summaryrefslogtreecommitdiff
path: root/src/buildstream/utils.py
blob: 872b5bd59f4d102191cf0d4ec5ac098f98fabfa1 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
#
#  Copyright (C) 2016-2018 Codethink Limited
#
#  This program is free software; you can redistribute it and/or
#  modify it under the terms of the GNU Lesser General Public
#  License as published by the Free Software Foundation; either
#  version 2 of the License, or (at your option) any later version.
#
#  This library is distributed in the hope that it will be useful,
#  but WITHOUT ANY WARRANTY; without even the implied warranty of
#  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.	 See the GNU
#  Lesser General Public License for more details.
#
#  You should have received a copy of the GNU Lesser General Public
#  License along with this library. If not, see <http://www.gnu.org/licenses/>.
#
#  Authors:
#        Tristan Van Berkom <tristan.vanberkom@codethink.co.uk>
"""
Utilities
=========
"""

import calendar
import errno
import hashlib
import os
import re
import shutil
import signal
import stat
from stat import S_ISDIR
import subprocess
import tempfile
import itertools
from contextlib import contextmanager
from pathlib import Path
from typing import Callable, IO, Iterable, Iterator, Optional, Tuple, Union

import psutil

from . import _signals
from ._exceptions import BstError, ErrorDomain
from ._protos.build.bazel.remote.execution.v2 import remote_execution_pb2

# Contains utils that have been rewritten in Cython for speed benefits
# This makes them available when importing from utils
from ._utils import url_directory_name  # pylint: disable=unused-import

# The magic number for timestamps: 2011-11-11 11:11:11
BST_ARBITRARY_TIMESTAMP = calendar.timegm((2011, 11, 11, 11, 11, 11))

# The separator we use for user specified aliases
_ALIAS_SEPARATOR = ':'
_URI_SCHEMES = ["http", "https", "ftp", "file", "git", "sftp", "ssh"]

# Main process pid
_MAIN_PID = os.getpid()

# The number of threads in the main process at startup.
# This is 1 except for certain test environments (xdist/execnet).
_INITIAL_NUM_THREADS_IN_MAIN_PROCESS = 1


class UtilError(BstError):
    """Raised by utility functions when system calls fail.

    This will be handled internally by the BuildStream core,
    if you need to handle this error, then it should be reraised,
    or either of the :class:`.ElementError` or :class:`.SourceError`
    exceptions should be raised from this error.
    """
    def __init__(self, message, reason=None):
        super().__init__(message, domain=ErrorDomain.UTIL, reason=reason)


class ProgramNotFoundError(BstError):
    """Raised if a required program is not found.

    It is normally unneeded to handle this exception from plugin code.
    """
    def __init__(self, message, reason=None):
        super().__init__(message, domain=ErrorDomain.PROG_NOT_FOUND, reason=reason)


class DirectoryExistsError(OSError):
    """Raised when a `os.rename` is attempted but the destination is an existing directory.
    """


class FileListResult():
    """An object which stores the result of one of the operations
    which run on a list of files.
    """

    def __init__(self):

        self.overwritten = []
        """List of files which were overwritten in the target directory"""

        self.ignored = []
        """List of files which were ignored, because they would have
        replaced a non empty directory"""

        self.failed_attributes = []
        """List of files for which attributes could not be copied over"""

        self.files_written = []
        """List of files that were written."""

    def combine(self, other: 'FileListResult') -> 'FileListResult':
        """Create a new FileListResult that contains the results of both.
        """
        ret = FileListResult()

        ret.overwritten = self.overwritten + other.overwritten
        ret.ignored = self.ignored + other.ignored
        ret.failed_attributes = self.failed_attributes + other.failed_attributes
        ret.files_written = self.files_written + other.files_written

        return ret


def list_relative_paths(directory: str) -> Iterator[str]:
    """A generator for walking directory relative paths

    This generator is useful for checking the full manifest of
    a directory.

    Symbolic links will not be followed, but will be included
    in the manifest.

    Args:
       directory: The directory to list files in

    Yields:
       Relative filenames in `directory`
    """
    for (dirpath, dirnames, filenames) in os.walk(directory):

        # os.walk does not decend into symlink directories, which
        # makes sense because otherwise we might have redundant
        # directories, or end up descending into directories outside
        # of the walk() directory.
        #
        # But symlinks to directories are still identified as
        # subdirectories in the walked `dirpath`, so we extract
        # these symlinks from `dirnames` and add them to `filenames`.
        #
        for d in dirnames:
            fullpath = os.path.join(dirpath, d)
            if os.path.islink(fullpath):
                filenames.append(d)

        # Modifying the dirnames directly ensures that the os.walk() generator
        # allows us to specify the order in which they will be iterated.
        dirnames.sort()
        filenames.sort()

        relpath = os.path.relpath(dirpath, directory)

        # We don't want "./" pre-pended to all the entries in the root of
        # `directory`, prefer to have no prefix in that case.
        basepath = relpath if relpath != '.' and dirpath != directory else ''

        # First yield the walked directory itself, except for the root
        if basepath != '':
            yield basepath

        # List the filenames in the walked directory
        for f in filenames:
            yield os.path.join(basepath, f)


# pylint: disable=anomalous-backslash-in-string
def glob(paths: Iterable[str], pattern: str) -> Iterator[str]:
    """A generator to yield paths which match the glob pattern

    Args:
       paths (iterable): The paths to check
       pattern (str): A glob pattern

    This generator will iterate over the passed *paths* and
    yield only the filenames which matched the provided *pattern*.

