Show More
@@ -1,404 +1,404 | |||
|
1 | 1 | # parsers.py - Python implementation of parsers.c |
|
2 | 2 | # |
|
3 | 3 | # Copyright 2009 Olivia Mackall <olivia@selenic.com> and others |
|
4 | 4 | # |
|
5 | 5 | # This software may be used and distributed according to the terms of the |
|
6 | 6 | # GNU General Public License version 2 or any later version. |
|
7 | 7 | |
|
8 | 8 | from __future__ import absolute_import |
|
9 | 9 | |
|
10 | 10 | import struct |
|
11 | 11 | import zlib |
|
12 | 12 | |
|
13 | 13 | from ..node import ( |
|
14 | 14 | nullrev, |
|
15 | 15 | sha1nodeconstants, |
|
16 | 16 | ) |
|
17 | 17 | from .. import ( |
|
18 | 18 | error, |
|
19 | 19 | pycompat, |
|
20 | 20 | util, |
|
21 | 21 | ) |
|
22 | 22 | |
|
23 | 23 | from ..revlogutils import nodemap as nodemaputil |
|
24 | 24 | from ..revlogutils import constants as revlog_constants |
|
25 | 25 | |
|
26 | 26 | stringio = pycompat.bytesio |
|
27 | 27 | |
|
28 | 28 | |
|
29 | 29 | _pack = struct.pack |
|
30 | 30 | _unpack = struct.unpack |
|
31 | 31 | _compress = zlib.compress |
|
32 | 32 | _decompress = zlib.decompress |
|
33 | 33 | |
|
34 | 34 | # Some code below makes tuples directly because it's more convenient. However, |
|
35 | 35 | # code outside this module should always use dirstatetuple. |
|
36 | 36 | def dirstatetuple(*x): |
|
37 | 37 | # x is a tuple |
|
38 | 38 | return x |
|
39 | 39 | |
|
40 | 40 | |
|
41 | 41 | def gettype(q): |
|
42 | 42 | return int(q & 0xFFFF) |
|
43 | 43 | |
|
44 | 44 | |
|
45 | 45 | def offset_type(offset, type): |
|
46 | 46 | return int(int(offset) << 16 | type) |
|
47 | 47 | |
|
48 | 48 | |
|
49 | 49 | class BaseIndexObject(object): |
|
50 | 50 | # Format of an index entry according to Python's `struct` language |
|
51 | 51 | index_format = revlog_constants.INDEX_ENTRY_V1 |
|
52 | 52 | # Size of a C unsigned long long int, platform independent |
|
53 | 53 | big_int_size = struct.calcsize(b'>Q') |
|
54 | 54 | # Size of a C long int, platform independent |
|
55 | 55 | int_size = struct.calcsize(b'>i') |
|
56 | 56 | # An empty index entry, used as a default value to be overridden, or nullrev |
|
57 | 57 | null_item = (0, 0, 0, -1, -1, -1, -1, sha1nodeconstants.nullid, 0, 0) |
|
58 | 58 | |
|
59 | 59 | @util.propertycache |
|
60 | 60 | def entry_size(self): |
|
61 | 61 | return self.index_format.size |
|
62 | 62 | |
|
63 | 63 | @property |
|
64 | 64 | def nodemap(self): |
|
65 | 65 | msg = b"index.nodemap is deprecated, use index.[has_node|rev|get_rev]" |
|
66 | 66 | util.nouideprecwarn(msg, b'5.3', stacklevel=2) |
|
67 | 67 | return self._nodemap |
|
68 | 68 | |
|
69 | 69 | @util.propertycache |
|
70 | 70 | def _nodemap(self): |
|
71 | 71 | nodemap = nodemaputil.NodeMap({sha1nodeconstants.nullid: nullrev}) |
|
72 | 72 | for r in range(0, len(self)): |
|
73 | 73 | n = self[r][7] |
|
74 | 74 | nodemap[n] = r |
|
75 | 75 | return nodemap |
|
76 | 76 | |
|
77 | 77 | def has_node(self, node): |
|
78 | 78 | """return True if the node exist in the index""" |
|
79 | 79 | return node in self._nodemap |
|
80 | 80 | |
|
81 | 81 | def rev(self, node): |
|
82 | 82 | """return a revision for a node |
|
83 | 83 | |
|
84 | 84 | If the node is unknown, raise a RevlogError""" |
|
85 | 85 | return self._nodemap[node] |
|
86 | 86 | |
|
87 | 87 | def get_rev(self, node): |
|
88 | 88 | """return a revision for a node |
|
89 | 89 | |
|
