720
767
reader._parse_dirblocks()
721
768
state._dirblock_state = DirState.IN_MEMORY_UNMODIFIED
771
cdef int minikind_from_mode(int mode):
772
# in order of frequency:
782
_encode = binascii.b2a_base64
785
from struct import pack
786
cdef _pack_stat(stat_value):
787
"""return a string representing the stat value's key fields.
789
:param stat_value: A stat oject with st_size, st_mtime, st_ctime, st_dev,
790
st_ino and st_mode fields.
792
cdef char result[6*4] # 6 long ints
794
aliased = <int *>result
795
aliased[0] = htonl(stat_value.st_size)
796
aliased[1] = htonl(int(stat_value.st_mtime))
797
aliased[2] = htonl(int(stat_value.st_ctime))
798
aliased[3] = htonl(stat_value.st_dev)
799
aliased[4] = htonl(stat_value.st_ino & 0xFFFFFFFF)
800
aliased[5] = htonl(stat_value.st_mode)
801
packed = PyString_FromStringAndSize(result, 6*4)
802
return _encode(packed)[:-1]
805
def update_entry(self, entry, abspath, stat_value):
806
"""Update the entry based on what is actually on disk.
808
This function only calculates the sha if it needs to - if the entry is
809
uncachable, or clearly different to the first parent's entry, no sha
810
is calculated, and None is returned.
812
:param entry: This is the dirblock entry for the file in question.
813
:param abspath: The path on disk for this file.
814
:param stat_value: (optional) if we already have done a stat on the
816
:return: None, or The sha1 hexdigest of the file (40 bytes) or link
819
return _update_entry(self, entry, abspath, stat_value)
822
cdef _update_entry(self, entry, abspath, stat_value):
823
"""Update the entry based on what is actually on disk.
825
This function only calculates the sha if it needs to - if the entry is
826
uncachable, or clearly different to the first parent's entry, no sha
827
is calculated, and None is returned.
829
:param self: The dirstate object this is operating on.
830
:param entry: This is the dirblock entry for the file in question.
831
:param abspath: The path on disk for this file.
832
:param stat_value: The stat value done on the path.
833
:return: None, or The sha1 hexdigest of the file (40 bytes) or link
836
# TODO - require pyrex 0.9.8, then use a pyd file to define access to the
837
# _st mode of the compiled stat objects.
838
cdef int minikind, saved_minikind
840
minikind = minikind_from_mode(stat_value.st_mode)
843
packed_stat = _pack_stat(stat_value)
844
details = PyList_GetItem_void_void(PyTuple_GetItem_void_void(<void *>entry, 1), 0)
845
saved_minikind = PyString_AsString_obj(<PyObject *>PyTuple_GetItem_void_void(details, 0))[0]
846
if minikind == c'd' and saved_minikind == c't':
848
saved_link_or_sha1 = PyTuple_GetItem_void_object(details, 1)
849
saved_file_size = PyTuple_GetItem_void_object(details, 2)
850
saved_executable = PyTuple_GetItem_void_object(details, 3)
851
saved_packed_stat = PyTuple_GetItem_void_object(details, 4)
852
# Deal with pyrex decrefing the objects
853
Py_INCREF(saved_link_or_sha1)
854
Py_INCREF(saved_file_size)
855
Py_INCREF(saved_executable)
856
Py_INCREF(saved_packed_stat)
857
#(saved_minikind, saved_link_or_sha1, saved_file_size,
858
# saved_executable, saved_packed_stat) = entry[1][0]
860
if (minikind == saved_minikind
861
and packed_stat == saved_packed_stat):
862
# The stat hasn't changed since we saved, so we can re-use the
867
# size should also be in packed_stat
868
if saved_file_size == stat_value.st_size:
869
return saved_link_or_sha1
871
# If we have gotten this far, that means that we need to actually
872
# process this entry.
875
executable = self._is_executable(stat_value.st_mode,
877
if self._cutoff_time is None:
878
self._sha_cutoff_time()
879
if (stat_value.st_mtime < self._cutoff_time
880
and stat_value.st_ctime < self._cutoff_time
881
and len(entry[1]) > 1
882
and entry[1][1][0] != 'a'):
883
# Could check for size changes for further optimised
884
# avoidance of sha1's. However the most prominent case of
885
# over-shaing is during initial add, which this catches.
886
link_or_sha1 = self._sha1_file(abspath)
887
entry[1][0] = ('f', link_or_sha1, stat_value.st_size,
888
executable, packed_stat)
890
entry[1][0] = ('f', '', stat_value.st_size,
891
executable, DirState.NULLSTAT)
892
elif minikind == c'd':
894
entry[1][0] = ('d', '', 0, False, packed_stat)
895
if saved_minikind != c'd':
