1
# Copyright (C) 2008, 2009, 2010 Canonical Ltd
1
# Copyright (C) 2009 Canonical Ltd
3
3
# This program is free software; you can redistribute it and/or modify
4
4
# it under the terms of the GNU General Public License as published by
24
24
cdef extern from "Python.h":
25
ctypedef struct PyObject:
27
25
ctypedef int Py_ssize_t # Required for older pyrex versions
28
26
int PyString_CheckExact(object)
29
27
char * PyString_AS_STRING(object)
46
44
unsigned long agg_offset
47
45
struct delta_index:
49
ctypedef enum delta_result:
57
delta_result create_delta_index(source_info *src,
60
int max_entries) nogil
61
delta_result create_delta_index_from_delta(source_info *delta,
63
delta_index **fresh) nogil
47
delta_index * create_delta_index(source_info *src, delta_index *old) nogil
48
delta_index * create_delta_index_from_delta(source_info *delta,
49
delta_index *old) nogil
64
50
void free_delta_index(delta_index *index) nogil
65
delta_result create_delta(delta_index *indexes,
66
void *buf, unsigned long bufsize,
67
unsigned long *delta_size,
68
unsigned long max_delta_size,
69
void **delta_data) nogil
51
void *create_delta(delta_index *indexes,
52
void *buf, unsigned long bufsize,
53
unsigned long *delta_size, unsigned long max_delta_size) nogil
70
54
unsigned long get_delta_hdr_size(unsigned char **datap,
71
55
unsigned char *top) nogil
72
unsigned long sizeof_delta_index(delta_index *index)
73
56
Py_ssize_t DELTA_SIZE_MIN
74
int get_hash_offset(delta_index *index, int pos, unsigned int *hash_offset)
75
int get_entry_summary(delta_index *index, int pos,
76
unsigned int *global_offset, unsigned int *hash_val)
77
unsigned int rabin_hash (unsigned char *data)
80
59
cdef void *safe_malloc(size_t count) except NULL:
104
83
return DeltaIndex(source)
107
cdef object _translate_delta_failure(delta_result result):
108
if result == DELTA_OUT_OF_MEMORY:
109
return MemoryError("Delta function failed to allocate memory")
110
elif result == DELTA_INDEX_NEEDED:
111
return ValueError("Delta function requires delta_index param")
112
elif result == DELTA_SOURCE_EMPTY:
113
return ValueError("Delta function given empty source_info param")
114
elif result == DELTA_SOURCE_BAD:
115
return RuntimeError("Delta function given invalid source_info param")
116
elif result == DELTA_BUFFER_EMPTY:
117
return ValueError("Delta function given empty buffer params")
118
return AssertionError("Unrecognised delta result code: %d" % result)
121
def _rabin_hash(content):
122
if not PyString_CheckExact(content):
123
raise ValueError('content must be a string')
124
if len(content) < 16:
125
raise ValueError('content must be at least 16 bytes long')
126
# Try to cast it to an int, if it can fit
127
return int(rabin_hash(<unsigned char*>(PyString_AS_STRING(content))))
130
86
cdef class DeltaIndex:
132
88
# We need Pyrex 0.9.8+ to understand a 'list' definition, and this object
135
91
cdef readonly object _sources
136
92
cdef source_info *_source_infos
137
93
cdef delta_index *_index
94
cdef readonly unsigned int _max_num_sources
138
95
cdef public unsigned long _source_offset
139
cdef readonly unsigned int _max_num_sources
140
cdef public int _max_bytes_to_index
142
def __init__(self, source=None, max_bytes_to_index=None):
97
def __init__(self, source=None):
143
98
self._sources = []
144
99
self._index = NULL
145
100
self._max_num_sources = 65000
146
101
self._source_infos = <source_info *>safe_malloc(sizeof(source_info)
147
102
* self._max_num_sources)
148
103
self._source_offset = 0
149
self._max_bytes_to_index = 0
150
if max_bytes_to_index is not None:
151
self._max_bytes_to_index = max_bytes_to_index
153
105
if source is not None:
154
106
self.add_source(source, 0)
156
def __sizeof__(self):
157
# We want to track the _source_infos allocations, but the referenced
158
# void* are actually tracked in _sources itself.
159
# XXX: Cython is capable of doing sizeof(class) and returning the size
160
# of the underlying struct. Pyrex (<= 0.9.9) refuses, so we need
161
# to do it manually. *sigh* Note that we might get it wrong
162
# because of alignment issues.
164
# PyObject start, vtable *, 3 object pointers, 2 C ints
165
size = ((sizeof(PyObject) + sizeof(void*) + 3*sizeof(PyObject*)
166
+ sizeof(unsigned long)
167
+ sizeof(unsigned int))
168
+ (sizeof(source_info) * self._max_num_sources)
169
+ sizeof_delta_index(self._index))
172
108
def __repr__(self):
173
109
return '%s(%d, %d)' % (self.__class__.__name__,
174
110
len(self._sources), self._source_offset)
182
118
def _has_index(self):
183
119
return (self._index != NULL)
185
def _dump_index(self):
186
"""Dump the pointers in the index.
