nfa(mut self) -> Self {
        self.match_type = Some(MatchType::Nfa(MatchNfaType::PikeVM));
        self
    }

    /// Sets the matching engine to use a bounded backtracking engine no
    /// matter what optimizations are possible.
    ///
    /// One must use this with care, since the bounded backtracking engine
    /// uses memory proportion to `len(regex) * len(text)`.
    ///
    /// This overrides whatever was previously set via the `automatic` or
    /// `nfa` methods.
    pub fn bounded_backtracking(mut self) -> Self {
        self.match_type = Some(MatchType::Nfa(MatchNfaType::Backtrack));
        self
    }

    /// Compiles byte based programs for use with the NFA matching engines.
    ///
    /// By default, the NFA engines match on Unicode scalar values. They can
    /// be made to use byte based programs instead. In general, the byte based
    /// programs are slower because of a less efficient encoding of character
    /// classes.
    ///
    /// Note that this does not impact DFA matching engines, which always
    /// execute on bytes.
    pub fn bytes(mut self, yes: bool) -> Self {
        self.bytes = yes;
        self
    }

    /// When disabled, the program compiled may match arbitrary bytes.
    ///
    /// When enabled (the default), all compiled programs exclusively match
    /// valid UTF-8 bytes.
    pub fn only_utf8(mut self, yes: bool) -> Self {
        self.only_utf8 = yes;
        self
    }

    /// Set the Unicode flag.
    pub fn unicode(mut self, yes: bool) -> Self {
        self.options.unicode = yes;
        self
    }

    /// Parse the current set of patterns into their AST and extract literals.
    fn parse(&self) -> Result<Parsed, Error> {
        let mut exprs = Vec::with_capacity(self.options.pats.len());
        let mut prefixes = Some(Literals::empty());
        let mut suffixes = Some(Literals::empty());
        let mut bytes = false;
        let is_set = self.options.pats.len() > 1;
        // If we're compiling a regex set and that set has any anchored
        // expressions, then disable all literal optimizations.
        for pat in &self.options.pats {
            let mut parser = ParserBuilder::new()
                .octal(self.options.octal)
                .case_insensitive(self.options.case_insensitive)
                .multi_line(self.options.multi_line)
                .dot_matches_new_line(self.options.dot_matches_new_line)
                .swap_greed(self.options.swap_greed)
                .ignore_whitespace(self.options.ignore_whitespace)
                .unicode(self.options.unicode)
                .allow_invalid_utf8(!self.only_utf8)
                .nest_limit(self.options.nest_limit)
                .build();
            let expr =
                parser.parse(pat).map_err(|e| Error::Syntax(e.to_string()))?;
            bytes = bytes || !expr.is_always_utf8();

            if cfg!(feature = "perf-literal") {
                if !expr.is_anchored_start() && expr.is_any_anchored_start() {
                    // Partial anchors unfortunately make it hard to use
                    // prefixes, so disable them.
                    prefixes = None;
                } else if is_set && expr.is_anchored_start() {
                    // Regex sets with anchors do not go well with literal
                    // optimizations.
                    prefixes = None;
                }
                prefixes = prefixes.and_then(|mut prefixes| {
                    if !prefixes.union_prefixes(&expr) {
                        None
                    } else {
                        Some(prefixes)
                    }
                });

                if !expr.is_anchored_end() && expr.is_any_anchored_end() {
                    // Partial anchors unfortunately make it hard to use
                    // suffixes, so disable them.
                    suffixes = None;
                } else if is_set && expr.is_anchored_end() {
                    // Regex sets with anchors do not go well with literal
                    // optimizations.
                    suffixes = None;
                }
                suffixes = suffixes.and_then(|mut suffixes| {
                    if !suffixes.union_suffixes(&expr) {
                        None
                    } else {
                        Some(suffixes)
                    }
                });
            }
            exprs.push(expr);
        }
        Ok(Parsed {
            exprs: exprs,
            prefixes: prefixes.unwrap_or_else(Literals::empty),
            suffixes: suffixes.unwrap_or_else(Literals::empty),
            bytes: bytes,
        })
    }

    /// Build an executor that can run a regular expression.
    pub fn build(self) -> Result<Exec, Error> {
        // Special case when we have no patterns to compile.
        // This can happen when compiling a regex set.
        if self.options.pats.is_empty() {
            let ro = Arc::new(ExecReadOnly {
                res: vec![],
                nfa: Program::new(),
                dfa: Program::new(),
                dfa_reverse: Program::new(),
                suffixes: LiteralSearcher::empty(),
                #[cfg(feature = "perf-literal")]
                ac: None,
                match_type: MatchType::Nothing,
            });
            let pool = ExecReadOnly::new_pool(&ro);
            return Ok(Exec { ro: ro, pool });
        }
        let parsed = self.parse()?;
        let mut nfa = Compiler::new()
            .size_limit(self.options.size_limit)
            .bytes(self.bytes || parsed.bytes)
            .only_utf8(self.only_utf8)
            .compile(&parsed.exprs)?;
        let mut dfa = Compiler::new()
            .size_limit(self.options.size_limit)
            .dfa(true)
            .only_utf8(self.only_utf8)
            .compile(&parsed.exprs)?;
        let mut dfa_reverse = Compiler::new()
            .size_limit(self.options.size_limit)
            .dfa(true)
            .only_utf8(self.only_utf8)
            .reverse(true)
            .compile(&parsed.exprs)?;

        #[cfg(feature = "perf-literal")]
        let ac = self.build_aho_corasick(&parsed);
        nfa.prefixes = LiteralSearcher::prefixes(parsed.prefixes);
        dfa.prefixes = nfa.prefixes.clone();
        dfa.dfa_size_limit = self.options.dfa_size_limit;
        dfa_reverse.dfa_size_limit = self.options.dfa_size_limit;

        let mut ro = ExecReadOnly {
            res: self.options.pats,
            nfa: nfa,
            dfa: dfa,
            dfa_reverse: dfa_reverse,
            suffixes: LiteralSearcher::suffixes(parsed.suffixes),
            #[cfg(feature = "perf-literal")]
            ac: ac,
            match_type: MatchType::Nothing,
        };
        ro.match_type = ro.choose_match_type(self.match_type);

        let ro = Arc::new(ro);
        let pool = ExecReadOnly::new_pool(&ro);
        Ok(Exec { ro, pool })
    }

