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
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
2077
2078
2079
2080
2081
2082
2083
2084
2085
2086
2087
2088
2089
2090
2091
2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
2110
2111
2112
2113
2114
2115
2116
//! Source positions and related helper functions.
//!
//! Important concepts in this module include:
//!
//! - the *span*, represented by [`SpanData`] and related types;
//! - source code as represented by a [`SourceMap`]; and
//! - interned strings, represented by [`Symbol`]s, with some common symbols available statically in the [`sym`] module.
//!
//! Unlike most compilers, the span contains not only the position in the source code, but also various other metadata,
//! such as the edition and macro hygiene. This metadata is stored in [`SyntaxContext`] and [`ExpnData`].
//!
//! ## Note
//!
//! This API is completely unstable and subject to change.

#![doc(html_root_url = "https://doc.rust-lang.org/nightly/nightly-rustc/")]
#![feature(array_windows)]
#![feature(let_else)]
#![feature(if_let_guard)]
#![feature(negative_impls)]
#![feature(min_specialization)]
#![feature(rustc_attrs)]
#![allow(rustc::potential_query_instability)]

#[macro_use]
extern crate rustc_macros;

#[macro_use]
extern crate tracing;

use rustc_data_structures::AtomicRef;
use rustc_macros::HashStable_Generic;
use rustc_serialize::{Decodable, Decoder, Encodable, Encoder};

mod caching_source_map_view;
pub mod source_map;
pub use self::caching_source_map_view::CachingSourceMapView;
use source_map::SourceMap;

pub mod edition;
use edition::Edition;
pub mod hygiene;
use hygiene::Transparency;
pub use hygiene::{DesugaringKind, ExpnKind, MacroKind};
pub use hygiene::{ExpnData, ExpnHash, ExpnId, LocalExpnId, SyntaxContext};
use rustc_data_structures::stable_hasher::HashingControls;
pub mod def_id;
use def_id::{CrateNum, DefId, DefPathHash, LocalDefId, LOCAL_CRATE};
pub mod lev_distance;
mod span_encoding;
pub use span_encoding::{Span, DUMMY_SP};

pub mod symbol;
pub use symbol::{sym, Symbol};

mod analyze_source_file;
pub mod fatal_error;

pub mod profiling;

use rustc_data_structures::stable_hasher::{HashStable, StableHasher};
use rustc_data_structures::sync::{Lock, Lrc};

use std::borrow::Cow;
use std::cmp::{self, Ordering};
use std::fmt;
use std::hash::Hash;
use std::ops::{Add, Range, Sub};
use std::path::{Path, PathBuf};
use std::str::FromStr;
use std::sync::Arc;

use md5::Digest;
use md5::Md5;
use sha1::Sha1;
use sha2::Sha256;

use tracing::debug;

#[cfg(test)]
mod tests;

// Per-session global variables: this struct is stored in thread-local storage
// in such a way that it is accessible without any kind of handle to all
// threads within the compilation session, but is not accessible outside the
// session.
pub struct SessionGlobals {
    symbol_interner: symbol::Interner,
    span_interner: Lock<span_encoding::SpanInterner>,
    hygiene_data: Lock<hygiene::HygieneData>,
    source_map: Lock<Option<Lrc<SourceMap>>>,
}

impl SessionGlobals {
    pub fn new(edition: Edition) -> SessionGlobals {
        SessionGlobals {
            symbol_interner: symbol::Interner::fresh(),
            span_interner: Lock::new(span_encoding::SpanInterner::default()),
            hygiene_data: Lock::new(hygiene::HygieneData::new(edition)),
            source_map: Lock::new(None),
        }
    }
}

#[inline]
pub fn create_session_globals_then<R>(edition: Edition, f: impl FnOnce() -> R) -> R {
    assert!(
        !SESSION_GLOBALS.is_set(),
        "SESSION_GLOBALS should never be overwritten! \
         Use another thread if you need another SessionGlobals"
    );
    let session_globals = SessionGlobals::new(edition);
    SESSION_GLOBALS.set(&session_globals, f)
}

#[inline]
pub fn set_session_globals_then<R>(session_globals: &SessionGlobals, f: impl FnOnce() -> R) -> R {
    assert!(
        !SESSION_GLOBALS.is_set(),
        "SESSION_GLOBALS should never be overwritten! \
         Use another thread if you need another SessionGlobals"
    );
    SESSION_GLOBALS.set(session_globals, f)
}

#[inline]
pub fn create_default_session_if_not_set_then<R, F>(f: F) -> R
where
    F: FnOnce(&SessionGlobals) -> R,
{
    create_session_if_not_set_then(edition::DEFAULT_EDITION, f)
}

#[inline]
pub fn create_session_if_not_set_then<R, F>(edition: Edition, f: F) -> R
where
    F: FnOnce(&SessionGlobals) -> R,
{
    if !SESSION_GLOBALS.is_set() {
        let session_globals = SessionGlobals::new(edition);
        SESSION_GLOBALS.set(&session_globals, || SESSION_GLOBALS.with(f))
    } else {
        SESSION_GLOBALS.with(f)
    }
}

#[inline]
pub fn with_session_globals<R, F>(f: F) -> R
where
    F: FnOnce(&SessionGlobals) -> R,
{
    SESSION_GLOBALS.with(f)
}

#[inline]
pub fn create_default_session_globals_then<R>(f: impl FnOnce() -> R) -> R {
    create_session_globals_then(edition::DEFAULT_EDITION, f)
}

// If this ever becomes non thread-local, `decode_syntax_context`
// and `decode_expn_id` will need to be updated to handle concurrent
// deserialization.
scoped_tls::scoped_thread_local!(static SESSION_GLOBALS: SessionGlobals);

// FIXME: We should use this enum or something like it to get rid of the
// use of magic `/rust/1.x/...` paths across the board.
#[derive(Debug, Eq, PartialEq, Clone, Ord, PartialOrd)]
#[derive(Decodable)]
pub enum RealFileName {
    LocalPath(PathBuf),
    /// For remapped paths (namely paths into libstd that have been mapped
    /// to the appropriate spot on the local host's file system, and local file
    /// system paths that have been remapped with `FilePathMapping`),
    Remapped {
        /// `local_path` is the (host-dependent) local path to the file. This is
        /// None if the file was imported from another crate
        local_path: Option<PathBuf>,
        /// `virtual_name` is the stable path rustc will store internally within
        /// build artifacts.
        virtual_name: PathBuf,
    },
}

impl Hash for RealFileName {
    fn hash<H: std::hash::Hasher>(&self, state: &mut H) {
        // To prevent #70924 from happening again we should only hash the
        // remapped (virtualized) path if that exists. This is because
        // virtualized paths to sysroot crates (/rust/$hash or /rust/$version)
        // remain stable even if the corresponding local_path changes
        self.remapped_path_if_available().hash(state)
    }
}

// This is functionally identical to #[derive(Encodable)], with the exception of
// an added assert statement
impl<S: Encoder> Encodable<S> for RealFileName {
    fn encode(&self, encoder: &mut S) {
        match *self {
            RealFileName::LocalPath(ref local_path) => encoder.emit_enum_variant(0, |encoder| {
                local_path.encode(encoder);
            }),

            RealFileName::Remapped { ref local_path, ref virtual_name } => encoder
                .emit_enum_variant(1, |encoder| {
                    // For privacy and build reproducibility, we must not embed host-dependant path in artifacts
                    // if they have been remapped by --remap-path-prefix
                    assert!(local_path.is_none());
                    local_path.encode(encoder);
                    virtual_name.encode(encoder);
                }),
        }
    }
}

impl RealFileName {
    /// Returns the path suitable for reading from the file system on the local host,
    /// if this information exists.
    /// Avoid embedding this in build artifacts; see `remapped_path_if_available()` for that.
    pub fn local_path(&self) -> Option<&Path> {
        match self {
            RealFileName::LocalPath(p) => Some(p),
            RealFileName::Remapped { local_path: p, virtual_name: _ } => {
                p.as_ref().map(PathBuf::as_path)
            }
        }
    }

    /// Returns the path suitable for reading from the file system on the local host,
    /// if this information exists.
    /// Avoid embedding this in build artifacts; see `remapped_path_if_available()` for that.
    pub fn into_local_path(self) -> Option<PathBuf> {
        match self {
            RealFileName::LocalPath(p) => Some(p),
            RealFileName::Remapped { local_path: p, virtual_name: _ } => p,
        }
    }

    /// Returns the path suitable for embedding into build artifacts. This would still
    /// be a local path if it has not been remapped. A remapped path will not correspond
    /// to a valid file system path: see `local_path_if_available()` for something that
    /// is more likely to return paths into the local host file system.
    pub fn remapped_path_if_available(&self) -> &Path {
        match self {
            RealFileName::LocalPath(p)
            | RealFileName::Remapped { local_path: _, virtual_name: p } => &p,
        }
    }