90 | 90 | If the node is unknown, return None""" |
|
91 | 91 | return self._nodemap.get(node) |
|
92 | 92 | |
|
93 | 93 | def _stripnodes(self, start): |
|
94 | 94 | if '_nodemap' in vars(self): |
|
95 | 95 | for r in range(start, len(self)): |
|
96 | 96 | n = self[r][7] |
|
97 | 97 | del self._nodemap[n] |
|
98 | 98 | |
|
99 | 99 | def clearcaches(self): |
|
100 | 100 | self.__dict__.pop('_nodemap', None) |
|
101 | 101 | |
|
102 | 102 | def __len__(self): |
|
103 | 103 | return self._lgt + len(self._extra) |
|
104 | 104 | |
|
105 | 105 | def append(self, tup): |
|
106 | 106 | if '_nodemap' in vars(self): |
|
107 | 107 | self._nodemap[tup[7]] = len(self) |
|
108 | 108 | data = self._pack_entry(tup) |
|
109 | 109 | self._extra.append(data) |
|
110 | 110 | |
|
111 | 111 | def _pack_entry(self, entry): |
|
112 | 112 | assert entry[8] == 0 |
|
113 | 113 | assert entry[9] == 0 |
|
114 | 114 | return self.index_format.pack(*entry[:8]) |
|
115 | 115 | |
|
116 | 116 | def _check_index(self, i): |
|
117 | 117 | if not isinstance(i, int): |
|
118 | 118 | raise TypeError(b"expecting int indexes") |
|
119 | 119 | if i < 0 or i >= len(self): |
|
120 | 120 | raise IndexError |
|
121 | 121 | |
|
122 | 122 | def __getitem__(self, i): |
|
123 | 123 | if i == -1: |
|
124 | 124 | return self.null_item |
|
125 | 125 | self._check_index(i) |
|
126 | 126 | if i >= self._lgt: |
|
127 | 127 | data = self._extra[i - self._lgt] |
|
128 | 128 | else: |
|
129 | 129 | index = self._calculate_index(i) |
|
130 | 130 | data = self._data[index : index + self.entry_size] |
|
131 | 131 | r = self._unpack_entry(data) |
|
132 | 132 | if self._lgt and i == 0: |
|
133 | 133 | r = (offset_type(0, gettype(r[0])),) + r[1:] |
|
134 | 134 | return r |
|
135 | 135 | |
|
136 | 136 | def _unpack_entry(self, data): |
|
137 | 137 | r = self.index_format.unpack(data) |
|
138 | 138 | r = r + (0, 0) |
|
139 | 139 | return r |
|
140 | 140 | |
|
141 | 141 | def pack_header(self, header): |
|
142 | 142 | """pack header information as binary""" |
|
143 | 143 | v_fmt = revlog_constants.INDEX_HEADER |
|
144 | 144 | return v_fmt.pack(header) |
|
145 | 145 | |
|
146 | 146 | def entry_binary(self, rev): |
|
147 | 147 | """return the raw binary string representing a revision""" |
|
148 | 148 | entry = self[rev] |
|
149 | 149 | p = revlog_constants.INDEX_ENTRY_V1.pack(*entry[:8]) |
|
150 | 150 | if rev == 0: |
|
151 | 151 | p = p[revlog_constants.INDEX_HEADER.size :] |
|
152 | 152 | return p |
|
153 | 153 | |
|
154 | 154 | |
|
155 | 155 | class IndexObject(BaseIndexObject): |
|
156 | 156 | def __init__(self, data): |
|
157 | 157 | assert len(data) % self.entry_size == 0, ( |
|
158 | 158 | len(data), |
|
159 | 159 | self.entry_size, |
|
160 | 160 | len(data) % self.entry_size, |
|
161 | 161 | ) |
|
162 | 162 | self._data = data |
|
163 | 163 | self._lgt = len(data) // self.entry_size |
|
164 | 164 | self._extra = [] |
|
165 | 165 | |
|
166 | 166 | def _calculate_index(self, i): |
|
167 | 167 | return i * self.entry_size |
|
168 | 168 | |
|
169 | 169 | def __delitem__(self, i): |
|
170 | 170 | if not isinstance(i, slice) or not i.stop == -1 or i.step is not None: |
|
171 | 171 | raise ValueError(b"deleting slices only supports a:-1 with step 1") |
|
172 | 172 | i = i.start |
|
173 | 173 | self._check_index(i) |
|
174 | 174 | self._stripnodes(i) |
|
175 | 175 | if i < self._lgt: |
|