    #[cfg(feature = "perf-literal")]
    fn build_aho_corasick(&self, parsed: &Parsed) -> Option<AhoCorasick<u32>> {
        if parsed.exprs.len() != 1 {
            return None;
        }
        let lits = match alternation_literals(&parsed.exprs[0]) {
            None => return None,
            Some(lits) => lits,
        };
        // If we have a small number of literals, then let Teddy handle
        // things (see literal/mod.rs).
        if lits.len() <= 32 {
            return None;
        }
        Some(
            AhoCorasickBuilder::new()
                .match_kind(MatchKind::LeftmostFirst)
                .auto_configure(&lits)
                // We always want this to reduce size, regardless
                // of what auto-configure does.
                .byte_classes(true)
                .build_with_size::<u32, _, _>(&lits)
                // This should never happen because we'd long exceed the
                // compilation limit for regexes first.
                .expect("AC automaton too big"),
        )
    }
}

impl<'c> RegularExpression for ExecNoSyncStr<'c> {
    type Text = str;

    fn slots_len(&self) -> usize {
        self.0.slots_len()
    }

    fn next_after_empty(&self, text: &str, i: usize) -> usize {
        next_utf8(text.as_bytes(), i)
    }

    #[cfg_attr(feature = "perf-inline", inline(always))]
    fn shortest_match_at(&self, text: &str, start: usize) -> Option<usize> {
        self.0.shortest_match_at(text.as_bytes(), start)
    }

    #[cfg_attr(feature = "perf-inline", inline(always))]
    fn is_match_at(&self, text: &str, start: usize) -> bool {
        self.0.is_match_at(text.as_bytes(), start)
    }

    #[cfg_attr(feature = "perf-inline", inline(always))]
    fn find_at(&self, text: &str, start: usize) -> Option<(usize, usize)> {
        self.0.find_at(text.as_bytes(), start)
    }

    #[cfg_attr(feature = "perf-inline", inline(always))]
    fn captures_read_at(
        &self,
        locs: &mut Locations,
        text: &str,
        start: usize,
    ) -> Option<(usize, usize)> {
        self.0.captures_read_at(locs, text.as_bytes(), start)
    }
}

impl<'c> RegularExpression for ExecNoSync<'c> {
    type Text = [u8];

    /// Returns the number of capture slots in the regular expression. (There
    /// are two slots for every capture group, corresponding to possibly empty
    /// start and end locations of the capture.)
    fn slots_len(&self) -> usize {
        self.ro.nfa.captures.len() * 2
    }

    fn next_after_empty(&self, _text: &[u8], i: usize) -> usize {
        i + 1
    }

    /// Returns the end of a match location, possibly occurring before the
    /// end location of the correct leftmost-first match.
    #[cfg_attr(feature = "perf-inline", inline(always))]
    fn shortest_match_at(&self, text: &[u8], start: usize) -> Option<usize> {
        if !self.is_anchor_end_match(text) {
            return None;
        }
        match self.ro.match_type {
            #[cfg(feature = "perf-literal")]
            MatchType::Literal(ty) => {
                self.find_literals(ty, text, start).map(|(_, e)| e)
            }
            #[cfg(feature = "perf-dfa")]
            MatchType::Dfa | MatchType::DfaMany => {
                match self.shortest_dfa(text, start) {
                    dfa::Result::Match(end) => Some(end),
                    dfa::Result::NoMatch(_) => None,
                    dfa::Result::Quit => self.shortest_nfa(text, start),
                }
            }
            #[cfg(feature = "perf-dfa")]
            MatchType::DfaAnchoredReverse => {
                match dfa::Fsm::reverse(
                    &self.ro.dfa_reverse,
                    self.cache.value(),
                    true,
                    &text[start..],
                    text.len(),
                ) {
                    dfa::Result::Match(_) => Some(text.len()),
                    dfa::Result::NoMatch(_) => None,
                    dfa::Result::Quit => self.shortest_nfa(text, start),
                }
            }
            #[cfg(all(feature = "perf-dfa", feature = "perf-literal"))]
            MatchType::DfaSuffix => {
                match self.shortest_dfa_reverse_suffix(text, start) {
                    dfa::Result::Match(e) => Some(e),
                    dfa::Result::NoMatch(_) => None,
                    dfa::Result::Quit => self.shortest_nfa(text, start),
                }
            }
            MatchType::Nfa(ty) => self.shortest_nfa_type(ty, text, start),
            MatchType::Nothing => None,
        }
    }

    /// Returns true if and only if the regex matches text.
    ///
    /// For single regular expressions, this is equivalent to calling
    /// shortest_match(...).is_some().
    #[cfg_attr(feature = "perf-inline", inline(always))]
    fn is_match_at(&self, text: &[u8], start: usize) -> bool {
        if !self.is_anchor_end_match(text) {
            return false;
        }
        // We need to do this dance because shortest_match relies on the NFA
        // filling in captures[1], but a RegexSet has no captures. In other
        // words, a RegexSet can't (currently) use shortest_match. ---AG
        match self.ro.match_type {
            #[cfg(feature = "perf-literal")]
            MatchType::Literal(ty) => {
                self.find_literals(ty, text, start).is_some()
            }
            #[cfg(feature = "perf-dfa")]
            MatchType::Dfa | MatchType::DfaMany => {
                match self.shortest_dfa(text, start) {
                    dfa::Result::Match(_) => true,
                    dfa::Result::NoMatch(_) => false,
                    dfa::Result::Quit => self.match_nfa(text, start),
                }
            }
            #[cfg(feature = "perf-dfa")]
            MatchType::DfaAnchoredReverse => {
                match dfa::Fsm::reverse(
                    &self.ro.dfa_reverse,
                    self.cache.value(),
                    true,
                    &text[start..],
                    text.len(),
                ) {
                    dfa::Result::Match(_) => true,
                    dfa::Result::NoMatch(_) => false,
                    dfa::Result::Quit => self.match_nfa(text, start),
                }
            }
            #[cfg(all(feature = "perf-dfa", feature = "perf-literal"))]
            MatchType::DfaSuffix => {
                match self.shortest_dfa_reverse_suffix(text, start) {
                    dfa::Result::Match(_) => true,
                    dfa::Result::NoMatch(_) => false,
                    dfa::Result::Quit => self.match_nfa(text, start),
                }
            }
            MatchType::Nfa(ty) => self.match_nfa_type(ty, text, start),
            MatchType::Nothing => false,
        }
    }

