36
43
def add_copy(self, start_byte, end_byte):
37
44
# The data stream allows >64kB in a copy, but to match the compiled
38
45
# code, we will also limit it to a 64kB copy
39
for start_byte in xrange(start_byte, end_byte, 64*1024):
40
num_bytes = min(64*1024, end_byte - start_byte)
46
for start_byte in range(start_byte, end_byte, 64 * 1024):
47
num_bytes = min(64 * 1024, end_byte - start_byte)
41
48
copy_bytes = encode_copy_instruction(start_byte, num_bytes)
42
49
self.out_lines.append(copy_bytes)
43
50
self.index_lines.append(False)
48
55
if self.cur_insert_len > 127:
49
56
raise AssertionError('We cannot insert more than 127 bytes'
51
self.out_lines.append(chr(self.cur_insert_len))
58
self.out_lines.append(int2byte(self.cur_insert_len))
52
59
self.index_lines.append(False)
53
60
self.out_lines.extend(self.cur_insert_lines)
54
61
if self.cur_insert_len < self.min_len_to_index:
55
self.index_lines.extend([False]*len(self.cur_insert_lines))
62
self.index_lines.extend([False] * len(self.cur_insert_lines))
57
self.index_lines.extend([True]*len(self.cur_insert_lines))
64
self.index_lines.extend([True] * len(self.cur_insert_lines))
58
65
self.cur_insert_lines = []
59
66
self.cur_insert_len = 0
62
69
# Flush out anything pending
63
70
self._flush_insert()
64
71
line_len = len(line)
65
for start_index in xrange(0, line_len, 127):
72
for start_index in range(0, line_len, 127):
66
73
next_len = min(127, line_len - start_index)
67
self.out_lines.append(chr(next_len))
74
self.out_lines.append(int2byte(next_len))
68
75
self.index_lines.append(False)
69
self.out_lines.append(line[start_index:start_index+next_len])
76
self.out_lines.append(line[start_index:start_index + next_len])
70
77
# We don't index long lines, because we won't be able to match
71
78
# a line split across multiple inserts anway
72
79
self.index_lines.append(False)
109
116
self.line_offsets = []
110
117
self.endpoint = 0
111
118
self._matching_lines = {}
112
self.extend_lines(lines, [True]*len(lines))
119
self.extend_lines(lines, [True] * len(lines))
114
121
def _update_matching_lines(self, new_lines, index):
115
122
matches = self._matching_lines
116
123
start_idx = len(self.lines)
117
124
if len(new_lines) != len(index):
118
125
raise AssertionError('The number of lines to be indexed does'
119
' not match the index/don\'t index flags: %d != %d'
120
% (len(new_lines), len(index)))
126
' not match the index/don\'t index flags: %d != %d'
127
% (len(new_lines), len(index)))
121
128
for idx, do_index in enumerate(index):
126
133
matches[line].add(start_idx + idx)
128
matches[line] = set([start_idx + idx])
135
matches[line] = {start_idx + idx}
130
137
def get_matches(self, line):
131
138
"""Return the lines which match the line in right."""
165
172
# This is the first match in a range
166
173
prev_locations = locations
168
locations = None # Consumed
175
locations = None # Consumed
170
177
# We have a match started, compare to see if any of the
171
178
# current matches can be continued
175
182
# At least one of the regions continues to match
176
183
prev_locations = set(next_locations)
178
locations = None # Consumed
185
locations = None # Consumed
180
187
# All current regions no longer match.
181
188
# This line does still match something, just not at the
222
229
# This block may be a 'short' block, check
223
230
old_start, new_start, range_len = block
224
231
matched_bytes = sum(map(len,
225
lines[new_start:new_start + range_len]))
232
lines[new_start:new_start + range_len]))
226
233
if matched_bytes < min_match_bytes:
228
235
if block is not None:
245
252
self.line_offsets.append(endpoint)
246
253
if len(self.line_offsets) != len(self.lines):
247
254
raise AssertionError('Somehow the line offset indicator'
248
' got out of sync with the line counter.')
255
' got out of sync with the line counter.')
249
256
self.endpoint = endpoint
251
258
def _flush_insert(self, start_linenum, end_linenum,
252
259
new_lines, out_lines, index_lines):
253
260
"""Add an 'insert' request to the data stream."""