    /// Returns the path suitable for reading from the file system on the local host,
    /// if this information exists. Otherwise returns the remapped name.
    /// Avoid embedding this in build artifacts; see `remapped_path_if_available()` for that.
    pub fn local_path_if_available(&self) -> &Path {
        match self {
            RealFileName::LocalPath(path)
            | RealFileName::Remapped { local_path: None, virtual_name: path }
            | RealFileName::Remapped { local_path: Some(path), virtual_name: _ } => path,
        }
    }

    pub fn to_string_lossy(&self, display_pref: FileNameDisplayPreference) -> Cow<'_, str> {
        match display_pref {
            FileNameDisplayPreference::Local => self.local_path_if_available().to_string_lossy(),
            FileNameDisplayPreference::Remapped => {
                self.remapped_path_if_available().to_string_lossy()
            }
        }
    }
}

/// Differentiates between real files and common virtual files.
#[derive(Debug, Eq, PartialEq, Clone, Ord, PartialOrd, Hash)]
#[derive(Decodable, Encodable)]
pub enum FileName {
    Real(RealFileName),
    /// Call to `quote!`.
    QuoteExpansion(u64),
    /// Command line.
    Anon(u64),
    /// Hack in `src/librustc_ast/parse.rs`.
    // FIXME(jseyfried)
    MacroExpansion(u64),
    ProcMacroSourceCode(u64),
    /// Strings provided as `--cfg [cfgspec]` stored in a `crate_cfg`.
    CfgSpec(u64),
    /// Strings provided as crate attributes in the CLI.
    CliCrateAttr(u64),
    /// Custom sources for explicit parser calls from plugins and drivers.
    Custom(String),
    DocTest(PathBuf, isize),
    /// Post-substitution inline assembly from LLVM.
    InlineAsm(u64),
}

impl From<PathBuf> for FileName {
    fn from(p: PathBuf) -> Self {
        assert!(!p.to_string_lossy().ends_with('>'));
        FileName::Real(RealFileName::LocalPath(p))
    }
}

#[derive(Clone, Copy, Eq, PartialEq, Hash, Debug)]
pub enum FileNameDisplayPreference {
    Remapped,
    Local,
}

pub struct FileNameDisplay<'a> {
    inner: &'a FileName,
    display_pref: FileNameDisplayPreference,
}

impl fmt::Display for FileNameDisplay<'_> {
    fn fmt(&self, fmt: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
        use FileName::*;
        match *self.inner {
            Real(ref name) => {
                write!(fmt, "{}", name.to_string_lossy(self.display_pref))
            }
            QuoteExpansion(_) => write!(fmt, "<quote expansion>"),
            MacroExpansion(_) => write!(fmt, "<macro expansion>"),
            Anon(_) => write!(fmt, "<anon>"),
            ProcMacroSourceCode(_) => write!(fmt, "<proc-macro source code>"),
            CfgSpec(_) => write!(fmt, "<cfgspec>"),
            CliCrateAttr(_) => write!(fmt, "<crate attribute>"),
            Custom(ref s) => write!(fmt, "<{}>", s),
            DocTest(ref path, _) => write!(fmt, "{}", path.display()),
            InlineAsm(_) => write!(fmt, "<inline asm>"),
        }
    }
}

impl<'a> FileNameDisplay<'a> {
    pub fn to_string_lossy(&self) -> Cow<'a, str> {
        match self.inner {
            FileName::Real(ref inner) => inner.to_string_lossy(self.display_pref),
            _ => Cow::from(self.to_string()),
        }
    }
}

impl FileName {
    pub fn is_real(&self) -> bool {
        use FileName::*;
        match *self {
            Real(_) => true,
            Anon(_)
            | MacroExpansion(_)
            | ProcMacroSourceCode(_)
            | CfgSpec(_)
            | CliCrateAttr(_)
            | Custom(_)
            | QuoteExpansion(_)
            | DocTest(_, _)
            | InlineAsm(_) => false,
        }
    }

    pub fn prefer_remapped(&self) -> FileNameDisplay<'_> {
        FileNameDisplay { inner: self, display_pref: FileNameDisplayPreference::Remapped }
    }

    // This may include transient local filesystem information.
    // Must not be embedded in build outputs.
    pub fn prefer_local(&self) -> FileNameDisplay<'_> {
        FileNameDisplay { inner: self, display_pref: FileNameDisplayPreference::Local }
    }

    pub fn display(&self, display_pref: FileNameDisplayPreference) -> FileNameDisplay<'_> {
        FileNameDisplay { inner: self, display_pref }
    }

    pub fn macro_expansion_source_code(src: &str) -> FileName {
        let mut hasher = StableHasher::new();
        src.hash(&mut hasher);
        FileName::MacroExpansion(hasher.finish())
    }

    pub fn anon_source_code(src: &str) -> FileName {
        let mut hasher = StableHasher::new();
        src.hash(&mut hasher);
        FileName::Anon(hasher.finish())
    }

    pub fn proc_macro_source_code(src: &str) -> FileName {
        let mut hasher = StableHasher::new();
        src.hash(&mut hasher);
        FileName::ProcMacroSourceCode(hasher.finish())
    }

    pub fn cfg_spec_source_code(src: &str) -> FileName {
        let mut hasher = StableHasher::new();
        src.hash(&mut hasher);
        FileName::QuoteExpansion(hasher.finish())
    }

    pub fn cli_crate_attr_source_code(src: &str) -> FileName {
        let mut hasher = StableHasher::new();
        src.hash(&mut hasher);
        FileName::CliCrateAttr(hasher.finish())
    }

    pub fn doc_test_source_code(path: PathBuf, line: isize) -> FileName {
        FileName::DocTest(path, line)
    }

    pub fn inline_asm_source_code(src: &str) -> FileName {
        let mut hasher = StableHasher::new();
        src.hash(&mut hasher);
        FileName::InlineAsm(hasher.finish())
    }
}

/// Represents a span.
///
/// Spans represent a region of code, used for error reporting. Positions in spans
/// are *absolute* positions from the beginning of the [`SourceMap`], not positions
/// relative to [`SourceFile`]s. Methods on the `SourceMap` can be used to relate spans back
/// to the original source.
///
/// You must be careful if the span crosses more than one file, since you will not be
/// able to use many of the functions on spans in source_map and you cannot assume
/// that the length of the span is equal to `span.hi - span.lo`; there may be space in the
/// [`BytePos`] range between files.
///
/// `SpanData` is public because `Span` uses a thread-local interner and can't be
/// sent to other threads, but some pieces of performance infra run in a separate thread.
/// Using `Span` is generally preferred.
#[derive(Clone, Copy, Hash, PartialEq, Eq)]
pub struct SpanData {
    pub lo: BytePos,
    pub hi: BytePos,
    /// Information about where the macro came from, if this piece of
    /// code was created by a macro expansion.
    pub ctxt: SyntaxContext,
    pub parent: Option<LocalDefId>,
}

// Order spans by position in the file.
impl Ord for SpanData {
    fn cmp(&self, other: &Self) -> Ordering {
        let SpanData {
            lo: s_lo,
            hi: s_hi,
            ctxt: s_ctxt,
            // `LocalDefId` does not implement `Ord`.
            // The other fields are enough to determine in-file order.
            parent: _,
        } = self;
        let SpanData {
            lo: o_lo,
            hi: o_hi,
            ctxt: o_ctxt,
            // `LocalDefId` does not implement `Ord`.
            // The other fields are enough to determine in-file order.
            parent: _,
        } = other;

        (s_lo, s_hi, s_ctxt).cmp(&(o_lo, o_hi, o_ctxt))
    }
}

impl PartialOrd for SpanData {
    fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
        Some(self.cmp(other))
    }
}

impl SpanData {
    #[inline]
    pub fn span(&self) -> Span {
        Span::new(self.lo, self.hi, self.ctxt, self.parent)
    }
    #[inline]
    pub fn with_lo(&self, lo: BytePos) -> Span {
        Span::new(lo, self.hi, self.ctxt, self.parent)
    }
    #[inline]
    pub fn with_hi(&self, hi: BytePos) -> Span {
        Span::new(self.lo, hi, self.ctxt, self.parent)
    }
    #[inline]
    pub fn with_ctxt(&self, ctxt: SyntaxContext) -> Span {
        Span::new(self.lo, self.hi, ctxt, self.parent)
    }
    #[inline]
    pub fn with_parent(&self, parent: Option<LocalDefId>) -> Span {
        Span::new(self.lo, self.hi, self.ctxt, parent)
    }
    /// Returns `true` if this is a dummy span with any hygienic context.
    #[inline]
    pub fn is_dummy(self) -> bool {
        self.lo.0 == 0 && self.hi.0 == 0
    }
    /// Returns `true` if `self` fully encloses `other`.
    pub fn contains(self, other: Self) -> bool {
        self.lo <= other.lo && other.hi <= self.hi
    }
}

