1
# Copyright (C) 2007, 2008 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
"""Helper functions for DirState.
19
This is the python implementation for DirState functions.
28
from bzrlib import cache_utf8, errors, osutils
29
from bzrlib.dirstate import DirState
30
from bzrlib.osutils import pathjoin, splitpath
33
# This is the Windows equivalent of ENOTDIR
34
# It is defined in pywin32.winerror, but we don't want a strong dependency for
36
# XXX: Perhaps we could get it from a windows header ?
37
cdef int ERROR_PATH_NOT_FOUND
38
ERROR_PATH_NOT_FOUND = 3
39
cdef int ERROR_DIRECTORY
42
#python2.4 support, and other platform-dependent includes
43
cdef extern from "python-compat.h":
44
unsigned long htonl(unsigned long)
46
# Give Pyrex some function definitions for it to understand.
47
# All of these are just hints to Pyrex, so that it can try to convert python
48
# objects into similar C objects. (such as PyInt => int).
49
# In anything defined 'cdef extern from XXX' the real C header will be
50
# imported, and the real definition will be used from there. So these are just
51
# hints, and do not need to match exactly to the C definitions.
54
ctypedef unsigned long size_t
56
cdef extern from "_dirstate_helpers_c.h":
61
cdef extern from "stdlib.h":
62
unsigned long int strtoul(char *nptr, char **endptr, int base)
65
cdef extern from 'sys/stat.h':
68
# On win32, this actually comes from "python-compat.h"
72
# These functions allow us access to a bit of the 'bare metal' of python
73
# objects, rather than going through the object abstraction. (For example,
74
# PyList_Append, rather than getting the 'append' attribute of the object, and
75
# creating a tuple, and then using PyCallObject).
76
# Functions that return (or take) a void* are meant to grab a C PyObject*. This
77
# differs from the Pyrex 'object'. If you declare a variable as 'object' Pyrex
78
# will automatically Py_INCREF and Py_DECREF when appropriate. But for some
79
# inner loops, we don't need to do that at all, as the reference only lasts for
81
# Note that the C API GetItem calls borrow references, so pyrex does the wrong
82
# thing if you declare e.g. object PyList_GetItem(object lst, int index) - you
83
# need to manually Py_INCREF yourself.
84
cdef extern from "Python.h":
85
ctypedef int Py_ssize_t
86
ctypedef struct PyObject:
88
int PyList_Append(object lst, object item) except -1
89
void *PyList_GetItem_object_void "PyList_GET_ITEM" (object lst, int index)
90
void *PyList_GetItem_void_void "PyList_GET_ITEM" (void * lst, int index)
91
object PyList_GET_ITEM(object lst, Py_ssize_t index)
92
int PyList_CheckExact(object)
93
Py_ssize_t PyList_GET_SIZE (object p)
95
void *PyTuple_GetItem_void_void "PyTuple_GET_ITEM" (void* tpl, int index)
96
object PyTuple_GetItem_void_object "PyTuple_GET_ITEM" (void* tpl, int index)
97
object PyTuple_GET_ITEM(object tpl, Py_ssize_t index)
100
char *PyString_AsString(object p)
101
char *PyString_AsString_obj "PyString_AsString" (PyObject *string)
102
char *PyString_AS_STRING_void "PyString_AS_STRING" (void *p)
103
int PyString_AsStringAndSize(object str, char **buffer, Py_ssize_t *length) except -1
104
object PyString_FromString(char *)
105
object PyString_FromStringAndSize(char *, Py_ssize_t)
106
int PyString_Size(object p)
107
int PyString_GET_SIZE_void "PyString_GET_SIZE" (void *p)
108
int PyString_CheckExact(object p)
109
void Py_INCREF(object o)
110
void Py_DECREF(object o)
113
cdef extern from "string.h":
114
int strncmp(char *s1, char *s2, int len)
115
void *memchr(void *s, int c, size_t len)
116
int memcmp(void *b1, void *b2, size_t len)
117
# ??? memrchr is a GNU extension :(
118
# void *memrchr(void *s, int c, size_t len)
121
cdef void* _my_memrchr(void *s, int c, size_t n):
122
# memrchr seems to be a GNU extension, so we have to implement it ourselves
135
def _py_memrchr(s, c):
136
"""Just to expose _my_memrchr for testing.
138
:param s: The Python string to search
139
:param c: The character to search for
140
:return: The offset to the last instance of 'c' in s
147
_s = PyString_AsString(s)
148
length = PyString_Size(s)
150
_c = PyString_AsString(c)
151
assert PyString_Size(c) == 1,\
152
'Must be a single character string, not %s' % (c,)
153
found = _my_memrchr(_s, _c[0], length)
156
return <char*>found - <char*>_s
158
cdef object safe_string_from_size(char *s, Py_ssize_t size):
160
# XXX: On 64-bit machines the <int> cast causes a C compiler warning.
161
raise AssertionError(
162
'tried to create a string with an invalid size: %d @0x%x'
164
return PyString_FromStringAndSize(s, size)
167
cdef int _is_aligned(void *ptr):
168
"""Is this pointer aligned to an integer size offset?
170
:return: 1 if this pointer is aligned, 0 otherwise.
172
return ((<intptr_t>ptr) & ((sizeof(int))-1)) == 0
175
cdef int _cmp_by_dirs(char *path1, int size1, char *path2, int size2):
176
cdef unsigned char *cur1
177
cdef unsigned char *cur2
178
cdef unsigned char *end1
179
cdef unsigned char *end2
185
if path1 == path2 and size1 == size2:
188
end1 = <unsigned char*>path1+size1
189
end2 = <unsigned char*>path2+size2
191
# Use 32-bit comparisons for the matching portion of the string.
192
# Almost all CPU's are faster at loading and comparing 32-bit integers,
193
# than they are at 8-bit integers.
194
# 99% of the time, these will be aligned, but in case they aren't just skip
196
if _is_aligned(path1) and _is_aligned(path2):
197
cur_int1 = <int*>path1
198
cur_int2 = <int*>path2
199
end_int1 = <int*>(path1 + size1 - (size1 % sizeof(int)))
200
end_int2 = <int*>(path2 + size2 - (size2 % sizeof(int)))
202
while cur_int1 < end_int1 and cur_int2 < end_int2:
203
if cur_int1[0] != cur_int2[0]:
205
cur_int1 = cur_int1 + 1
206
cur_int2 = cur_int2 + 1
208
cur1 = <unsigned char*>cur_int1
209
cur2 = <unsigned char*>cur_int2
211
cur1 = <unsigned char*>path1
212
cur2 = <unsigned char*>path2
214
while cur1 < end1 and cur2 < end2:
215
if cur1[0] == cur2[0]:
216
# This character matches, just go to the next one
220
# The current characters do not match
222
return -1 # Reached the end of path1 segment first
223
elif cur2[0] == c'/':
224
return 1 # Reached the end of path2 segment first
225
elif cur1[0] < cur2[0]:
230
# We reached the end of at least one of the strings
232
return 1 # Not at the end of cur1, must be at the end of cur2
234
return -1 # At the end of cur1, but not at cur2
235
# We reached the end of both strings
239
def cmp_by_dirs_c(path1, path2):
240
"""Compare two paths directory by directory.