254
bytes_to_insert = ''.join(new_lines[start_linenum:end_linenum])
261
bytes_to_insert = b''.join(new_lines[start_linenum:end_linenum])
255
262
insert_length = len(bytes_to_insert)
256
263
# Each insert instruction is at most 127 bytes long
257
for start_byte in xrange(0, insert_length, 127):
264
for start_byte in range(0, insert_length, 127):
258
265
insert_count = min(insert_length - start_byte, 127)
259
out_lines.append(chr(insert_count))
266
out_lines.append(int2byte(insert_count))
260
267
# Don't index the 'insert' instruction
261
268
index_lines.append(False)
262
insert = bytes_to_insert[start_byte:start_byte+insert_count]
269
insert = bytes_to_insert[start_byte:start_byte + insert_count]
263
270
as_lines = osutils.split_lines(insert)
264
271
out_lines.extend(as_lines)
265
index_lines.extend([True]*len(as_lines))
272
index_lines.extend([True] * len(as_lines))
267
274
def _flush_copy(self, old_start_linenum, num_lines,
268
275
out_lines, index_lines):
274
281
num_bytes = stop_byte - first_byte
275
282
# The data stream allows >64kB in a copy, but to match the compiled
276
283
# code, we will also limit it to a 64kB copy
277
for start_byte in xrange(first_byte, stop_byte, 64*1024):
278
num_bytes = min(64*1024, stop_byte - start_byte)
284
for start_byte in range(first_byte, stop_byte, 64 * 1024):
285
num_bytes = min(64 * 1024, stop_byte - start_byte)
279
286
copy_bytes = encode_copy_instruction(start_byte, num_bytes)
280
287
out_lines.append(copy_bytes)
281
288
index_lines.append(False)
285
292
if bytes_length is None:
286
293
bytes_length = sum(map(len, new_lines))
287
294
# reserved for content type, content length
288
out_lines = ['', '', encode_base128_int(bytes_length)]
295
out_lines = [b'', b'', encode_base128_int(bytes_length)]
289
296
index_lines = [False, False, False]
290
297
output_handler = _OutputHandler(out_lines, index_lines,
291
298
self._MIN_MATCH_BYTES)
296
303
for old_start, new_start, range_len in blocks:
297
304
if new_start != current_line_num:
298
305
# non-matching region, insert the content
299
output_handler.add_insert(new_lines[current_line_num:new_start])
306
output_handler.add_insert(
307
new_lines[current_line_num:new_start])
300
308
current_line_num = new_start + range_len
302
310
# Convert the line based offsets into byte based offsets
312
320
def encode_base128_int(val):
313
321
"""Convert an integer into a 7-bit lsb encoding."""
316
324
while val >= 0x80:
317
bytes.append(chr((val | 0x80) & 0xFF))
325
data.append((val | 0x80) & 0xFF)
319
bytes.append(chr(val))
320
return ''.join(bytes)
323
def decode_base128_int(bytes):
331
def decode_base128_int(data):
324
332
"""Decode an integer from a 7-bit lsb encoding."""
328
bval = ord(bytes[offset])
336
bval = indexbytes(data, offset)
329
337
while bval >= 0x80:
330
338
val |= (bval & 0x7F) << shift
333
bval = ord(bytes[offset])
341
bval = indexbytes(data, offset)
334
342
val |= bval << shift
336
344
return val, offset
360
368
base_byte = length & 0xff
362
370
copy_command |= copy_bit
363
copy_bytes.append(chr(base_byte))
371
copy_bytes.append(int2byte(base_byte))
365
copy_bytes[0] = chr(copy_command)
366
return ''.join(copy_bytes)
373
copy_bytes[0] = int2byte(copy_command)
374
return b''.join(copy_bytes)
369
377
def decode_copy_instruction(bytes, cmd, pos):
388
offset = ord(bytes[pos])
396
offset = indexbytes(bytes, pos)
391
offset = offset | (ord(bytes[pos]) << 8)
399
offset = offset | (indexbytes(bytes, pos) << 8)
394
offset = offset | (ord(bytes[pos]) << 16)
402
offset = offset | (indexbytes(bytes, pos) << 16)
397
offset = offset | (ord(bytes[pos]) << 24)
405
offset = offset | (indexbytes(bytes, pos) << 24)
400
length = ord(bytes[pos])
408
length = indexbytes(bytes, pos)
403
length = length | (ord(bytes[pos]) << 8)
411
length = length | (indexbytes(bytes, pos) << 8)
406
length = length | (ord(bytes[pos]) << 16)
414
length = length | (indexbytes(bytes, pos) << 16)
413
421
def make_delta(source_bytes, target_bytes):
414
422
"""Create a delta from source to target."""
415
if type(source_bytes) is not str:
416
raise TypeError('source is not a str')
417
if type(target_bytes) is not str:
418
raise TypeError('target is not a str')
423
if not isinstance(source_bytes, bytes):
424
raise TypeError('source is not bytes')
425
if not isinstance(target_bytes, bytes):
426
raise TypeError('target is not bytes')
419
427
line_locations = LinesDeltaIndex(osutils.split_lines(source_bytes))
420
428
delta, _ = line_locations.make_delta(osutils.split_lines(target_bytes),
421
429
bytes_length=len(target_bytes))
422
return ''.join(delta)
430
return b''.join(delta)
425
433
def apply_delta(basis, delta):
426
434
"""Apply delta to this object to become new_version_id."""
427
if type(basis) is not str:
428
raise TypeError('basis is not a str')
429
if type(delta) is not str:
430
raise TypeError('delta is not a str')
435
if not isinstance(basis, bytes):
436
raise TypeError('basis is not bytes')
437
if not isinstance(delta, bytes):
438
raise TypeError('delta is not bytes')
431
439
target_length, pos = decode_base128_int(delta)
433
441
len_delta = len(delta)
434
442
while pos < len_delta:
435
cmd = ord(delta[pos])
443
cmd = indexbytes(delta, pos)
438
446
offset, length, pos = decode_copy_instruction(delta, cmd, pos)
441
449
raise ValueError('data would copy bytes past the'
443
451
lines.append(basis[offset:last])
444
else: # Insert of 'cmd' bytes
452
else: # Insert of 'cmd' bytes
446
454
raise ValueError('Command == 0 not supported yet')
447
lines.append(delta[pos:pos+cmd])
455
lines.append(delta[pos:pos + cmd])
449
bytes = ''.join(lines)
450
if len(bytes) != target_length:
457
data = b''.join(lines)
458
if len(data) != target_length:
451
459
raise ValueError('Delta claimed to be %d long, but ended up'
452
460
' %d long' % (target_length, len(bytes)))
456
464
def apply_delta_to_source(source, delta_start, delta_end):