| 1 | # -*- coding: utf-8 -*- |
|---|
| 2 | # |
|---|
| 3 | # Copyright (C) 2005 Edgewall Software |
|---|
| 4 | # Copyright (C) 2006 K.S.Sreeram <sreeram@tachyontech.net> |
|---|
| 5 | # Copyright (C) 2007-2010 Lele Gaifax <lele@metapensiero.it> |
|---|
| 6 | # |
|---|
| 7 | # This software is licensed as described in the file COPYING, which |
|---|
| 8 | # you should have received as part of this distribution. The terms |
|---|
| 9 | # are also available at http://trac.edgewall.com/license.html. |
|---|
| 10 | # |
|---|
| 11 | # This software consists of voluntary contributions made by many |
|---|
| 12 | # individuals. For the exact contribution history, see the revision |
|---|
| 13 | # history and logs, available at http://projects.edgewall.com/trac/. |
|---|
| 14 | # |
|---|
| 15 | # Author: K.S.Sreeram <sreeram@tachyontech.net> |
|---|
| 16 | |
|---|
| 17 | ''' |
|---|
| 18 | This module implements the trac versioncontrol backend API. |
|---|
| 19 | The API consists of 3 classes: DarcsRepository, DarcsNode |
|---|
| 20 | and DarcsChangeset. |
|---|
| 21 | |
|---|
| 22 | Please see the docs in trac.versioncontrol.api for the interface |
|---|
| 23 | which is implemented by this module. |
|---|
| 24 | ''' |
|---|
| 25 | |
|---|
| 26 | import os, StringIO, mimetypes |
|---|
| 27 | from datetime import tzinfo, timedelta, datetime |
|---|
| 28 | import time |
|---|
| 29 | |
|---|
| 30 | from trac.util import TracError |
|---|
| 31 | from trac.util.datefmt import to_timestamp, utc |
|---|
| 32 | from trac.versioncontrol import (Repository, Node, Changeset, |
|---|
| 33 | NoSuchChangeset, NoSuchNode) |
|---|
| 34 | |
|---|
| 35 | from command import DarcsCommand |
|---|
| 36 | from updatedb import update_darcsdb |
|---|
| 37 | from dbutil import (CHANGE_ADDED, CHANGE_EDITED, CHANGE_MOVED, |
|---|
| 38 | CHANGE_MOVED_EDITED, CHANGE_REMOVED, |
|---|
| 39 | IS_TRAC_0_10_X, IS_TRAC_0_11_X, TimedDB, |
|---|
| 40 | IS_TRAC_0_12_OR_BETTER, NODE_DIR_TYPE, |
|---|
| 41 | NODE_FILE_TYPE, get_node_type, get_prev_path_rev, |
|---|
| 42 | get_repository_id, query_nodes_for_revision) |
|---|
| 43 | |
|---|
| 44 | # mapping from node types used by the darcs backend and the types |
|---|
| 45 | # used by the trac api |
|---|
| 46 | _node_type_map = { |
|---|
| 47 | NODE_FILE_TYPE : Node.FILE, |
|---|
| 48 | NODE_DIR_TYPE : Node.DIRECTORY |
|---|
| 49 | } |
|---|
| 50 | |
|---|
| 51 | # mapping from change types used by the darcs backend and the types |
|---|
| 52 | # used by the trac api |
|---|
| 53 | _change_map = { |
|---|
| 54 | CHANGE_ADDED : Changeset.ADD, |
|---|
| 55 | CHANGE_REMOVED : Changeset.DELETE, |
|---|
| 56 | CHANGE_MOVED : Changeset.MOVE, |
|---|
| 57 | CHANGE_EDITED : Changeset.EDIT, |
|---|
| 58 | #FIXME: treat moved&edited as just moved? |
|---|
| 59 | CHANGE_MOVED_EDITED : Changeset.MOVE |
|---|
| 60 | } |
|---|
| 61 | |
|---|