242
This is equivalent to doing::
244
cmp(path1.split('/'), path2.split('/'))
246
The idea is that you should compare path components separately. This
247
differs from plain ``cmp(path1, path2)`` for paths like ``'a-b'`` and
248
``a/b``. "a-b" comes after "a" but would come before "a/b" lexically.
250
:param path1: first path
251
:param path2: second path
252
:return: negative number if ``path1`` comes first,
253
0 if paths are equal,
254
and positive number if ``path2`` sorts first
256
if not PyString_CheckExact(path1):
257
raise TypeError("'path1' must be a plain string, not %s: %r"
258
% (type(path1), path1))
259
if not PyString_CheckExact(path2):
260
raise TypeError("'path2' must be a plain string, not %s: %r"
261
% (type(path2), path2))
262
return _cmp_by_dirs(PyString_AsString(path1),
263
PyString_Size(path1),
264
PyString_AsString(path2),
265
PyString_Size(path2))
268
def _cmp_path_by_dirblock_c(path1, path2):
269
"""Compare two paths based on what directory they are in.
271
This generates a sort order, such that all children of a directory are
272
sorted together, and grandchildren are in the same order as the
273
children appear. But all grandchildren come after all children.
275
In other words, all entries in a directory are sorted together, and
276
directorys are sorted in cmp_by_dirs order.
278
:param path1: first path
279
:param path2: the second path
280
:return: negative number if ``path1`` comes first,
282
and a positive number if ``path2`` sorts first
284
if not PyString_CheckExact(path1):
285
raise TypeError("'path1' must be a plain string, not %s: %r"
286
% (type(path1), path1))
287
if not PyString_CheckExact(path2):
288
raise TypeError("'path2' must be a plain string, not %s: %r"
289
% (type(path2), path2))
290
return _cmp_path_by_dirblock(PyString_AsString(path1),
291
PyString_Size(path1),
292
PyString_AsString(path2),
293
PyString_Size(path2))
296
cdef int _cmp_path_by_dirblock(char *path1, int path1_len,
297
char *path2, int path2_len):
298
"""Compare two paths by what directory they are in.
300
see ``_cmp_path_by_dirblock_c`` for details.
303
cdef int dirname1_len
305
cdef int dirname2_len
307
cdef int basename1_len
309
cdef int basename2_len
313
if path1_len == 0 and path2_len == 0:
316
if path1 == path2 and path1_len == path2_len:
325
basename1 = <char*>_my_memrchr(path1, c'/', path1_len)
327
if basename1 == NULL:
329
basename1_len = path1_len
334
dirname1_len = basename1 - path1
335
basename1 = basename1 + 1
336
basename1_len = path1_len - dirname1_len - 1
338
basename2 = <char*>_my_memrchr(path2, c'/', path2_len)
340
if basename2 == NULL:
342
basename2_len = path2_len
347
dirname2_len = basename2 - path2
348
basename2 = basename2 + 1
349
basename2_len = path2_len - dirname2_len - 1
351
cmp_val = _cmp_by_dirs(dirname1, dirname1_len,
352
dirname2, dirname2_len)
356
cur_len = basename1_len
357
if basename2_len < basename1_len:
358
cur_len = basename2_len
360
cmp_val = memcmp(basename1, basename2, cur_len)
363
if basename1_len == basename2_len:
365
if basename1_len < basename2_len:
370
def _bisect_path_left_c(paths, path):
371
"""Return the index where to insert path into paths.
373
This uses a path-wise comparison so we get::
383
:param paths: A list of paths to search through
384
:param path: A single path to insert
385
:return: An offset where 'path' can be inserted.
386
:seealso: bisect.bisect_left
397
if not PyList_CheckExact(paths):
398
raise TypeError("you must pass a python list for 'paths' not: %s %r"
399
% (type(paths), paths))
400
if not PyString_CheckExact(path):
401
raise TypeError("you must pass a string for 'path' not: %s %r"
402
% (type(path), path))
407
path_cstr = PyString_AsString(path)
408
path_size = PyString_Size(path)
411
_mid = (_lo + _hi) / 2
412
cur = PyList_GetItem_object_void(paths, _mid)
413
cur_cstr = PyString_AS_STRING_void(cur)
414
cur_size = PyString_GET_SIZE_void(cur)
415
if _cmp_path_by_dirblock(cur_cstr, cur_size, path_cstr, path_size) < 0:
422
def _bisect_path_right_c(paths, path):
423
"""Return the index where to insert path into paths.
425
This uses a path-wise comparison so we get::
435
:param paths: A list of paths to search through
436
:param path: A single path to insert
437
:return: An offset where 'path' can be inserted.
438
:seealso: bisect.bisect_right
449
if not PyList_CheckExact(paths):
450
raise TypeError("you must pass a python list for 'paths' not: %s %r"
451
% (type(paths), paths))
452
if not PyString_CheckExact(path):
453
raise TypeError("you must pass a string for 'path' not: %s %r"
454
% (type(path), path))
459
path_cstr = PyString_AsString(path)
460
path_size = PyString_Size(path)
463
_mid = (_lo + _hi) / 2
464
cur = PyList_GetItem_object_void(paths, _mid)
465
cur_cstr = PyString_AS_STRING_void(cur)
466
cur_size = PyString_GET_SIZE_void(cur)
467
if _cmp_path_by_dirblock(path_cstr, path_size, cur_cstr, cur_size) < 0:
474
def bisect_dirblock_c(dirblocks, dirname, lo=0, hi=None, cache=None):
475
"""Return the index where to insert dirname into the dirblocks.
477
The return value idx is such that all directories blocks in dirblock[:idx]
478
have names < dirname, and all blocks in dirblock[idx:] have names >=
481
Optional args lo (default 0) and hi (default len(dirblocks)) bound the
482
slice of a to be searched.
487
cdef char *dirname_cstr
488
cdef int dirname_size
493
if not PyList_CheckExact(dirblocks):
494
raise TypeError("you must pass a python list for 'dirblocks' not: %s %r"
495
% (type(dirblocks), dirblocks))
496
if not PyString_CheckExact(dirname):
497
raise TypeError("you must pass a string for dirname not: %s %r"
498
% (type(dirname), dirname))
505
dirname_cstr = PyString_AsString(dirname)
506
dirname_size = PyString_Size(dirname)
509
_mid = (_lo + _hi) / 2
510
# Grab the dirname for the current dirblock
511
# cur = dirblocks[_mid][0]
512
cur = PyTuple_GetItem_void_void(
513
PyList_GetItem_object_void(dirblocks, _mid), 0)
514
cur_cstr = PyString_AS_STRING_void(cur)
515
cur_size = PyString_GET_SIZE_void(cur)
516
if _cmp_by_dirs(cur_cstr, cur_size, dirname_cstr, dirname_size) < 0:
524
"""Maintain the current location, and return fields as you parse them."""