    /// Finds the start and end location of the leftmost-first match, starting
    /// at the given location.
    #[cfg_attr(feature = "perf-inline", inline(always))]
    fn find_at(&self, text: &[u8], start: usize) -> Option<(usize, usize)> {
        if !self.is_anchor_end_match(text) {
            return None;
        }
        match self.ro.match_type {
            #[cfg(feature = "perf-literal")]
            MatchType::Literal(ty) => self.find_literals(ty, text, start),
            #[cfg(feature = "perf-dfa")]
            MatchType::Dfa => match self.find_dfa_forward(text, start) {
                dfa::Result::Match((s, e)) => Some((s, e)),
                dfa::Result::NoMatch(_) => None,
                dfa::Result::Quit => {
                    self.find_nfa(MatchNfaType::Auto, text, start)
                }
            },
            #[cfg(feature = "perf-dfa")]
            MatchType::DfaAnchoredReverse => {
                match self.find_dfa_anchored_reverse(text, start) {
                    dfa::Result::Match((s, e)) => Some((s, e)),
                    dfa::Result::NoMatch(_) => None,
                    dfa::Result::Quit => {
                        self.find_nfa(MatchNfaType::Auto, text, start)
                    }
                }
            }
            #[cfg(all(feature = "perf-dfa", feature = "perf-literal"))]
            MatchType::DfaSuffix => {
                match self.find_dfa_reverse_suffix(text, start) {
                    dfa::Result::Match((s, e)) => Some((s, e)),
                    dfa::Result::NoMatch(_) => None,
                    dfa::Result::Quit => {
                        self.find_nfa(MatchNfaType::Auto, text, start)
                    }
                }
            }
            MatchType::Nfa(ty) => self.find_nfa(ty, text, start),
            MatchType::Nothing => None,
            #[cfg(feature = "perf-dfa")]
            MatchType::DfaMany => {
                unreachable!("BUG: RegexSet cannot be used with find")
            }
        }
    }

    /// Finds the start and end location of the leftmost-first match and also
    /// fills in all matching capture groups.
    ///
    /// The number of capture slots given should be equal to the total number
    /// of capture slots in the compiled program.
    ///
    /// Note that the first two slots always correspond to the start and end
    /// locations of the overall match.
    fn captures_read_at(
        &self,
        locs: &mut Locations,
        text: &[u8],
        start: usize,
    ) -> Option<(usize, usize)> {
        let slots = locs.as_slots();
        for slot in slots.iter_mut() {
            *slot = None;
        }
        // If the caller unnecessarily uses this, then we try to save them
        // from themselves.
        match slots.len() {
            0 => return self.find_at(text, start),
            2 => {
                return self.find_at(text, start).map(|(s, e)| {
                    slots[0] = Some(s);
                    slots[1] = Some(e);
                    (s, e)
                });
            }
            _ => {} // fallthrough
        }
        if !self.is_anchor_end_match(text) {
            return None;
        }
        match self.ro.match_type {
            #[cfg(feature = "perf-literal")]
            MatchType::Literal(ty) => {
                self.find_literals(ty, text, start).and_then(|(s, e)| {
                    self.captures_nfa_type(
                        MatchNfaType::Auto,
                        slots,
                        text,
                        s,
                        e,
                    )
                })
            }
            #[cfg(feature = "perf-dfa")]
            MatchType::Dfa => {
                if self.ro.nfa.is_anchored_start {
                    self.captures_nfa(slots, text, start)
                } else {
                    match self.find_dfa_forward(text, start) {
                        dfa::Result::Match((s, e)) => self.captures_nfa_type(
                            MatchNfaType::Auto,
                            slots,
                            text,
                            s,
                            e,
                        ),
                        dfa::Result::NoMatch(_) => None,
                        dfa::Result::Quit => {
                            self.captures_nfa(slots, text, start)
                        }
                    }
                }
            }
            #[cfg(feature = "perf-dfa")]
            MatchType::DfaAnchoredReverse => {
                match self.find_dfa_anchored_reverse(text, start) {
                    dfa::Result::Match((s, e)) => self.captures_nfa_type(
                        MatchNfaType::Auto,
                        slots,
                        text,
                        s,
                        e,
                    ),
                    dfa::Result::NoMatch(_) => None,
                    dfa::Result::Quit => self.captures_nfa(slots, text, start),
                }
            }
            #[cfg(all(feature = "perf-dfa", feature = "perf-literal"))]
            MatchType::DfaSuffix => {
                match self.find_dfa_reverse_suffix(text, start) {
                    dfa::Result::Match((s, e)) => self.captures_nfa_type(
                        MatchNfaType::Auto,
                        slots,
                        text,
                        s,
                        e,
                    ),
                    dfa::Result::NoMatch(_) => None,
                    dfa::Result::Quit => self.captures_nfa(slots, text, start),
                }
            }
            MatchType::Nfa(ty) => {
                self.captures_nfa_type(ty, slots, text, start, text.len())
            }
            MatchType::Nothing => None,
            #[cfg(feature = "perf-dfa")]
            MatchType::DfaMany => {
                unreachable!("BUG: RegexSet cannot be used with captures")
            }
        }
    }
}

