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., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
17
"""Helper functions for DirState.
19
This is the python implementation for DirState functions.
29
from bzrlib import cache_utf8, errors, osutils
30
from bzrlib.dirstate import DirState
31
from bzrlib.osutils import pathjoin, splitpath
34
# This is the Windows equivalent of ENOTDIR
35
# It is defined in pywin32.winerror, but we don't want a strong dependency for
37
# XXX: Perhaps we could get it from a windows header ?
38
cdef int ERROR_PATH_NOT_FOUND
39
ERROR_PATH_NOT_FOUND = 3
40
cdef int ERROR_DIRECTORY
43
#python2.4 support, and other platform-dependent includes
44
cdef extern from "python-compat.h":
45
unsigned long htonl(unsigned long)
47
# Give Pyrex some function definitions for it to understand.
48
# All of these are just hints to Pyrex, so that it can try to convert python
49
# objects into similar C objects. (such as PyInt => int).
50
# In anything defined 'cdef extern from XXX' the real C header will be
51
# imported, and the real definition will be used from there. So these are just
52
# hints, and do not need to match exactly to the C definitions.
55
ctypedef unsigned long size_t
57
cdef extern from "_dirstate_helpers_pyx.h":
62
cdef extern from "stdlib.h":
63
unsigned long int strtoul(char *nptr, char **endptr, int base)
66
cdef extern from 'sys/stat.h':
69
# On win32, this actually comes from "python-compat.h"
73
# These functions allow us access to a bit of the 'bare metal' of python
74
# objects, rather than going through the object abstraction. (For example,
75
# PyList_Append, rather than getting the 'append' attribute of the object, and
76
# creating a tuple, and then using PyCallObject).
77
# Functions that return (or take) a void* are meant to grab a C PyObject*. This
78
# differs from the Pyrex 'object'. If you declare a variable as 'object' Pyrex
79
# will automatically Py_INCREF and Py_DECREF when appropriate. But for some
80
# inner loops, we don't need to do that at all, as the reference only lasts for
82
# Note that the C API GetItem calls borrow references, so pyrex does the wrong
83
# thing if you declare e.g. object PyList_GetItem(object lst, int index) - you
84
# need to manually Py_INCREF yourself.
85
cdef extern from "Python.h":
86
ctypedef int Py_ssize_t
87
ctypedef struct PyObject:
89
int PyList_Append(object lst, object item) except -1
90
void *PyList_GetItem_object_void "PyList_GET_ITEM" (object lst, int index)
91
void *PyList_GetItem_void_void "PyList_GET_ITEM" (void * lst, int index)
92
object PyList_GET_ITEM(object lst, Py_ssize_t index)
93
int PyList_CheckExact(object)
94
Py_ssize_t PyList_GET_SIZE (object p)
96
void *PyTuple_GetItem_void_void "PyTuple_GET_ITEM" (void* tpl, int index)
97
object PyTuple_GetItem_void_object "PyTuple_GET_ITEM" (void* tpl, int index)
98
object PyTuple_GET_ITEM(object tpl, Py_ssize_t index)
101
char *PyString_AsString(object p)
102
char *PyString_AsString_obj "PyString_AsString" (PyObject *string)
103
char *PyString_AS_STRING_void "PyString_AS_STRING" (void *p)
104
int PyString_AsStringAndSize(object str, char **buffer, Py_ssize_t *length) except -1
105
object PyString_FromString(char *)
106
object PyString_FromStringAndSize(char *, Py_ssize_t)
107
int PyString_Size(object p)
108
int PyString_GET_SIZE_void "PyString_GET_SIZE" (void *p)
109
int PyString_CheckExact(object p)
110
void Py_INCREF(object o)
111
void Py_DECREF(object o)
114
cdef extern from "string.h":
115
int strncmp(char *s1, char *s2, int len)
116
void *memchr(void *s, int c, size_t len)
117
int memcmp(void *b1, void *b2, size_t len)
118
# ??? memrchr is a GNU extension :(
119
# void *memrchr(void *s, int c, size_t len)
122
cdef void* _my_memrchr(void *s, int c, size_t n):
123
# memrchr seems to be a GNU extension, so we have to implement it ourselves
136
def _py_memrchr(s, c):
137
"""Just to expose _my_memrchr for testing.
139
:param s: The Python string to search
140
:param c: The character to search for
141
:return: The offset to the last instance of 'c' in s
148
_s = PyString_AsString(s)
149
length = PyString_Size(s)
151
_c = PyString_AsString(c)
152
assert PyString_Size(c) == 1,\
153
'Must be a single character string, not %s' % (c,)
154
found = _my_memrchr(_s, _c[0], length)
157
return <char*>found - <char*>_s
159
cdef object safe_string_from_size(char *s, Py_ssize_t size):
161
# XXX: On 64-bit machines the <int> cast causes a C compiler warning.
162
raise AssertionError(
163
'tried to create a string with an invalid size: %d @0x%x'
165
return PyString_FromStringAndSize(s, size)
168
cdef int _is_aligned(void *ptr):
169
"""Is this pointer aligned to an integer size offset?
171
:return: 1 if this pointer is aligned, 0 otherwise.
173
return ((<intptr_t>ptr) & ((sizeof(int))-1)) == 0
176
cdef int _cmp_by_dirs(char *path1, int size1, char *path2, int size2):
177
cdef unsigned char *cur1
178
cdef unsigned char *cur2
179
cdef unsigned char *end1
180
cdef unsigned char *end2
186
if path1 == path2 and size1 == size2:
189
end1 = <unsigned char*>path1+size1
190
end2 = <unsigned char*>path2+size2
192
# Use 32-bit comparisons for the matching portion of the string.
193
# Almost all CPU's are faster at loading and comparing 32-bit integers,
194
# than they are at 8-bit integers.
195
# 99% of the time, these will be aligned, but in case they aren't just skip
197
if _is_aligned(path1) and _is_aligned(path2):
198
cur_int1 = <int*>path1
199
cur_int2 = <int*>path2
200
end_int1 = <int*>(path1 + size1 - (size1 % sizeof(int)))
201
end_int2 = <int*>(path2 + size2 - (size2 % sizeof(int)))
203
while cur_int1 < end_int1 and cur_int2 < end_int2:
204
if cur_int1[0] != cur_int2[0]:
206
cur_int1 = cur_int1 + 1
207
cur_int2 = cur_int2 + 1
209
cur1 = <unsigned char*>cur_int1
210
cur2 = <unsigned char*>cur_int2
212
cur1 = <unsigned char*>path1
213
cur2 = <unsigned char*>path2
215
while cur1 < end1 and cur2 < end2:
216
if cur1[0] == cur2[0]:
217
# This character matches, just go to the next one
221
# The current characters do not match
223
return -1 # Reached the end of path1 segment first
224
elif cur2[0] == c'/':
225
return 1 # Reached the end of path2 segment first
226
elif cur1[0] < cur2[0]:
231
# We reached the end of at least one of the strings
233
return 1 # Not at the end of cur1, must be at the end of cur2
235
return -1 # At the end of cur1, but not at cur2
236
# We reached the end of both strings
240
def cmp_by_dirs(path1, path2):
241
"""Compare two paths directory by directory.