526
cdef object state # The DirState object
527
cdef object text # The overall string object
528
cdef char *text_cstr # Pointer to the beginning of text
529
cdef int text_size # Length of text
531
cdef char *end_cstr # End of text
532
cdef char *cur_cstr # Pointer to the current record
533
cdef char *next # Pointer to the end of this record
535
def __init__(self, text, state):
538
self.text_cstr = PyString_AsString(text)
539
self.text_size = PyString_Size(text)
540
self.end_cstr = self.text_cstr + self.text_size
541
self.cur_cstr = self.text_cstr
543
cdef char *get_next(self, int *size) except NULL:
544
"""Return a pointer to the start of the next field."""
546
cdef Py_ssize_t extra_len
548
if self.cur_cstr == NULL:
549
raise AssertionError('get_next() called when cur_str is NULL')
550
elif self.cur_cstr >= self.end_cstr:
551
raise AssertionError('get_next() called when there are no chars'
554
self.cur_cstr = <char*>memchr(next, c'\0', self.end_cstr - next)
555
if self.cur_cstr == NULL:
556
extra_len = self.end_cstr - next
557
raise errors.DirstateCorrupt(self.state,
558
'failed to find trailing NULL (\\0).'
559
' Trailing garbage: %r'
560
% safe_string_from_size(next, extra_len))
561
size[0] = self.cur_cstr - next
562
self.cur_cstr = self.cur_cstr + 1
565
cdef object get_next_str(self):
566
"""Get the next field as a Python string."""
569
next = self.get_next(&size)
570
return safe_string_from_size(next, size)
572
cdef int _init(self) except -1:
573
"""Get the pointer ready.
575
This assumes that the dirstate header has already been read, and we
576
already have the dirblock string loaded into memory.
577
This just initializes our memory pointers, etc for parsing of the
582
# The first field should be an empty string left over from the Header
583
first = self.get_next(&size)
584
if first[0] != c'\0' and size == 0:
585
raise AssertionError('First character should be null not: %s'
589
cdef object _get_entry(self, int num_trees, void **p_current_dirname,
591
"""Extract the next entry.
593
This parses the next entry based on the current location in
595
Each entry can be considered a "row" in the total table. And each row
596
has a fixed number of columns. It is generally broken up into "key"
597
columns, then "current" columns, and then "parent" columns.
599
:param num_trees: How many parent trees need to be parsed
600
:param p_current_dirname: A pointer to the current PyString
601
representing the directory name.
602
We pass this in as a void * so that pyrex doesn't have to
603
increment/decrement the PyObject reference counter for each
605
We use a pointer so that _get_entry can update it with the new
607
:param new_block: This is to let the caller know that it needs to
608
create a new directory block to store the next entry.
610
cdef object path_name_file_id_key
611
cdef char *entry_size_cstr
612
cdef unsigned long int entry_size
613
cdef char* executable_cstr
614
cdef int is_executable
615
cdef char* dirname_cstr
620
cdef object fingerprint
623
# Read the 'key' information (dirname, name, file_id)
624
dirname_cstr = self.get_next(&cur_size)
625
# Check to see if we have started a new directory block.
626
# If so, then we need to create a new dirname PyString, so that it can
627
# be used in all of the tuples. This saves time and memory, by re-using
628
# the same object repeatedly.
630
# Do the cheap 'length of string' check first. If the string is a
631
# different length, then we *have* to be a different directory.
632
if (cur_size != PyString_GET_SIZE_void(p_current_dirname[0])
633
or strncmp(dirname_cstr,
634
# Extract the char* from our current dirname string. We
635
# know it is a PyString, so we can use
636
# PyString_AS_STRING, we use the _void version because
637
# we are tricking Pyrex by using a void* rather than an
639
PyString_AS_STRING_void(p_current_dirname[0]),
641
dirname = safe_string_from_size(dirname_cstr, cur_size)
642
p_current_dirname[0] = <void*>dirname
647
# Build up the key that will be used.
648
# By using <object>(void *) Pyrex will automatically handle the
649
# Py_INCREF that we need.
650
path_name_file_id_key = (<object>p_current_dirname[0],
655
# Parse all of the per-tree information. current has the information in
656
# the same location as parent trees. The only difference is that 'info'
657
# is a 'packed_stat' for current, while it is a 'revision_id' for
659
# minikind, fingerprint, and info will be returned as regular python
661
# entry_size and is_executable will be parsed into a python Long and
662
# python Boolean, respectively.
663
# TODO: jam 20070718 Consider changin the entry_size conversion to
664
# prefer python Int when possible. They are generally faster to
665
# work with, and it will be rare that we have a file >2GB.
666
# Especially since this code is pretty much fixed at a max of
669
for i from 0 <= i < num_trees:
670
minikind = self.get_next_str()
671
fingerprint = self.get_next_str()
672
entry_size_cstr = self.get_next(&cur_size)
673
entry_size = strtoul(entry_size_cstr, NULL, 10)
674
executable_cstr = self.get_next(&cur_size)
675
is_executable = (executable_cstr[0] == c'y')
676
info = self.get_next_str()
677
PyList_Append(trees, (
679
fingerprint, # fingerprint
681
is_executable,# executable
682
info, # packed_stat or revision_id
685
# The returned tuple is (key, [trees])
686
ret = (path_name_file_id_key, trees)
687
# Ignore the trailing newline, but assert that it does exist, this
688
# ensures that we always finish parsing a line on an end-of-entry
690
trailing = self.get_next(&cur_size)
691
if cur_size != 1 or trailing[0] != c'\n':
692
raise errors.DirstateCorrupt(self.state,
693
'Bad parse, we expected to end on \\n, not: %d %s: %s'
694
% (cur_size, safe_string_from_size(trailing, cur_size),
698
def _parse_dirblocks(self):
699
"""Parse all dirblocks in the state file."""
701
cdef object current_block
703
cdef void * current_dirname
705
cdef int expected_entry_count
708
num_trees = self.state._num_present_parents() + 1
709
expected_entry_count = self.state._num_entries
711
# Ignore the first record
715
dirblocks = [('', current_block), ('', [])]
716
self.state._dirblocks = dirblocks
718
current_dirname = <void*>obj
722
# TODO: jam 2007-05-07 Consider pre-allocating some space for the
723
# members, and then growing and shrinking from there. If most
724
# directories have close to 10 entries in them, it would save a
725
# few mallocs if we default our list size to something
726
# reasonable. Or we could malloc it to something large (100 or
727
# so), and then truncate. That would give us a malloc + realloc,
728
# rather than lots of reallocs.
729
while self.cur_cstr < self.end_cstr:
730
entry = self._get_entry(num_trees, ¤t_dirname, &new_block)
732
# new block - different dirname
734
PyList_Append(dirblocks,
735
(<object>current_dirname, current_block))
736
PyList_Append(current_block, entry)
737
entry_count = entry_count + 1
738
if entry_count != expected_entry_count:
739
raise errors.DirstateCorrupt(self.state,
740
'We read the wrong number of entries.'
741
' We expected to read %s, but read %s'
742
% (expected_entry_count, entry_count))
743
self.state._split_root_dirblock_into_contents()
746
def _read_dirblocks_c(state):
747
"""Read in the dirblocks for the given DirState object.
749
This is tightly bound to the DirState internal representation. It should be
750
thought of as a member function, which is only separated out so that we can
751
re-write it in pyrex.
753
:param state: A DirState object.