impl<'c> ExecNoSync<'c> {
    /// Finds the leftmost-first match using only literal search.
    #[cfg(feature = "perf-literal")]
    #[cfg_attr(feature = "perf-inline", inline(always))]
    fn find_literals(
        &self,
        ty: MatchLiteralType,
        text: &[u8],
        start: usize,
    ) -> Option<(usize, usize)> {
        use self::MatchLiteralType::*;
        match ty {
            Unanchored => {
                let lits = &self.ro.nfa.prefixes;
                lits.find(&text[start..]).map(|(s, e)| (start + s, start + e))
            }
            AnchoredStart => {
                let lits = &self.ro.nfa.prefixes;
                if start == 0 || !self.ro.nfa.is_anchored_start {
                    lits.find_start(&text[start..])
                        .map(|(s, e)| (start + s, start + e))
                } else {
                    None
                }
            }
            AnchoredEnd => {
                let lits = &self.ro.suffixes;
                lits.find_end(&text[start..])
                    .map(|(s, e)| (start + s, start + e))
            }
            AhoCorasick => self
                .ro
                .ac
                .as_ref()
                .unwrap()
                .find(&text[start..])
                .map(|m| (start + m.start(), start + m.end())),
        }
    }

    /// Finds the leftmost-first match (start and end) using only the DFA.
    ///
    /// If the result returned indicates that the DFA quit, then another
    /// matching engine should be used.
    #[cfg(feature = "perf-dfa")]
    #[cfg_attr(feature = "perf-inline", inline(always))]
    fn find_dfa_forward(
        &self,
        text: &[u8],
        start: usize,
    ) -> dfa::Result<(usize, usize)> {
        use dfa::Result::*;
        let end = match dfa::Fsm::forward(
            &self.ro.dfa,
            self.cache.value(),
            false,
            text,
            start,
        ) {
            NoMatch(i) => return NoMatch(i),
            Quit => return Quit,
            Match(end) if start == end => return Match((start, start)),
            Match(end) => end,
        };
        // Now run the DFA in reverse to find the start of the match.
        match dfa::Fsm::reverse(
            &self.ro.dfa_reverse,
            self.cache.value(),
            false,
            &text[start..],
            end - start,
        ) {
            Match(s) => Match((start + s, end)),
            NoMatch(i) => NoMatch(i),
            Quit => Quit,
        }
    }

    /// Finds the leftmost-first match (start and end) using only the DFA,
    /// but assumes the regex is anchored at the end and therefore starts at
    /// the end of the regex and matches in reverse.
    ///
    /// If the result returned indicates that the DFA quit, then another
    /// matching engine should be used.
    #[cfg(feature = "perf-dfa")]
    #[cfg_attr(feature = "perf-inline", inline(always))]
    fn find_dfa_anchored_reverse(
        &self,
        text: &[u8],
        start: usize,
    ) -> dfa::Result<(usize, usize)> {
        use dfa::Result::*;
        match dfa::Fsm::reverse(
            &self.ro.dfa_reverse,
            self.cache.value(),
            false,
            &text[start..],
            text.len() - start,
        ) {
            Match(s) => Match((start + s, text.len())),
            NoMatch(i) => NoMatch(i),
            Quit => Quit,
        }
    }

    /// Finds the end of the shortest match using only the DFA.
    #[cfg(feature = "perf-dfa")]
    #[cfg_attr(feature = "perf-inline", inline(always))]
    fn shortest_dfa(&self, text: &[u8], start: usize) -> dfa::Result<usize> {
        dfa::Fsm::forward(&self.ro.dfa, self.cache.value(), true, text, start)
    }

    /// Finds the end of the shortest match using only the DFA by scanning for
    /// suffix literals.
    #[cfg(all(feature = "perf-dfa", feature = "perf-literal"))]
    #[cfg_attr(feature = "perf-inline", inline(always))]
    fn shortest_dfa_reverse_suffix(
        &self,
        text: &[u8],
        start: usize,
    ) -> dfa::Result<usize> {
        match self.exec_dfa_reverse_suffix(text, start) {
            None => self.shortest_dfa(text, start),
            Some(r) => r.map(|(_, end)| end),
        }
    }

    /// Finds the end of the shortest match using only the DFA by scanning for
    /// suffix literals. It also reports the start of the match.
    ///
    /// Note that if None is returned, then the optimization gave up to avoid
    /// worst case quadratic behavior. A forward scanning DFA should be tried
    /// next.
    ///
    /// If a match is returned and the full leftmost-first match is desired,
    /// then a forward scan starting from the beginning of the match must be
    /// done.
    ///
    /// If the result returned indicates that the DFA quit, then another
    /// matching engine should be used.
    #[cfg(all(feature = "perf-dfa", feature = "perf-literal"))]
    #[cfg_attr(feature = "perf-inline", inline(always))]
    fn exec_dfa_reverse_suffix(
        &self,
        text: &[u8],
        original_start: usize,
    ) -> Option<dfa::Result<(usize, usize)>> {
        use dfa::Result::*;

        let lcs = self.ro.suffixes.lcs();
        debug_assert!(lcs.len() >= 1);
        let mut start = original_start;
        let mut end = start;
        let mut last_literal = start;
        while end <= text.len() {
            last_literal += match lcs.find(&text[last_literal..]) {
                None => return Some(NoMatch(text.len())),
                Some(i) => i,
            };
            end = last_literal + lcs.len();
            match dfa::Fsm::reverse(
                &self.ro.dfa_reverse,
                self.cache.value(),
                false,
                &text[start..end],
                end - start,
            ) {
                Match(0) | NoMatch(0) => return None,
                Match(i) => return Some(Match((start + i, end))),
                NoMatch(i) => {
                    start += i;
                    last_literal += 1;
                    continue;
                }
                Quit => return Some(Quit),
            };
        }
        Some(NoMatch(text.len()))
    }