896
# This changed from something into a directory. Make sure we
897
# have a directory block for it. This doesn't happen very
898
# often, so this doesn't have to be super fast.
899
block_index, entry_index, dir_present, file_present = \
900
self._get_block_entry_index(entry[0][0], entry[0][1], 0)
901
self._ensure_block(block_index, entry_index,
902
pathjoin(entry[0][0], entry[0][1]))
903
elif minikind == c'l':
904
link_or_sha1 = self._read_link(abspath, saved_link_or_sha1)
905
if self._cutoff_time is None:
906
self._sha_cutoff_time()
907
if (stat_value.st_mtime < self._cutoff_time
908
and stat_value.st_ctime < self._cutoff_time):
909
entry[1][0] = ('l', link_or_sha1, stat_value.st_size,
912
entry[1][0] = ('l', '', stat_value.st_size,
913
False, DirState.NULLSTAT)
914
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
918
cdef char _minikind_from_string(object string):
919
"""Convert a python string to a char."""
920
return PyString_AsString(string)[0]
923
cdef object _kind_absent
924
cdef object _kind_file
925
cdef object _kind_directory
926
cdef object _kind_symlink
927
cdef object _kind_relocated
928
cdef object _kind_tree_reference
929
_kind_absent = "absent"
931
_kind_directory = "directory"
932
_kind_symlink = "symlink"
933
_kind_relocated = "relocated"
934
_kind_tree_reference = "tree-reference"
937
cdef object _minikind_to_kind(char minikind):
938
"""Create a string kind for minikind."""
939
cdef char _minikind[1]
942
elif minikind == c'd':
943
return _kind_directory
944
elif minikind == c'a':
946
elif minikind == c'r':
947
return _kind_relocated
948
elif minikind == c'l':
950
elif minikind == c't':
951
return _kind_tree_reference
952
_minikind[0] = minikind
953
raise KeyError(PyString_FromStringAndSize(_minikind, 1))
956
cdef int _versioned_minikind(char minikind):
957
"""Return non-zero if minikind is in fltd"""
958
return (minikind == c'f' or
964
cdef class ProcessEntryC:
966
cdef object old_dirname_to_file_id # dict
967
cdef object new_dirname_to_file_id # dict
968
cdef object last_source_parent
969
cdef object last_target_parent
970
cdef object include_unchanged
971
cdef object use_filesystem_for_exec
972
cdef object utf8_decode
973
cdef readonly object searched_specific_files
974
cdef object search_specific_files
976
# Current iteration variables:
977
cdef object current_root
978
cdef object current_root_unicode
979
cdef object root_entries
980
cdef int root_entries_pos, root_entries_len
981
cdef object root_abspath
982
cdef int source_index, target_index
983
cdef int want_unversioned
985
cdef object dir_iterator
987
cdef object current_block
988
cdef int current_block_pos
989
cdef object current_block_list
990
cdef object current_dir_info
991
cdef object current_dir_list
993
cdef object root_dir_info
994
cdef object bisect_left
999
def __init__(self, include_unchanged, use_filesystem_for_exec,
1000
search_specific_files, state, source_index, target_index,
1001
want_unversioned, tree):
1002
self.old_dirname_to_file_id = {}
1003
self.new_dirname_to_file_id = {}
1004
# Using a list so that we can access the values and change them in
1005
# nested scope. Each one is [path, file_id, entry]
1006
self.last_source_parent = [None, None]
1007
self.last_target_parent = [None, None]
1008
self.include_unchanged = include_unchanged
1009
self.use_filesystem_for_exec = use_filesystem_for_exec
1010
self.utf8_decode = cache_utf8._utf8_decode
1011
# for all search_indexs in each path at or under each element of
1012
# search_specific_files, if the detail is relocated: add the id, and add the
1013
# relocated path as one to search if its not searched already. If the
1014
# detail is not relocated, add the id.
1015
self.searched_specific_files = set()
1016
self.search_specific_files = search_specific_files
1018
self.current_root = None
1019
self.current_root_unicode = None
1020
self.root_entries = None
1021
self.root_entries_pos = 0
1022
self.root_entries_len = 0
1023
self.root_abspath = None
1024
if source_index is None:
1025
self.source_index = -1
1027
self.source_index = source_index
1028
self.target_index = target_index
1029
self.want_unversioned = want_unversioned
1031
self.dir_iterator = None
1032
self.block_index = -1
1033
self.current_block = None
1034
self.current_block_list = None
1035
self.current_block_pos = -1
1036
self.current_dir_info = None
1037
self.current_dir_list = None
1039
self.root_dir_info = None
1040
self.bisect_left = bisect.bisect_left
1041
self.pathjoin = osutils.pathjoin
1042
self.fstat = os.fstat
1043
self.sha_file = osutils.sha_file
1045
cdef _process_entry(self, entry, path_info):
1046
"""Compare an entry and real disk to generate delta information.