755
:postcondition: The dirblocks will be loaded into the appropriate fields in
758
state._state_file.seek(state._end_of_header)
759
text = state._state_file.read()
760
# TODO: check the crc checksums. crc_measured = zlib.crc32(text)
762
reader = Reader(text, state)
764
reader._parse_dirblocks()
765
state._dirblock_state = DirState.IN_MEMORY_UNMODIFIED
768
cdef int minikind_from_mode(int mode):
769
# in order of frequency:
779
_encode = binascii.b2a_base64
782
from struct import pack
783
cdef _pack_stat(stat_value):
784
"""return a string representing the stat value's key fields.
786
:param stat_value: A stat oject with st_size, st_mtime, st_ctime, st_dev,
787
st_ino and st_mode fields.
789
cdef char result[6*4] # 6 long ints
791
aliased = <int *>result
792
aliased[0] = htonl(stat_value.st_size)
793
aliased[1] = htonl(int(stat_value.st_mtime))
794
aliased[2] = htonl(int(stat_value.st_ctime))
795
aliased[3] = htonl(stat_value.st_dev)
796
aliased[4] = htonl(stat_value.st_ino & 0xFFFFFFFF)
797
aliased[5] = htonl(stat_value.st_mode)
798
packed = PyString_FromStringAndSize(result, 6*4)
799
return _encode(packed)[:-1]
802
def update_entry(self, entry, abspath, stat_value):
803
"""Update the entry based on what is actually on disk.
805
:param entry: This is the dirblock entry for the file in question.
806
:param abspath: The path on disk for this file.
807
:param stat_value: (optional) if we already have done a stat on the
809
:return: The sha1 hexdigest of the file (40 bytes) or link target of a
812
return _update_entry(self, entry, abspath, stat_value)
815
cdef _update_entry(self, entry, abspath, stat_value):
816
"""Update the entry based on what is actually on disk.
818
:param self: The dirstate object this is operating on.
819
:param entry: This is the dirblock entry for the file in question.
820
:param abspath: The path on disk for this file.
821
:param stat_value: The stat value done on the path.
822
:return: The sha1 hexdigest of the file (40 bytes) or link target of a
825
# TODO - require pyrex 0.9.8, then use a pyd file to define access to the
826
# _st mode of the compiled stat objects.
827
cdef int minikind, saved_minikind
829
minikind = minikind_from_mode(stat_value.st_mode)
832
packed_stat = _pack_stat(stat_value)
833
details = PyList_GetItem_void_void(PyTuple_GetItem_void_void(<void *>entry, 1), 0)
834
saved_minikind = PyString_AsString_obj(<PyObject *>PyTuple_GetItem_void_void(details, 0))[0]
835
saved_link_or_sha1 = PyTuple_GetItem_void_object(details, 1)
836
saved_file_size = PyTuple_GetItem_void_object(details, 2)
837
saved_executable = PyTuple_GetItem_void_object(details, 3)
838
saved_packed_stat = PyTuple_GetItem_void_object(details, 4)
839
# Deal with pyrex decrefing the objects
840
Py_INCREF(saved_link_or_sha1)
841
Py_INCREF(saved_file_size)
842
Py_INCREF(saved_executable)
843
Py_INCREF(saved_packed_stat)
844
#(saved_minikind, saved_link_or_sha1, saved_file_size,
845
# saved_executable, saved_packed_stat) = entry[1][0]
847
if (minikind == saved_minikind
848
and packed_stat == saved_packed_stat):
849
# The stat hasn't changed since we saved, so we can re-use the
854
# size should also be in packed_stat
855
if saved_file_size == stat_value.st_size:
856
return saved_link_or_sha1
858
# If we have gotten this far, that means that we need to actually
859
# process this entry.
862
link_or_sha1 = self._sha1_file(abspath)
863
executable = self._is_executable(stat_value.st_mode,
865
if self._cutoff_time is None:
866
self._sha_cutoff_time()
867
if (stat_value.st_mtime < self._cutoff_time
868
and stat_value.st_ctime < self._cutoff_time):
869
entry[1][0] = ('f', link_or_sha1, stat_value.st_size,
870
executable, packed_stat)
872
entry[1][0] = ('f', '', stat_value.st_size,
873
executable, DirState.NULLSTAT)
874
elif minikind == c'd':
876
entry[1][0] = ('d', '', 0, False, packed_stat)
877
if saved_minikind != c'd':
878
# This changed from something into a directory. Make sure we
879
# have a directory block for it. This doesn't happen very
880
# often, so this doesn't have to be super fast.
881
block_index, entry_index, dir_present, file_present = \
882
self._get_block_entry_index(entry[0][0], entry[0][1], 0)
883
self._ensure_block(block_index, entry_index,
884
pathjoin(entry[0][0], entry[0][1]))
885
elif minikind == c'l':
886
link_or_sha1 = self._read_link(abspath, saved_link_or_sha1)
887
if self._cutoff_time is None:
888
self._sha_cutoff_time()
889
if (stat_value.st_mtime < self._cutoff_time
890
and stat_value.st_ctime < self._cutoff_time):
891
entry[1][0] = ('l', link_or_sha1, stat_value.st_size,
894
entry[1][0] = ('l', '', stat_value.st_size,
895
False, DirState.NULLSTAT)
896
self._dirblock_state = DirState.IN_MEMORY_MODIFIED
900
cdef char _minikind_from_string(object string):
901
"""Convert a python string to a char."""
902
return PyString_AsString(string)[0]
905
cdef object _kind_absent
906
cdef object _kind_file
907
cdef object _kind_directory
908
cdef object _kind_symlink
909
cdef object _kind_relocated
910
cdef object _kind_tree_reference
911
_kind_absent = "absent"
913
_kind_directory = "directory"
914
_kind_symlink = "symlink"
915
_kind_relocated = "relocated"
916
_kind_tree_reference = "tree-reference"
919
cdef object _minikind_to_kind(char minikind):
920
"""Create a string kind for minikind."""
921
cdef char _minikind[1]
924
elif minikind == c'd':
925
return _kind_directory
926
elif minikind == c'a':
928
elif minikind == c'r':
929
return _kind_relocated
930
elif minikind == c'l':
932
elif minikind == c't':
933
return _kind_tree_reference
934
_minikind[0] = minikind
935
raise KeyError(PyString_FromStringAndSize(_minikind, 1))
938
cdef int _versioned_minikind(char minikind):
939
"""Return non-zero if minikind is in fltd"""
940
return (minikind == c'f' or
946
cdef class ProcessEntryC:
948
cdef object old_dirname_to_file_id # dict
949
cdef object new_dirname_to_file_id # dict
950
cdef readonly object uninteresting
951
cdef object last_source_parent
952
cdef object last_target_parent
953
cdef object include_unchanged
954
cdef object use_filesystem_for_exec
955
cdef object utf8_decode
956
cdef readonly object searched_specific_files
957
cdef object search_specific_files
959
# Current iteration variables:
960
cdef object current_root
961
cdef object current_root_unicode
962
cdef object root_entries
963
cdef int root_entries_pos, root_entries_len
964
cdef object root_abspath
965
cdef int source_index, target_index
966
cdef int want_unversioned
968
cdef object dir_iterator
970
cdef object current_block
971
cdef int current_block_pos
972
cdef object current_block_list
973
cdef object current_dir_info
974
cdef object current_dir_list
976
cdef object root_dir_info
977
cdef object bisect_left
980
def __init__(self, include_unchanged, use_filesystem_for_exec,
981
search_specific_files, state, source_index, target_index,
982
want_unversioned, tree):
983
self.old_dirname_to_file_id = {}
984
self.new_dirname_to_file_id = {}
985
# Just a sentry, so that _process_entry can say that this
986
# record is handled, but isn't interesting to process (unchanged)
987
self.uninteresting = object()
988
# Using a list so that we can access the values and change them in
989
# nested scope. Each one is [path, file_id, entry]
990
self.last_source_parent = [None, None]
991
self.last_target_parent = [None, None]