176 | 176 | self._data = self._data[: i * self.entry_size] |
|
177 | 177 | self._lgt = i |
|
178 | 178 | self._extra = [] |
|
179 | 179 | else: |
|
180 | 180 | self._extra = self._extra[: i - self._lgt] |
|
181 | 181 | |
|
182 | 182 | |
|
183 | 183 | class PersistentNodeMapIndexObject(IndexObject): |
|
184 | 184 | """a Debug oriented class to test persistent nodemap |
|
185 | 185 | |
|
186 | 186 | We need a simple python object to test API and higher level behavior. See |
|
187 | 187 | the Rust implementation for more serious usage. This should be used only |
|
188 | 188 | through the dedicated `devel.persistent-nodemap` config. |
|
189 | 189 | """ |
|
190 | 190 | |
|
191 | 191 | def nodemap_data_all(self): |
|
192 | 192 | """Return bytes containing a full serialization of a nodemap |
|
193 | 193 | |
|
194 | 194 | The nodemap should be valid for the full set of revisions in the |
|
195 | 195 | index.""" |
|
196 | 196 | return nodemaputil.persistent_data(self) |
|
197 | 197 | |
|
198 | 198 | def nodemap_data_incremental(self): |
|
199 | 199 | """Return bytes containing a incremental update to persistent nodemap |
|
200 | 200 | |
|
201 | 201 | This containst the data for an append-only update of the data provided |
|
202 | 202 | in the last call to `update_nodemap_data`. |
|
203 | 203 | """ |
|
204 | 204 | if self._nm_root is None: |
|
205 | 205 | return None |
|
206 | 206 | docket = self._nm_docket |
|
207 | 207 | changed, data = nodemaputil.update_persistent_data( |
|
208 | 208 | self, self._nm_root, self._nm_max_idx, self._nm_docket.tip_rev |
|
209 | 209 | ) |
|
210 | 210 | |
|
211 | 211 | self._nm_root = self._nm_max_idx = self._nm_docket = None |
|
212 | 212 | return docket, changed, data |
|
213 | 213 | |
|
214 | 214 | def update_nodemap_data(self, docket, nm_data): |
|
215 | 215 | """provide full block of persisted binary data for a nodemap |
|
216 | 216 | |
|
217 | 217 | The data are expected to come from disk. See `nodemap_data_all` for a |
|
218 | 218 | produceur of such data.""" |
|
219 | 219 | if nm_data is not None: |
|
220 | 220 | self._nm_root, self._nm_max_idx = nodemaputil.parse_data(nm_data) |
|
221 | 221 | if self._nm_root: |
|
222 | 222 | self._nm_docket = docket |
|
223 | 223 | else: |
|
224 | 224 | self._nm_root = self._nm_max_idx = self._nm_docket = None |
|
225 | 225 | |
|
226 | 226 | |
|
227 | 227 | class InlinedIndexObject(BaseIndexObject): |
|
228 | 228 | def __init__(self, data, inline=0): |
|
229 | 229 | self._data = data |
|
230 | 230 | self._lgt = self._inline_scan(None) |
|
231 | 231 | self._inline_scan(self._lgt) |
|
232 | 232 | self._extra = [] |
|
233 | 233 | |
|
234 | 234 | def _inline_scan(self, lgt): |
|
235 | 235 | off = 0 |
|
236 | 236 | if lgt is not None: |
|
237 | 237 | self._offsets = [0] * lgt |
|
238 | 238 | count = 0 |
|
239 | 239 | while off <= len(self._data) - self.entry_size: |
|
240 | 240 | start = off + self.big_int_size |
|
241 | 241 | (s,) = struct.unpack( |
|
242 | 242 | b'>i', |
|
243 | 243 | self._data[start : start + self.int_size], |
|
244 | 244 | ) |
|
245 | 245 | if lgt is not None: |
|
246 | 246 | self._offsets[count] = off |
|
247 | 247 | count += 1 |
|
248 | 248 | off += self.entry_size + s |
|
249 | 249 | if off != len(self._data): |
|
250 | 250 | raise ValueError(b"corrupted data") |
|
251 | 251 | return count |
|
252 | 252 | |
|
253 | 253 | def __delitem__(self, i): |
|