243
This is equivalent to doing::
245
cmp(path1.split('/'), path2.split('/'))
247
The idea is that you should compare path components separately. This
248
differs from plain ``cmp(path1, path2)`` for paths like ``'a-b'`` and
249
``a/b``. "a-b" comes after "a" but would come before "a/b" lexically.
251
:param path1: first path
252
:param path2: second path
253
:return: negative number if ``path1`` comes first,
254
0 if paths are equal,
255
and positive number if ``path2`` sorts first
257
if not PyString_CheckExact(path1):
258
raise TypeError("'path1' must be a plain string, not %s: %r"
259
% (type(path1), path1))
260
if not PyString_CheckExact(path2):
261
raise TypeError("'path2' must be a plain string, not %s: %r"
262
% (type(path2), path2))
263
return _cmp_by_dirs(PyString_AsString(path1),
264
PyString_Size(path1),
265
PyString_AsString(path2),
266
PyString_Size(path2))
269
def _cmp_path_by_dirblock(path1, path2):
270
"""Compare two paths based on what directory they are in.
272
This generates a sort order, such that all children of a directory are
273
sorted together, and grandchildren are in the same order as the
274
children appear. But all grandchildren come after all children.
276
In other words, all entries in a directory are sorted together, and
277
directorys are sorted in cmp_by_dirs order.
279
:param path1: first path
280
:param path2: the second path
281
:return: negative number if ``path1`` comes first,
283
and a positive number if ``path2`` sorts first
285
if not PyString_CheckExact(path1):
286
raise TypeError("'path1' must be a plain string, not %s: %r"
287
% (type(path1), path1))
288
if not PyString_CheckExact(path2):
289
raise TypeError("'path2' must be a plain string, not %s: %r"
290
% (type(path2), path2))
291
return _cmp_path_by_dirblock_intern(PyString_AsString(path1),
292
PyString_Size(path1),
293
PyString_AsString(path2),
294
PyString_Size(path2))
297
cdef int _cmp_path_by_dirblock_intern(char *path1, int path1_len,
298
char *path2, int path2_len):
299
"""Compare two paths by what directory they are in.
301
see ``_cmp_path_by_dirblock`` for details.
304
cdef int dirname1_len
306
cdef int dirname2_len
308
cdef int basename1_len
310
cdef int basename2_len
314
if path1_len == 0 and path2_len == 0:
317
if path1 == path2 and path1_len == path2_len:
326
basename1 = <char*>_my_memrchr(path1, c'/', path1_len)
328
if basename1 == NULL:
330
basename1_len = path1_len
335
dirname1_len = basename1 - path1
336
basename1 = basename1 + 1
337
basename1_len = path1_len - dirname1_len - 1
339
basename2 = <char*>_my_memrchr(path2, c'/', path2_len)
341
if basename2 == NULL:
343
basename2_len = path2_len
348
dirname2_len = basename2 - path2
349
basename2 = basename2 + 1
350
basename2_len = path2_len - dirname2_len - 1
352
cmp_val = _cmp_by_dirs(dirname1, dirname1_len,
353
dirname2, dirname2_len)
357
cur_len = basename1_len
358
if basename2_len < basename1_len:
359
cur_len = basename2_len
361
cmp_val = memcmp(basename1, basename2, cur_len)
364
if basename1_len == basename2_len:
366
if basename1_len < basename2_len:
371
def _bisect_path_left(paths, path):
372
"""Return the index where to insert path into paths.
374
This uses a path-wise comparison so we get::
384
:param paths: A list of paths to search through
385
:param path: A single path to insert
386
:return: An offset where 'path' can be inserted.
387
:seealso: bisect.bisect_left
398
if not PyList_CheckExact(paths):
399
raise TypeError("you must pass a python list for 'paths' not: %s %r"
400
% (type(paths), paths))
401
if not PyString_CheckExact(path):
402
raise TypeError("you must pass a string for 'path' not: %s %r"
403
% (type(path), path))
408
path_cstr = PyString_AsString(path)
409
path_size = PyString_Size(path)
412
_mid = (_lo + _hi) / 2
413
cur = PyList_GetItem_object_void(paths, _mid)
414
cur_cstr = PyString_AS_STRING_void(cur)
415
cur_size = PyString_GET_SIZE_void(cur)
416
if _cmp_path_by_dirblock_intern(cur_cstr, cur_size,
417
path_cstr, path_size) < 0:
424
def _bisect_path_right(paths, path):
425
"""Return the index where to insert path into paths.
427
This uses a path-wise comparison so we get::
437
:param paths: A list of paths to search through
438
:param path: A single path to insert
439
:return: An offset where 'path' can be inserted.
440
:seealso: bisect.bisect_right
451
if not PyList_CheckExact(paths):
452
raise TypeError("you must pass a python list for 'paths' not: %s %r"
453
% (type(paths), paths))
454
if not PyString_CheckExact(path):
455
raise TypeError("you must pass a string for 'path' not: %s %r"
456
% (type(path), path))
461
path_cstr = PyString_AsString(path)
462
path_size = PyString_Size(path)
465
_mid = (_lo + _hi) / 2
466
cur = PyList_GetItem_object_void(paths, _mid)
467
cur_cstr = PyString_AS_STRING_void(cur)
468
cur_size = PyString_GET_SIZE_void(cur)
469
if _cmp_path_by_dirblock_intern(path_cstr, path_size,
470
cur_cstr, cur_size) < 0:
477
def bisect_dirblock(dirblocks, dirname, lo=0, hi=None, cache=None):
478
"""Return the index where to insert dirname into the dirblocks.
480
The return value idx is such that all directories blocks in dirblock[:idx]
481
have names < dirname, and all blocks in dirblock[idx:] have names >=
484
Optional args lo (default 0) and hi (default len(dirblocks)) bound the
485
slice of a to be searched.