992
self.include_unchanged = include_unchanged
993
self.use_filesystem_for_exec = use_filesystem_for_exec
994
self.utf8_decode = cache_utf8._utf8_decode
995
# for all search_indexs in each path at or under each element of
996
# search_specific_files, if the detail is relocated: add the id, and add the
997
# relocated path as one to search if its not searched already. If the
998
# detail is not relocated, add the id.
999
self.searched_specific_files = set()
1000
self.search_specific_files = search_specific_files
1002
self.current_root = None
1003
self.current_root_unicode = None
1004
self.root_entries = None
1005
self.root_entries_pos = 0
1006
self.root_entries_len = 0
1007
self.root_abspath = None
1008
if source_index is None:
1009
self.source_index = -1
1011
self.source_index = source_index
1012
self.target_index = target_index
1013
self.want_unversioned = want_unversioned
1015
self.dir_iterator = None
1016
self.block_index = -1
1017
self.current_block = None
1018
self.current_block_list = None
1019
self.current_block_pos = -1
1020
self.current_dir_info = None
1021
self.current_dir_list = None
1023
self.root_dir_info = None
1024
self.bisect_left = bisect.bisect_left
1025
self.pathjoin = osutils.pathjoin
1027
cdef _process_entry(self, entry, path_info):
1028
"""Compare an entry and real disk to generate delta information.
1030
:param path_info: top_relpath, basename, kind, lstat, abspath for
1031
the path of entry. If None, then the path is considered absent.
1032
(Perhaps we should pass in a concrete entry for this ?)
1033
Basename is returned as a utf8 string because we expect this
1034
tuple will be ignored, and don't want to take the time to
1036
:return: None if the these don't match
1037
A tuple of information about the change, or
1038
the object 'uninteresting' if these match, but are
1039
basically identical.
1041
cdef char target_minikind
1042
cdef char source_minikind
1044
cdef int content_change
1045
cdef object details_list
1047
details_list = entry[1]
1048
if -1 == self.source_index:
1049
source_details = DirState.NULL_PARENT_DETAILS
1051
source_details = details_list[self.source_index]
1052
target_details = details_list[self.target_index]
1053
target_minikind = _minikind_from_string(target_details[0])
1054
if path_info is not None and _versioned_minikind(target_minikind):
1055
if self.target_index != 0:
1056
raise AssertionError("Unsupported target index %d" % target_index)
1057
link_or_sha1 = _update_entry(self.state, entry, path_info[4], path_info[3])
1058
# The entry may have been modified by update_entry
1059
target_details = details_list[self.target_index]
1060
target_minikind = _minikind_from_string(target_details[0])
1063
# the rest of this function is 0.3 seconds on 50K paths, or
1064
# 0.000006 seconds per call.
1065
source_minikind = _minikind_from_string(source_details[0])
1066
if ((_versioned_minikind(source_minikind) or source_minikind == c'r')
1067
and _versioned_minikind(target_minikind)):
1068
# claimed content in both: diff
1069
# r | fdlt | | add source to search, add id path move and perform
1070
# | | | diff check on source-target
1071
# r | fdlt | a | dangling file that was present in the basis.
1073
if source_minikind != c'r':
1074
old_dirname = entry[0][0]
1075
old_basename = entry[0][1]
1076
old_path = path = None
1078
# add the source to the search path to find any children it
1079
# has. TODO ? : only add if it is a container ?
1080
if not osutils.is_inside_any(self.searched_specific_files,
1082
self.search_specific_files.add(source_details[1])
1083
# generate the old path; this is needed for stating later
1085
old_path = source_details[1]
1086
old_dirname, old_basename = os.path.split(old_path)
1087
path = self.pathjoin(entry[0][0], entry[0][1])
1088
old_entry = self.state._get_entry(self.source_index,
1090
# update the source details variable to be the real
1092
if old_entry == (None, None):
1093
raise errors.CorruptDirstate(self.state._filename,
1094
"entry '%s/%s' is considered renamed from %r"
1095
" but source does not exist\n"
1096
"entry: %s" % (entry[0][0], entry[0][1], old_path, entry))
1097
source_details = old_entry[1][self.source_index]
1098
source_minikind = _minikind_from_string(source_details[0])
1099
if path_info is None:
1100
# the file is missing on disk, show as removed.
1105
# source and target are both versioned and disk file is present.
1106
target_kind = path_info[2]
1107
if target_kind == 'directory':
1109
old_path = path = self.pathjoin(old_dirname, old_basename)
1110
file_id = entry[0][2]
1111
self.new_dirname_to_file_id[path] = file_id
1112
if source_minikind != c'd':
1115
# directories have no fingerprint
1118
elif target_kind == 'file':
1119
if source_minikind != c'f':
1122
# We could check the size, but we already have the
1124
content_change = (link_or_sha1 != source_details[1])
1125
# Target details is updated at update_entry time
1126
if self.use_filesystem_for_exec:
1127
# We don't need S_ISREG here, because we are sure
1128
# we are dealing with a file.
1129
target_exec = bool(S_IXUSR & path_info[3].st_mode)
1131
target_exec = target_details[3]
1132
elif target_kind == 'symlink':
1133
if source_minikind != c'l':
1136
content_change = (link_or_sha1 != source_details[1])
1138
elif target_kind == 'tree-reference':
1139
if source_minikind != c't':
1145
raise Exception, "unknown kind %s" % path_info[2]
1146
if source_minikind == c'd':
1148
old_path = path = self.pathjoin(old_dirname, old_basename)
1150
file_id = entry[0][2]
1151
self.old_dirname_to_file_id[old_path] = file_id
1152
# parent id is the entry for the path in the target tree
1153
if old_dirname == self.last_source_parent[0]:
1154
source_parent_id = self.last_source_parent[1]
1157
source_parent_id = self.old_dirname_to_file_id[old_dirname]
1159
source_parent_entry = self.state._get_entry(self.source_index,
1160
path_utf8=old_dirname)
1161
source_parent_id = source_parent_entry[0][2]
1162
if source_parent_id == entry[0][2]:
1163
# This is the root, so the parent is None
1164
source_parent_id = None
1166
self.last_source_parent[0] = old_dirname
1167
self.last_source_parent[1] = source_parent_id
1168
new_dirname = entry[0][0]
1169
if new_dirname == self.last_target_parent[0]:
1170
target_parent_id = self.last_target_parent[1]
1173
target_parent_id = self.new_dirname_to_file_id[new_dirname]