1048
:param path_info: top_relpath, basename, kind, lstat, abspath for
1049
the path of entry. If None, then the path is considered absent.
1050
(Perhaps we should pass in a concrete entry for this ?)
1051
Basename is returned as a utf8 string because we expect this
1052
tuple will be ignored, and don't want to take the time to
1054
:return: (iter_changes_result, changed). If the entry has not been
1055
handled then changed is None. Otherwise it is False if no content
1056
or metadata changes have occured, and None if any content or
1057
metadata change has occured. If self.include_unchanged is True then
1058
if changed is not None, iter_changes_result will always be a result
1059
tuple. Otherwise, iter_changes_result is None unless changed is
1062
cdef char target_minikind
1063
cdef char source_minikind
1065
cdef int content_change
1066
cdef object details_list
1068
details_list = entry[1]
1069
if -1 == self.source_index:
1070
source_details = DirState.NULL_PARENT_DETAILS
1072
source_details = details_list[self.source_index]
1073
target_details = details_list[self.target_index]
1074
target_minikind = _minikind_from_string(target_details[0])
1075
if path_info is not None and _versioned_minikind(target_minikind):
1076
if self.target_index != 0:
1077
raise AssertionError("Unsupported target index %d" %
1079
link_or_sha1 = _update_entry(self.state, entry, path_info[4], path_info[3])
1080
# The entry may have been modified by update_entry
1081
target_details = details_list[self.target_index]
1082
target_minikind = _minikind_from_string(target_details[0])
1085
# the rest of this function is 0.3 seconds on 50K paths, or
1086
# 0.000006 seconds per call.
1087
source_minikind = _minikind_from_string(source_details[0])
1088
if ((_versioned_minikind(source_minikind) or source_minikind == c'r')
1089
and _versioned_minikind(target_minikind)):
1090
# claimed content in both: diff
1091
# r | fdlt | | add source to search, add id path move and perform
1092
# | | | diff check on source-target
1093
# r | fdlt | a | dangling file that was present in the basis.
1095
if source_minikind != c'r':
1096
old_dirname = entry[0][0]
1097
old_basename = entry[0][1]
1098
old_path = path = None
1100
# add the source to the search path to find any children it
1101
# has. TODO ? : only add if it is a container ?
1102
if not osutils.is_inside_any(self.searched_specific_files,
1104
self.search_specific_files.add(source_details[1])
1105
# generate the old path; this is needed for stating later
1107
old_path = source_details[1]
1108
old_dirname, old_basename = os.path.split(old_path)
1109
path = self.pathjoin(entry[0][0], entry[0][1])
1110
old_entry = self.state._get_entry(self.source_index,
1112
# update the source details variable to be the real
1114
if old_entry == (None, None):
1115
raise errors.CorruptDirstate(self.state._filename,
1116
"entry '%s/%s' is considered renamed from %r"
1117
" but source does not exist\n"
1118
"entry: %s" % (entry[0][0], entry[0][1], old_path, entry))
1119
source_details = old_entry[1][self.source_index]
1120
source_minikind = _minikind_from_string(source_details[0])
1121
if path_info is None:
1122
# the file is missing on disk, show as removed.
1127
# source and target are both versioned and disk file is present.
1128
target_kind = path_info[2]
1129
if target_kind == 'directory':
1131
old_path = path = self.pathjoin(old_dirname, old_basename)
1132
file_id = entry[0][2]
1133
self.new_dirname_to_file_id[path] = file_id
1134
if source_minikind != c'd':
1137
# directories have no fingerprint
1140
elif target_kind == 'file':
1141
if source_minikind != c'f':
1144
# Check the sha. We can't just rely on the size as
1145
# content filtering may mean differ sizes actually
1146
# map to the same content
1147
if link_or_sha1 is None:
1149
statvalue, link_or_sha1 = \
1150
self.state._sha1_provider.stat_and_sha1(
1152
self.state._observed_sha1(entry, link_or_sha1,
1154
content_change = (link_or_sha1 != source_details[1])
1155
# Target details is updated at update_entry time
1156
if self.use_filesystem_for_exec:
1157
# We don't need S_ISREG here, because we are sure
1158
# we are dealing with a file.
1159
target_exec = bool(S_IXUSR & path_info[3].st_mode)
1161
target_exec = target_details[3]
1162
elif target_kind == 'symlink':
1163
if source_minikind != c'l':
1166
content_change = (link_or_sha1 != source_details[1])
1168
elif target_kind == 'tree-reference':
1169
if source_minikind != c't':
1175
raise Exception, "unknown kind %s" % path_info[2]
1176
if source_minikind == c'd':
1178
old_path = path = self.pathjoin(old_dirname, old_basename)
1180
file_id = entry[0][2]
1181
self.old_dirname_to_file_id[old_path] = file_id
1182
# parent id is the entry for the path in the target tree
1183
if old_dirname == self.last_source_parent[0]:
1184
source_parent_id = self.last_source_parent[1]
1187
source_parent_id = self.old_dirname_to_file_id[old_dirname]
1189
source_parent_entry = self.state._get_entry(self.source_index,
1190
path_utf8=old_dirname)
1191
source_parent_id = source_parent_entry[0][2]
1192
if source_parent_id == entry[0][2]:
1193
# This is the root, so the parent is None
1194
source_parent_id = None
1196
self.last_source_parent[0] = old_dirname
1197
self.last_source_parent[1] = source_parent_id
1198
new_dirname = entry[0][0]
1199
if new_dirname == self.last_target_parent[0]:
1200
target_parent_id = self.last_target_parent[1]
1203
target_parent_id = self.new_dirname_to_file_id[new_dirname]
1205
# TODO: We don't always need to do the lookup, because the
1206
# parent entry will be the same as the source entry.
1207
target_parent_entry = self.state._get_entry(self.target_index,
1208
path_utf8=new_dirname)
1209
if target_parent_entry == (None, None):
1210
raise AssertionError(
1211
"Could not find target parent in wt: %s\nparent of: %s"
1212
% (new_dirname, entry))
1213
target_parent_id = target_parent_entry[0][2]
1214
if target_parent_id == entry[0][2]:
1215
# This is the root, so the parent is None
1216
target_parent_id = None
1218
self.last_target_parent[0] = new_dirname
1219
self.last_target_parent[1] = target_parent_id
1221
source_exec = source_details[3]
1222
changed = (content_change
1223
or source_parent_id != target_parent_id
1224
or old_basename != entry[0][1]
1225
or source_exec != target_exec
1227
if not changed and not self.include_unchanged:
1230
if old_path is None:
1231
path = self.pathjoin(old_dirname, old_basename)
1233
old_path_u = self.utf8_decode(old_path)[0]
1236
old_path_u = self.utf8_decode(old_path)[0]
1237
if old_path == path:
1240
path_u = self.utf8_decode(path)[0]
1241
source_kind = _minikind_to_kind(source_minikind)
1242
return (entry[0][2],
1243
(old_path_u, path_u),
1246
(source_parent_id, target_parent_id),
1247
(self.utf8_decode(old_basename)[0], self.utf8_decode(entry[0][1])[0]),
1248
(source_kind, target_kind),
1249
(source_exec, target_exec)), changed
1250
elif source_minikind == c'a' and _versioned_minikind(target_minikind):
1251
# looks like a new file
1252
path = self.pathjoin(entry[0][0], entry[0][1])
1253
# parent id is the entry for the path in the target tree
1254
# TODO: these are the same for an entire directory: cache em.
1255
parent_entry = self.state._get_entry(self.target_index,
1256
path_utf8=entry[0][0])
1257
if parent_entry is None:
1258
raise errors.DirstateCorrupt(self.state,
1259
"We could not find the parent entry in index %d"
1260
" for the entry: %s"
1261
% (self.target_index, entry[0]))
1262
parent_id = parent_entry[0][2]
1263
if parent_id == entry[0][2]:
1265
if path_info is not None:
1267
if self.use_filesystem_for_exec:
1268
# We need S_ISREG here, because we aren't sure if this
1271
S_ISREG(path_info[3].st_mode)
1272
and S_IXUSR & path_info[3].st_mode)
1274
target_exec = target_details[3]
1275
return (entry[0][2],
1276
(None, self.utf8_decode(path)[0]),
1280
(None, self.utf8_decode(entry[0][1])[0]),
1281
(None, path_info[2]),
1282
(None, target_exec)), True
1284
# Its a missing file, report it as such.
1285
return (entry[0][2],
1286
(None, self.utf8_decode(path)[0]),
1290
(None, self.utf8_decode(entry[0][1])[0]),
1292
(None, False)), True
1293
elif _versioned_minikind(source_minikind) and target_minikind == c'a':