254 | 254 | if not isinstance(i, slice) or not i.stop == -1 or i.step is not None: |
|
255 | 255 | raise ValueError(b"deleting slices only supports a:-1 with step 1") |
|
256 | 256 | i = i.start |
|
257 | 257 | self._check_index(i) |
|
258 | 258 | self._stripnodes(i) |
|
259 | 259 | if i < self._lgt: |
|
260 | 260 | self._offsets = self._offsets[:i] |
|
261 | 261 | self._lgt = i |
|
262 | 262 | self._extra = [] |
|
263 | 263 | else: |
|
264 | 264 | self._extra = self._extra[: i - self._lgt] |
|
265 | 265 | |
|
266 | 266 | def _calculate_index(self, i): |
|
267 | 267 | return self._offsets[i] |
|
268 | 268 | |
|
269 | 269 | |
|
270 | 270 | def parse_index2(data, inline, revlogv2=False): |
|
271 | 271 | if not inline: |
|
272 | 272 | cls = IndexObject2 if revlogv2 else IndexObject |
|
273 | 273 | return cls(data), None |
|
274 | 274 | cls = InlinedIndexObject2 if revlogv2 else InlinedIndexObject |
|
275 | 275 | return cls(data, inline), (0, data) |
|
276 | 276 | |
|
277 | 277 | |
|
278 | 278 | class Index2Mixin(object): |
|
279 | 279 | index_format = revlog_constants.INDEX_ENTRY_V2 |
|
280 | 280 | |
|
281 | 281 | def replace_sidedata_info( |
|
282 |
self, |
|
|
282 | self, rev, sidedata_offset, sidedata_length, offset_flags | |
|
283 | 283 | ): |
|
284 | 284 | """ |
|
285 | 285 | Replace an existing index entry's sidedata offset and length with new |
|
286 | 286 | ones. |
|
287 | 287 | This cannot be used outside of the context of sidedata rewriting, |
|
288 |
inside the transaction that creates the revision ` |
|
|
288 | inside the transaction that creates the revision `rev`. | |
|
289 | 289 | """ |
|
290 |
if |
|
|
290 | if rev < 0: | |
|
291 | 291 | raise KeyError |
|
292 |
self._check_index( |
|
|
292 | self._check_index(rev) | |
|
293 | 293 | sidedata_format = b">Qi" |
|
294 | 294 | packed_size = struct.calcsize(sidedata_format) |
|
295 |
if |
|
|
295 | if rev >= self._lgt: | |
|
296 | 296 | packed = _pack(sidedata_format, sidedata_offset, sidedata_length) |
|
297 |
old = self._extra[ |
|
|
297 | old = self._extra[rev - self._lgt] | |
|
298 | 298 | offset_flags = struct.pack(b">Q", offset_flags) |
|
299 | 299 | new = offset_flags + old[8:64] + packed + old[64 + packed_size :] |
|
300 |
self._extra[ |
|
|
300 | self._extra[rev - self._lgt] = new | |
|
301 | 301 | else: |
|
302 | 302 | msg = b"cannot rewrite entries outside of this transaction" |
|
303 | 303 | raise KeyError(msg) |
|
304 | 304 | |
|
305 | 305 | def _unpack_entry(self, data): |
|
306 | 306 | return self.index_format.unpack(data) |
|
307 | 307 | |
|
308 | 308 | def _pack_entry(self, entry): |
|
309 | 309 | return self.index_format.pack(*entry) |
|
310 | 310 | |
|
311 | 311 | def entry_binary(self, rev): |
|
312 | 312 | """return the raw binary string representing a revision""" |
|
313 | 313 | entry = self[rev] |
|
314 | 314 | p = revlog_constants.INDEX_ENTRY_V2.pack(*entry) |
|
315 | 315 | return p |
|
316 | 316 | |
|
317 | 317 | def pack_header(self, header): |
|
318 | 318 | """pack header information as binary""" |
|
319 | 319 | msg = 'version header should go in the docket, not the index: %d' |
|
320 | 320 | msg %= header |
|
321 | 321 | raise error.ProgrammingError(msg) |
|
322 | 322 | |
|
323 | 323 | |
|
324 | 324 | class IndexObject2(Index2Mixin, IndexObject): |
|
325 | 325 | pass |
|
326 | 326 | |
|
327 | 327 | |
|
328 | 328 | class InlinedIndexObject2(Index2Mixin, InlinedIndexObject): |