1175
# TODO: We don't always need to do the lookup, because the
1176
# parent entry will be the same as the source entry.
1177
target_parent_entry = self.state._get_entry(self.target_index,
1178
path_utf8=new_dirname)
1179
if target_parent_entry == (None, None):
1180
raise AssertionError(
1181
"Could not find target parent in wt: %s\nparent of: %s"
1182
% (new_dirname, entry))
1183
target_parent_id = target_parent_entry[0][2]
1184
if target_parent_id == entry[0][2]:
1185
# This is the root, so the parent is None
1186
target_parent_id = None
1188
self.last_target_parent[0] = new_dirname
1189
self.last_target_parent[1] = target_parent_id
1191
source_exec = source_details[3]
1192
if (self.include_unchanged
1194
or source_parent_id != target_parent_id
1195
or old_basename != entry[0][1]
1196
or source_exec != target_exec
1198
if old_path is None:
1199
path = self.pathjoin(old_dirname, old_basename)
1201
old_path_u = self.utf8_decode(old_path)[0]
1204
old_path_u = self.utf8_decode(old_path)[0]
1205
if old_path == path:
1208
path_u = self.utf8_decode(path)[0]
1209
source_kind = _minikind_to_kind(source_minikind)
1210
return (entry[0][2],
1211
(old_path_u, path_u),
1214
(source_parent_id, target_parent_id),
1215
(self.utf8_decode(old_basename)[0], self.utf8_decode(entry[0][1])[0]),
1216
(source_kind, target_kind),
1217
(source_exec, target_exec))
1219
return self.uninteresting
1220
elif source_minikind == c'a' and _versioned_minikind(target_minikind):
1221
# looks like a new file
1222
path = self.pathjoin(entry[0][0], entry[0][1])
1223
# parent id is the entry for the path in the target tree
1224
# TODO: these are the same for an entire directory: cache em.
1225
parent_id = self.state._get_entry(self.target_index,
1226
path_utf8=entry[0][0])[0][2]
1227
if parent_id == entry[0][2]:
1229
if path_info is not None:
1231
if self.use_filesystem_for_exec:
1232
# We need S_ISREG here, because we aren't sure if this
1235
S_ISREG(path_info[3].st_mode)
1236
and S_IXUSR & path_info[3].st_mode)
1238
target_exec = target_details[3]
1239
return (entry[0][2],
1240
(None, self.utf8_decode(path)[0]),
1244
(None, self.utf8_decode(entry[0][1])[0]),
1245
(None, path_info[2]),
1246
(None, target_exec))
1248
# Its a missing file, report it as such.
1249
return (entry[0][2],
1250
(None, self.utf8_decode(path)[0]),
1254
(None, self.utf8_decode(entry[0][1])[0]),
1257
elif _versioned_minikind(source_minikind) and target_minikind == c'a':
1258
# unversioned, possibly, or possibly not deleted: we dont care.
1259
# if its still on disk, *and* theres no other entry at this
1260
# path [we dont know this in this routine at the moment -
1261
# perhaps we should change this - then it would be an unknown.
1262
old_path = self.pathjoin(entry[0][0], entry[0][1])
1263
# parent id is the entry for the path in the target tree
1264
parent_id = self.state._get_entry(self.source_index, path_utf8=entry[0][0])[0][2]
1265
if parent_id == entry[0][2]:
1267
return (entry[0][2],
1268
(self.utf8_decode(old_path)[0], None),
1272
(self.utf8_decode(entry[0][1])[0], None),
1273
(_minikind_to_kind(source_minikind), None),
1274
(source_details[3], None))
1275
elif _versioned_minikind(source_minikind) and target_minikind == c'r':
1276
# a rename; could be a true rename, or a rename inherited from
1277
# a renamed parent. TODO: handle this efficiently. Its not
1278
# common case to rename dirs though, so a correct but slow
1279
# implementation will do.
1280
if not osutils.is_inside_any(self.searched_specific_files, target_details[1]):
1281
self.search_specific_files.add(target_details[1])
1282
elif ((source_minikind == c'r' or source_minikind == c'a') and
1283
(target_minikind == c'r' or target_minikind == c'a')):
1284
# neither of the selected trees contain this path,
1285
# so skip over it. This is not currently directly tested, but
1286
# is indirectly via test_too_much.TestCommands.test_conflicts.
1289
raise AssertionError("don't know how to compare "
1290
"source_minikind=%r, target_minikind=%r"
1291
% (source_minikind, target_minikind))
1292
## import pdb;pdb.set_trace()
1298
def iter_changes(self):
1301
cdef void _update_current_block(self):
1302
if (self.block_index < len(self.state._dirblocks) and
1303
osutils.is_inside(self.current_root, self.state._dirblocks[self.block_index][0])):
1304
self.current_block = self.state._dirblocks[self.block_index]
1305
self.current_block_list = self.current_block[1]
1306
self.current_block_pos = 0
1308
self.current_block = None
1309
self.current_block_list = None
1312
# Simple thunk to allow tail recursion without pyrex confusion
1313
return self._iter_next()
1315
cdef _iter_next(self):
1316
"""Iterate over the changes."""