// The interner is pointed to by a thread local value which is only set on the main thread
// with parallelization is disabled. So we don't allow `Span` to transfer between threads
// to avoid panics and other errors, even though it would be memory safe to do so.
#[cfg(not(parallel_compiler))]
impl !Send for Span {}
#[cfg(not(parallel_compiler))]
impl !Sync for Span {}

impl PartialOrd for Span {
    fn partial_cmp(&self, rhs: &Self) -> Option<Ordering> {
        PartialOrd::partial_cmp(&self.data(), &rhs.data())
    }
}
impl Ord for Span {
    fn cmp(&self, rhs: &Self) -> Ordering {
        Ord::cmp(&self.data(), &rhs.data())
    }
}

impl Span {
    #[inline]
    pub fn lo(self) -> BytePos {
        self.data().lo
    }
    #[inline]
    pub fn with_lo(self, lo: BytePos) -> Span {
        self.data().with_lo(lo)
    }
    #[inline]
    pub fn hi(self) -> BytePos {
        self.data().hi
    }
    #[inline]
    pub fn with_hi(self, hi: BytePos) -> Span {
        self.data().with_hi(hi)
    }
    #[inline]
    pub fn ctxt(self) -> SyntaxContext {
        self.data_untracked().ctxt
    }
    pub fn eq_ctxt(self, other: Span) -> bool {
        self.data_untracked().ctxt == other.data_untracked().ctxt
    }
    #[inline]
    pub fn with_ctxt(self, ctxt: SyntaxContext) -> Span {
        self.data_untracked().with_ctxt(ctxt)
    }
    #[inline]
    pub fn parent(self) -> Option<LocalDefId> {
        self.data().parent
    }
    #[inline]
    pub fn with_parent(self, ctxt: Option<LocalDefId>) -> Span {
        self.data().with_parent(ctxt)
    }

    /// Returns `true` if this is a dummy span with any hygienic context.
    #[inline]
    pub fn is_dummy(self) -> bool {
        self.data_untracked().is_dummy()
    }

    /// Returns `true` if this span comes from a macro or desugaring.
    #[inline]
    pub fn from_expansion(self) -> bool {
        self.ctxt() != SyntaxContext::root()
    }

    /// Returns `true` if `span` originates in a derive-macro's expansion.
    pub fn in_derive_expansion(self) -> bool {
        matches!(self.ctxt().outer_expn_data().kind, ExpnKind::Macro(MacroKind::Derive, _))
    }

    /// Gate suggestions that would not be appropriate in a context the user didn't write.
    pub fn can_be_used_for_suggestions(self) -> bool {
        !self.from_expansion()
        // FIXME: If this span comes from a `derive` macro but it points at code the user wrote,
        // the callsite span and the span will be pointing at different places. It also means that
        // we can safely provide suggestions on this span.
            || (matches!(self.ctxt().outer_expn_data().kind, ExpnKind::Macro(MacroKind::Derive, _))
                && self.parent_callsite().map(|p| (p.lo(), p.hi())) != Some((self.lo(), self.hi())))
    }

    #[inline]
    pub fn with_root_ctxt(lo: BytePos, hi: BytePos) -> Span {
        Span::new(lo, hi, SyntaxContext::root(), None)
    }

    /// Returns a new span representing an empty span at the beginning of this span.
    #[inline]
    pub fn shrink_to_lo(self) -> Span {
        let span = self.data_untracked();
        span.with_hi(span.lo)
    }
    /// Returns a new span representing an empty span at the end of this span.
    #[inline]
    pub fn shrink_to_hi(self) -> Span {
        let span = self.data_untracked();
        span.with_lo(span.hi)
    }

    #[inline]
    /// Returns `true` if `hi == lo`.
    pub fn is_empty(self) -> bool {
        let span = self.data_untracked();
        span.hi == span.lo
    }

    /// Returns `self` if `self` is not the dummy span, and `other` otherwise.
    pub fn substitute_dummy(self, other: Span) -> Span {
        if self.is_dummy() { other } else { self }
    }

    /// Returns `true` if `self` fully encloses `other`.
    pub fn contains(self, other: Span) -> bool {
        let span = self.data();
        let other = other.data();
        span.contains(other)
    }

    /// Returns `true` if `self` touches `other`.
    pub fn overlaps(self, other: Span) -> bool {
        let span = self.data();
        let other = other.data();
        span.lo < other.hi && other.lo < span.hi
    }

    /// Returns `true` if the spans are equal with regards to the source text.
    ///
    /// Use this instead of `==` when either span could be generated code,
    /// and you only care that they point to the same bytes of source text.
    pub fn source_equal(self, other: Span) -> bool {
        let span = self.data();
        let other = other.data();
        span.lo == other.lo && span.hi == other.hi
    }

    /// Returns `Some(span)`, where the start is trimmed by the end of `other`.
    pub fn trim_start(self, other: Span) -> Option<Span> {
        let span = self.data();
        let other = other.data();
        if span.hi > other.hi { Some(span.with_lo(cmp::max(span.lo, other.hi))) } else { None }
    }

    /// Returns the source span -- this is either the supplied span, or the span for
    /// the macro callsite that expanded to it.
    pub fn source_callsite(self) -> Span {
        let expn_data = self.ctxt().outer_expn_data();
        if !expn_data.is_root() { expn_data.call_site.source_callsite() } else { self }
    }

    /// The `Span` for the tokens in the previous macro expansion from which `self` was generated,
    /// if any.
    pub fn parent_callsite(self) -> Option<Span> {
        let expn_data = self.ctxt().outer_expn_data();
        if !expn_data.is_root() { Some(expn_data.call_site) } else { None }
    }

    /// Walk down the expansion ancestors to find a span that's contained within `outer`.
    pub fn find_ancestor_inside(mut self, outer: Span) -> Option<Span> {
        while !outer.contains(self) {
            self = self.parent_callsite()?;
        }
        Some(self)
    }

    /// Edition of the crate from which this span came.
    pub fn edition(self) -> edition::Edition {
        self.ctxt().edition()
    }

    #[inline]
    pub fn rust_2015(self) -> bool {
        self.edition() == edition::Edition::Edition2015
    }

    #[inline]
    pub fn rust_2018(self) -> bool {
        self.edition() >= edition::Edition::Edition2018
    }

    #[inline]
    pub fn rust_2021(self) -> bool {
        self.edition() >= edition::Edition::Edition2021
    }

    #[inline]
    pub fn rust_2024(self) -> bool {
        self.edition() >= edition::Edition::Edition2024
    }

    /// Returns the source callee.
    ///
    /// Returns `None` if the supplied span has no expansion trace,
    /// else returns the `ExpnData` for the macro definition
    /// corresponding to the source callsite.
    pub fn source_callee(self) -> Option<ExpnData> {
        fn source_callee(expn_data: ExpnData) -> ExpnData {
            let next_expn_data = expn_data.call_site.ctxt().outer_expn_data();
            if !next_expn_data.is_root() { source_callee(next_expn_data) } else { expn_data }
        }
        let expn_data = self.ctxt().outer_expn_data();
        if !expn_data.is_root() { Some(source_callee(expn_data)) } else { None }
    }

    /// Checks if a span is "internal" to a macro in which `#[unstable]`
    /// items can be used (that is, a macro marked with
    /// `#[allow_internal_unstable]`).
    pub fn allows_unstable(self, feature: Symbol) -> bool {
        self.ctxt()
            .outer_expn_data()
            .allow_internal_unstable
            .map_or(false, |features| features.iter().any(|&f| f == feature))
    }

    /// Checks if this span arises from a compiler desugaring of kind `kind`.
    pub fn is_desugaring(self, kind: DesugaringKind) -> bool {
        match self.ctxt().outer_expn_data().kind {
            ExpnKind::Desugaring(k) => k == kind,
            _ => false,
        }
    }

    /// Returns the compiler desugaring that created this span, or `None`
    /// if this span is not from a desugaring.
    pub fn desugaring_kind(self) -> Option<DesugaringKind> {
        match self.ctxt().outer_expn_data().kind {
            ExpnKind::Desugaring(k) => Some(k),
            _ => None,
        }
    }

    /// Checks if a span is "internal" to a macro in which `unsafe`
    /// can be used without triggering the `unsafe_code` lint.
    //  (that is, a macro marked with `#[allow_internal_unsafe]`).
    pub fn allows_unsafe(self) -> bool {
        self.ctxt().outer_expn_data().allow_internal_unsafe
    }

    pub fn macro_backtrace(mut self) -> impl Iterator<Item = ExpnData> {
        let mut prev_span = DUMMY_SP;
        std::iter::from_fn(move || {
            loop {
                let expn_data = self.ctxt().outer_expn_data();
                if expn_data.is_root() {
                    return None;
                }

                let is_recursive = expn_data.call_site.source_equal(prev_span);

                prev_span = self;
                self = expn_data.call_site;

                // Don't print recursive invocations.
                if !is_recursive {
                    return Some(expn_data);
                }
            }
        })
    }

    /// Returns a `Span` that would enclose both `self` and `end`.
    ///
    /// ```text
    ///     ____             ___
    ///     self lorem ipsum end
    ///     ^^^^^^^^^^^^^^^^^^^^
    /// ```
    pub fn to(self, end: Span) -> Span {
        let span_data = self.data();
        let end_data = end.data();
        // FIXME(jseyfried): `self.ctxt` should always equal `end.ctxt` here (cf. issue #23480).
        // Return the macro span on its own to avoid weird diagnostic output. It is preferable to
        // have an incomplete span than a completely nonsensical one.
        if span_data.ctxt != end_data.ctxt {
            if span_data.ctxt == SyntaxContext::root() {
                return end;
            } else if end_data.ctxt == SyntaxContext::root() {
                return self;
            }
            // Both spans fall within a macro.
            // FIXME(estebank): check if it is the *same* macro.
        }
        Span::new(
            cmp::min(span_data.lo, end_data.lo),
            cmp::max(span_data.hi, end_data.hi),
            if span_data.ctxt == SyntaxContext::root() { end_data.ctxt } else { span_data.ctxt },
            if span_data.parent == end_data.parent { span_data.parent } else { None },
        )
    }