490
cdef char *dirname_cstr
491
cdef int dirname_size
496
if not PyList_CheckExact(dirblocks):
497
raise TypeError("you must pass a python list for 'dirblocks' not: %s %r"
498
% (type(dirblocks), dirblocks))
499
if not PyString_CheckExact(dirname):
500
raise TypeError("you must pass a string for dirname not: %s %r"
501
% (type(dirname), dirname))
508
dirname_cstr = PyString_AsString(dirname)
509
dirname_size = PyString_Size(dirname)
512
_mid = (_lo + _hi) / 2
513
# Grab the dirname for the current dirblock
514
# cur = dirblocks[_mid][0]
515
cur = PyTuple_GetItem_void_void(
516
PyList_GetItem_object_void(dirblocks, _mid), 0)
517
cur_cstr = PyString_AS_STRING_void(cur)
518
cur_size = PyString_GET_SIZE_void(cur)
519
if _cmp_by_dirs(cur_cstr, cur_size, dirname_cstr, dirname_size) < 0:
527
"""Maintain the current location, and return fields as you parse them."""
529
cdef object state # The DirState object
530
cdef object text # The overall string object
531
cdef char *text_cstr # Pointer to the beginning of text
532
cdef int text_size # Length of text
534
cdef char *end_cstr # End of text
535
cdef char *cur_cstr # Pointer to the current record
536
cdef char *next # Pointer to the end of this record
538
def __init__(self, text, state):
541
self.text_cstr = PyString_AsString(text)
542
self.text_size = PyString_Size(text)
543
self.end_cstr = self.text_cstr + self.text_size
544
self.cur_cstr = self.text_cstr
546
cdef char *get_next(self, int *size) except NULL:
547
"""Return a pointer to the start of the next field."""
549
cdef Py_ssize_t extra_len
551
if self.cur_cstr == NULL:
552
raise AssertionError('get_next() called when cur_str is NULL')
553
elif self.cur_cstr >= self.end_cstr:
554
raise AssertionError('get_next() called when there are no chars'
557
self.cur_cstr = <char*>memchr(next, c'\0', self.end_cstr - next)
558
if self.cur_cstr == NULL:
559
extra_len = self.end_cstr - next
560
raise errors.DirstateCorrupt(self.state,
561
'failed to find trailing NULL (\\0).'
562
' Trailing garbage: %r'
563
% safe_string_from_size(next, extra_len))
564
size[0] = self.cur_cstr - next
565
self.cur_cstr = self.cur_cstr + 1
568
cdef object get_next_str(self):
569
"""Get the next field as a Python string."""
572
next = self.get_next(&size)
573
return safe_string_from_size(next, size)
575
cdef int _init(self) except -1:
576
"""Get the pointer ready.
578
This assumes that the dirstate header has already been read, and we
579
already have the dirblock string loaded into memory.
580
This just initializes our memory pointers, etc for parsing of the
585
# The first field should be an empty string left over from the Header
586
first = self.get_next(&size)
587
if first[0] != c'\0' and size == 0:
588
raise AssertionError('First character should be null not: %s'
592
cdef object _get_entry(self, int num_trees, void **p_current_dirname,
594
"""Extract the next entry.
596
This parses the next entry based on the current location in
598
Each entry can be considered a "row" in the total table. And each row
599
has a fixed number of columns. It is generally broken up into "key"
600
columns, then "current" columns, and then "parent" columns.
602
:param num_trees: How many parent trees need to be parsed
603
:param p_current_dirname: A pointer to the current PyString
604
representing the directory name.
605
We pass this in as a void * so that pyrex doesn't have to
606
increment/decrement the PyObject reference counter for each
608
We use a pointer so that _get_entry can update it with the new
610
:param new_block: This is to let the caller know that it needs to
611
create a new directory block to store the next entry.
613
cdef object path_name_file_id_key
614
cdef char *entry_size_cstr
615
cdef unsigned long int entry_size
616
cdef char* executable_cstr
617
cdef int is_executable
618
cdef char* dirname_cstr
623
cdef object fingerprint
626
# Read the 'key' information (dirname, name, file_id)
627
dirname_cstr = self.get_next(&cur_size)
628
# Check to see if we have started a new directory block.
629
# If so, then we need to create a new dirname PyString, so that it can
630
# be used in all of the tuples. This saves time and memory, by re-using
631
# the same object repeatedly.
633
# Do the cheap 'length of string' check first. If the string is a
634
# different length, then we *have* to be a different directory.
635
if (cur_size != PyString_GET_SIZE_void(p_current_dirname[0])
636
or strncmp(dirname_cstr,
637
# Extract the char* from our current dirname string. We
638
# know it is a PyString, so we can use
639
# PyString_AS_STRING, we use the _void version because
640
# we are tricking Pyrex by using a void* rather than an
642
PyString_AS_STRING_void(p_current_dirname[0]),
644
dirname = safe_string_from_size(dirname_cstr, cur_size)
645
p_current_dirname[0] = <void*>dirname
650
# Build up the key that will be used.
651
# By using <object>(void *) Pyrex will automatically handle the
652
# Py_INCREF that we need.
653
path_name_file_id_key = (<object>p_current_dirname[0],
658
# Parse all of the per-tree information. current has the information in
659
# the same location as parent trees. The only difference is that 'info'
660
# is a 'packed_stat' for current, while it is a 'revision_id' for
662
# minikind, fingerprint, and info will be returned as regular python
664
# entry_size and is_executable will be parsed into a python Long and
665
# python Boolean, respectively.
666
# TODO: jam 20070718 Consider changin the entry_size conversion to
667
# prefer python Int when possible. They are generally faster to
668
# work with, and it will be rare that we have a file >2GB.
669
# Especially since this code is pretty much fixed at a max of
672
for i from 0 <= i < num_trees:
673
minikind = self.get_next_str()
674
fingerprint = self.get_next_str()
675
entry_size_cstr = self.get_next(&cur_size)
676
entry_size = strtoul(entry_size_cstr, NULL, 10)
677
executable_cstr = self.get_next(&cur_size)
678
is_executable = (executable_cstr[0] == c'y')
679
info = self.get_next_str()
680
PyList_Append(trees, (
682
fingerprint, # fingerprint
684
is_executable,# executable
685
info, # packed_stat or revision_id
688
# The returned tuple is (key, [trees])
689
ret = (path_name_file_id_key, trees)
690
# Ignore the trailing newline, but assert that it does exist, this
691
# ensures that we always finish parsing a line on an end-of-entry
693
trailing = self.get_next(&cur_size)
694
if cur_size != 1 or trailing[0] != c'\n':
695
raise errors.DirstateCorrupt(self.state,
696
'Bad parse, we expected to end on \\n, not: %d %s: %s'
697
% (cur_size, safe_string_from_size(trailing, cur_size),
701
def _parse_dirblocks(self):
702
"""Parse all dirblocks in the state file."""