    +--------+------------------------------------------------------------------+
    | Meta   | Description                                                      |
    +========+==================================================================+
    | \*     | Zero or more of any character, excepting path separators         |
    +--------+------------------------------------------------------------------+
    | \**    | Zero or more of any character, including path separators         |
    +--------+------------------------------------------------------------------+
    | ?      | One of any character, except for path separators                 |
    +--------+------------------------------------------------------------------+
    | [abc]  | One of any of the specified characters                           |
    +--------+------------------------------------------------------------------+
    | [a-z]  | One of the characters in the specified range                     |
    +--------+------------------------------------------------------------------+
    | [!abc] | Any single character, except the specified characters            |
    +--------+------------------------------------------------------------------+
    | [!a-z] | Any single character, except those in the specified range        |
    +--------+------------------------------------------------------------------+

    .. note::

       Escaping of the metacharacters is not possible

    """
    # Ensure leading slash, just because we want patterns
    # to match file lists regardless of whether the patterns
    # or file lists had a leading slash or not.
    if not pattern.startswith(os.sep):
        pattern = os.sep + pattern

    expression = _glob2re(pattern)
    regexer = re.compile(expression)

    for filename in paths:
        filename_try = filename
        if not filename_try.startswith(os.sep):
            filename_try = os.sep + filename_try

        if regexer.match(filename_try):
            yield filename


def sha256sum(filename: str) -> str:
    """Calculate the sha256sum of a file

    Args:
       filename: A path to a file on disk

    Returns:
      An sha256 checksum string

    Raises:
       UtilError: In the case there was an issue opening
                  or reading `filename`
    """
    try:
        h = hashlib.sha256()
        with open(filename, "rb") as f:
            for chunk in iter(lambda: f.read(65536), b""):
                h.update(chunk)

    except OSError as e:
        raise UtilError("Failed to get a checksum of file '{}': {}"
                        .format(filename, e)) from e

    return h.hexdigest()


def safe_copy(src: str, dest: str, *, result: Optional[FileListResult] = None) -> None:
    """Copy a file while preserving attributes

    Args:
       src: The source filename
       dest: The destination filename
       result: An optional collective result

    Raises:
       UtilError: In the case of unexpected system call failures

    This is almost the same as shutil.copy2(), except that
    we unlink *dest* before overwriting it if it exists, just
    incase *dest* is a hardlink to a different file.
    """
    # First unlink the target if it exists
    try:
        os.unlink(dest)
    except OSError as e:
        if e.errno != errno.ENOENT:
            raise UtilError("Failed to remove destination file '{}': {}"
                            .format(dest, e)) from e

    shutil.copyfile(src, dest)
    try:
        shutil.copystat(src, dest)
    except PermissionError:
        # If we failed to copy over some file stats, dont treat
        # it as an unrecoverable error, but provide some feedback
        # we can use for a warning.
        #
        # This has a tendency of happening when attempting to copy
        # over extended file attributes.
        if result:
            result.failed_attributes.append(dest)

    except shutil.Error as e:
        raise UtilError("Failed to copy '{} -> {}': {}"
                        .format(src, dest, e)) from e


def safe_link(src: str, dest: str, *, result: Optional[FileListResult] = None, _unlink=False) -> None:
    """Try to create a hardlink, but resort to copying in the case of cross device links.

    Args:
       src: The source filename
       dest: The destination filename
       result: An optional collective result

    Raises:
       UtilError: In the case of unexpected system call failures
    """

    if _unlink:
        # First unlink the target if it exists
        try:
            os.unlink(dest)
        except OSError as e:
            if e.errno != errno.ENOENT:
                raise UtilError("Failed to remove destination file '{}': {}"
                                .format(dest, e)) from e

    # If we can't link it due to cross-device hardlink, copy
    try:
        os.link(src, dest)
    except OSError as e:
        if e.errno == errno.EEXIST and not _unlink:
            # Target exists already, unlink and try again
            safe_link(src, dest, result=result, _unlink=True)
        elif e.errno == errno.EXDEV:
            safe_copy(src, dest)
        else:
            raise UtilError("Failed to link '{} -> {}': {}"
                            .format(src, dest, e)) from e


def safe_remove(path: str) -> bool:
    """Removes a file or directory

    This will remove a file if it exists, and will
    remove a directory if the directory is empty.

    Args:
       path: The path to remove

    Returns:
       True if `path` was removed or did not exist, False
       if `path` was a non empty directory.