    /// Returns a `Span` between the end of `self` to the beginning of `end`.
    ///
    /// ```text
    ///     ____             ___
    ///     self lorem ipsum end
    ///         ^^^^^^^^^^^^^
    /// ```
    pub fn between(self, end: Span) -> Span {
        let span = self.data();
        let end = end.data();
        Span::new(
            span.hi,
            end.lo,
            if end.ctxt == SyntaxContext::root() { end.ctxt } else { span.ctxt },
            if span.parent == end.parent { span.parent } else { None },
        )
    }

    /// Returns a `Span` from the beginning of `self` until the beginning of `end`.
    ///
    /// ```text
    ///     ____             ___
    ///     self lorem ipsum end
    ///     ^^^^^^^^^^^^^^^^^
    /// ```
    pub fn until(self, end: Span) -> Span {
        // Most of this function's body is copied from `to`.
        // We can't just do `self.to(end.shrink_to_lo())`,
        // because to also does some magic where it uses min/max so
        // it can handle overlapping spans. Some advanced mis-use of
        // `until` with different ctxts makes this visible.
        let span_data = self.data();
        let end_data = end.data();
        // FIXME(jseyfried): `self.ctxt` should always equal `end.ctxt` here (cf. issue #23480).
        // Return the macro span on its own to avoid weird diagnostic output. It is preferable to
        // have an incomplete span than a completely nonsensical one.
        if span_data.ctxt != end_data.ctxt {
            if span_data.ctxt == SyntaxContext::root() {
                return end;
            } else if end_data.ctxt == SyntaxContext::root() {
                return self;
            }
            // Both spans fall within a macro.
            // FIXME(estebank): check if it is the *same* macro.
        }
        Span::new(
            span_data.lo,
            end_data.lo,
            if end_data.ctxt == SyntaxContext::root() { end_data.ctxt } else { span_data.ctxt },
            if span_data.parent == end_data.parent { span_data.parent } else { None },
        )
    }

    pub fn from_inner(self, inner: InnerSpan) -> Span {
        let span = self.data();
        Span::new(
            span.lo + BytePos::from_usize(inner.start),
            span.lo + BytePos::from_usize(inner.end),
            span.ctxt,
            span.parent,
        )
    }

    /// Equivalent of `Span::def_site` from the proc macro API,
    /// except that the location is taken from the `self` span.
    pub fn with_def_site_ctxt(self, expn_id: ExpnId) -> Span {
        self.with_ctxt_from_mark(expn_id, Transparency::Opaque)
    }

    /// Equivalent of `Span::call_site` from the proc macro API,
    /// except that the location is taken from the `self` span.
    pub fn with_call_site_ctxt(self, expn_id: ExpnId) -> Span {
        self.with_ctxt_from_mark(expn_id, Transparency::Transparent)
    }

    /// Equivalent of `Span::mixed_site` from the proc macro API,
    /// except that the location is taken from the `self` span.
    pub fn with_mixed_site_ctxt(self, expn_id: ExpnId) -> Span {
        self.with_ctxt_from_mark(expn_id, Transparency::SemiTransparent)
    }

    /// Produces a span with the same location as `self` and context produced by a macro with the
    /// given ID and transparency, assuming that macro was defined directly and not produced by
    /// some other macro (which is the case for built-in and procedural macros).
    pub fn with_ctxt_from_mark(self, expn_id: ExpnId, transparency: Transparency) -> Span {
        self.with_ctxt(SyntaxContext::root().apply_mark(expn_id, transparency))
    }

    #[inline]
    pub fn apply_mark(self, expn_id: ExpnId, transparency: Transparency) -> Span {
        let span = self.data();
        span.with_ctxt(span.ctxt.apply_mark(expn_id, transparency))
    }

    #[inline]
    pub fn remove_mark(&mut self) -> ExpnId {
        let mut span = self.data();
        let mark = span.ctxt.remove_mark();
        *self = Span::new(span.lo, span.hi, span.ctxt, span.parent);
        mark
    }

    #[inline]
    pub fn adjust(&mut self, expn_id: ExpnId) -> Option<ExpnId> {
        let mut span = self.data();
        let mark = span.ctxt.adjust(expn_id);
        *self = Span::new(span.lo, span.hi, span.ctxt, span.parent);
        mark
    }

    #[inline]
    pub fn normalize_to_macros_2_0_and_adjust(&mut self, expn_id: ExpnId) -> Option<ExpnId> {
        let mut span = self.data();
        let mark = span.ctxt.normalize_to_macros_2_0_and_adjust(expn_id);
        *self = Span::new(span.lo, span.hi, span.ctxt, span.parent);
        mark
    }

    #[inline]
    pub fn glob_adjust(&mut self, expn_id: ExpnId, glob_span: Span) -> Option<Option<ExpnId>> {
        let mut span = self.data();
        let mark = span.ctxt.glob_adjust(expn_id, glob_span);
        *self = Span::new(span.lo, span.hi, span.ctxt, span.parent);
        mark
    }

    #[inline]
    pub fn reverse_glob_adjust(
        &mut self,
        expn_id: ExpnId,
        glob_span: Span,
    ) -> Option<Option<ExpnId>> {
        let mut span = self.data();
        let mark = span.ctxt.reverse_glob_adjust(expn_id, glob_span);
        *self = Span::new(span.lo, span.hi, span.ctxt, span.parent);
        mark
    }

    #[inline]
    pub fn normalize_to_macros_2_0(self) -> Span {
        let span = self.data();
        span.with_ctxt(span.ctxt.normalize_to_macros_2_0())
    }

    #[inline]
    pub fn normalize_to_macro_rules(self) -> Span {
        let span = self.data();
        span.with_ctxt(span.ctxt.normalize_to_macro_rules())
    }
}

impl Default for Span {
    fn default() -> Self {
        DUMMY_SP
    }
}

impl<E: Encoder> Encodable<E> for Span {
    default fn encode(&self, s: &mut E) {
        let span = self.data();
        span.lo.encode(s);
        span.hi.encode(s);
    }
}
impl<D: Decoder> Decodable<D> for Span {
    default fn decode(s: &mut D) -> Span {
        let lo = Decodable::decode(s);
        let hi = Decodable::decode(s);

        Span::new(lo, hi, SyntaxContext::root(), None)
    }
}

/// Calls the provided closure, using the provided `SourceMap` to format
/// any spans that are debug-printed during the closure's execution.
///
/// Normally, the global `TyCtxt` is used to retrieve the `SourceMap`
/// (see `rustc_interface::callbacks::span_debug1`). However, some parts
/// of the compiler (e.g. `rustc_parse`) may debug-print `Span`s before
/// a `TyCtxt` is available. In this case, we fall back to
/// the `SourceMap` provided to this function. If that is not available,
/// we fall back to printing the raw `Span` field values.
pub fn with_source_map<T, F: FnOnce() -> T>(source_map: Lrc<SourceMap>, f: F) -> T {
    with_session_globals(|session_globals| {
        *session_globals.source_map.borrow_mut() = Some(source_map);
    });
    struct ClearSourceMap;
    impl Drop for ClearSourceMap {
        fn drop(&mut self) {
            with_session_globals(|session_globals| {
                session_globals.source_map.borrow_mut().take();
            });
        }
    }

    let _guard = ClearSourceMap;
    f()
}

impl fmt::Debug for Span {
    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
        with_session_globals(|session_globals| {
            if let Some(source_map) = &*session_globals.source_map.borrow() {
                write!(f, "{} ({:?})", source_map.span_to_diagnostic_string(*self), self.ctxt())
            } else {
                f.debug_struct("Span")
                    .field("lo", &self.lo())
                    .field("hi", &self.hi())
                    .field("ctxt", &self.ctxt())
                    .finish()
            }
        })
    }
}

impl fmt::Debug for SpanData {
    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
        fmt::Debug::fmt(&Span::new(self.lo, self.hi, self.ctxt, self.parent), f)
    }
}

/// Identifies an offset of a multi-byte character in a `SourceFile`.
#[derive(Copy, Clone, Encodable, Decodable, Eq, PartialEq, Debug)]
pub struct MultiByteChar {
    /// The absolute offset of the character in the `SourceMap`.
    pub pos: BytePos,
    /// The number of bytes, `>= 2`.
    pub bytes: u8,
}

/// Identifies an offset of a non-narrow character in a `SourceFile`.
#[derive(Copy, Clone, Encodable, Decodable, Eq, PartialEq, Debug)]
pub enum NonNarrowChar {
    /// Represents a zero-width character.
    ZeroWidth(BytePos),
    /// Represents a wide (full-width) character.
    Wide(BytePos),
    /// Represents a tab character, represented visually with a width of 4 characters.
    Tab(BytePos),
}

impl NonNarrowChar {
    fn new(pos: BytePos, width: usize) -> Self {
        match width {
            0 => NonNarrowChar::ZeroWidth(pos),
            2 => NonNarrowChar::Wide(pos),
            4 => NonNarrowChar::Tab(pos),
            _ => panic!("width {} given for non-narrow character", width),
        }
    }