1317
# This function single steps through an iterator. As such while loops
1318
# are often exited by 'return' - the code is structured so that the
1319
# next call into the function will return to the same while loop. Note
1320
# that all flow control needed to re-reach that step is reexecuted,
1321
# which can be a performance problem. It has not yet been tuned to
1322
# minimise this; a state machine is probably the simplest restructuring
1323
# to both minimise this overhead and make the code considerably more
1327
# compare source_index and target_index at or under each element of search_specific_files.
1328
# follow the following comparison table. Note that we only want to do diff operations when
1329
# the target is fdl because thats when the walkdirs logic will have exposed the pathinfo
1333
# Source | Target | disk | action
1334
# r | fdlt | | add source to search, add id path move and perform
1335
# | | | diff check on source-target
1336
# r | fdlt | a | dangling file that was present in the basis.
1338
# r | a | | add source to search
1340
# r | r | | this path is present in a non-examined tree, skip.
1341
# r | r | a | this path is present in a non-examined tree, skip.
1342
# a | fdlt | | add new id
1343
# a | fdlt | a | dangling locally added file, skip
1344
# a | a | | not present in either tree, skip
1345
# a | a | a | not present in any tree, skip
1346
# a | r | | not present in either tree at this path, skip as it
1347
# | | | may not be selected by the users list of paths.
1348
# a | r | a | not present in either tree at this path, skip as it
1349
# | | | may not be selected by the users list of paths.
1350
# fdlt | fdlt | | content in both: diff them
1351
# fdlt | fdlt | a | deleted locally, but not unversioned - show as deleted ?
1352
# fdlt | a | | unversioned: output deleted id for now
1353
# fdlt | a | a | unversioned and deleted: output deleted id
1354
# fdlt | r | | relocated in this tree, so add target to search.
1355
# | | | Dont diff, we will see an r,fd; pair when we reach
1356
# | | | this id at the other path.
1357
# fdlt | r | a | relocated in this tree, so add target to search.
1358
# | | | Dont diff, we will see an r,fd; pair when we reach
1359
# | | | this id at the other path.
1361
# TODO: jam 20070516 - Avoid the _get_entry lookup overhead by
1362
# keeping a cache of directories that we have seen.
1363
cdef object current_dirname, current_blockname
1364
cdef char * current_dirname_c, * current_blockname_c
1365
cdef int advance_entry, advance_path
1366
cdef int path_handled
1367
uninteresting = self.uninteresting
1368
searched_specific_files = self.searched_specific_files
1369
# Are we walking a root?
1370
while self.root_entries_pos < self.root_entries_len:
1371
entry = self.root_entries[self.root_entries_pos]
1372
self.root_entries_pos = self.root_entries_pos + 1
1373
result = self._process_entry(entry, self.root_dir_info)
1374
if result is not None and result is not self.uninteresting:
1376
# Have we finished the prior root, or never started one ?
1377
if self.current_root is None:
1378
# TODO: the pending list should be lexically sorted? the
1379
# interface doesn't require it.
1381
self.current_root = self.search_specific_files.pop()
1383
raise StopIteration()
1384
self.current_root_unicode = self.current_root.decode('utf8')
1385
self.searched_specific_files.add(self.current_root)
1386
# process the entries for this containing directory: the rest will be
1387
# found by their parents recursively.
1388
self.root_entries = self.state._entries_for_path(self.current_root)
1389
self.root_entries_len = len(self.root_entries)
1390
self.root_abspath = self.tree.abspath(self.current_root_unicode)
1392
root_stat = os.lstat(self.root_abspath)
1394
if e.errno == errno.ENOENT:
1395
# the path does not exist: let _process_entry know that.
1396
self.root_dir_info = None
1398
# some other random error: hand it up.
1401
self.root_dir_info = ('', self.current_root,
1402
osutils.file_kind_from_stat_mode(root_stat.st_mode), root_stat,
1404
if self.root_dir_info[2] == 'directory':
1405
if self.tree._directory_is_tree_reference(
1406
self.current_root_unicode):
1407
self.root_dir_info = self.root_dir_info[:2] + \
1408
('tree-reference',) + self.root_dir_info[3:]
1409
if not self.root_entries and not self.root_dir_info:
1410
# this specified path is not present at all, skip it.
1411
# (tail recursion, can do a loop once the full structure is
1413
return self._iter_next()
1415
self.root_entries_pos = 0
1416
# XXX Clarity: This loop is duplicated a out the self.current_root
1417
# is None guard above: if we return from it, it completes there
1418
# (and the following if block cannot trigger because
1419
# path_handled must be true, so the if block is not # duplicated.
1420
while self.root_entries_pos < self.root_entries_len:
1421
entry = self.root_entries[self.root_entries_pos]
1422
self.root_entries_pos = self.root_entries_pos + 1
1423
result = self._process_entry(entry, self.root_dir_info)
1424
if result is not None:
1426
if result is not self.uninteresting:
1428
# handle unversioned specified paths:
1429
if self.want_unversioned and not path_handled and self.root_dir_info:
1430
new_executable = bool(
1431
stat.S_ISREG(self.root_dir_info[3].st_mode)
1432
and stat.S_IEXEC & self.root_dir_info[3].st_mode)
1434
(None, self.current_root_unicode),
1438
(None, splitpath(self.current_root_unicode)[-1]),
1439
(None, self.root_dir_info[2]),
1440
(None, new_executable)
1442
# If we reach here, the outer flow continues, which enters into the
1443
# per-root setup logic.
1444
if self.current_dir_info is None and self.current_block is None:
1445
# setup iteration of this root:
1446
self.current_dir_list = None
1447
if self.root_dir_info and self.root_dir_info[2] == 'tree-reference':
1448
self.current_dir_info = None
1450
self.dir_iterator = osutils._walkdirs_utf8(self.root_abspath,
1451
prefix=self.current_root)
1454
self.current_dir_info = self.dir_iterator.next()
1455
self.current_dir_list = self.current_dir_info[1]
1457
# there may be directories in the inventory even though
1458
# this path is not a file on disk: so mark it as end of
1460
if e.errno in (errno.ENOENT, errno.ENOTDIR, errno.EINVAL):
1461
self.current_dir_info = None
1462
elif sys.platform == 'win32':
1463
# on win32, python2.4 has e.errno == ERROR_DIRECTORY, but
1464
# python 2.5 has e.errno == EINVAL,
1465
# and e.winerror == ERROR_DIRECTORY
1467
e_winerror = e.winerror
1468
except AttributeError:
1470
win_errors = (ERROR_DIRECTORY, ERROR_PATH_NOT_FOUND)
1471
if (e.errno in win_errors or e_winerror in win_errors):
1472
self.current_dir_info = None
1474
# Will this really raise the right exception ?
1479
if self.current_dir_info[0][0] == '':
1480
# remove .bzr from iteration
1481
bzr_index = self.bisect_left(self.current_dir_list, ('.bzr',))
1482
if self.current_dir_list[bzr_index][0] != '.bzr':
1483
raise AssertionError()
1484
del self.current_dir_list[bzr_index]
1485
initial_key = (self.current_root, '', '')
1486
self.block_index, _ = self.state._find_block_index_from_key(initial_key)
1487
if self.block_index == 0:
1488
# we have processed the total root already, but because the
1489
# initial key matched it we should skip it here.
1490
self.block_index = self.block_index + 1
1491
self._update_current_block()
1492
# walk until both the directory listing and the versioned metadata
1494
while (self.current_dir_info is not None
1495
or self.current_block is not None):
1496
# Uncommon case - a missing directory or an unversioned directory:
1497
if (self.current_dir_info and self.current_block
1498
and self.current_dir_info[0][0] != self.current_block[0]):
1499
# Work around pyrex broken heuristic - current_dirname has
1500
# the same scope as current_dirname_c
1501
current_dirname = self.current_dir_info[0][0]
1502
current_dirname_c = PyString_AS_STRING_void(
1503
<void *>current_dirname)
1504
current_blockname = self.current_block[0]
1505
current_blockname_c = PyString_AS_STRING_void(
1506
<void *>current_blockname)
1507
# In the python generator we evaluate this if block once per
1508
# dir+block; because we reenter in the pyrex version its being
1509
# evaluated once per path: we could cache the result before
1510
# doing the while loop and probably save time.
1511
if _cmp_by_dirs(current_dirname_c,
1512
PyString_Size(current_dirname),
1513
current_blockname_c,
1514
PyString_Size(current_blockname)) < 0:
1515
# filesystem data refers to paths not covered by the
1516
# dirblock. this has two possibilities:
1517
# A) it is versioned but empty, so there is no block for it
1518
# B) it is not versioned.
1520
# if (A) then we need to recurse into it to check for
1521
# new unknown files or directories.
1522
# if (B) then we should ignore it, because we don't
1523
# recurse into unknown directories.
1524
# We are doing a loop
1525
while self.path_index < len(self.current_dir_list):
1526
current_path_info = self.current_dir_list[self.path_index]
1527
# dont descend into this unversioned path if it is
1529
if current_path_info[2] in ('directory',
1531
del self.current_dir_list[self.path_index]
1532
self.path_index = self.path_index - 1
1533
self.path_index = self.path_index + 1
1534
if self.want_unversioned:
1535
if current_path_info[2] == 'directory':
1536
if self.tree._directory_is_tree_reference(
1537
self.utf8_decode(current_path_info[0])[0]):
1538
current_path_info = current_path_info[:2] + \
1539
('tree-reference',) + current_path_info[3:]
1540
new_executable = bool(
1541
stat.S_ISREG(current_path_info[3].st_mode)
1542
and stat.S_IEXEC & current_path_info[3].st_mode)
1544
(None, self.utf8_decode(current_path_info[0])[0]),
1548
(None, self.utf8_decode(current_path_info[1])[0]),
1549
(None, current_path_info[2]),
1550
(None, new_executable))
1551
# This dir info has been handled, go to the next
1553
self.current_dir_list = None
1555
self.current_dir_info = self.dir_iterator.next()
1556
self.current_dir_list = self.current_dir_info[1]
1557
except StopIteration:
1558
self.current_dir_info = None
1560
# We have a dirblock entry for this location, but there
1561
# is no filesystem path for this. This is most likely
1562
# because a directory was removed from the disk.
1563
# We don't have to report the missing directory,
1564
# because that should have already been handled, but we
1565
# need to handle all of the files that are contained
1567
while self.current_block_pos < len(self.current_block_list):
1568
current_entry = self.current_block_list[self.current_block_pos]
1569
self.current_block_pos = self.current_block_pos + 1
1570
# entry referring to file not present on disk.
1571
# advance the entry only, after processing.
1572
result = self._process_entry(current_entry, None)
1573
if result is not None:
1574
if result is not self.uninteresting:
1576
self.block_index = self.block_index + 1
1577
self._update_current_block()
1578
continue # next loop-on-block/dir
1579
result = self._loop_one_block()
1580
if result is not None:
1582
if len(self.search_specific_files):
1583
# More supplied paths to process
1584
self.current_root = None
1585
return self._iter_next()
1586
raise StopIteration()
1588
cdef object _maybe_tree_ref(self, current_path_info):
1589
if self.tree._directory_is_tree_reference(
1590
self.utf8_decode(current_path_info[0])[0]):
1591
return current_path_info[:2] + \
1592
('tree-reference',) + current_path_info[3:]
1594
return current_path_info
1596
cdef object _loop_one_block(self):
1597
# current_dir_info and current_block refer to the same directory -
1598
# this is the common case code.
1599
# Assign local variables for current path and entry:
1600
cdef object current_entry
1601
cdef object current_path_info
1602
cdef int path_handled
1605
# cdef char * temp_str
1606
# cdef Py_ssize_t temp_str_length
1607
# PyString_AsStringAndSize(disk_kind, &temp_str, &temp_str_length)
1608
# if not strncmp(temp_str, "directory", temp_str_length):
1609
if (self.current_block is not None and
1610
self.current_block_pos < PyList_GET_SIZE(self.current_block_list)):
1611
current_entry = PyList_GET_ITEM(self.current_block_list,
1612
self.current_block_pos)
1614
Py_INCREF(current_entry)
1616
current_entry = None
1617
if (self.current_dir_info is not None and
1618
self.path_index < PyList_GET_SIZE(self.current_dir_list)):
1619
current_path_info = PyList_GET_ITEM(self.current_dir_list,
1622
Py_INCREF(current_path_info)
1623
disk_kind = PyTuple_GET_ITEM(current_path_info, 2)
1625
Py_INCREF(disk_kind)
1626
if disk_kind == "directory":
1627
current_path_info = self._maybe_tree_ref(current_path_info)
1629
current_path_info = None
1630
while (current_entry is not None or current_path_info is not None):
1635
if current_entry is None:
1636
# unversioned - the check for path_handled when the path
1637
# is advanced will yield this path if needed.
1639
elif current_path_info is None:
1640
# no path is fine: the per entry code will handle it.
1641
result = self._process_entry(current_entry, current_path_info)
1642
if result is not None:
1643
if result is self.uninteresting:
1646
minikind = _minikind_from_string(
1647
current_entry[1][self.target_index][0])
1648
cmp_result = cmp(current_path_info[1], current_entry[0][1])
1649
if (cmp_result or minikind == c'a' or minikind == c'r'):
1650
# The current path on disk doesn't match the dirblock
1651
# record. Either the dirblock record is marked as
1652
# absent/renamed, or the file on disk is not present at all
1653
# in the dirblock. Either way, report about the dirblock
1654
# entry, and let other code handle the filesystem one.
1656
# Compare the basename for these files to determine
1659
# extra file on disk: pass for now, but only
1660
# increment the path, not the entry
1663
# entry referring to file not present on disk.
1664
# advance the entry only, after processing.
1665
result = self._process_entry(current_entry, None)
1666
if result is not None:
1667
if result is self.uninteresting:
1671
# paths are the same,and the dirstate entry is not
1672
# absent or renamed.
1673
result = self._process_entry(current_entry, current_path_info)
1674
if result is not None:
1676
if result is self.uninteresting:
1678
# >- loop control starts here:
1680
if advance_entry and current_entry is not None:
1681
self.current_block_pos = self.current_block_pos + 1
1682
if self.current_block_pos < PyList_GET_SIZE(self.current_block_list):
1683
current_entry = self.current_block_list[self.current_block_pos]
1685
current_entry = None
1687
if advance_path and current_path_info is not None:
1688
if not path_handled:
1689
# unversioned in all regards
1690
if self.want_unversioned:
1691
new_executable = bool(
1692
stat.S_ISREG(current_path_info[3].st_mode)
1693
and stat.S_IEXEC & current_path_info[3].st_mode)
1695
relpath_unicode = self.utf8_decode(current_path_info[0])[0]
1696
except UnicodeDecodeError:
1697
raise errors.BadFilenameEncoding(
1698
current_path_info[0], osutils._fs_enc)
1699
if result is not None:
1700
raise AssertionError(
1701
"result is not None: %r" % result)
1703
(None, relpath_unicode),
1707
(None, self.utf8_decode(current_path_info[1])[0]),
1708
(None, current_path_info[2]),
1709
(None, new_executable))
1710
# dont descend into this unversioned path if it is
1712
if current_path_info[2] in ('directory'):
1713
del self.current_dir_list[self.path_index]
1714
self.path_index = self.path_index - 1
1715
# dont descend the disk iterator into any tree
1717
if current_path_info[2] == 'tree-reference':
1718
del self.current_dir_list[self.path_index]
1719
self.path_index = self.path_index - 1
1720
self.path_index = self.path_index + 1
1721
if self.path_index < len(self.current_dir_list):
1722
current_path_info = self.current_dir_list[self.path_index]
1723
if current_path_info[2] == 'directory':
1724
current_path_info = self._maybe_tree_ref(
1727
current_path_info = None
1728
if result is not None:
1729
# Found a result on this pass, yield it
1731
if self.current_block is not None:
1732
self.block_index = self.block_index + 1
1733
self._update_current_block()
1734
if self.current_dir_info is not None:
1736
self.current_dir_list = None
1738
self.current_dir_info = self.dir_iterator.next()
1739
self.current_dir_list = self.current_dir_info[1]
1740
except StopIteration:
1741
self.current_dir_info = None
added
removed
bzrlib/_dirstate_helpers_c.pyx