1294
# unversioned, possibly, or possibly not deleted: we dont care.
1295
# if its still on disk, *and* theres no other entry at this
1296
# path [we dont know this in this routine at the moment -
1297
# perhaps we should change this - then it would be an unknown.
1298
old_path = self.pathjoin(entry[0][0], entry[0][1])
1299
# parent id is the entry for the path in the target tree
1300
parent_id = self.state._get_entry(self.source_index, path_utf8=entry[0][0])[0][2]
1301
if parent_id == entry[0][2]:
1303
return (entry[0][2],
1304
(self.utf8_decode(old_path)[0], None),
1308
(self.utf8_decode(entry[0][1])[0], None),
1309
(_minikind_to_kind(source_minikind), None),
1310
(source_details[3], None)), True
1311
elif _versioned_minikind(source_minikind) and target_minikind == c'r':
1312
# a rename; could be a true rename, or a rename inherited from
1313
# a renamed parent. TODO: handle this efficiently. Its not
1314
# common case to rename dirs though, so a correct but slow
1315
# implementation will do.
1316
if not osutils.is_inside_any(self.searched_specific_files, target_details[1]):
1317
self.search_specific_files.add(target_details[1])
1318
elif ((source_minikind == c'r' or source_minikind == c'a') and
1319
(target_minikind == c'r' or target_minikind == c'a')):
1320
# neither of the selected trees contain this path,
1321
# so skip over it. This is not currently directly tested, but
1322
# is indirectly via test_too_much.TestCommands.test_conflicts.
1325
raise AssertionError("don't know how to compare "
1326
"source_minikind=%r, target_minikind=%r"
1327
% (source_minikind, target_minikind))
1328
## import pdb;pdb.set_trace()
1334
def iter_changes(self):
1337
cdef void _update_current_block(self):
1338
if (self.block_index < len(self.state._dirblocks) and
1339
osutils.is_inside(self.current_root, self.state._dirblocks[self.block_index][0])):
1340
self.current_block = self.state._dirblocks[self.block_index]
1341
self.current_block_list = self.current_block[1]
1342
self.current_block_pos = 0
1344
self.current_block = None
1345
self.current_block_list = None
1348
# Simple thunk to allow tail recursion without pyrex confusion
1349
return self._iter_next()
1351
cdef _iter_next(self):
1352
"""Iterate over the changes."""