    /// Finds the leftmost-first match (start and end) using only the DFA
    /// by scanning for suffix literals.
    ///
    /// If the result returned indicates that the DFA quit, then another
    /// matching engine should be used.
    #[cfg(all(feature = "perf-dfa", feature = "perf-literal"))]
    #[cfg_attr(feature = "perf-inline", inline(always))]
    fn find_dfa_reverse_suffix(
        &self,
        text: &[u8],
        start: usize,
    ) -> dfa::Result<(usize, usize)> {
        use dfa::Result::*;

        let match_start = match self.exec_dfa_reverse_suffix(text, start) {
            None => return self.find_dfa_forward(text, start),
            Some(Match((start, _))) => start,
            Some(r) => return r,
        };
        // At this point, we've found a match. The only way to quit now
        // without a match is if the DFA gives up (seems unlikely).
        //
        // Now run the DFA forwards to find the proper end of the match.
        // (The suffix literal match can only indicate the earliest
        // possible end location, which may appear before the end of the
        // leftmost-first match.)
        match dfa::Fsm::forward(
            &self.ro.dfa,
            self.cache.value(),
            false,
            text,
            match_start,
        ) {
            NoMatch(_) => panic!("BUG: reverse match implies forward match"),
            Quit => Quit,
            Match(e) => Match((match_start, e)),
        }
    }

    /// Executes the NFA engine to return whether there is a match or not.
    ///
    /// Ideally, we could use shortest_nfa(...).is_some() and get the same
    /// performance characteristics, but regex sets don't have captures, which
    /// shortest_nfa depends on.
    #[cfg(feature = "perf-dfa")]
    fn match_nfa(&self, text: &[u8], start: usize) -> bool {
        self.match_nfa_type(MatchNfaType::Auto, text, start)
    }

    /// Like match_nfa, but allows specification of the type of NFA engine.
    fn match_nfa_type(
        &self,
        ty: MatchNfaType,
        text: &[u8],
        start: usize,
    ) -> bool {
        self.exec_nfa(
            ty,
            &mut [false],
            &mut [],
            true,
            false,
            text,
            start,
            text.len(),
        )
    }

    /// Finds the shortest match using an NFA.
    #[cfg(feature = "perf-dfa")]
    fn shortest_nfa(&self, text: &[u8], start: usize) -> Option<usize> {
        self.shortest_nfa_type(MatchNfaType::Auto, text, start)
    }

    /// Like shortest_nfa, but allows specification of the type of NFA engine.
    fn shortest_nfa_type(
        &self,
        ty: MatchNfaType,
        text: &[u8],
        start: usize,
    ) -> Option<usize> {
        let mut slots = [None, None];
        if self.exec_nfa(
            ty,
            &mut [false],
            &mut slots,
            true,
            true,
            text,
            start,
            text.len(),
        ) {
            slots[1]
        } else {
            None
        }
    }

    /// Like find, but executes an NFA engine.
    fn find_nfa(
        &self,
        ty: MatchNfaType,
        text: &[u8],
        start: usize,
    ) -> Option<(usize, usize)> {
        let mut slots = [None, None];
        if self.exec_nfa(
            ty,
            &mut [false],
            &mut slots,
            false,
            false,
            text,
            start,
            text.len(),
        ) {
            match (slots[0], slots[1]) {
                (Some(s), Some(e)) => Some((s, e)),
                _ => None,
            }
        } else {
            None
        }
    }

    /// Like find_nfa, but fills in captures.
    ///
    /// `slots` should have length equal to `2 * nfa.captures.len()`.
    #[cfg(feature = "perf-dfa")]
    fn captures_nfa(
        &self,
        slots: &mut [Slot],
        text: &[u8],
        start: usize,
    ) -> Option<(usize, usize)> {
        self.captures_nfa_type(
            MatchNfaType::Auto,
            slots,
            text,
            start,
            text.len(),
        )
    }

    /// Like captures_nfa, but allows specification of type of NFA engine.
    fn captures_nfa_type(
        &self,
        ty: MatchNfaType,
        slots: &mut [Slot],
        text: &[u8],
        start: usize,
        end: usize,
    ) -> Option<(usize, usize)> {
        if self.exec_nfa(
            ty,
            &mut [false],
            slots,
            false,
            false,
            text,
            start,
            end,
        ) {
            match (slots[0], slots[1]) {
                (Some(s), Some(e)) => Some((s, e)),
                _ => None,
            }
        } else {
            None
        }
    }

    fn exec_nfa(
        &self,
        mut ty: MatchNfaType,
        matches: &mut [bool],
        slots: &mut [Slot],
        quit_after_match: bool,
        quit_after_match_with_pos: bool,
        text: &[u8],
        start: usize,
        end: usize,
    ) -> bool {
        use self::MatchNfaType::*;
        if let Auto = ty {
            if backtrack::should_exec(self.ro.nfa.len(), text.len()) {
                ty = Backtrack;
            } else {
                ty = PikeVM;
            }
        }
        // The backtracker can't return the shortest match position as it is
        // implemented today. So if someone calls `shortest_match` and we need
        // to run an NFA, then use the PikeVM.
        if quit_after_match_with_pos || ty == PikeVM {
            self.exec_pikevm(
                matches,
                slots,
                quit_after_match,
                text,
                start,
                end,
            )
        } else {
            self.exec_backtrack(matches, slots, text, start, end)
        }
    }

    /// Always run the NFA algorithm.
    fn exec_pikevm(
        &self,
        matches: &mut [bool],
        slots: &mut [Slot],
        quit_after_match: bool,
        text: &[u8],
        start: usize,
        end: usize,
    ) -> bool {
        if self.ro.nfa.uses_bytes() {
            pikevm::Fsm::exec(
                &self.ro.nfa,
                self.cache.value(),
                matches,
                slots,
                quit_after_match,
                ByteInput::new(text, self.ro.nfa.only_utf8),
                start,
                end,
            )
        } else {
            pikevm::Fsm::exec(
                &self.ro.nfa,
                self.cache.value(),
                matches,
                slots,
                quit_after_match,
                CharInput::new(text),
                start,
                end,
            )
        }
    }