    Raises:
       UtilError: In the case of unexpected system call failures
    """
    try:
        if S_ISDIR(os.lstat(path).st_mode):
            os.rmdir(path)
        else:
            os.unlink(path)

        # File removed/unlinked successfully
        return True

    except OSError as e:
        if e.errno == errno.ENOTEMPTY:
            # Path is non-empty directory
            return False
        elif e.errno == errno.ENOENT:
            # Path does not exist
            return True

        raise UtilError("Failed to remove '{}': {}"
                        .format(path, e))


def copy_files(src: str,
               dest: str,
               *,
               filter_callback: Optional[Callable[[str], bool]] = None,
               ignore_missing: bool = False,
               report_written: bool = False) -> FileListResult:
    """Copy files from source to destination.

    Args:
       src: The source directory
       dest: The destination directory
       filter_callback: Optional filter callback. Called with the relative path as
                        argument for every file in the source directory. The file is
                        copied only if the callable returns True. If no filter callback
                        is specified, all files will be copied.
       ignore_missing: Dont raise any error if a source file is missing
       report_written: Add to the result object the full list of files written

    Returns:
       The result describing what happened during this file operation

    Raises:
       UtilError: In the case of unexpected system call failures

    .. note::

       Directories in `dest` are replaced with files from `src`,
       unless the existing directory in `dest` is not empty in which
       case the path will be reported in the return value.

       UNIX domain socket files from `src` are ignored.
    """
    result = FileListResult()
    try:
        _process_list(src, dest, safe_copy, result,
                      filter_callback=filter_callback,
                      ignore_missing=ignore_missing,
                      report_written=report_written)
    except OSError as e:
        raise UtilError("Failed to copy '{} -> {}': {}"
                        .format(src, dest, e))
    return result


def link_files(src: str,
               dest: str,
               *,
               filter_callback: Optional[Callable[[str], bool]] = None,
               ignore_missing: bool = False,
               report_written: bool = False) -> FileListResult:
    """Hardlink files from source to destination.

    Args:
       src: The source directory
       dest: The destination directory
       filter_callback: Optional filter callback. Called with the relative path as
                        argument for every file in the source directory. The file is
                        hardlinked only if the callable returns True. If no filter
                        callback is specified, all files will be hardlinked.
       ignore_missing: Dont raise any error if a source file is missing
       report_written: Add to the result object the full list of files written

    Returns:
       The result describing what happened during this file operation

    Raises:
       UtilError: In the case of unexpected system call failures

    .. note::

       Directories in `dest` are replaced with files from `src`,
       unless the existing directory in `dest` is not empty in which
       case the path will be reported in the return value.

    .. note::

       If a hardlink cannot be created due to crossing filesystems,
       then the file will be copied instead.

       UNIX domain socket files from `src` are ignored.
    """
    result = FileListResult()
    try:
        _process_list(src, dest, safe_link, result,
                      filter_callback=filter_callback,
                      ignore_missing=ignore_missing,
                      report_written=report_written)
    except OSError as e:
        raise UtilError("Failed to link '{} -> {}': {}"
                        .format(src, dest, e))

    return result


def get_host_tool(name: str) -> str:
    """Get the full path of a host tool

    Args:
       name (str): The name of the program to search for

    Returns:
       The full path to the program, if found

    Raises:
       :class:`.ProgramNotFoundError`
    """
    search_path = os.environ.get('PATH')
    program_path = shutil.which(name, path=search_path)

    if not program_path:
        raise ProgramNotFoundError("Did not find '{}' in PATH: {}".format(name, search_path))

    return program_path


def get_bst_version() -> Tuple[int, int]:
    """Gets the major, minor release portion of the
    BuildStream version.

    Returns:
       A 2-tuple of form (major version, minor version)
    """
    # Import this only conditionally, it's not resolved at bash complete time
    from . import __version__  # pylint: disable=cyclic-import
    versions = __version__.split('.')[:2]

    if versions[0] == '0+untagged':
        raise UtilError("Your git repository has no tags - BuildStream can't "
                        "determine its version. Please run `git fetch --tags`.")

    try:
        return (int(versions[0]), int(versions[1]))
    except IndexError:
        raise UtilError("Cannot detect Major and Minor parts of the version\n"
                        "Version: {} not in XX.YY.whatever format"
                        .format(__version__))
    except ValueError:
        raise UtilError("Cannot convert version to integer numbers\n"
                        "Version: {} not in Integer.Integer.whatever format"
                        .format(__version__))


def move_atomic(source: Union[Path, str], destination: Union[Path, str], *, ensure_parents: bool = True) -> None:
    """Move the source to the destination using atomic primitives.

    This uses `os.rename` to move a file or directory to a new destination.
    It wraps some `OSError` thrown errors to ensure their handling is correct.

    The main reason for this to exist is that rename can throw different errors
    for the same symptom (https://www.unix.com/man-page/POSIX/3posix/rename/)
    when we are moving a directory.

    We are especially interested here in the case when the destination already
    exists, is a directory and is not empty. In this case, either EEXIST or
    ENOTEMPTY can be thrown.