    /// Returns the absolute offset of the character in the `SourceMap`.
    pub fn pos(&self) -> BytePos {
        match *self {
            NonNarrowChar::ZeroWidth(p) | NonNarrowChar::Wide(p) | NonNarrowChar::Tab(p) => p,
        }
    }

    /// Returns the width of the character, 0 (zero-width) or 2 (wide).
    pub fn width(&self) -> usize {
        match *self {
            NonNarrowChar::ZeroWidth(_) => 0,
            NonNarrowChar::Wide(_) => 2,
            NonNarrowChar::Tab(_) => 4,
        }
    }
}

impl Add<BytePos> for NonNarrowChar {
    type Output = Self;

    fn add(self, rhs: BytePos) -> Self {
        match self {
            NonNarrowChar::ZeroWidth(pos) => NonNarrowChar::ZeroWidth(pos + rhs),
            NonNarrowChar::Wide(pos) => NonNarrowChar::Wide(pos + rhs),
            NonNarrowChar::Tab(pos) => NonNarrowChar::Tab(pos + rhs),
        }
    }
}

impl Sub<BytePos> for NonNarrowChar {
    type Output = Self;

    fn sub(self, rhs: BytePos) -> Self {
        match self {
            NonNarrowChar::ZeroWidth(pos) => NonNarrowChar::ZeroWidth(pos - rhs),
            NonNarrowChar::Wide(pos) => NonNarrowChar::Wide(pos - rhs),
            NonNarrowChar::Tab(pos) => NonNarrowChar::Tab(pos - rhs),
        }
    }
}

/// Identifies an offset of a character that was normalized away from `SourceFile`.
#[derive(Copy, Clone, Encodable, Decodable, Eq, PartialEq, Debug)]
pub struct NormalizedPos {
    /// The absolute offset of the character in the `SourceMap`.
    pub pos: BytePos,
    /// The difference between original and normalized string at position.
    pub diff: u32,
}

#[derive(PartialEq, Eq, Clone, Debug)]
pub enum ExternalSource {
    /// No external source has to be loaded, since the `SourceFile` represents a local crate.
    Unneeded,
    Foreign {
        kind: ExternalSourceKind,
        /// This SourceFile's byte-offset within the source_map of its original crate.
        original_start_pos: BytePos,
        /// The end of this SourceFile within the source_map of its original crate.
        original_end_pos: BytePos,
    },
}

/// The state of the lazy external source loading mechanism of a `SourceFile`.
#[derive(PartialEq, Eq, Clone, Debug)]
pub enum ExternalSourceKind {
    /// The external source has been loaded already.
    Present(Lrc<String>),
    /// No attempt has been made to load the external source.
    AbsentOk,
    /// A failed attempt has been made to load the external source.
    AbsentErr,
    Unneeded,
}

impl ExternalSource {
    pub fn get_source(&self) -> Option<&Lrc<String>> {
        match self {
            ExternalSource::Foreign { kind: ExternalSourceKind::Present(ref src), .. } => Some(src),
            _ => None,
        }
    }
}

#[derive(Debug)]
pub struct OffsetOverflowError;

#[derive(Copy, Clone, Debug, PartialEq, Eq, PartialOrd, Ord, Hash, Encodable, Decodable)]
#[derive(HashStable_Generic)]
pub enum SourceFileHashAlgorithm {
    Md5,
    Sha1,
    Sha256,
}

impl FromStr for SourceFileHashAlgorithm {
    type Err = ();

    fn from_str(s: &str) -> Result<SourceFileHashAlgorithm, ()> {
        match s {
            "md5" => Ok(SourceFileHashAlgorithm::Md5),
            "sha1" => Ok(SourceFileHashAlgorithm::Sha1),
            "sha256" => Ok(SourceFileHashAlgorithm::Sha256),
            _ => Err(()),
        }
    }
}

/// The hash of the on-disk source file used for debug info.
#[derive(Copy, Clone, PartialEq, Eq, Debug, Hash)]
#[derive(HashStable_Generic, Encodable, Decodable)]
pub struct SourceFileHash {
    pub kind: SourceFileHashAlgorithm,
    value: [u8; 32],
}

impl SourceFileHash {
    pub fn new(kind: SourceFileHashAlgorithm, src: &str) -> SourceFileHash {
        let mut hash = SourceFileHash { kind, value: Default::default() };
        let len = hash.hash_len();
        let value = &mut hash.value[..len];
        let data = src.as_bytes();
        match kind {
            SourceFileHashAlgorithm::Md5 => {
                value.copy_from_slice(&Md5::digest(data));
            }
            SourceFileHashAlgorithm::Sha1 => {
                value.copy_from_slice(&Sha1::digest(data));
            }
            SourceFileHashAlgorithm::Sha256 => {
                value.copy_from_slice(&Sha256::digest(data));
            }
        }
        hash
    }

    /// Check if the stored hash matches the hash of the string.
    pub fn matches(&self, src: &str) -> bool {
        Self::new(self.kind, src) == *self
    }

    /// The bytes of the hash.
    pub fn hash_bytes(&self) -> &[u8] {
        let len = self.hash_len();
        &self.value[..len]
    }

    fn hash_len(&self) -> usize {
        match self.kind {
            SourceFileHashAlgorithm::Md5 => 16,
            SourceFileHashAlgorithm::Sha1 => 20,
            SourceFileHashAlgorithm::Sha256 => 32,
        }
    }
}

#[derive(HashStable_Generic)]
#[derive(Copy, PartialEq, PartialOrd, Clone, Ord, Eq, Hash, Debug, Encodable, Decodable)]
pub enum DebuggerVisualizerType {
    Natvis,
    GdbPrettyPrinter,
}

/// A single debugger visualizer file.
#[derive(HashStable_Generic)]
#[derive(Clone, Debug, Hash, PartialEq, Eq, PartialOrd, Ord, Encodable, Decodable)]
pub struct DebuggerVisualizerFile {
    /// The complete debugger visualizer source.
    pub src: Arc<[u8]>,
    /// Indicates which visualizer type this targets.
    pub visualizer_type: DebuggerVisualizerType,
}

impl DebuggerVisualizerFile {
    pub fn new(src: Arc<[u8]>, visualizer_type: DebuggerVisualizerType) -> Self {
        DebuggerVisualizerFile { src, visualizer_type }
    }
}

#[derive(Clone)]
pub enum SourceFileLines {
    /// The source file lines, in decoded (random-access) form.
    Lines(Vec<BytePos>),

    /// The source file lines, in undecoded difference list form.
    Diffs(SourceFileDiffs),
}

impl SourceFileLines {
    pub fn is_lines(&self) -> bool {
        matches!(self, SourceFileLines::Lines(_))
    }
}

/// The source file lines in difference list form. This matches the form
/// used within metadata, which saves space by exploiting the fact that the
/// lines list is sorted and individual lines are usually not that long.
///
/// We read it directly from metadata and only decode it into `Lines` form
/// when necessary. This is a significant performance win, especially for
/// small crates where very little of `std`'s metadata is used.
#[derive(Clone)]
pub struct SourceFileDiffs {
    /// Position of the first line. Note that this is always encoded as a
    /// `BytePos` because it is often much larger than any of the
    /// differences.
    line_start: BytePos,

    /// Always 1, 2, or 4. Always as small as possible, while being big
    /// enough to hold the length of the longest line in the source file.
    /// The 1 case is by far the most common.
    bytes_per_diff: usize,

    /// The number of diffs encoded in `raw_diffs`. Always one less than
    /// the number of lines in the source file.
    num_diffs: usize,

    /// The diffs in "raw" form. Each segment of `bytes_per_diff` length
    /// encodes one little-endian diff. Note that they aren't LEB128
    /// encoded. This makes for much faster decoding. Besides, the
    /// bytes_per_diff==1 case is by far the most common, and LEB128
    /// encoding has no effect on that case.
    raw_diffs: Vec<u8>,
}

/// A single source in the [`SourceMap`].
#[derive(Clone)]
pub struct SourceFile {
    /// The name of the file that the source came from. Source that doesn't
    /// originate from files has names between angle brackets by convention
    /// (e.g., `<anon>`).
    pub name: FileName,
    /// The complete source code.
    pub src: Option<Lrc<String>>,
    /// The source code's hash.
    pub src_hash: SourceFileHash,
    /// The external source code (used for external crates, which will have a `None`
    /// value as `self.src`.
    pub external_src: Lock<ExternalSource>,
    /// The start position of this source in the `SourceMap`.
    pub start_pos: BytePos,
    /// The end position of this source in the `SourceMap`.
    pub end_pos: BytePos,
    /// Locations of lines beginnings in the source code.
    pub lines: Lock<SourceFileLines>,
    /// Locations of multi-byte characters in the source code.
    pub multibyte_chars: Vec<MultiByteChar>,
    /// Width of characters that are not narrow in the source code.
    pub non_narrow_chars: Vec<NonNarrowChar>,
    /// Locations of characters removed during normalization.
    pub normalized_pos: Vec<NormalizedPos>,
    /// A hash of the filename, used for speeding up hashing in incremental compilation.
    pub name_hash: u128,
    /// Indicates which crate this `SourceFile` was imported from.
    pub cnum: CrateNum,
}

impl<S: Encoder> Encodable<S> for SourceFile {
    fn encode(&self, s: &mut S) {
        self.name.encode(s);
        self.src_hash.encode(s);
        self.start_pos.encode(s);
        self.end_pos.encode(s);