188
This is an arbitrary layout, used for testing. It is not meant to be
189
used in production code.
191
:return: (hash_list, entry_list)
192
hash_list A list of offsets, so hash[i] points to the 'hash
193
bucket' starting at the given offset and going until
195
entry_list A list of (text_offset, hash_val). text_offset is the
196
offset in the "source" texts, and hash_val is the RABIN
197
hash for that offset.
198
Note that the entry should be in the hash bucket
200
hash[(hash_val & mask)] && hash[(hash_val & mask) + 1]
203
cdef unsigned int text_offset
204
cdef unsigned int hash_val
205
cdef unsigned int hash_offset
206
if self._index == NULL:
210
while get_hash_offset(self._index, pos, &hash_offset):
211
hash_list.append(int(hash_offset))
215
while get_entry_summary(self._index, pos, &text_offset, &hash_val):
216
# Map back using 'int' so that we don't get Long everywhere, when
217
# almost everything is <2**31.
218
val = tuple(map(int, [text_offset, hash_val]))
219
entry_list.append(val)
221
return hash_list, entry_list
223
121
def add_delta_source(self, delta, unadded_bytes):
224
122
"""Add a new delta to the source texts.
249
146
src.size = c_delta_size
250
147
src.agg_offset = self._source_offset + unadded_bytes
252
res = create_delta_index_from_delta(src, self._index, &index)
254
raise _translate_delta_failure(res)
149
index = create_delta_index_from_delta(src, self._index)
255
150
self._source_offset = src.agg_offset + src.size
256
if index != self._index:
257
152
free_delta_index(self._index)
258
153
self._index = index
263
158
:param source: The text in question, this must be a byte string
264
159
:param unadded_bytes: Assume there are this many bytes that didn't get
265
160
added between this source and the end of the previous source.
266
:param max_pointers: Add no more than this many entries to the index.
267
By default, we sample every 16 bytes, if that would require more
268
than max_entries, we will reduce the sampling rate.
269
A value of 0 means unlimited, None means use the default limit.
271
162
cdef char *c_source
272
163
cdef Py_ssize_t c_source_size
273
164
cdef delta_index *index
274
cdef delta_result res
275
165
cdef unsigned int source_location
276
166
cdef source_info *src
277
167
cdef unsigned int num_indexes
278
cdef int max_num_entries
280
169
if not PyString_CheckExact(source):
281
170
raise TypeError('source is not a str')
298
187
# We delay creating the index on the first insert
299
188
if source_location != 0:
301
res = create_delta_index(src, self._index, &index,
302
self._max_bytes_to_index)
304
raise _translate_delta_failure(res)
305
if index != self._index:
190
index = create_delta_index(src, self._index)
306
192
free_delta_index(self._index)
307
193
self._index = index
309
195
cdef _populate_first_index(self):
310
196
cdef delta_index *index
311
cdef delta_result res
312
197
if len(self._sources) != 1 or self._index != NULL:
313
198
raise AssertionError('_populate_first_index should only be'
314
199
' called when we have a single source and no index yet')
316
# We know that self._index is already NULL, so create_delta_index
317
# will always create a new index unless there's a malloc failure
201
# We know that self._index is already NULL, so whatever
202
# create_delta_index returns is fine
319
res = create_delta_index(&self._source_infos[0], NULL, &index,
320
self._max_bytes_to_index)
322
raise _translate_delta_failure(res)
204
self._index = create_delta_index(&self._source_infos[0], NULL)
205
assert self._index != NULL
325
207
cdef _expand_sources(self):
326
208
raise RuntimeError('if we move self._source_infos, then we need to'
356
237
# allocate the bytes into the final string
357
238
c_max_delta_size = max_delta_size
359
res = create_delta(self._index, target, target_size,
360
&delta_size, c_max_delta_size, &delta)
240
delta = create_delta(self._index,
242
&delta_size, c_max_delta_size)
363
245
result = PyString_FromStringAndSize(<char *>delta, delta_size)
365
elif res != DELTA_SIZE_TOO_BIG:
366
raise _translate_delta_failure(res)
470
350
# Copy instruction
471
351
data = _decode_copy_instruction(data, cmd, &cp_off, &cp_size)
472
352
if (cp_off + cp_size < cp_size or
473
cp_off + cp_size > <unsigned int>source_size or
474
cp_size > <unsigned int>size):
353
cp_off + cp_size > source_size or
477
357
memcpy(out, source + cp_off, cp_size)
added
removed
bzrlib/_groupcompress_pyx.pyx