    /// Always runs the NFA using bounded backtracking.
    fn exec_backtrack(
        &self,
        matches: &mut [bool],
        slots: &mut [Slot],
        text: &[u8],
        start: usize,
        end: usize,
    ) -> bool {
        if self.ro.nfa.uses_bytes() {
            backtrack::Bounded::exec(
                &self.ro.nfa,
                self.cache.value(),
                matches,
                slots,
                ByteInput::new(text, self.ro.nfa.only_utf8),
                start,
                end,
            )
        } else {
            backtrack::Bounded::exec(
                &self.ro.nfa,
                self.cache.value(),
                matches,
                slots,
                CharInput::new(text),
                start,
                end,
            )
        }
    }

    /// Finds which regular expressions match the given text.
    ///
    /// `matches` should have length equal to the number of regexes being
    /// searched.
    ///
    /// This is only useful when one wants to know which regexes in a set
    /// match some text.
    pub fn many_matches_at(
        &self,
        matches: &mut [bool],
        text: &[u8],
        start: usize,
    ) -> bool {
        use self::MatchType::*;
        if !self.is_anchor_end_match(text) {
            return false;
        }
        match self.ro.match_type {
            #[cfg(feature = "perf-literal")]
            Literal(ty) => {
                debug_assert_eq!(matches.len(), 1);
                matches[0] = self.find_literals(ty, text, start).is_some();
                matches[0]
            }
            #[cfg(feature = "perf-dfa")]
            Dfa | DfaAnchoredReverse | DfaMany => {
                match dfa::Fsm::forward_many(
                    &self.ro.dfa,
                    self.cache.value(),
                    matches,
                    text,
                    start,
                ) {
                    dfa::Result::Match(_) => true,
                    dfa::Result::NoMatch(_) => false,
                    dfa::Result::Quit => self.exec_nfa(
                        MatchNfaType::Auto,
                        matches,
                        &mut [],
                        false,
                        false,
                        text,
                        start,
                        text.len(),
                    ),
                }
            }
            #[cfg(all(feature = "perf-dfa", feature = "perf-literal"))]
            DfaSuffix => {
                match dfa::Fsm::forward_many(
                    &self.ro.dfa,
                    self.cache.value(),
                    matches,
                    text,
                    start,
                ) {
                    dfa::Result::Match(_) => true,
                    dfa::Result::NoMatch(_) => false,
                    dfa::Result::Quit => self.exec_nfa(
                        MatchNfaType::Auto,
                        matches,
                        &mut [],
                        false,
                        false,
                        text,
                        start,
                        text.len(),
                    ),
                }
            }
            Nfa(ty) => self.exec_nfa(
                ty,
                matches,
                &mut [],
                false,
                false,
                text,
                start,
                text.len(),
            ),
            Nothing => false,
        }
    }

    #[cfg_attr(feature = "perf-inline", inline(always))]
    fn is_anchor_end_match(&self, text: &[u8]) -> bool {
        #[cfg(not(feature = "perf-literal"))]
        fn imp(_: &ExecReadOnly, _: &[u8]) -> bool {
            true
        }

        #[cfg(feature = "perf-literal")]
        fn imp(ro: &ExecReadOnly, text: &[u8]) -> bool {
            // Only do this check if the haystack is big (>1MB).
            if text.len() > (1 << 20) && ro.nfa.is_anchored_end {
                let lcs = ro.suffixes.lcs();
                if lcs.len() >= 1 && !lcs.is_suffix(text) {
                    return false;
                }
            }
            true
        }

        imp(&self.ro, text)
    }

    pub fn capture_name_idx(&self) -> &Arc<HashMap<String, usize>> {
        &self.ro.nfa.capture_name_idx
    }
}

impl<'c> ExecNoSyncStr<'c> {
    pub fn capture_name_idx(&self) -> &Arc<HashMap<String, usize>> {
        self.0.capture_name_idx()
    }
}

impl Exec {
    /// Get a searcher that isn't Sync.
    #[cfg_attr(feature = "perf-inline", inline(always))]
    pub fn searcher(&self) -> ExecNoSync {
        ExecNoSync {
            ro: &self.ro, // a clone is too expensive here! (and not needed)
            cache: self.pool.get(),
        }
    }

    /// Get a searcher that isn't Sync and can match on &str.
    #[cfg_attr(feature = "perf-inline", inline(always))]
    pub fn searcher_str(&self) -> ExecNoSyncStr {
        ExecNoSyncStr(self.searcher())
    }

    /// Build a Regex from this executor.
    pub fn into_regex(self) -> re_unicode::Regex {
        re_unicode::Regex::from(self)
    }

    /// Build a RegexSet from this executor.
    pub fn into_regex_set(self) -> re_set::unicode::RegexSet {
        re_set::unicode::RegexSet::from(self)
    }

    /// Build a Regex from this executor that can match arbitrary bytes.
    pub fn into_byte_regex(self) -> re_bytes::Regex {
        re_bytes::Regex::from(self)
    }

    /// Build a RegexSet from this executor that can match arbitrary bytes.
    pub fn into_byte_regex_set(self) -> re_set::bytes::RegexSet {
        re_set::bytes::RegexSet::from(self)
    }

    /// The original regular expressions given by the caller that were
    /// compiled.
    pub fn regex_strings(&self) -> &[String] {
        &self.ro.res
    }

    /// Return a slice of capture names.
    ///
    /// Any capture that isn't named is None.
    pub fn capture_names(&self) -> &[Option<String>] {
        &self.ro.nfa.captures
    }