        // We are always in `Lines` form by the time we reach here.
        assert!(self.lines.borrow().is_lines());
        self.lines(|lines| {
            // Store the length.
            s.emit_u32(lines.len() as u32);

            // Compute and store the difference list.
            if lines.len() != 0 {
                let max_line_length = if lines.len() == 1 {
                    0
                } else {
                    lines
                        .array_windows()
                        .map(|&[fst, snd]| snd - fst)
                        .map(|bp| bp.to_usize())
                        .max()
                        .unwrap()
                };

                let bytes_per_diff: usize = match max_line_length {
                    0..=0xFF => 1,
                    0x100..=0xFFFF => 2,
                    _ => 4,
                };

                // Encode the number of bytes used per diff.
                s.emit_u8(bytes_per_diff as u8);

                // Encode the first element.
                lines[0].encode(s);

                // Encode the difference list.
                let diff_iter = lines.array_windows().map(|&[fst, snd]| snd - fst);
                let num_diffs = lines.len() - 1;
                let mut raw_diffs;
                match bytes_per_diff {
                    1 => {
                        raw_diffs = Vec::with_capacity(num_diffs);
                        for diff in diff_iter {
                            raw_diffs.push(diff.0 as u8);
                        }
                    }
                    2 => {
                        raw_diffs = Vec::with_capacity(bytes_per_diff * num_diffs);
                        for diff in diff_iter {
                            raw_diffs.extend_from_slice(&(diff.0 as u16).to_le_bytes());
                        }
                    }
                    4 => {
                        raw_diffs = Vec::with_capacity(bytes_per_diff * num_diffs);
                        for diff in diff_iter {
                            raw_diffs.extend_from_slice(&(diff.0 as u32).to_le_bytes());
                        }
                    }
                    _ => unreachable!(),
                }
                s.emit_raw_bytes(&raw_diffs);
            }
        });

        self.multibyte_chars.encode(s);
        self.non_narrow_chars.encode(s);
        self.name_hash.encode(s);
        self.normalized_pos.encode(s);
        self.cnum.encode(s);
    }
}

impl<D: Decoder> Decodable<D> for SourceFile {
    fn decode(d: &mut D) -> SourceFile {
        let name: FileName = Decodable::decode(d);
        let src_hash: SourceFileHash = Decodable::decode(d);
        let start_pos: BytePos = Decodable::decode(d);
        let end_pos: BytePos = Decodable::decode(d);
        let lines = {
            let num_lines: u32 = Decodable::decode(d);
            if num_lines > 0 {
                // Read the number of bytes used per diff.
                let bytes_per_diff = d.read_u8() as usize;

                // Read the first element.
                let line_start: BytePos = Decodable::decode(d);

                // Read the difference list.
                let num_diffs = num_lines as usize - 1;
                let raw_diffs = d.read_raw_bytes(bytes_per_diff * num_diffs).to_vec();
                SourceFileLines::Diffs(SourceFileDiffs {
                    line_start,
                    bytes_per_diff,
                    num_diffs,
                    raw_diffs,
                })
            } else {
                SourceFileLines::Lines(vec![])
            }
        };
        let multibyte_chars: Vec<MultiByteChar> = Decodable::decode(d);
        let non_narrow_chars: Vec<NonNarrowChar> = Decodable::decode(d);
        let name_hash: u128 = Decodable::decode(d);
        let normalized_pos: Vec<NormalizedPos> = Decodable::decode(d);
        let cnum: CrateNum = Decodable::decode(d);
        SourceFile {
            name,
            start_pos,
            end_pos,
            src: None,
            src_hash,
            // Unused - the metadata decoder will construct
            // a new SourceFile, filling in `external_src` properly
            external_src: Lock::new(ExternalSource::Unneeded),
            lines: Lock::new(lines),
            multibyte_chars,
            non_narrow_chars,
            normalized_pos,
            name_hash,
            cnum,
        }
    }
}

impl fmt::Debug for SourceFile {
    fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result {
        write!(fmt, "SourceFile({:?})", self.name)
    }
}

impl SourceFile {
    pub fn new(
        name: FileName,
        mut src: String,
        start_pos: BytePos,
        hash_kind: SourceFileHashAlgorithm,
    ) -> Self {
        // Compute the file hash before any normalization.
        let src_hash = SourceFileHash::new(hash_kind, &src);
        let normalized_pos = normalize_src(&mut src, start_pos);

        let name_hash = {
            let mut hasher: StableHasher = StableHasher::new();
            name.hash(&mut hasher);
            hasher.finish::<u128>()
        };
        let end_pos = start_pos.to_usize() + src.len();
        assert!(end_pos <= u32::MAX as usize);

        let (lines, multibyte_chars, non_narrow_chars) =
            analyze_source_file::analyze_source_file(&src, start_pos);

        SourceFile {
            name,
            src: Some(Lrc::new(src)),
            src_hash,
            external_src: Lock::new(ExternalSource::Unneeded),
            start_pos,
            end_pos: Pos::from_usize(end_pos),
            lines: Lock::new(SourceFileLines::Lines(lines)),
            multibyte_chars,
            non_narrow_chars,
            normalized_pos,
            name_hash,
            cnum: LOCAL_CRATE,
        }
    }

    pub fn lines<F, R>(&self, f: F) -> R
    where
        F: FnOnce(&[BytePos]) -> R,
    {
        let mut guard = self.lines.borrow_mut();
        match &*guard {
            SourceFileLines::Lines(lines) => f(lines),
            SourceFileLines::Diffs(SourceFileDiffs {
                mut line_start,
                bytes_per_diff,
                num_diffs,
                raw_diffs,
            }) => {
                // Convert from "diffs" form to "lines" form.
                let num_lines = num_diffs + 1;
                let mut lines = Vec::with_capacity(num_lines);
                lines.push(line_start);

                assert_eq!(*num_diffs, raw_diffs.len() / bytes_per_diff);
                match bytes_per_diff {
                    1 => {
                        lines.extend(raw_diffs.into_iter().map(|&diff| {
                            line_start = line_start + BytePos(diff as u32);
                            line_start
                        }));
                    }
                    2 => {
                        lines.extend((0..*num_diffs).map(|i| {
                            let pos = bytes_per_diff * i;
                            let bytes = [raw_diffs[pos], raw_diffs[pos + 1]];
                            let diff = u16::from_le_bytes(bytes);
                            line_start = line_start + BytePos(diff as u32);
                            line_start
                        }));
                    }
                    4 => {
                        lines.extend((0..*num_diffs).map(|i| {
                            let pos = bytes_per_diff * i;
                            let bytes = [
                                raw_diffs[pos],
                                raw_diffs[pos + 1],
                                raw_diffs[pos + 2],
                                raw_diffs[pos + 3],
                            ];
                            let diff = u32::from_le_bytes(bytes);
                            line_start = line_start + BytePos(diff);
                            line_start
                        }));
                    }
                    _ => unreachable!(),
                }
                let res = f(&lines);
                *guard = SourceFileLines::Lines(lines);
                res
            }
        }
    }

    /// Returns the `BytePos` of the beginning of the current line.
    pub fn line_begin_pos(&self, pos: BytePos) -> BytePos {
        let line_index = self.lookup_line(pos).unwrap();
        self.lines(|lines| lines[line_index])
    }

    /// Add externally loaded source.
    /// If the hash of the input doesn't match or no input is supplied via None,
    /// it is interpreted as an error and the corresponding enum variant is set.
    /// The return value signifies whether some kind of source is present.
    pub fn add_external_src<F>(&self, get_src: F) -> bool
    where
        F: FnOnce() -> Option<String>,
    {
        if matches!(
            *self.external_src.borrow(),
            ExternalSource::Foreign { kind: ExternalSourceKind::AbsentOk, .. }
        ) {
            let src = get_src();
            let mut external_src = self.external_src.borrow_mut();
            // Check that no-one else have provided the source while we were getting it
            if let ExternalSource::Foreign {
                kind: src_kind @ ExternalSourceKind::AbsentOk, ..
            } = &mut *external_src
            {
                if let Some(mut src) = src {
                    // The src_hash needs to be computed on the pre-normalized src.
                    if self.src_hash.matches(&src) {
                        normalize_src(&mut src, BytePos::from_usize(0));
                        *src_kind = ExternalSourceKind::Present(Lrc::new(src));
                        return true;
                    }
                } else {
                    *src_kind = ExternalSourceKind::AbsentErr;
                }

                false
            } else {
                self.src.is_some() || external_src.get_source().is_some()
            }
        } else {
            self.src.is_some() || self.external_src.borrow().get_source().is_some()
        }
    }