|
329 | 329 | def _inline_scan(self, lgt): |
|
330 | 330 | sidedata_length_pos = 72 |
|
331 | 331 | off = 0 |
|
332 | 332 | if lgt is not None: |
|
333 | 333 | self._offsets = [0] * lgt |
|
334 | 334 | count = 0 |
|
335 | 335 | while off <= len(self._data) - self.entry_size: |
|
336 | 336 | start = off + self.big_int_size |
|
337 | 337 | (data_size,) = struct.unpack( |
|
338 | 338 | b'>i', |
|
339 | 339 | self._data[start : start + self.int_size], |
|
340 | 340 | ) |
|
341 | 341 | start = off + sidedata_length_pos |
|
342 | 342 | (side_data_size,) = struct.unpack( |
|
343 | 343 | b'>i', self._data[start : start + self.int_size] |
|
344 | 344 | ) |
|
345 | 345 | if lgt is not None: |
|
346 | 346 | self._offsets[count] = off |
|
347 | 347 | count += 1 |
|
348 | 348 | off += self.entry_size + data_size + side_data_size |
|
349 | 349 | if off != len(self._data): |
|
350 | 350 | raise ValueError(b"corrupted data") |
|
351 | 351 | return count |
|
352 | 352 | |
|
353 | 353 | |
|
354 | 354 | def parse_index_devel_nodemap(data, inline): |
|
355 | 355 | """like parse_index2, but alway return a PersistentNodeMapIndexObject""" |
|
356 | 356 | return PersistentNodeMapIndexObject(data), None |
|
357 | 357 | |
|
358 | 358 | |
|
359 | 359 | def parse_dirstate(dmap, copymap, st): |
|
360 | 360 | parents = [st[:20], st[20:40]] |
|
361 | 361 | # dereference fields so they will be local in loop |
|
362 | 362 | format = b">cllll" |
|
363 | 363 | e_size = struct.calcsize(format) |
|
364 | 364 | pos1 = 40 |
|
365 | 365 | l = len(st) |
|
366 | 366 | |
|
367 | 367 | # the inner loop |
|
368 | 368 | while pos1 < l: |
|
369 | 369 | pos2 = pos1 + e_size |
|
370 | 370 | e = _unpack(b">cllll", st[pos1:pos2]) # a literal here is faster |
|
371 | 371 | pos1 = pos2 + e[4] |
|
372 | 372 | f = st[pos2:pos1] |
|
373 | 373 | if b'\0' in f: |
|
374 | 374 | f, c = f.split(b'\0') |
|
375 | 375 | copymap[f] = c |
|
376 | 376 | dmap[f] = e[:4] |
|
377 | 377 | return parents |
|
378 | 378 | |
|
379 | 379 | |
|
380 | 380 | def pack_dirstate(dmap, copymap, pl, now): |
|
381 | 381 | now = int(now) |
|
382 | 382 | cs = stringio() |
|
383 | 383 | write = cs.write |
|
384 | 384 | write(b"".join(pl)) |
|
385 | 385 | for f, e in pycompat.iteritems(dmap): |
|
386 | 386 | if e[0] == b'n' and e[3] == now: |
|
387 | 387 | # The file was last modified "simultaneously" with the current |
|
388 | 388 | # write to dirstate (i.e. within the same second for file- |
|
389 | 389 | # systems with a granularity of 1 sec). This commonly happens |
|
390 | 390 | # for at least a couple of files on 'update'. |
|
391 | 391 | # The user could change the file without changing its size |
|
392 | 392 | # within the same second. Invalidate the file's mtime in |
|
393 | 393 | # dirstate, forcing future 'status' calls to compare the |
|
394 | 394 | # contents of the file if the size is the same. This prevents |
|
395 | 395 | # mistakenly treating such files as clean. |
|
396 | 396 | e = dirstatetuple(e[0], e[1], e[2], -1) |
|
397 | 397 | dmap[f] = e |
|
398 | 398 | |
|
399 | 399 | if f in copymap: |
|
400 | 400 | f = b"%s\0%s" % (f, copymap[f]) |
|
401 | 401 | e = _pack(b">cllll", e[0], e[1], e[2], e[3], len(f)) |
|
402 | 402 | write(e) |
|
403 | 403 | write(f) |
|
404 | 404 | return cs.getvalue() |
General Comments 0
You need to be logged in to leave comments.
Login now