    In order to ensure consistent handling of these exceptions, this function
    should be used instead of `os.rename`

    Args:
      source: source to rename
      destination: destination to which to move the source
      ensure_parents: Whether or not to create the parent's directories
                      of the destination (default: True)
    Raises:
      DirectoryExistsError: if the destination directory already exists and is
                            not empty
      OSError: if another filesystem level error occured
    """
    if ensure_parents:
        os.makedirs(os.path.dirname(str(destination)), exist_ok=True)

    try:
        os.rename(str(source), str(destination))
    except OSError as exc:
        if exc.errno in (errno.EEXIST, errno.ENOTEMPTY):
            raise DirectoryExistsError(*exc.args) from exc
        raise


@contextmanager
def save_file_atomic(filename: str,
                     mode: str = 'w',
                     *,
                     buffering: int = -1,
                     encoding: Optional[str] = None,
                     errors: Optional[str] = None,
                     newline: Optional[str] = None,
                     closefd: bool = True,
                     opener: Optional[Callable[[str, int], int]] = None,
                     tempdir: Optional[str] = None) -> Iterator[IO]:
    """Save a file with a temporary name and rename it into place when ready.

    This is a context manager which is meant for saving data to files.
    The data is written to a temporary file, which gets renamed to the target
    name when the context is closed. This avoids readers of the file from
    getting an incomplete file.

    **Example:**

    .. code:: python

      with save_file_atomic('/path/to/foo', 'w') as f:
          f.write(stuff)

    The file will be called something like ``tmpCAFEBEEF`` until the
    context block ends, at which point it gets renamed to ``foo``. The
    temporary file will be created in the same directory as the output file.
    The ``filename`` parameter must be an absolute path.

    If an exception occurs or the process is terminated, the temporary file will
    be deleted.
    """
    # This feature has been proposed for upstream Python in the past, e.g.:
    # https://bugs.python.org/issue8604

    assert os.path.isabs(filename), "The utils.save_file_atomic() parameter ``filename`` must be an absolute path"
    if tempdir is None:
        tempdir = os.path.dirname(filename)
    fd, tempname = tempfile.mkstemp(dir=tempdir)
    os.close(fd)

    f = open(tempname, mode=mode, buffering=buffering, encoding=encoding,
             errors=errors, newline=newline, closefd=closefd, opener=opener)

    def cleanup_tempfile():
        f.close()
        try:
            os.remove(tempname)
        except FileNotFoundError:
            pass
        except OSError as e:
            raise UtilError("Failed to cleanup temporary file {}: {}".format(tempname, e)) from e

    try:
        with _signals.terminator(cleanup_tempfile):
            # Disable type-checking since "IO[Any]" has no attribute "real_filename"
            f.real_filename = filename		# type: ignore
            yield f
            f.close()
            # This operation is atomic, at least on platforms we care about:
            # https://bugs.python.org/issue8828
            os.replace(tempname, filename)
    except Exception:
        cleanup_tempfile()
        raise


# _get_dir_size():
#
# Get the disk usage of a given directory in bytes.
#
# This function assumes that files do not inadvertantly
# disappear while this function is running.
#
# Arguments:
#     (str) The path whose size to check.
#
# Returns:
#     (int) The size on disk in bytes.
#
def _get_dir_size(path):
    path = os.path.abspath(path)

    def get_size(path):
        total = 0

        for f in os.scandir(path):
            total += f.stat(follow_symlinks=False).st_size

            if f.is_dir(follow_symlinks=False):
                total += get_size(f.path)

        return total

    return get_size(path)


# _get_volume_size():
#
# Gets the overall usage and total size of a mounted filesystem in bytes.
#
# Args:
#    path (str): The path to check
#
# Returns:
#    (int): The total number of bytes on the volume
#    (int): The number of available bytes on the volume
#
def _get_volume_size(path):
    try:
        usage = shutil.disk_usage(path)
    except OSError as e:
        raise UtilError("Failed to retrieve stats on volume for path '{}': {}"
                        .format(path, e)) from e
    return usage.total, usage.free


# _parse_size():
#
# Convert a string representing data size to a number of
# bytes. E.g. "2K" -> 2048.
#
# This uses the same format as systemd's
# [resource-control](https://www.freedesktop.org/software/systemd/man/systemd.resource-control.html#).
#
# Arguments:
#     size (str) The string to parse
#     volume (str) A path on the volume to consider for percentage
#                  specifications
#
# Returns:
#     (int|None) The number of bytes, or None if 'infinity' was specified.
#
# Raises:
#     UtilError if the string is not a valid data size.
#
def _parse_size(size, volume):
    if size == 'infinity':
        return None

    matches = re.fullmatch(r'([0-9]+\.?[0-9]*)([KMGT%]?)', size)
    if matches is None:
        raise UtilError("{} is not a valid data size.".format(size))

    num, unit = matches.groups()

    if unit == '%':
        num = float(num)
        if num > 100:
            raise UtilError("{}% is not a valid percentage value.".format(num))

        disk_size, _ = _get_volume_size(volume)

        return disk_size * (num / 100)

    units = ('', 'K', 'M', 'G', 'T')
    return int(num) * 1024**units.index(unit)