704
cdef object current_block
706
cdef void * current_dirname
708
cdef int expected_entry_count
711
num_trees = self.state._num_present_parents() + 1
712
expected_entry_count = self.state._num_entries
714
# Ignore the first record
718
dirblocks = [('', current_block), ('', [])]
719
self.state._dirblocks = dirblocks
721
current_dirname = <void*>obj
725
# TODO: jam 2007-05-07 Consider pre-allocating some space for the
726
# members, and then growing and shrinking from there. If most
727
# directories have close to 10 entries in them, it would save a
728
# few mallocs if we default our list size to something
729
# reasonable. Or we could malloc it to something large (100 or
730
# so), and then truncate. That would give us a malloc + realloc,
731
# rather than lots of reallocs.
732
while self.cur_cstr < self.end_cstr:
733
entry = self._get_entry(num_trees, ¤t_dirname, &new_block)
735
# new block - different dirname
737
PyList_Append(dirblocks,
738
(<object>current_dirname, current_block))
739
PyList_Append(current_block, entry)
740
entry_count = entry_count + 1
741
if entry_count != expected_entry_count:
742
raise errors.DirstateCorrupt(self.state,
743
'We read the wrong number of entries.'
744
' We expected to read %s, but read %s'
745
% (expected_entry_count, entry_count))
746
self.state._split_root_dirblock_into_contents()
749
def _read_dirblocks(state):
750
"""Read in the dirblocks for the given DirState object.
752
This is tightly bound to the DirState internal representation. It should be
753
thought of as a member function, which is only separated out so that we can
754
re-write it in pyrex.
756
:param state: A DirState object.
758
:postcondition: The dirblocks will be loaded into the appropriate fields in
761
state._state_file.seek(state._end_of_header)
762
text = state._state_file.read()
763
# TODO: check the crc checksums. crc_measured = zlib.crc32(text)
765
reader = Reader(text, state)
767
reader._parse_dirblocks()
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 readonly object uninteresting
969
cdef object last_source_parent
970
cdef object last_target_parent
971
cdef object include_unchanged
972
cdef object use_filesystem_for_exec
973
cdef object utf8_decode
974
cdef readonly object searched_specific_files
975
cdef object search_specific_files
977
# Current iteration variables:
978
cdef object current_root
979
cdef object current_root_unicode
980
cdef object root_entries
981
cdef int root_entries_pos, root_entries_len
982
cdef object root_abspath
983
cdef int source_index, target_index
984
cdef int want_unversioned
986
cdef object dir_iterator
988
cdef object current_block
989
cdef int current_block_pos
990
cdef object current_block_list
991
cdef object current_dir_info
992
cdef object current_dir_list
994
cdef object root_dir_info
995
cdef object bisect_left
1000
def __init__(self, include_unchanged, use_filesystem_for_exec,
1001
search_specific_files, state, source_index, target_index,
1002
want_unversioned, tree):
1003
self.old_dirname_to_file_id = {}
1004
self.new_dirname_to_file_id = {}
1005
# Just a sentry, so that _process_entry can say that this
1006
# record is handled, but isn't interesting to process (unchanged)
1007
self.uninteresting = object()
1008
# Using a list so that we can access the values and change them in
1009
# nested scope. Each one is [path, file_id, entry]
1010
self.last_source_parent = [None, None]
1011
self.last_target_parent = [None, None]
1012
self.include_unchanged = include_unchanged
1013
self.use_filesystem_for_exec = use_filesystem_for_exec
1014
self.utf8_decode = cache_utf8._utf8_decode
1015
# for all search_indexs in each path at or under each element of
1016
# search_specific_files, if the detail is relocated: add the id, and add the
1017
# relocated path as one to search if its not searched already. If the
1018
# detail is not relocated, add the id.
1019
self.searched_specific_files = set()
1020
self.search_specific_files = search_specific_files
1022
self.current_root = None
1023
self.current_root_unicode = None
1024
self.root_entries = None
1025
self.root_entries_pos = 0
1026
self.root_entries_len = 0
1027
self.root_abspath = None
1028
if source_index is None:
1029
self.source_index = -1
1031
self.source_index = source_index
1032
self.target_index = target_index
1033
self.want_unversioned = want_unversioned
1035
self.dir_iterator = None
1036
self.block_index = -1
1037
self.current_block = None
1038
self.current_block_list = None
1039
self.current_block_pos = -1
1040
self.current_dir_info = None
1041
self.current_dir_list = None
1043
self.root_dir_info = None
1044
self.bisect_left = bisect.bisect_left
1045
self.pathjoin = osutils.pathjoin
1046
self.fstat = os.fstat
1047
self.sha_file = osutils.sha_file
1049
cdef _process_entry(self, entry, path_info):
1050
"""Compare an entry and real disk to generate delta information.
1052
:param path_info: top_relpath, basename, kind, lstat, abspath for
1053
the path of entry. If None, then the path is considered absent.
1054
(Perhaps we should pass in a concrete entry for this ?)
1055
Basename is returned as a utf8 string because we expect this
1056
tuple will be ignored, and don't want to take the time to
1058
:return: None if the these don't match
1059
A tuple of information about the change, or
1060
the object 'uninteresting' if these match, but are
1061
basically identical.