    /// Return a reference to named groups mapping (from group name to
    /// group position).
    pub fn capture_name_idx(&self) -> &Arc<HashMap<String, usize>> {
        &self.ro.nfa.capture_name_idx
    }
}

impl Clone for Exec {
    fn clone(&self) -> Exec {
        let pool = ExecReadOnly::new_pool(&self.ro);
        Exec { ro: self.ro.clone(), pool }
    }
}

impl ExecReadOnly {
    fn choose_match_type(&self, hint: Option<MatchType>) -> MatchType {
        if let Some(MatchType::Nfa(_)) = hint {
            return hint.unwrap();
        }
        // If the NFA is empty, then we'll never match anything.
        if self.nfa.insts.is_empty() {
            return MatchType::Nothing;
        }
        if let Some(literalty) = self.choose_literal_match_type() {
            return literalty;
        }
        if let Some(dfaty) = self.choose_dfa_match_type() {
            return dfaty;
        }
        // We're so totally hosed.
        MatchType::Nfa(MatchNfaType::Auto)
    }

    /// If a plain literal scan can be used, then a corresponding literal
    /// search type is returned.
    fn choose_literal_match_type(&self) -> Option<MatchType> {
        #[cfg(not(feature = "perf-literal"))]
        fn imp(_: &ExecReadOnly) -> Option<MatchType> {
            None
        }

        #[cfg(feature = "perf-literal")]
        fn imp(ro: &ExecReadOnly) -> Option<MatchType> {
            // If our set of prefixes is complete, then we can use it to find
            // a match in lieu of a regex engine. This doesn't quite work well
            // in the presence of multiple regexes, so only do it when there's
            // one.
            //
            // TODO(burntsushi): Also, don't try to match literals if the regex
            // is partially anchored. We could technically do it, but we'd need
            // to create two sets of literals: all of them and then the subset
            // that aren't anchored. We would then only search for all of them
            // when at the beginning of the input and use the subset in all
            // other cases.
            if ro.res.len() != 1 {
                return None;
            }
            if ro.ac.is_some() {
                return Some(MatchType::Literal(
                    MatchLiteralType::AhoCorasick,
                ));
            }
            if ro.nfa.prefixes.complete() {
                return if ro.nfa.is_anchored_start {
                    Some(MatchType::Literal(MatchLiteralType::AnchoredStart))
                } else {
                    Some(MatchType::Literal(MatchLiteralType::Unanchored))
                };
            }
            if ro.suffixes.complete() {
                return if ro.nfa.is_anchored_end {
                    Some(MatchType::Literal(MatchLiteralType::AnchoredEnd))
                } else {
                    // This case shouldn't happen. When the regex isn't
                    // anchored, then complete prefixes should imply complete
                    // suffixes.
                    Some(MatchType::Literal(MatchLiteralType::Unanchored))
                };
            }
            None
        }

        imp(self)
    }

    /// If a DFA scan can be used, then choose the appropriate DFA strategy.
    fn choose_dfa_match_type(&self) -> Option<MatchType> {
        #[cfg(not(feature = "perf-dfa"))]
        fn imp(_: &ExecReadOnly) -> Option<MatchType> {
            None
        }

        #[cfg(feature = "perf-dfa")]
        fn imp(ro: &ExecReadOnly) -> Option<MatchType> {
            if !dfa::can_exec(&ro.dfa) {
                return None;
            }
            // Regex sets require a slightly specialized path.
            if ro.res.len() >= 2 {
                return Some(MatchType::DfaMany);
            }
            // If the regex is anchored at the end but not the start, then
            // just match in reverse from the end of the haystack.
            if !ro.nfa.is_anchored_start && ro.nfa.is_anchored_end {
                return Some(MatchType::DfaAnchoredReverse);
            }
            #[cfg(feature = "perf-literal")]
            {
                // If there's a longish suffix literal, then it might be faster
                // to look for that first.
                if ro.should_suffix_scan() {
                    return Some(MatchType::DfaSuffix);
                }
            }
            // Fall back to your garden variety forward searching lazy DFA.
            Some(MatchType::Dfa)
        }

        imp(self)
    }

    /// Returns true if the program is amenable to suffix scanning.
    ///
    /// When this is true, as a heuristic, we assume it is OK to quickly scan
    /// for suffix literals and then do a *reverse* DFA match from any matches
    /// produced by the literal scan. (And then followed by a forward DFA
    /// search, since the previously found suffix literal maybe not actually be
    /// the end of a match.)
    ///
    /// This is a bit of a specialized optimization, but can result in pretty
    /// big performance wins if 1) there are no prefix literals and 2) the
    /// suffix literals are pretty rare in the text. (1) is obviously easy to
    /// account for but (2) is harder. As a proxy, we assume that longer
    /// strings are generally rarer, so we only enable this optimization when
    /// we have a meaty suffix.
    #[cfg(all(feature = "perf-dfa", feature = "perf-literal"))]
    fn should_suffix_scan(&self) -> bool {
        if self.suffixes.is_empty() {
            return false;
        }
        let lcs_len = self.suffixes.lcs().char_len();
        lcs_len >= 3 && lcs_len > self.dfa.prefixes.lcp().char_len()
    }

    fn new_pool(ro: &Arc<ExecReadOnly>) -> Box<Pool<ProgramCache>> {
        let ro = ro.clone();
        Box::new(Pool::new(Box::new(move || {
            AssertUnwindSafe(RefCell::new(ProgramCacheInner::new(&ro)))
        })))
    }
}

#[derive(Clone, Copy, Debug)]
enum MatchType {
    /// A single or multiple literal search. This is only used when the regex
    /// can be decomposed into a literal search.
    #[cfg(feature = "perf-literal")]
    Literal(MatchLiteralType),
    /// A normal DFA search.
    #[cfg(feature = "perf-dfa")]
    Dfa,
    /// A reverse DFA search starting from the end of a haystack.
    #[cfg(feature = "perf-dfa")]
    DfaAnchoredReverse,
    /// A reverse DFA search with suffix literal scanning.
    #[cfg(all(feature = "perf-dfa", feature = "perf-literal"))]
    DfaSuffix,
    /// Use the DFA on two or more regular expressions.
    #[cfg(feature = "perf-dfa")]
    DfaMany,
    /// An NFA variant.
    Nfa(MatchNfaType),
    /// No match is ever possible, so don't ever try to search.
    Nothing,
}