# _pretty_size()
#
# Converts a number of bytes into a string representation in KiB, MiB, GiB, TiB
# represented as K, M, G, T etc.
#
# Args:
#   size (int): The size to convert in bytes.
#   dec_places (int): The number of decimal places to output to.
#
# Returns:
#   (str): The string representation of the number of bytes in the largest
def _pretty_size(size, dec_places=0):
    psize = size
    unit = 'B'
    units = ('B', 'K', 'M', 'G', 'T')
    for unit in units:
        if psize < 1024:
            break
        elif unit != units[-1]:
            psize /= 1024
    return "{size:g}{unit}".format(size=round(psize, dec_places), unit=unit)


# _is_main_process()
#
# Return whether we are in the main process or not.
#
def _is_main_process():
    assert _MAIN_PID is not None
    return os.getpid() == _MAIN_PID


# Recursively remove directories, ignoring file permissions as much as
# possible.
def _force_rmtree(rootpath, **kwargs):

    os.chmod(rootpath, 0o755)
    for root, dirs, _ in os.walk(rootpath):
        for d in dirs:
            path = os.path.join(root, d.lstrip('/'))
            if os.path.exists(path) and not os.path.islink(path):
                try:
                    os.chmod(path, 0o755)
                except OSError as e:
                    raise UtilError("Failed to ensure write permission on file '{}': {}"
                                    .format(path, e))

    try:
        shutil.rmtree(rootpath, **kwargs)
    except OSError as e:
        raise UtilError("Failed to remove cache directory '{}': {}"
                        .format(rootpath, e))


# Recursively make directories in target area
def _copy_directories(srcdir, destdir, target):
    this_dir = os.path.dirname(target)
    new_dir = os.path.join(destdir, this_dir)

    if not os.path.lexists(new_dir):
        if this_dir:
            yield from _copy_directories(srcdir, destdir, this_dir)

        old_dir = os.path.join(srcdir, this_dir)
        if os.path.lexists(old_dir):
            dir_stat = os.lstat(old_dir)
            mode = dir_stat.st_mode

            if stat.S_ISDIR(mode) or stat.S_ISLNK(mode):
                os.makedirs(new_dir)
                yield (new_dir, mode)
            else:
                raise UtilError('Source directory tree has file where '
                                'directory expected: {}'.format(old_dir))


# _ensure_real_directory()
#
# Ensure `path` is a real directory and there are no symlink components.
#
# Symlink components are allowed in `root`.
#
def _ensure_real_directory(root, path):
    destpath = root
    for name in os.path.split(path):
        destpath = os.path.join(destpath, name)
        try:
            deststat = os.lstat(destpath)
            if not stat.S_ISDIR(deststat.st_mode):
                relpath = destpath[len(root):]

                if stat.S_ISLNK(deststat.st_mode):
                    filetype = 'symlink'
                elif stat.S_ISREG(deststat.st_mode):
                    filetype = 'regular file'
                else:
                    filetype = 'special file'

                raise UtilError('Destination is a {}, not a directory: {}'.format(filetype, relpath))
        except FileNotFoundError:
            os.makedirs(destpath)


# _process_list()
#
# Internal helper for copying/moving/linking file lists
#
# This will handle directories, symlinks and special files
# internally, the `actionfunc` will only be called for regular files.
#
# Args:
#    srcdir: The source base directory
#    destdir: The destination base directory
#    actionfunc: The function to call for regular files
#    result: The FileListResult
#    filter_callback: Optional callback to invoke for every directory entry
#    ignore_missing: Dont raise any error if a source file is missing
#
#
def _process_list(srcdir, destdir, actionfunc, result,
                  filter_callback=None,
                  ignore_missing=False, report_written=False):

    # Keep track of directory permissions, since these need to be set
    # *after* files have been written.
    permissions = []

    filelist = list_relative_paths(srcdir)

    if filter_callback:
        filelist = [path for path in filelist if filter_callback(path)]

    # Now walk the list
    for path in filelist:
        srcpath = os.path.join(srcdir, path)
        destpath = os.path.join(destdir, path)

        # Ensure that the parent of the destination path exists without symlink
        # components.
        _ensure_real_directory(destdir, os.path.dirname(path))

        # Add to the results the list of files written
        if report_written:
            result.files_written.append(path)

        # Collect overlaps
        if os.path.lexists(destpath) and not os.path.isdir(destpath):
            result.overwritten.append(path)

        # The destination directory may not have been created separately
        permissions.extend(_copy_directories(srcdir, destdir, path))

        try:
            file_stat = os.lstat(srcpath)
            mode = file_stat.st_mode

        except FileNotFoundError as e:
            # Skip this missing file
            if ignore_missing:
                continue
            else:
                raise UtilError("Source file is missing: {}".format(srcpath)) from e

        if stat.S_ISDIR(mode):
            # Ensure directory exists in destination
            _ensure_real_directory(destdir, path)
            permissions.append((destpath, os.stat(srcpath).st_mode))

        elif stat.S_ISLNK(mode):
            if not safe_remove(destpath):
                result.ignored.append(path)
                continue

            target = os.readlink(srcpath)
            os.symlink(target, destpath)

        elif stat.S_ISREG(mode):
            # Process the file.
            if not safe_remove(destpath):
                result.ignored.append(path)
                continue

            actionfunc(srcpath, destpath, result=result)

        elif stat.S_ISCHR(mode) or stat.S_ISBLK(mode):
            # Block or character device. Put contents of st_dev in a mknod.
            if not safe_remove(destpath):
                result.ignored.append(path)
                continue

            if os.path.lexists(destpath):
                os.remove(destpath)
            os.mknod(destpath, file_stat.st_mode, file_stat.st_rdev)
            os.chmod(destpath, file_stat.st_mode)

        elif stat.S_ISFIFO(mode):
            os.mkfifo(destpath, mode)

        elif stat.S_ISSOCK(mode):
            # We can't duplicate the process serving the socket anyway
            pass

        else:
            # Unsupported type.
            raise UtilError('Cannot extract {} into staging-area. Unsupported type.'.format(srcpath))