1353
# This function single steps through an iterator. As such while loops
1354
# are often exited by 'return' - the code is structured so that the
1355
# next call into the function will return to the same while loop. Note
1356
# that all flow control needed to re-reach that step is reexecuted,
1357
# which can be a performance problem. It has not yet been tuned to
1358
# minimise this; a state machine is probably the simplest restructuring
1359
# to both minimise this overhead and make the code considerably more
1363
# compare source_index and target_index at or under each element of search_specific_files.
1364
# follow the following comparison table. Note that we only want to do diff operations when
1365
# the target is fdl because thats when the walkdirs logic will have exposed the pathinfo
1369
# Source | Target | disk | action
1370
# r | fdlt | | add source to search, add id path move and perform
1371
# | | | diff check on source-target
1372
# r | fdlt | a | dangling file that was present in the basis.
1374
# r | a | | add source to search
1376
# r | r | | this path is present in a non-examined tree, skip.
1377
# r | r | a | this path is present in a non-examined tree, skip.
1378
# a | fdlt | | add new id
1379
# a | fdlt | a | dangling locally added file, skip
1380
# a | a | | not present in either tree, skip
1381
# a | a | a | not present in any tree, skip
1382
# a | r | | not present in either tree at this path, skip as it
1383
# | | | may not be selected by the users list of paths.
1384
# a | r | a | not present in either tree at this path, skip as it
1385
# | | | may not be selected by the users list of paths.
1386
# fdlt | fdlt | | content in both: diff them
1387
# fdlt | fdlt | a | deleted locally, but not unversioned - show as deleted ?
1388
# fdlt | a | | unversioned: output deleted id for now
1389
# fdlt | a | a | unversioned and deleted: output deleted id
1390
# fdlt | r | | relocated in this tree, so add target to search.
1391
# | | | Dont diff, we will see an r,fd; pair when we reach
1392
# | | | this id at the other path.
1393
# fdlt | r | a | relocated in this tree, so add target to search.
1394
# | | | Dont diff, we will see an r,fd; pair when we reach
1395
# | | | this id at the other path.
1397
# TODO: jam 20070516 - Avoid the _get_entry lookup overhead by
1398
# keeping a cache of directories that we have seen.
1399
cdef object current_dirname, current_blockname
1400
cdef char * current_dirname_c, * current_blockname_c
1401
cdef int advance_entry, advance_path
1402
cdef int path_handled
1403
searched_specific_files = self.searched_specific_files
1404
# Are we walking a root?
1405
while self.root_entries_pos < self.root_entries_len:
1406
entry = self.root_entries[self.root_entries_pos]
1407
self.root_entries_pos = self.root_entries_pos + 1
1408
result, changed = self._process_entry(entry, self.root_dir_info)
1409
if changed is not None and changed or self.include_unchanged:
1411
# Have we finished the prior root, or never started one ?
1412
if self.current_root is None:
1413
# TODO: the pending list should be lexically sorted? the
1414
# interface doesn't require it.
1416
self.current_root = self.search_specific_files.pop()
1418
raise StopIteration()
1419
self.current_root_unicode = self.current_root.decode('utf8')
1420
self.searched_specific_files.add(self.current_root)
1421
# process the entries for this containing directory: the rest will be
1422
# found by their parents recursively.
1423
self.root_entries = self.state._entries_for_path(self.current_root)
1424
self.root_entries_len = len(self.root_entries)
1425
self.root_abspath = self.tree.abspath(self.current_root_unicode)
1427
root_stat = os.lstat(self.root_abspath)
1429
if e.errno == errno.ENOENT:
1430
# the path does not exist: let _process_entry know that.
1431
self.root_dir_info = None
1433
# some other random error: hand it up.
1436
self.root_dir_info = ('', self.current_root,
1437
osutils.file_kind_from_stat_mode(root_stat.st_mode), root_stat,
1439
if self.root_dir_info[2] == 'directory':
1440
if self.tree._directory_is_tree_reference(
1441
self.current_root_unicode):
1442
self.root_dir_info = self.root_dir_info[:2] + \
1443
('tree-reference',) + self.root_dir_info[3:]
1444
if not self.root_entries and not self.root_dir_info:
1445
# this specified path is not present at all, skip it.
1446
# (tail recursion, can do a loop once the full structure is
1448
return self._iter_next()
1450
self.root_entries_pos = 0
1451
# XXX Clarity: This loop is duplicated a out the self.current_root
1452
# is None guard above: if we return from it, it completes there
1453
# (and the following if block cannot trigger because
1454
# path_handled must be true, so the if block is not # duplicated.
1455
while self.root_entries_pos < self.root_entries_len:
1456
entry = self.root_entries[self.root_entries_pos]
1457
self.root_entries_pos = self.root_entries_pos + 1
1458
result, changed = self._process_entry(entry, self.root_dir_info)
1459
if changed is not None:
1461
if changed or self.include_unchanged:
1463
# handle unversioned specified paths:
1464
if self.want_unversioned and not path_handled and self.root_dir_info:
1465
new_executable = bool(
1466
stat.S_ISREG(self.root_dir_info[3].st_mode)
1467
and stat.S_IEXEC & self.root_dir_info[3].st_mode)
1469
(None, self.current_root_unicode),
1473
(None, splitpath(self.current_root_unicode)[-1]),
1474
(None, self.root_dir_info[2]),
1475
(None, new_executable)
1477
# If we reach here, the outer flow continues, which enters into the
1478
# per-root setup logic.
1479
if self.current_dir_info is None and self.current_block is None:
1480
# setup iteration of this root:
1481
self.current_dir_list = None
1482
if self.root_dir_info and self.root_dir_info[2] == 'tree-reference':
1483
self.current_dir_info = None
1485
self.dir_iterator = osutils._walkdirs_utf8(self.root_abspath,
1486
prefix=self.current_root)
1489
self.current_dir_info = self.dir_iterator.next()
1490
self.current_dir_list = self.current_dir_info[1]
1492
# there may be directories in the inventory even though
1493
# this path is not a file on disk: so mark it as end of
1495
if e.errno in (errno.ENOENT, errno.ENOTDIR, errno.EINVAL):
1496
self.current_dir_info = None
1497
elif sys.platform == 'win32':
1498
# on win32, python2.4 has e.errno == ERROR_DIRECTORY, but
1499
# python 2.5 has e.errno == EINVAL,
1500
# and e.winerror == ERROR_DIRECTORY
1502
e_winerror = e.winerror
1503
except AttributeError:
1505
win_errors = (ERROR_DIRECTORY, ERROR_PATH_NOT_FOUND)
1506
if (e.errno in win_errors or e_winerror in win_errors):
1507
self.current_dir_info = None
1509
# Will this really raise the right exception ?
1514
if self.current_dir_info[0][0] == '':
1515
# remove .bzr from iteration
1516
bzr_index = self.bisect_left(self.current_dir_list, ('.bzr',))
1517
if self.current_dir_list[bzr_index][0] != '.bzr':
1518
raise AssertionError()
1519
del self.current_dir_list[bzr_index]
1520
initial_key = (self.current_root, '', '')
1521
self.block_index, _ = self.state._find_block_index_from_key(initial_key)
1522
if self.block_index == 0:
1523
# we have processed the total root already, but because the
1524
# initial key matched it we should skip it here.
1525
self.block_index = self.block_index + 1
1526
self._update_current_block()
1527
# walk until both the directory listing and the versioned metadata
1529
while (self.current_dir_info is not None
1530
or self.current_block is not None):
1531
# Uncommon case - a missing directory or an unversioned directory:
1532
if (self.current_dir_info and self.current_block
1533
and self.current_dir_info[0][0] != self.current_block[0]):
1534
# Work around pyrex broken heuristic - current_dirname has
1535
# the same scope as current_dirname_c
1536
current_dirname = self.current_dir_info[0][0]
1537
current_dirname_c = PyString_AS_STRING_void(
1538
<void *>current_dirname)
1539
current_blockname = self.current_block[0]
1540
current_blockname_c = PyString_AS_STRING_void(
1541
<void *>current_blockname)
1542
# In the python generator we evaluate this if block once per
1543
# dir+block; because we reenter in the pyrex version its being
1544
# evaluated once per path: we could cache the result before
1545
# doing the while loop and probably save time.
1546
if _cmp_by_dirs(current_dirname_c,
1547
PyString_Size(current_dirname),
1548
current_blockname_c,
1549
PyString_Size(current_blockname)) < 0:
1550
# filesystem data refers to paths not covered by the
1551
# dirblock. this has two possibilities:
1552
# A) it is versioned but empty, so there is no block for it
1553
# B) it is not versioned.
1555
# if (A) then we need to recurse into it to check for
1556
# new unknown files or directories.
1557
# if (B) then we should ignore it, because we don't
1558
# recurse into unknown directories.
1559
# We are doing a loop
1560
while self.path_index < len(self.current_dir_list):
1561
current_path_info = self.current_dir_list[self.path_index]
1562
# dont descend into this unversioned path if it is
1564
if current_path_info[2] in ('directory',
1566
del self.current_dir_list[self.path_index]
1567
self.path_index = self.path_index - 1
1568
self.path_index = self.path_index + 1
1569
if self.want_unversioned:
1570
if current_path_info[2] == 'directory':
1571
if self.tree._directory_is_tree_reference(
1572
self.utf8_decode(current_path_info[0])[0]):
1573
current_path_info = current_path_info[:2] + \
1574
('tree-reference',) + current_path_info[3:]
1575
new_executable = bool(
1576
stat.S_ISREG(current_path_info[3].st_mode)
1577
and stat.S_IEXEC & current_path_info[3].st_mode)
1579
(None, self.utf8_decode(current_path_info[0])[0]),
1583
(None, self.utf8_decode(current_path_info[1])[0]),
1584
(None, current_path_info[2]),
1585
(None, new_executable))
1586
# This dir info has been handled, go to the next
1588
self.current_dir_list = None
1590
self.current_dir_info = self.dir_iterator.next()
1591
self.current_dir_list = self.current_dir_info[1]
1592
except StopIteration:
1593
self.current_dir_info = None
1595
# We have a dirblock entry for this location, but there
1596
# is no filesystem path for this. This is most likely
1597
# because a directory was removed from the disk.
1598
# We don't have to report the missing directory,
1599
# because that should have already been handled, but we
1600
# need to handle all of the files that are contained
1602
while self.current_block_pos < len(self.current_block_list):
1603
current_entry = self.current_block_list[self.current_block_pos]
1604
self.current_block_pos = self.current_block_pos + 1
1605
# entry referring to file not present on disk.
1606
# advance the entry only, after processing.
1607
result, changed = self._process_entry(current_entry, None)
1608
if changed is not None:
1609
if changed or self.include_unchanged:
1611
self.block_index = self.block_index + 1
1612
self._update_current_block()
1613
continue # next loop-on-block/dir
1614
result = self._loop_one_block()
1615
if result is not None:
1617
if len(self.search_specific_files):
1618
# More supplied paths to process
1619
self.current_root = None
1620
return self._iter_next()
1621
raise StopIteration()
1623
cdef object _maybe_tree_ref(self, current_path_info):
1624
if self.tree._directory_is_tree_reference(
1625
self.utf8_decode(current_path_info[0])[0]):
1626
return current_path_info[:2] + \
1627
('tree-reference',) + current_path_info[3:]
1629
return current_path_info
1631
cdef object _loop_one_block(self):