    /// Gets a line from the list of pre-computed line-beginnings.
    /// The line number here is 0-based.
    pub fn get_line(&self, line_number: usize) -> Option<Cow<'_, str>> {
        fn get_until_newline(src: &str, begin: usize) -> &str {
            // We can't use `lines.get(line_number+1)` because we might
            // be parsing when we call this function and thus the current
            // line is the last one we have line info for.
            let slice = &src[begin..];
            match slice.find('\n') {
                Some(e) => &slice[..e],
                None => slice,
            }
        }

        let begin = {
            let line = self.lines(|lines| lines.get(line_number).copied())?;
            let begin: BytePos = line - self.start_pos;
            begin.to_usize()
        };

        if let Some(ref src) = self.src {
            Some(Cow::from(get_until_newline(src, begin)))
        } else if let Some(src) = self.external_src.borrow().get_source() {
            Some(Cow::Owned(String::from(get_until_newline(src, begin))))
        } else {
            None
        }
    }

    pub fn is_real_file(&self) -> bool {
        self.name.is_real()
    }

    pub fn is_imported(&self) -> bool {
        self.src.is_none()
    }

    pub fn count_lines(&self) -> usize {
        self.lines(|lines| lines.len())
    }

    /// Finds the line containing the given position. The return value is the
    /// index into the `lines` array of this `SourceFile`, not the 1-based line
    /// number. If the source_file is empty or the position is located before the
    /// first line, `None` is returned.
    pub fn lookup_line(&self, pos: BytePos) -> Option<usize> {
        self.lines(|lines| match lines.binary_search(&pos) {
            Ok(idx) => Some(idx),
            Err(0) => None,
            Err(idx) => Some(idx - 1),
        })
    }

    pub fn line_bounds(&self, line_index: usize) -> Range<BytePos> {
        if self.is_empty() {
            return self.start_pos..self.end_pos;
        }

        self.lines(|lines| {
            assert!(line_index < lines.len());
            if line_index == (lines.len() - 1) {
                lines[line_index]..self.end_pos
            } else {
                lines[line_index]..lines[line_index + 1]
            }
        })
    }

    /// Returns whether or not the file contains the given `SourceMap` byte
    /// position. The position one past the end of the file is considered to be
    /// contained by the file. This implies that files for which `is_empty`
    /// returns true still contain one byte position according to this function.
    #[inline]
    pub fn contains(&self, byte_pos: BytePos) -> bool {
        byte_pos >= self.start_pos && byte_pos <= self.end_pos
    }

    #[inline]
    pub fn is_empty(&self) -> bool {
        self.start_pos == self.end_pos
    }

    /// Calculates the original byte position relative to the start of the file
    /// based on the given byte position.
    pub fn original_relative_byte_pos(&self, pos: BytePos) -> BytePos {
        // Diff before any records is 0. Otherwise use the previously recorded
        // diff as that applies to the following characters until a new diff
        // is recorded.
        let diff = match self.normalized_pos.binary_search_by(|np| np.pos.cmp(&pos)) {
            Ok(i) => self.normalized_pos[i].diff,
            Err(i) if i == 0 => 0,
            Err(i) => self.normalized_pos[i - 1].diff,
        };

        BytePos::from_u32(pos.0 - self.start_pos.0 + diff)
    }

    /// Converts an absolute `BytePos` to a `CharPos` relative to the `SourceFile`.
    pub fn bytepos_to_file_charpos(&self, bpos: BytePos) -> CharPos {
        // The number of extra bytes due to multibyte chars in the `SourceFile`.
        let mut total_extra_bytes = 0;

        for mbc in self.multibyte_chars.iter() {
            debug!("{}-byte char at {:?}", mbc.bytes, mbc.pos);
            if mbc.pos < bpos {
                // Every character is at least one byte, so we only
                // count the actual extra bytes.
                total_extra_bytes += mbc.bytes as u32 - 1;
                // We should never see a byte position in the middle of a
                // character.
                assert!(bpos.to_u32() >= mbc.pos.to_u32() + mbc.bytes as u32);
            } else {
                break;
            }
        }

        assert!(self.start_pos.to_u32() + total_extra_bytes <= bpos.to_u32());
        CharPos(bpos.to_usize() - self.start_pos.to_usize() - total_extra_bytes as usize)
    }

    /// Looks up the file's (1-based) line number and (0-based `CharPos`) column offset, for a
    /// given `BytePos`.
    pub fn lookup_file_pos(&self, pos: BytePos) -> (usize, CharPos) {
        let chpos = self.bytepos_to_file_charpos(pos);
        match self.lookup_line(pos) {
            Some(a) => {
                let line = a + 1; // Line numbers start at 1
                let linebpos = self.lines(|lines| lines[a]);
                let linechpos = self.bytepos_to_file_charpos(linebpos);
                let col = chpos - linechpos;
                debug!("byte pos {:?} is on the line at byte pos {:?}", pos, linebpos);
                debug!("char pos {:?} is on the line at char pos {:?}", chpos, linechpos);
                debug!("byte is on line: {}", line);
                assert!(chpos >= linechpos);
                (line, col)
            }
            None => (0, chpos),
        }
    }

    /// Looks up the file's (1-based) line number, (0-based `CharPos`) column offset, and (0-based)
    /// column offset when displayed, for a given `BytePos`.
    pub fn lookup_file_pos_with_col_display(&self, pos: BytePos) -> (usize, CharPos, usize) {
        let (line, col_or_chpos) = self.lookup_file_pos(pos);
        if line > 0 {
            let col = col_or_chpos;
            let linebpos = self.lines(|lines| lines[line - 1]);
            let col_display = {
                let start_width_idx = self
                    .non_narrow_chars
                    .binary_search_by_key(&linebpos, |x| x.pos())
                    .unwrap_or_else(|x| x);
                let end_width_idx = self
                    .non_narrow_chars
                    .binary_search_by_key(&pos, |x| x.pos())
                    .unwrap_or_else(|x| x);
                let special_chars = end_width_idx - start_width_idx;
                let non_narrow: usize = self.non_narrow_chars[start_width_idx..end_width_idx]
                    .iter()
                    .map(|x| x.width())
                    .sum();
                col.0 - special_chars + non_narrow
            };
            (line, col, col_display)
        } else {
            let chpos = col_or_chpos;
            let col_display = {
                let end_width_idx = self
                    .non_narrow_chars
                    .binary_search_by_key(&pos, |x| x.pos())
                    .unwrap_or_else(|x| x);
                let non_narrow: usize =
                    self.non_narrow_chars[0..end_width_idx].iter().map(|x| x.width()).sum();
                chpos.0 - end_width_idx + non_narrow
            };
            (0, chpos, col_display)
        }
    }
}

/// Normalizes the source code and records the normalizations.
fn normalize_src(src: &mut String, start_pos: BytePos) -> Vec<NormalizedPos> {
    let mut normalized_pos = vec![];
    remove_bom(src, &mut normalized_pos);
    normalize_newlines(src, &mut normalized_pos);

    // Offset all the positions by start_pos to match the final file positions.
    for np in &mut normalized_pos {
        np.pos.0 += start_pos.0;
    }

    normalized_pos
}

/// Removes UTF-8 BOM, if any.
fn remove_bom(src: &mut String, normalized_pos: &mut Vec<NormalizedPos>) {
    if src.starts_with('\u{feff}') {
        src.drain(..3);
        normalized_pos.push(NormalizedPos { pos: BytePos(0), diff: 3 });
    }
}

/// Replaces `\r\n` with `\n` in-place in `src`.
///
/// Returns error if there's a lone `\r` in the string.
fn normalize_newlines(src: &mut String, normalized_pos: &mut Vec<NormalizedPos>) {
    if !src.as_bytes().contains(&b'\r') {
        return;
    }

    // We replace `\r\n` with `\n` in-place, which doesn't break utf-8 encoding.
    // While we *can* call `as_mut_vec` and do surgery on the live string
    // directly, let's rather steal the contents of `src`. This makes the code
    // safe even if a panic occurs.

    let mut buf = std::mem::replace(src, String::new()).into_bytes();
    let mut gap_len = 0;
    let mut tail = buf.as_mut_slice();
    let mut cursor = 0;
    let original_gap = normalized_pos.last().map_or(0, |l| l.diff);
    loop {
        let idx = match find_crlf(&tail[gap_len..]) {
            None => tail.len(),
            Some(idx) => idx + gap_len,
        };
        tail.copy_within(gap_len..idx, 0);
        tail = &mut tail[idx - gap_len..];
        if tail.len() == gap_len {
            break;
        }
        cursor += idx - gap_len;
        gap_len += 1;
        normalized_pos.push(NormalizedPos {
            pos: BytePos::from_usize(cursor + 1),
            diff: original_gap + gap_len as u32,
        });
    }

    // Account for removed `\r`.
    // After `set_len`, `buf` is guaranteed to contain utf-8 again.
    let new_len = buf.len() - gap_len;
    unsafe {
        buf.set_len(new_len);
        *src = String::from_utf8_unchecked(buf);
    }

    fn find_crlf(src: &[u8]) -> Option<usize> {
        let mut search_idx = 0;
        while let Some(idx) = find_cr(&src[search_idx..]) {
            if src[search_idx..].get(idx + 1) != Some(&b'\n') {
                search_idx += idx + 1;
                continue;
            }
            return Some(search_idx + idx);
        }
        None
    }

    fn find_cr(src: &[u8]) -> Option<usize> {
        src.iter().position(|&b| b == b'\r')
    }
}