    # Write directory permissions now that all files have been written
    for d, perms in permissions:
        os.chmod(d, perms)


# _set_deterministic_user()
#
# Set the uid/gid for every file in a directory tree to the process'
# euid/guid.
#
# Args:
#    directory (str): The directory to recursively set the uid/gid on
#
def _set_deterministic_user(directory):
    user = os.geteuid()
    group = os.getegid()

    for root, dirs, files in os.walk(directory.encode("utf-8"), topdown=False):
        for filename in files:
            os.chown(os.path.join(root, filename), user, group, follow_symlinks=False)

        for dirname in dirs:
            os.chown(os.path.join(root, dirname), user, group, follow_symlinks=False)


# _set_deterministic_mtime()
#
# Set the mtime for every file in a directory tree to the same.
#
# Args:
#    directory (str): The directory to recursively set the mtime on
#
def _set_deterministic_mtime(directory):
    for dirname, _, filenames in os.walk(directory.encode("utf-8"), topdown=False):
        for filename in filenames:
            pathname = os.path.join(dirname, filename)

            # Python's os.utime only ever modifies the timestamp
            # of the target, it is not acceptable to set the timestamp
            # of the target here, if we are staging the link target we
            # will also set its timestamp.
            #
            # We should however find a way to modify the actual link's
            # timestamp, this outdated python bug report claims that
            # it is impossible:
            #
            #   http://bugs.python.org/issue623782
            #
            # However, nowadays it is possible at least on gnuish systems
            # with with the lutimes glibc function.
            if not os.path.islink(pathname):
                os.utime(pathname, (BST_ARBITRARY_TIMESTAMP, BST_ARBITRARY_TIMESTAMP))

        os.utime(dirname, (BST_ARBITRARY_TIMESTAMP, BST_ARBITRARY_TIMESTAMP))


# _tempdir()
#
# A context manager for doing work in a temporary directory.
#
# Args:
#    dir (str): A path to a parent directory for the temporary directory
#    suffix (str): A suffix for the temproary directory name
#    prefix (str): A prefix for the temporary directory name
#
# Yields:
#    (str): The temporary directory
#
# In addition to the functionality provided by python's
# tempfile.TemporaryDirectory() context manager, this one additionally
# supports cleaning up the temp directory on SIGTERM.
#
@contextmanager
def _tempdir(suffix="", prefix="tmp", dir=None):  # pylint: disable=redefined-builtin
    tempdir = tempfile.mkdtemp(suffix=suffix, prefix=prefix, dir=dir)

    def cleanup_tempdir():
        if os.path.isdir(tempdir):
            _force_rmtree(tempdir)

    try:
        with _signals.terminator(cleanup_tempdir):
            yield tempdir
    finally:
        cleanup_tempdir()


# _tempnamedfile()
#
# A context manager for doing work on an open temporary file
# which is guaranteed to be named and have an entry in the filesystem.
#
# Args:
#    dir (str): A path to a parent directory for the temporary file
#    suffix (str): A suffix for the temproary file name
#    prefix (str): A prefix for the temporary file name
#
# Yields:
#    (tempfile.NamedTemporaryFile): The temporary file handle
#
# Do not use tempfile.NamedTemporaryFile() directly, as this will
# leak files on the filesystem when BuildStream exits a process
# on SIGTERM.
#
@contextmanager
def _tempnamedfile(suffix="", prefix="tmp", dir=None):  # pylint: disable=redefined-builtin
    temp = None

    def close_tempfile():
        if temp is not None:
            temp.close()

    with _signals.terminator(close_tempfile), \
            tempfile.NamedTemporaryFile(suffix=suffix, prefix=prefix, dir=dir) as temp:
        yield temp