1632
# current_dir_info and current_block refer to the same directory -
1633
# this is the common case code.
1634
# Assign local variables for current path and entry:
1635
cdef object current_entry
1636
cdef object current_path_info
1637
cdef int path_handled
1640
# cdef char * temp_str
1641
# cdef Py_ssize_t temp_str_length
1642
# PyString_AsStringAndSize(disk_kind, &temp_str, &temp_str_length)
1643
# if not strncmp(temp_str, "directory", temp_str_length):
1644
if (self.current_block is not None and
1645
self.current_block_pos < PyList_GET_SIZE(self.current_block_list)):
1646
current_entry = PyList_GET_ITEM(self.current_block_list,
1647
self.current_block_pos)
1649
Py_INCREF(current_entry)
1651
current_entry = None
1652
if (self.current_dir_info is not None and
1653
self.path_index < PyList_GET_SIZE(self.current_dir_list)):
1654
current_path_info = PyList_GET_ITEM(self.current_dir_list,
1657
Py_INCREF(current_path_info)
1658
disk_kind = PyTuple_GET_ITEM(current_path_info, 2)
1660
Py_INCREF(disk_kind)
1661
if disk_kind == "directory":
1662
current_path_info = self._maybe_tree_ref(current_path_info)
1664
current_path_info = None
1665
while (current_entry is not None or current_path_info is not None):
1670
if current_entry is None:
1671
# unversioned - the check for path_handled when the path
1672
# is advanced will yield this path if needed.
1674
elif current_path_info is None:
1675
# no path is fine: the per entry code will handle it.
1676
result, changed = self._process_entry(current_entry,
1679
minikind = _minikind_from_string(
1680
current_entry[1][self.target_index][0])
1681
cmp_result = cmp(current_path_info[1], current_entry[0][1])
1682
if (cmp_result or minikind == c'a' or minikind == c'r'):
1683
# The current path on disk doesn't match the dirblock
1684
# record. Either the dirblock record is marked as
1685
# absent/renamed, or the file on disk is not present at all
1686
# in the dirblock. Either way, report about the dirblock
1687
# entry, and let other code handle the filesystem one.
1689
# Compare the basename for these files to determine
1692
# extra file on disk: pass for now, but only
1693
# increment the path, not the entry
1696
# entry referring to file not present on disk.
1697
# advance the entry only, after processing.
1698
result, changed = self._process_entry(current_entry,
1702
# paths are the same,and the dirstate entry is not
1703
# absent or renamed.
1704
result, changed = self._process_entry(current_entry,
1706
if changed is not None:
1708
# >- loop control starts here:
1710
if advance_entry and current_entry is not None:
1711
self.current_block_pos = self.current_block_pos + 1
1712
if self.current_block_pos < PyList_GET_SIZE(self.current_block_list):
1713
current_entry = self.current_block_list[self.current_block_pos]
1715
current_entry = None
1717
if advance_path and current_path_info is not None:
1718
if not path_handled:
1719
# unversioned in all regards
1720
if self.want_unversioned:
1721
new_executable = bool(
1722
stat.S_ISREG(current_path_info[3].st_mode)
1723
and stat.S_IEXEC & current_path_info[3].st_mode)
1725
relpath_unicode = self.utf8_decode(current_path_info[0])[0]
1726
except UnicodeDecodeError:
1727
raise errors.BadFilenameEncoding(
1728
current_path_info[0], osutils._fs_enc)
1729
if result is not None:
1730
raise AssertionError(
1731
"result is not None: %r" % result)
1733
(None, relpath_unicode),
1737
(None, self.utf8_decode(current_path_info[1])[0]),
1738
(None, current_path_info[2]),
1739
(None, new_executable))
1740
# dont descend into this unversioned path if it is
1742
if current_path_info[2] in ('directory'):
1743
del self.current_dir_list[self.path_index]
1744
self.path_index = self.path_index - 1
1745
# dont descend the disk iterator into any tree
1747
if current_path_info[2] == 'tree-reference':
1748
del self.current_dir_list[self.path_index]
1749
self.path_index = self.path_index - 1
1750
self.path_index = self.path_index + 1
1751
if self.path_index < len(self.current_dir_list):
1752
current_path_info = self.current_dir_list[self.path_index]
1753
if current_path_info[2] == 'directory':
1754
current_path_info = self._maybe_tree_ref(
1757
current_path_info = None
1758
if result is not None:
1759
# Found a result on this pass, yield it
1761
if self.current_block is not None:
1762
self.block_index = self.block_index + 1
1763
self._update_current_block()
1764
if self.current_dir_info is not None:
1766
self.current_dir_list = None
1768
self.current_dir_info = self.dir_iterator.next()
1769
self.current_dir_list = self.current_dir_info[1]
1770
except StopIteration:
1771
self.current_dir_info = None
added
removed
bzrlib/_dirstate_helpers_pyx.pyx