// _____________________________________________________________________________
// Pos, BytePos, CharPos
//

pub trait Pos {
    fn from_usize(n: usize) -> Self;
    fn to_usize(&self) -> usize;
    fn from_u32(n: u32) -> Self;
    fn to_u32(&self) -> u32;
}

macro_rules! impl_pos {
    (
        $(
            $(#[$attr:meta])*
            $vis:vis struct $ident:ident($inner_vis:vis $inner_ty:ty);
        )*
    ) => {
        $(
            $(#[$attr])*
            $vis struct $ident($inner_vis $inner_ty);

            impl Pos for $ident {
                #[inline(always)]
                fn from_usize(n: usize) -> $ident {
                    $ident(n as $inner_ty)
                }

                #[inline(always)]
                fn to_usize(&self) -> usize {
                    self.0 as usize
                }

                #[inline(always)]
                fn from_u32(n: u32) -> $ident {
                    $ident(n as $inner_ty)
                }

                #[inline(always)]
                fn to_u32(&self) -> u32 {
                    self.0 as u32
                }
            }

            impl Add for $ident {
                type Output = $ident;

                #[inline(always)]
                fn add(self, rhs: $ident) -> $ident {
                    $ident(self.0 + rhs.0)
                }
            }

            impl Sub for $ident {
                type Output = $ident;

                #[inline(always)]
                fn sub(self, rhs: $ident) -> $ident {
                    $ident(self.0 - rhs.0)
                }
            }
        )*
    };
}

impl_pos! {
    /// A byte offset.
    ///
    /// Keep this small (currently 32-bits), as AST contains a lot of them.
    #[derive(Clone, Copy, PartialEq, Eq, Hash, PartialOrd, Ord, Debug)]
    pub struct BytePos(pub u32);

    /// A character offset.
    ///
    /// Because of multibyte UTF-8 characters, a byte offset
    /// is not equivalent to a character offset. The [`SourceMap`] will convert [`BytePos`]
    /// values to `CharPos` values as necessary.
    #[derive(Clone, Copy, PartialEq, Eq, PartialOrd, Ord, Debug)]
    pub struct CharPos(pub usize);
}

impl<S: Encoder> Encodable<S> for BytePos {
    fn encode(&self, s: &mut S) {
        s.emit_u32(self.0);
    }
}

impl<D: Decoder> Decodable<D> for BytePos {
    fn decode(d: &mut D) -> BytePos {
        BytePos(d.read_u32())
    }
}

// _____________________________________________________________________________
// Loc, SourceFileAndLine, SourceFileAndBytePos
//

/// A source code location used for error reporting.
#[derive(Debug, Clone)]
pub struct Loc {
    /// Information about the original source.
    pub file: Lrc<SourceFile>,
    /// The (1-based) line number.
    pub line: usize,
    /// The (0-based) column offset.
    pub col: CharPos,
    /// The (0-based) column offset when displayed.
    pub col_display: usize,
}

// Used to be structural records.
#[derive(Debug)]
pub struct SourceFileAndLine {
    pub sf: Lrc<SourceFile>,
    /// Index of line, starting from 0.
    pub line: usize,
}
#[derive(Debug)]
pub struct SourceFileAndBytePos {
    pub sf: Lrc<SourceFile>,
    pub pos: BytePos,
}

#[derive(Copy, Clone, Debug, PartialEq, Eq)]
pub struct LineInfo {
    /// Index of line, starting from 0.
    pub line_index: usize,

    /// Column in line where span begins, starting from 0.
    pub start_col: CharPos,

    /// Column in line where span ends, starting from 0, exclusive.
    pub end_col: CharPos,
}

pub struct FileLines {
    pub file: Lrc<SourceFile>,
    pub lines: Vec<LineInfo>,
}

pub static SPAN_TRACK: AtomicRef<fn(LocalDefId)> = AtomicRef::new(&((|_| {}) as fn(_)));

// _____________________________________________________________________________
// SpanLinesError, SpanSnippetError, DistinctSources, MalformedSourceMapPositions
//

pub type FileLinesResult = Result<FileLines, SpanLinesError>;

#[derive(Clone, PartialEq, Eq, Debug)]
pub enum SpanLinesError {
    DistinctSources(DistinctSources),
}

#[derive(Clone, PartialEq, Eq, Debug)]
pub enum SpanSnippetError {
    IllFormedSpan(Span),
    DistinctSources(DistinctSources),
    MalformedForSourcemap(MalformedSourceMapPositions),
    SourceNotAvailable { filename: FileName },
}

#[derive(Clone, PartialEq, Eq, Debug)]
pub struct DistinctSources {
    pub begin: (FileName, BytePos),
    pub end: (FileName, BytePos),
}

#[derive(Clone, PartialEq, Eq, Debug)]
pub struct MalformedSourceMapPositions {
    pub name: FileName,
    pub source_len: usize,
    pub begin_pos: BytePos,
    pub end_pos: BytePos,
}

/// Range inside of a `Span` used for diagnostics when we only have access to relative positions.
#[derive(Copy, Clone, PartialEq, Eq, Debug)]
pub struct InnerSpan {
    pub start: usize,
    pub end: usize,
}

impl InnerSpan {
    pub fn new(start: usize, end: usize) -> InnerSpan {
        InnerSpan { start, end }
    }
}

/// Requirements for a `StableHashingContext` to be used in this crate.
///
/// This is a hack to allow using the [`HashStable_Generic`] derive macro
/// instead of implementing everything in rustc_middle.
pub trait HashStableContext {
    fn def_path_hash(&self, def_id: DefId) -> DefPathHash;
    fn hash_spans(&self) -> bool;
    /// Accesses `sess.opts.debugging_opts.incremental_ignore_spans` since
    /// we don't have easy access to a `Session`
    fn debug_opts_incremental_ignore_spans(&self) -> bool;
    fn def_span(&self, def_id: LocalDefId) -> Span;
    fn span_data_to_lines_and_cols(
        &mut self,
        span: &SpanData,
    ) -> Option<(Lrc<SourceFile>, usize, BytePos, usize, BytePos)>;
    fn hashing_controls(&self) -> HashingControls;
}

impl<CTX> HashStable<CTX> for Span
where
    CTX: HashStableContext,
{
    /// Hashes a span in a stable way. We can't directly hash the span's `BytePos`
    /// fields (that would be similar to hashing pointers, since those are just
    /// offsets into the `SourceMap`). Instead, we hash the (file name, line, column)
    /// triple, which stays the same even if the containing `SourceFile` has moved
    /// within the `SourceMap`.
    ///
    /// Also note that we are hashing byte offsets for the column, not unicode
    /// codepoint offsets. For the purpose of the hash that's sufficient.
    /// Also, hashing filenames is expensive so we avoid doing it twice when the
    /// span starts and ends in the same file, which is almost always the case.
    fn hash_stable(&self, ctx: &mut CTX, hasher: &mut StableHasher) {
        const TAG_VALID_SPAN: u8 = 0;
        const TAG_INVALID_SPAN: u8 = 1;
        const TAG_RELATIVE_SPAN: u8 = 2;

        if !ctx.hash_spans() {
            return;
        }

        let span = self.data_untracked();
        span.ctxt.hash_stable(ctx, hasher);
        span.parent.hash_stable(ctx, hasher);

        if span.is_dummy() {
            Hash::hash(&TAG_INVALID_SPAN, hasher);
            return;
        }

        if let Some(parent) = span.parent {
            let def_span = ctx.def_span(parent).data_untracked();
            if def_span.contains(span) {
                // This span is enclosed in a definition: only hash the relative position.
                Hash::hash(&TAG_RELATIVE_SPAN, hasher);
                (span.lo - def_span.lo).to_u32().hash_stable(ctx, hasher);
                (span.hi - def_span.lo).to_u32().hash_stable(ctx, hasher);
                return;
            }
        }

        // If this is not an empty or invalid span, we want to hash the last
        // position that belongs to it, as opposed to hashing the first
        // position past it.
        let Some((file, line_lo, col_lo, line_hi, col_hi)) = ctx.span_data_to_lines_and_cols(&span) else {
            Hash::hash(&TAG_INVALID_SPAN, hasher);
            return;
        };

        Hash::hash(&TAG_VALID_SPAN, hasher);
        // We truncate the stable ID hash and line and column numbers. The chances
        // of causing a collision this way should be minimal.
        Hash::hash(&(file.name_hash as u64), hasher);

        // Hash both the length and the end location (line/column) of a span. If we
        // hash only the length, for example, then two otherwise equal spans with
        // different end locations will have the same hash. This can cause a problem
        // during incremental compilation wherein a previous result for a query that
        // depends on the end location of a span will be incorrectly reused when the
        // end location of the span it depends on has changed (see issue #74890). A
        // similar analysis applies if some query depends specifically on the length
        // of the span, but we only hash the end location. So hash both.

        let col_lo_trunc = (col_lo.0 as u64) & 0xFF;
        let line_lo_trunc = ((line_lo as u64) & 0xFF_FF_FF) << 8;
        let col_hi_trunc = (col_hi.0 as u64) & 0xFF << 32;
        let line_hi_trunc = ((line_hi as u64) & 0xFF_FF_FF) << 40;
        let col_line = col_lo_trunc | line_lo_trunc | col_hi_trunc | line_hi_trunc;
        let len = (span.hi - span.lo).0;
        Hash::hash(&col_line, hasher);
        Hash::hash(&len, hasher);
    }
}