1063
cdef char target_minikind
1064
cdef char source_minikind
1066
cdef int content_change
1067
cdef object details_list
1069
details_list = entry[1]
1070
if -1 == self.source_index:
1071
source_details = DirState.NULL_PARENT_DETAILS
1073
source_details = details_list[self.source_index]
1074
target_details = details_list[self.target_index]
1075
target_minikind = _minikind_from_string(target_details[0])
1076
if path_info is not None and _versioned_minikind(target_minikind):
1077
if self.target_index != 0:
1078
raise AssertionError("Unsupported target index %d" %
1080
link_or_sha1 = _update_entry(self.state, entry, path_info[4], path_info[3])
1081
# The entry may have been modified by update_entry
1082
target_details = details_list[self.target_index]
1083
target_minikind = _minikind_from_string(target_details[0])
1086
# the rest of this function is 0.3 seconds on 50K paths, or
1087
# 0.000006 seconds per call.
1088
source_minikind = _minikind_from_string(source_details[0])
1089
if ((_versioned_minikind(source_minikind) or source_minikind == c'r')
1090
and _versioned_minikind(target_minikind)):
1091
# claimed content in both: diff
1092
# r | fdlt | | add source to search, add id path move and perform
1093
# | | | diff check on source-target
1094
# r | fdlt | a | dangling file that was present in the basis.
1096
if source_minikind != c'r':
1097
old_dirname = entry[0][0]
1098
old_basename = entry[0][1]
1099
old_path = path = None
1101
# add the source to the search path to find any children it
1102
# has. TODO ? : only add if it is a container ?
1103
if not osutils.is_inside_any(self.searched_specific_files,
1105
self.search_specific_files.add(source_details[1])
1106
# generate the old path; this is needed for stating later
1108
old_path = source_details[1]
1109
old_dirname, old_basename = os.path.split(old_path)
1110
path = self.pathjoin(entry[0][0], entry[0][1])
1111
old_entry = self.state._get_entry(self.source_index,
1113
# update the source details variable to be the real
1115
if old_entry == (None, None):
1116
raise errors.CorruptDirstate(self.state._filename,
1117
"entry '%s/%s' is considered renamed from %r"
1118
" but source does not exist\n"
1119
"entry: %s" % (entry[0][0], entry[0][1], old_path, entry))
1120
source_details = old_entry[1][self.source_index]
1121
source_minikind = _minikind_from_string(source_details[0])
1122
if path_info is None:
1123
# the file is missing on disk, show as removed.
1128
# source and target are both versioned and disk file is present.
1129
target_kind = path_info[2]
1130
if target_kind == 'directory':
1132
old_path = path = self.pathjoin(old_dirname, old_basename)
1133
file_id = entry[0][2]
1134
self.new_dirname_to_file_id[path] = file_id
1135
if source_minikind != c'd':
1138
# directories have no fingerprint
1141
elif target_kind == 'file':
1142
if source_minikind != c'f':
1145
# Check the sha. We can't just rely on the size as
1146
# content filtering may mean differ sizes actually
1147
# map to the same content
1148
if link_or_sha1 is None:
1150
statvalue, link_or_sha1 = \
1151
self.state._sha1_provider.stat_and_sha1(
1153
self.state._observed_sha1(entry, link_or_sha1,
1155
content_change = (link_or_sha1 != source_details[1])
1156
# Target details is updated at update_entry time
1157
if self.use_filesystem_for_exec:
1158
# We don't need S_ISREG here, because we are sure
1159
# we are dealing with a file.
1160
target_exec = bool(S_IXUSR & path_info[3].st_mode)
1162
target_exec = target_details[3]
1163
elif target_kind == 'symlink':
1164
if source_minikind != c'l':
1167
content_change = (link_or_sha1 != source_details[1])
1169
elif target_kind == 'tree-reference':
1170
if source_minikind != c't':
1176
raise Exception, "unknown kind %s" % path_info[2]
1177
if source_minikind == c'd':
1179
old_path = path = self.pathjoin(old_dirname, old_basename)
1181
file_id = entry[0][2]
1182
self.old_dirname_to_file_id[old_path] = file_id
1183
# parent id is the entry for the path in the target tree
1184
if old_dirname == self.last_source_parent[0]:
1185
source_parent_id = self.last_source_parent[1]
1188
source_parent_id = self.old_dirname_to_file_id[old_dirname]
1190
source_parent_entry = self.state._get_entry(self.source_index,
1191
path_utf8=old_dirname)
1192
source_parent_id = source_parent_entry[0][2]
1193
if source_parent_id == entry[0][2]:
1194
# This is the root, so the parent is None
1195
source_parent_id = None
1197
self.last_source_parent[0] = old_dirname
1198
self.last_source_parent[1] = source_parent_id
1199
new_dirname = entry[0][0]
1200
if new_dirname == self.last_target_parent[0]:
1201
target_parent_id = self.last_target_parent[1]
1204
target_parent_id = self.new_dirname_to_file_id[new_dirname]
1206
# TODO: We don't always need to do the lookup, because the
1207
# parent entry will be the same as the source entry.
1208
target_parent_entry = self.state._get_entry(self.target_index,
1209
path_utf8=new_dirname)
1210
if target_parent_entry == (None, None):
1211
raise AssertionError(
1212
"Could not find target parent in wt: %s\nparent of: %s"
1213
% (new_dirname, entry))
1214
target_parent_id = target_parent_entry[0][2]
1215
if target_parent_id == entry[0][2]:
1216
# This is the root, so the parent is None
1217
target_parent_id = None
1219
self.last_target_parent[0] = new_dirname
1220
self.last_target_parent[1] = target_parent_id
1222
source_exec = source_details[3]
1223
if (self.include_unchanged
1225
or source_parent_id != target_parent_id
1226
or old_basename != entry[0][1]
1227
or source_exec != target_exec
1229
if old_path is None:
1230
path = self.pathjoin(old_dirname, old_basename)
1232
old_path_u = self.utf8_decode(old_path)[0]
1235
old_path_u = self.utf8_decode(old_path)[0]
1236
if old_path == path:
1239
path_u = self.utf8_decode(path)[0]
1240
source_kind = _minikind_to_kind(source_minikind)
1241
return (entry[0][2],
1242
(old_path_u, path_u),
1245
(source_parent_id, target_parent_id),
1246
(self.utf8_decode(old_basename)[0], self.utf8_decode(entry[0][1])[0]),
1247
(source_kind, target_kind),
1248
(source_exec, target_exec))
1250
return self.uninteresting
1251
elif source_minikind == c'a' and _versioned_minikind(target_minikind):
1252
# looks like a new file
1253
path = self.pathjoin(entry[0][0], entry[0][1])
1254
# parent id is the entry for the path in the target tree
1255
# TODO: these are the same for an entire directory: cache em.
1256
parent_entry = self.state._get_entry(self.target_index,
1257
path_utf8=entry[0][0])
1258
if parent_entry is None:
1259
raise errors.DirstateCorrupt(self.state,
1260
"We could not find the parent entry in index %d"
1261
" for the entry: %s"
1262
% (self.target_index, entry[0]))
1263
parent_id = parent_entry[0][2]
1264
if parent_id == entry[0][2]:
1266
if path_info is not None:
1268
if self.use_filesystem_for_exec:
1269
# We need S_ISREG here, because we aren't sure if this
1272
S_ISREG(path_info[3].st_mode)
1273
and S_IXUSR & path_info[3].st_mode)
1275
target_exec = target_details[3]
1276
return (entry[0][2],
1277
(None, self.utf8_decode(path)[0]),
1281
(None, self.utf8_decode(entry[0][1])[0]),
1282
(None, path_info[2]),
1283
(None, target_exec))