# _tempnamedfile_name()
#
# A context manager for doing work on a temporary file, via the filename only.
#
# Note that a Windows restriction prevents us from using both the file
# descriptor and the file name of tempfile.NamedTemporaryFile, the same file
# cannot be opened twice at the same time. This wrapper makes it easier to
# operate in a Windows-friendly way when only filenames are needed.
#
# Takes care to delete the file even if interrupted by SIGTERM. Note that there
# is a race-condition, so this is done on a best-effort basis.
#
# Args:
#    dir (str): A path to a parent directory for the temporary file, this is
#               not optional for security reasons, please see below.
#
# Note that 'dir' should not be a directory that may be shared with potential
# attackers that have rights to affect the directory, e.g. the system temp
# directory. This is because an attacker can replace the temporary file with
# e.g. a symlink to another location, that the BuildStream user has rights to
# but the attacker does not.
#
# See here for more information:
# https://www.owasp.org/index.php/Insecure_Temporary_File
#
# Yields:
#    (str): The temporary file name
#
@contextmanager
def _tempnamedfile_name(dir):  # pylint: disable=redefined-builtin
    filename = None

    def rm_tempfile():
        nonlocal filename
        if filename is not None:
            # Note that this code is re-entrant - it's possible for us to
            # "delete while we're deleting" if we are responding to SIGTERM.
            #
            # For simplicity, choose to leak the file in this unlikely case,
            # rather than introducing a scheme for handling the file sometimes
            # being missing. Note that we will still leak tempfiles on forced
            # termination anyway.
            #
            # Note that we must not try to delete the file more than once. If
            # we delete the file then another temporary file can be created
            # with the same name.
            #
            filename2 = filename
            filename = None
            os.remove(filename2)

    with _signals.terminator(rm_tempfile):
        fd, filename = tempfile.mkstemp(dir=dir)
        os.close(fd)
        try:
            yield filename
        finally:
            rm_tempfile()


# _kill_process_tree()
#
# Brutally murder a process and all of its children
#
# Args:
#    pid (int): Process ID
#
def _kill_process_tree(pid):
    proc = psutil.Process(pid)
    children = proc.children(recursive=True)

    def kill_proc(p):
        try:
            p.kill()
        except psutil.AccessDenied:
            # Ignore this error, it can happen with
            # some setuid bwrap processes.
            pass
        except psutil.NoSuchProcess:
            # It is certain that this has already been sent
            # SIGTERM, so there is a window where the process
            # could have exited already.
            pass

    # Bloody Murder
    for child in children:
        kill_proc(child)
    kill_proc(proc)


# _call()
#
# A wrapper for subprocess.call() supporting suspend and resume
#
# Args:
#    popenargs (list): Popen() arguments
#    terminate (bool): Whether to attempt graceful termination before killing
#    rest_of_args (kwargs): Remaining arguments to subprocess.call()
#
# Returns:
#    (int): The process exit code.
#    (str): The program output.
#
def _call(*popenargs, terminate=False, **kwargs):

    kwargs['start_new_session'] = True

    process = None

    old_preexec_fn = kwargs.get('preexec_fn')
    if 'preexec_fn' in kwargs:
        del kwargs['preexec_fn']

    def preexec_fn():
        os.umask(stat.S_IWGRP | stat.S_IWOTH)
        if old_preexec_fn is not None:
            old_preexec_fn()

    # Handle termination, suspend and resume
    def kill_proc():
        if process:

            # Some callers know that their subprocess can be
            # gracefully terminated, make an attempt first
            if terminate:
                proc = psutil.Process(process.pid)
                proc.terminate()

                try:
                    proc.wait(20)
                except psutil.TimeoutExpired:
                    # Did not terminate within the timeout: murder
                    _kill_process_tree(process.pid)

            else:
                # FIXME: This is a brutal but reliable approach
                #
                # Other variations I've tried which try SIGTERM first
                # and then wait for child processes to exit gracefully
                # have not reliably cleaned up process trees and have
                # left orphaned git or ssh processes alive.
                #
                # This cleans up the subprocesses reliably but may
                # cause side effects such as possibly leaving stale
                # locks behind. Hopefully this should not be an issue
                # as long as any child processes only interact with
                # the temp directories which we control and cleanup
                # ourselves.
                #
                _kill_process_tree(process.pid)

    def suspend_proc():
        if process:
            group_id = os.getpgid(process.pid)
            os.killpg(group_id, signal.SIGSTOP)

    def resume_proc():
        if process:
            group_id = os.getpgid(process.pid)
            os.killpg(group_id, signal.SIGCONT)

    with _signals.suspendable(suspend_proc, resume_proc), _signals.terminator(kill_proc):
        process = subprocess.Popen(  # pylint: disable=subprocess-popen-preexec-fn
            *popenargs, preexec_fn=preexec_fn, universal_newlines=True, **kwargs)
        output, _ = process.communicate()
        exit_code = process.poll()

    return (exit_code, output)