#[derive(Clone, Copy, Debug)]
#[cfg(feature = "perf-literal")]
enum MatchLiteralType {
    /// Match literals anywhere in text.
    Unanchored,
    /// Match literals only at the start of text.
    AnchoredStart,
    /// Match literals only at the end of text.
    AnchoredEnd,
    /// Use an Aho-Corasick automaton. This requires `ac` to be Some on
    /// ExecReadOnly.
    AhoCorasick,
}

#[derive(Clone, Copy, Debug, Eq, PartialEq)]
enum MatchNfaType {
    /// Choose between Backtrack and PikeVM.
    Auto,
    /// NFA bounded backtracking.
    ///
    /// (This is only set by tests, since it never makes sense to always want
    /// backtracking.)
    Backtrack,
    /// The Pike VM.
    ///
    /// (This is only set by tests, since it never makes sense to always want
    /// the Pike VM.)
    PikeVM,
}

/// `ProgramCache` maintains reusable allocations for each matching engine
/// available to a particular program.
///
/// We declare this as unwind safe since it's a cache that's only used for
/// performance purposes. If a panic occurs, it is (or should be) always safe
/// to continue using the same regex object.
pub type ProgramCache = AssertUnwindSafe<RefCell<ProgramCacheInner>>;

#[derive(Debug)]
pub struct ProgramCacheInner {
    pub pikevm: pikevm::Cache,
    pub backtrack: backtrack::Cache,
    #[cfg(feature = "perf-dfa")]
    pub dfa: dfa::Cache,
    #[cfg(feature = "perf-dfa")]
    pub dfa_reverse: dfa::Cache,
}

impl ProgramCacheInner {
    fn new(ro: &ExecReadOnly) -> Self {
        ProgramCacheInner {
            pikevm: pikevm::Cache::new(&ro.nfa),
            backtrack: backtrack::Cache::new(&ro.nfa),
            #[cfg(feature = "perf-dfa")]
            dfa: dfa::Cache::new(&ro.dfa),
            #[cfg(feature = "perf-dfa")]
            dfa_reverse: dfa::Cache::new(&ro.dfa_reverse),
        }
    }
}

/// Alternation literals checks if the given HIR is a simple alternation of
/// literals, and if so, returns them. Otherwise, this returns None.
#[cfg(feature = "perf-literal")]
fn alternation_literals(expr: &Hir) -> Option<Vec<Vec<u8>>> {
    use syntax::hir::{HirKind, Literal};

    // This is pretty hacky, but basically, if `is_alternation_literal` is
    // true, then we can make several assumptions about the structure of our
    // HIR. This is what justifies the `unreachable!` statements below.
    //
    // This code should be refactored once we overhaul this crate's
    // optimization pipeline, because this is a terribly inflexible way to go
    // about things.

    if !expr.is_alternation_literal() {
        return None;
    }
    let alts = match *expr.kind() {
        HirKind::Alternation(ref alts) => alts,
        _ => return None, // one literal isn't worth it
    };

    let extendlit = |lit: &Literal, dst: &mut Vec<u8>| match *lit {
        Literal::Unicode(c) => {
            let mut buf = [0; 4];
            dst.extend_from_slice(c.encode_utf8(&mut buf).as_bytes());
        }
        Literal::Byte(b) => {
            dst.push(b);
        }
    };

    let mut lits = vec![];
    for alt in alts {
        let mut lit = vec![];
        match *alt.kind() {
            HirKind::Literal(ref x) => extendlit(x, &mut lit),
            HirKind::Concat(ref exprs) => {
                for e in exprs {
                    match *e.kind() {
                        HirKind::Literal(ref x) => extendlit(x, &mut lit),
                        _ => unreachable!("expected literal, got {:?}", e),
                    }
                }
            }
            _ => unreachable!("expected literal or concat, got {:?}", alt),
        }
        lits.push(lit);
    }
    Some(lits)
}

#[cfg(test)]
mod test {
    #[test]
    fn uppercut_s_backtracking_bytes_default_bytes_mismatch() {
        use internal::ExecBuilder;

        let backtrack_bytes_re = ExecBuilder::new("^S")
            .bounded_backtracking()
            .only_utf8(false)
            .build()
            .map(|exec| exec.into_byte_regex())
            .map_err(|err| format!("{}", err))
            .unwrap();

        let default_bytes_re = ExecBuilder::new("^S")
            .only_utf8(false)
            .build()
            .map(|exec| exec.into_byte_regex())
            .map_err(|err| format!("{}", err))
            .unwrap();

        let input = vec![83, 83];

        let s1 = backtrack_bytes_re.split(&input);
        let s2 = default_bytes_re.split(&input);
        for (chunk1, chunk2) in s1.zip(s2) {
            assert_eq!(chunk1, chunk2);
        }
    }

    #[test]
    fn unicode_lit_star_backtracking_utf8bytes_default_utf8bytes_mismatch() {
        use internal::ExecBuilder;

        let backtrack_bytes_re = ExecBuilder::new(r"^(?u:\*)")
            .bounded_backtracking()
            .bytes(true)
            .build()
            .map(|exec| exec.into_regex())
            .map_err(|err| format!("{}", err))
            .unwrap();

        let default_bytes_re = ExecBuilder::new(r"^(?u:\*)")
            .bytes(true)
            .build()
            .map(|exec| exec.into_regex())
            .map_err(|err| format!("{}", err))
            .unwrap();

        let input = "**";

        let s1 = backtrack_bytes_re.split(input);
        let s2 = default_bytes_re.split(input);
        for (chunk1, chunk2) in s1.zip(s2) {
            assert_eq!(chunk1, chunk2);
        }
    }
}