1285
# Its a missing file, report it as such.
1286
return (entry[0][2],
1287
(None, self.utf8_decode(path)[0]),
1291
(None, self.utf8_decode(entry[0][1])[0]),
1294
elif _versioned_minikind(source_minikind) and target_minikind == c'a':
1295
# unversioned, possibly, or possibly not deleted: we dont care.
1296
# if its still on disk, *and* theres no other entry at this
1297
# path [we dont know this in this routine at the moment -
1298
# perhaps we should change this - then it would be an unknown.
1299
old_path = self.pathjoin(entry[0][0], entry[0][1])
1300
# parent id is the entry for the path in the target tree
1301
parent_id = self.state._get_entry(self.source_index, path_utf8=entry[0][0])[0][2]
1302
if parent_id == entry[0][2]:
1304
return (entry[0][2],
1305
(self.utf8_decode(old_path)[0], None),
1309
(self.utf8_decode(entry[0][1])[0], None),
1310
(_minikind_to_kind(source_minikind), None),
1311
(source_details[3], None))
1312
elif _versioned_minikind(source_minikind) and target_minikind == c'r':
1313
# a rename; could be a true rename, or a rename inherited from
1314
# a renamed parent. TODO: handle this efficiently. Its not
1315
# common case to rename dirs though, so a correct but slow
1316
# implementation will do.
1317
if not osutils.is_inside_any(self.searched_specific_files, target_details[1]):
1318
self.search_specific_files.add(target_details[1])
1319
elif ((source_minikind == c'r' or source_minikind == c'a') and
1320
(target_minikind == c'r' or target_minikind == c'a')):
1321
# neither of the selected trees contain this path,
1322
# so skip over it. This is not currently directly tested, but
1323
# is indirectly via test_too_much.TestCommands.test_conflicts.
1326
raise AssertionError("don't know how to compare "
1327
"source_minikind=%r, target_minikind=%r"
1328
% (source_minikind, target_minikind))
1329
## import pdb;pdb.set_trace()
1335
def iter_changes(self):
1338
cdef void _update_current_block(self):
1339
if (self.block_index < len(self.state._dirblocks) and
1340
osutils.is_inside(self.current_root, self.state._dirblocks[self.block_index][0])):
1341
self.current_block = self.state._dirblocks[self.block_index]
1342
self.current_block_list = self.current_block[1]
1343
self.current_block_pos = 0
1345
self.current_block = None
1346
self.current_block_list = None
1349
# Simple thunk to allow tail recursion without pyrex confusion
1350
return self._iter_next()
1352
cdef _iter_next(self):
1353
"""Iterate over the changes."""
1354
# This function single steps through an iterator. As such while loops
1355
# are often exited by 'return' - the code is structured so that the
1356
# next call into the function will return to the same while loop. Note
1357
# that all flow control needed to re-reach that step is reexecuted,
1358
# which can be a performance problem. It has not yet been tuned to
1359
# minimise this; a state machine is probably the simplest restructuring
1360
# to both minimise this overhead and make the code considerably more
1364
# compare source_index and target_index at or under each element of search_specific_files.
1365
# follow the following comparison table. Note that we only want to do diff operations when
1366
# the target is fdl because thats when the walkdirs logic will have exposed the pathinfo
1370
# Source | Target | disk | action
1371
# r | fdlt | | add source to search, add id path move and perform
1372
# | | | diff check on source-target
1373
# r | fdlt | a | dangling file that was present in the basis.
1375
# r | a | | add source to search
1377
# r | r | | this path is present in a non-examined tree, skip.
1378
# r | r | a | this path is present in a non-examined tree, skip.
1379
# a | fdlt | | add new id
1380
# a | fdlt | a | dangling locally added file, skip
1381
# a | a | | not present in either tree, skip
1382
# a | a | a | not present in any tree, skip
1383
# a | r | | not present in either tree at this path, skip as it
1384
# | | | may not be selected by the users list of paths.
1385
# a | r | a | not present in either tree at this path, skip as it
1386
# | | | may not be selected by the users list of paths.
1387
# fdlt | fdlt | | content in both: diff them
1388
# fdlt | fdlt | a | deleted locally, but not unversioned - show as deleted ?
1389
# fdlt | a | | unversioned: output deleted id for now
1390
# fdlt | a | a | unversioned and deleted: output deleted id
1391
# fdlt | r | | relocated in this tree, so add target to search.
1392
# | | | Dont diff, we will see an r,fd; pair when we reach
1393
# | | | this id at the other path.
1394
# fdlt | r | a | relocated in this tree, so add target to search.
1395
# | | | Dont diff, we will see an r,fd; pair when we reach
1396
# | | | this id at the other path.
1398
# TODO: jam 20070516 - Avoid the _get_entry lookup overhead by
1399
# keeping a cache of directories that we have seen.
1400
cdef object current_dirname, current_blockname
1401
cdef char * current_dirname_c, * current_blockname_c
1402
cdef int advance_entry, advance_path
1403
cdef int path_handled
1404
uninteresting = self.uninteresting
1405
searched_specific_files = self.searched_specific_files
1406
# Are we walking a root?
1407
while self.root_entries_pos < self.root_entries_len:
1408
entry = self.root_entries[self.root_entries_pos]
1409
self.root_entries_pos = self.root_entries_pos + 1
1410
result = self._process_entry(entry, self.root_dir_info)
1411
if result is not None and result is not self.uninteresting:
1413
# Have we finished the prior root, or never started one ?
1414
if self.current_root is None:
1415
# TODO: the pending list should be lexically sorted? the
1416
# interface doesn't require it.
1418
self.current_root = self.search_specific_files.pop()
1420
raise StopIteration()
1421
self.current_root_unicode = self.current_root.decode('utf8')
1422
self.searched_specific_files.add(self.current_root)
1423
# process the entries for this containing directory: the rest will be
1424
# found by their parents recursively.
1425
self.root_entries = self.state._entries_for_path(self.current_root)
1426
self.root_entries_len = len(self.root_entries)
1427
self.root_abspath = self.tree.abspath(self.current_root_unicode)
1429
root_stat = os.lstat(self.root_abspath)
1431
if e.errno == errno.ENOENT:
1432
# the path does not exist: let _process_entry know that.
1433
self.root_dir_info = None
1435
# some other random error: hand it up.
1438
self.root_dir_info = ('', self.current_root,
1439
osutils.file_kind_from_stat_mode(root_stat.st_mode), root_stat,
1441
if self.root_dir_info[2] == 'directory':
1442
if self.tree._directory_is_tree_reference(
1443
self.current_root_unicode):
1444
self.root_dir_info = self.root_dir_info[:2] + \
1445
('tree-reference',) + self.root_dir_info[3:]
1446
if not self.root_entries and not self.root_dir_info:
1447
# this specified path is not present at all, skip it.
1448
# (tail recursion, can do a loop once the full structure is
1450
return self._iter_next()
1452
self.root_entries_pos = 0
1453
# XXX Clarity: This loop is duplicated a out the self.current_root
1454
# is None guard above: if we return from it, it completes there
1455
# (and the following if block cannot trigger because
1456
# path_handled must be true, so the if block is not # duplicated.
1457
while self.root_entries_pos < self.root_entries_len:
1458
entry = self.root_entries[self.root_entries_pos]
1459
self.root_entries_pos = self.root_entries_pos + 1
1460
result = self._process_entry(entry, self.root_dir_info)
1461
if result is not None:
1463
if result is not self.uninteresting:
1465
# handle unversioned specified paths:
1466
if self.want_unversioned and not path_handled and self.root_dir_info:
1467
new_executable = bool(
1468
stat.S_ISREG(self.root_dir_info[3].st_mode)
1469
and stat.S_IEXEC & self.root_dir_info[3].st_mode)
1471
(None, self.current_root_unicode),
1475
(None, splitpath(self.current_root_unicode)[-1]),
1476
(None, self.root_dir_info[2]),
1477
(None, new_executable)
1479
# If we reach here, the outer flow continues, which enters into the
1480
# per-root setup logic.
1481
if self.current_dir_info is None and self.current_block is None:
1482
# setup iteration of this root:
1483
self.current_dir_list = None
1484
if self.root_dir_info and self.root_dir_info[2] == 'tree-reference':
1485
self.current_dir_info = None
1487
self.dir_iterator = osutils._walkdirs_utf8(self.root_abspath,
1488
prefix=self.current_root)
1491
self.current_dir_info = self.dir_iterator.next()
1492
self.current_dir_list = self.current_dir_info[1]
1494
# there may be directories in the inventory even though
1495
# this path is not a file on disk: so mark it as end of
1497
if e.errno in (errno.ENOENT, errno.ENOTDIR, errno.EINVAL):
1498
self.current_dir_info = None
1499
elif sys.platform == 'win32':
1500
# on win32, python2.4 has e.errno == ERROR_DIRECTORY, but
1501
# python 2.5 has e.errno == EINVAL,
1502
# and e.winerror == ERROR_DIRECTORY
1504
e_winerror = e.winerror
1505
except AttributeError:
1507
win_errors = (ERROR_DIRECTORY, ERROR_PATH_NOT_FOUND)
1508
if (e.errno in win_errors or e_winerror in win_errors):
1509
self.current_dir_info = None
1511
# Will this really raise the right exception ?
1516
if self.current_dir_info[0][0] == '':
1517
# remove .bzr from iteration
1518
bzr_index = self.bisect_left(self.current_dir_list, ('.bzr',))
1519
if self.current_dir_list[bzr_index][0] != '.bzr':
1520
raise AssertionError()
1521
del self.current_dir_list[bzr_index]
1522
initial_key = (self.current_root, '', '')
1523
self.block_index, _ = self.state._find_block_index_from_key(initial_key)
1524
if self.block_index == 0:
1525
# we have processed the total root already, but because the
1526
# initial key matched it we should skip it here.
1527
self.block_index = self.block_index + 1
1528
self._update_current_block()
1529
# walk until both the directory listing and the versioned metadata
1531
while (self.current_dir_info is not None
1532
or self.current_block is not None):
1533
# Uncommon case - a missing directory or an unversioned directory:
1534
if (self.current_dir_info and self.current_block
1535
and self.current_dir_info[0][0] != self.current_block[0]):
1536
# Work around pyrex broken heuristic - current_dirname has
1537
# the same scope as current_dirname_c
1538
current_dirname = self.current_dir_info[0][0]
1539
current_dirname_c = PyString_AS_STRING_void(
1540
<void *>current_dirname)
1541
current_blockname = self.current_block[0]
1542
current_blockname_c = PyString_AS_STRING_void(
1543
<void *>current_blockname)
1544
# In the python generator we evaluate this if block once per
1545
# dir+block; because we reenter in the pyrex version its being
1546
# evaluated once per path: we could cache the result before
1547
# doing the while loop and probably save time.
1548
if _cmp_by_dirs(current_dirname_c,
1549
PyString_Size(current_dirname),
1550
current_blockname_c,
1551
PyString_Size(current_blockname)) < 0:
1552
# filesystem data refers to paths not covered by the
1553
# dirblock. this has two possibilities:
1554
# A) it is versioned but empty, so there is no block for it
1555
# B) it is not versioned.
1557
# if (A) then we need to recurse into it to check for
1558
# new unknown files or directories.
1559
# if (B) then we should ignore it, because we don't
1560
# recurse into unknown directories.
1561
# We are doing a loop
1562
while self.path_index < len(self.current_dir_list):
1563
current_path_info = self.current_dir_list[self.path_index]
1564
# dont descend into this unversioned path if it is
1566
if current_path_info[2] in ('directory',
1568
del self.current_dir_list[self.path_index]
1569
self.path_index = self.path_index - 1
1570
self.path_index = self.path_index + 1
1571
if self.want_unversioned:
1572
if current_path_info[2] == 'directory':
1573
if self.tree._directory_is_tree_reference(
1574
self.utf8_decode(current_path_info[0])[0]):
1575
current_path_info = current_path_info[:2] + \
1576
('tree-reference',) + current_path_info[3:]
1577
new_executable = bool(
1578
stat.S_ISREG(current_path_info[3].st_mode)
1579
and stat.S_IEXEC & current_path_info[3].st_mode)
1581
(None, self.utf8_decode(current_path_info[0])[0]),
1585
(None, self.utf8_decode(current_path_info[1])[0]),
1586
(None, current_path_info[2]),
1587
(None, new_executable))
1588
# This dir info has been handled, go to the next
1590
self.current_dir_list = None
1592
self.current_dir_info = self.dir_iterator.next()
1593
self.current_dir_list = self.current_dir_info[1]
1594
except StopIteration:
1595
self.current_dir_info = None
1597
# We have a dirblock entry for this location, but there
1598
# is no filesystem path for this. This is most likely
1599
# because a directory was removed from the disk.
1600
# We don't have to report the missing directory,
1601
# because that should have already been handled, but we
1602
# need to handle all of the files that are contained
1604
while self.current_block_pos < len(self.current_block_list):
1605
current_entry = self.current_block_list[self.current_block_pos]
1606
self.current_block_pos = self.current_block_pos + 1
1607
# entry referring to file not present on disk.
1608
# advance the entry only, after processing.
1609
result = self._process_entry(current_entry, None)
1610
if result is not None:
1611
if result is not self.uninteresting:
1613
self.block_index = self.block_index + 1
1614
self._update_current_block()
1615
continue # next loop-on-block/dir
1616
result = self._loop_one_block()
1617
if result is not None:
1619
if len(self.search_specific_files):
1620
# More supplied paths to process
1621
self.current_root = None
1622
return self._iter_next()
1623
raise StopIteration()
1625
cdef object _maybe_tree_ref(self, current_path_info):
1626
if self.tree._directory_is_tree_reference(
1627
self.utf8_decode(current_path_info[0])[0]):
1628
return current_path_info[:2] + \
1629
('tree-reference',) + current_path_info[3:]
1631
return current_path_info
1633
cdef object _loop_one_block(self):
1634
# current_dir_info and current_block refer to the same directory -
1635
# this is the common case code.
1636
# Assign local variables for current path and entry:
1637
cdef object current_entry
1638
cdef object current_path_info
1639
cdef int path_handled
1642
# cdef char * temp_str
1643
# cdef Py_ssize_t temp_str_length
1644
# PyString_AsStringAndSize(disk_kind, &temp_str, &temp_str_length)
1645
# if not strncmp(temp_str, "directory", temp_str_length):
1646
if (self.current_block is not None and
1647
self.current_block_pos < PyList_GET_SIZE(self.current_block_list)):
1648
current_entry = PyList_GET_ITEM(self.current_block_list,
1649
self.current_block_pos)
1651
Py_INCREF(current_entry)
1653
current_entry = None
1654
if (self.current_dir_info is not None and
1655
self.path_index < PyList_GET_SIZE(self.current_dir_list)):
1656
current_path_info = PyList_GET_ITEM(self.current_dir_list,
1659
Py_INCREF(current_path_info)
1660
disk_kind = PyTuple_GET_ITEM(current_path_info, 2)
1662
Py_INCREF(disk_kind)
1663
if disk_kind == "directory":
1664
current_path_info = self._maybe_tree_ref(current_path_info)
1666
current_path_info = None
1667
while (current_entry is not None or current_path_info is not None):
1672
if current_entry is None:
1673
# unversioned - the check for path_handled when the path
1674
# is advanced will yield this path if needed.
1676
elif current_path_info is None:
1677
# no path is fine: the per entry code will handle it.
1678
result = self._process_entry(current_entry, current_path_info)
1679
if result is not None:
1680
if result is self.uninteresting:
1683
minikind = _minikind_from_string(
1684
current_entry[1][self.target_index][0])
1685
cmp_result = cmp(current_path_info[1], current_entry[0][1])
1686
if (cmp_result or minikind == c'a' or minikind == c'r'):
1687
# The current path on disk doesn't match the dirblock
1688
# record. Either the dirblock record is marked as
1689
# absent/renamed, or the file on disk is not present at all
1690
# in the dirblock. Either way, report about the dirblock
1691
# entry, and let other code handle the filesystem one.
1693
# Compare the basename for these files to determine
1696
# extra file on disk: pass for now, but only
1697
# increment the path, not the entry
1700
# entry referring to file not present on disk.
1701
# advance the entry only, after processing.
1702
result = self._process_entry(current_entry, None)
1703
if result is not None:
1704
if result is self.uninteresting:
1708
# paths are the same,and the dirstate entry is not
1709
# absent or renamed.
1710
result = self._process_entry(current_entry, current_path_info)
1711
if result is not None:
1713
if result is self.uninteresting:
1715
# >- loop control starts here:
1717
if advance_entry and current_entry is not None:
1718
self.current_block_pos = self.current_block_pos + 1
1719
if self.current_block_pos < PyList_GET_SIZE(self.current_block_list):
1720
current_entry = self.current_block_list[self.current_block_pos]
1722
current_entry = None
1724
if advance_path and current_path_info is not None:
1725
if not path_handled:
1726
# unversioned in all regards
1727
if self.want_unversioned:
1728
new_executable = bool(
1729
stat.S_ISREG(current_path_info[3].st_mode)
1730
and stat.S_IEXEC & current_path_info[3].st_mode)
1732
relpath_unicode = self.utf8_decode(current_path_info[0])[0]
1733
except UnicodeDecodeError:
1734
raise errors.BadFilenameEncoding(
1735
current_path_info[0], osutils._fs_enc)
1736
if result is not None:
1737
raise AssertionError(
1738
"result is not None: %r" % result)
1740
(None, relpath_unicode),
1744
(None, self.utf8_decode(current_path_info[1])[0]),
1745
(None, current_path_info[2]),
1746
(None, new_executable))
1747
# dont descend into this unversioned path if it is
1749
if current_path_info[2] in ('directory'):
1750
del self.current_dir_list[self.path_index]
1751
self.path_index = self.path_index - 1
1752
# dont descend the disk iterator into any tree
1754
if current_path_info[2] == 'tree-reference':
1755
del self.current_dir_list[self.path_index]
1756
self.path_index = self.path_index - 1
1757
self.path_index = self.path_index + 1
1758
if self.path_index < len(self.current_dir_list):
1759
current_path_info = self.current_dir_list[self.path_index]
1760
if current_path_info[2] == 'directory':
1761
current_path_info = self._maybe_tree_ref(
1764
current_path_info = None
1765
if result is not None:
1766
# Found a result on this pass, yield it
1768
if self.current_block is not None:
1769
self.block_index = self.block_index + 1
1770
self._update_current_block()
1771
if self.current_dir_info is not None:
1773
self.current_dir_list = None
1775
self.current_dir_info = self.dir_iterator.next()
1776
self.current_dir_list = self.current_dir_info[1]
1777
except StopIteration:
1778
self.current_dir_info = None
added
removed
bzrlib/_dirstate_helpers_pyx.pyx