# _glob2re()
#
# Function to translate a glob style pattern into a regex
#
# Args:
#    pat (str): The glob pattern
#
# This is a modified version of the python standard library's
# fnmatch.translate() function which supports path like globbing
# a bit more correctly, and additionally supports recursive glob
# patterns with double asterisk.
#
# Note that this will only support the most basic of standard
# glob patterns, and additionally the recursive double asterisk.
#
# Support includes:
#
#   *          Match any pattern except a path separator
#   **         Match any pattern, including path separators
#   ?          Match any single character
#   [abc]      Match one of the specified characters
#   [A-Z]      Match one of the characters in the specified range
#   [!abc]     Match any single character, except the specified characters
#   [!A-Z]     Match any single character, except those in the specified range
#
def _glob2re(pat):
    i, n = 0, len(pat)
    res = '(?ms)'
    while i < n:
        c = pat[i]
        i = i + 1
        if c == '*':
            # fnmatch.translate() simply uses the '.*' separator here,
            # we only want that for double asterisk (bash 'globstar' behavior)
            #
            if i < n and pat[i] == '*':
                res = res + '.*'
                i = i + 1
            else:
                res = res + '[^/]*'
        elif c == '?':
            # fnmatch.translate() simply uses the '.' wildcard here, but
            # we dont want to match path separators here
            res = res + '[^/]'
        elif c == '[':
            j = i
            if j < n and pat[j] == '!':
                j = j + 1
            if j < n and pat[j] == ']':
                j = j + 1
            while j < n and pat[j] != ']':
                j = j + 1
            if j >= n:
                res = res + '\\['
            else:
                stuff = pat[i:j].replace('\\', '\\\\')
                i = j + 1
                if stuff[0] == '!':
                    stuff = '^' + stuff[1:]
                elif stuff[0] == '^':
                    stuff = '\\' + stuff
                res = '{}[{}]'.format(res, stuff)
        else:
            res = res + re.escape(c)
    return res + r'\Z'


# _deduplicate()
#
# Remove duplicate entries in a list or other iterable.
#
# Copied verbatim from the unique_everseen() example at
# https://docs.python.org/3/library/itertools.html#itertools-recipes
#
# Args:
#    iterable (iterable): What to deduplicate
#    key (callable): Optional function to map from list entry to value
#
# Returns:
#    (generator): Generator that produces a deduplicated version of 'iterable'
#
def _deduplicate(iterable, key=None):
    seen = set()
    seen_add = seen.add
    if key is None:
        for element in itertools.filterfalse(seen.__contains__, iterable):
            seen_add(element)
            yield element
    else:
        for element in iterable:
            k = key(element)
            if k not in seen:
                seen_add(k)
                yield element


# Like os.path.getmtime(), but returns the mtime of a link rather than
# the target, if the filesystem supports that.
#
def _get_link_mtime(path):
    path_stat = os.lstat(path)
    return path_stat.st_mtime


# _message_digest()
#
# Args:
#    message_buffer (str): String to create digest of
#
# Returns:
#    (remote_execution_pb2.Digest): Content digest
#
def _message_digest(message_buffer):
    sha = hashlib.sha256(message_buffer)
    digest = remote_execution_pb2.Digest()
    digest.hash = sha.hexdigest()
    digest.size_bytes = len(message_buffer)
    return digest


# _search_upward_for_files()
#
# Searches upwards (from directory, then directory's parent directory...)
# for any of the files listed in `filenames`.
#
# If multiple filenames are specified, and present in the same directory,
# the first filename in the list will be returned.
#
# Args:
#    directory (str): The directory to begin searching for files from
#    filenames (list of str): The names of files to search for
#
# Returns:
#    (str): The directory a file was found in, or None
#    (str): The name of the first file that was found in that directory, or None
#
def _search_upward_for_files(directory, filenames):
    directory = os.path.abspath(directory)
    while True:
        for filename in filenames:
            file_path = os.path.join(directory, filename)
            if os.path.isfile(file_path):
                return directory, filename

        parent_dir = os.path.dirname(directory)
        if directory == parent_dir:
            # i.e. we've reached the root of the filesystem
            return None, None
        directory = parent_dir


# _deterministic_umask()
#
# Context managed to apply a umask to a section that may be affected by a users
# umask. Restores old mask afterwards.
#
@contextmanager
def _deterministic_umask():
    old_umask = os.umask(0o022)

    try:
        yield
    finally:
        os.umask(old_umask)


# _get_compression:
#
# Given a file name infer the compression
#
# Args:
#    tar (str): The file name from which to determine compression
#
# Returns:
#    (str): One from '', 'gz', 'xz', 'bz2'
#
# Raises:
#    UtilError: In the case where an unsupported file extension has been provided,
#               expecting compression.
#
#
def _get_compression(tar):
    mapped_extensions = {'.tar': '', '.gz': 'gz', '.xz': 'xz', '.bz2': 'bz2'}

    name, ext = os.path.splitext(tar)

    try:
        return mapped_extensions[ext]
    except KeyError:
        # If ext not in mapped_extensions, find out if inner ext is .tar
        # If so, we assume we have been given an unsupported extension,
        # which expects compression. Raise an error
        _, suffix = os.path.splitext(name)
        if suffix == '.tar':
            raise UtilError("Expected compression with unknown file extension ('{}'), "
                            "supported extensions are ('.tar'), ('.gz'), ('.xz'), ('.bz2')".format(ext))
        else:
            # Assume just an unconventional name was provided, default to uncompressed
            return ''


# _is_single_threaded()
#
# Return whether the current Process is single-threaded. Don't count threads
# in the main process that were created by a test environment (xdist/execnet)
# before BuildStream was executed.
#
def _is_single_threaded():
    # Use psutil as threading.active_count() doesn't include gRPC threads.
    process = psutil.Process()
    num_threads = process.num_threads()
    if process.pid == _MAIN_PID:
        return num_threads == _INITIAL_NUM_THREADS_IN_MAIN_PROCESS
    else:
        return num_threads == 1