| 62 | class DarcsRepository(Repository): |
|---|
| 63 | def __init__(self, db, path, log, darcscmd, possible_encodings, |
|---|
| 64 | params, eager_annotations): |
|---|
| 65 | if IS_TRAC_0_12_OR_BETTER: |
|---|
| 66 | Repository.__init__(self, 'darcs:%s' % path, params, log) |
|---|
| 67 | else: |
|---|
| 68 | Repository.__init__(self, 'darcs:%s' % path, params, None, log) |
|---|
| 69 | self.db = TimedDB(db, log) |
|---|
| 70 | self.path = path |
|---|
| 71 | self.__cmd = DarcsCommand(darcscmd, path, log, possible_encodings) |
|---|
| 72 | if not IS_TRAC_0_12_OR_BETTER: |
|---|
| 73 | self.log = log |
|---|
| 74 | self.id = get_repository_id(db, path) or 0 |
|---|
| 75 | self.eager_annotations = eager_annotations |
|---|
| 76 | if IS_TRAC_0_10_X: |
|---|
| 77 | self.sync() |
|---|
| 78 | |
|---|
| 79 | def close(self): |
|---|
| 80 | pass |
|---|
| 81 | |
|---|
| 82 | def get_changeset(self, rev): |
|---|
| 83 | rev = self.normalize_rev(rev) |
|---|
| 84 | return DarcsChangeset(self, rev) |
|---|
| 85 | |
|---|
| 86 | def get_node(self, path, rev=None): |
|---|
| 87 | path = self.normalize_path(path) |
|---|
| 88 | rev = self.normalize_rev(rev) |
|---|
| 89 | # compute node_id, node_type and last_rev and then |
|---|
| 90 | # create a DarcsNode object. |
|---|
| 91 | # 'last_rev' is the last revision <= rev where this |
|---|
| 92 | # node was modified. |
|---|
| 93 | if path == '/': |
|---|
| 94 | node_id = None |
|---|
| 95 | node_type = NODE_DIR_TYPE |
|---|
| 96 | last_rev = rev |
|---|
| 97 | else: |
|---|
| 98 | c = self.db.cursor() |
|---|
| 99 | q,args = query_nodes_for_revision(self.id, rev, 'dnc.path = %s') |
|---|
| 100 | args.append(path) |
|---|
| 101 | c.execute(q, args) |
|---|
| 102 | row = c.fetchone() |
|---|
| 103 | if row is None: |
|---|
| 104 | raise NoSuchNode(path, rev) |
|---|
| 105 | node_id,last_rev = row[:2] |
|---|
| 106 | node_type = get_node_type(self.db, self.id, node_id) |
|---|
| 107 | return DarcsNode(node_id, node_type, path, last_rev, |
|---|
| 108 | self, self.__cmd, self.log) |
|---|
| 109 | |
|---|
| 110 | def get_oldest_rev(self): |
|---|
| 111 | if self.get_youngest_rev() is None: |
|---|
| 112 | return None |
|---|
| 113 | return 1 |
|---|
| 114 | |
|---|
| 115 | def get_youngest_rev(self): |
|---|
| 116 | c = self.db.cursor() |
|---|
| 117 | c.execute('SELECT max(rev) FROM darcs_changesets ' |
|---|
| 118 | 'WHERE repo_id = %s', (self.id,)) |
|---|
| 119 | row = c.fetchone() |
|---|
| 120 | return row and row[0] or None |
|---|
| 121 | |
|---|
| 122 | def previous_rev(self, rev, path=''): |
|---|
| 123 | rev = self.normalize_rev(rev) |
|---|
| 124 | if rev > 1: |
|---|
| 125 | return rev-1 |
|---|
| 126 | return None |
|---|
| 127 | |
|---|
| 128 | def next_rev(self, rev, path=''): |
|---|
| 129 | rev = self.normalize_rev(rev) |
|---|
| 130 | if rev < self.get_youngest_rev(): |
|---|
| 131 | return rev+1 |
|---|
| 132 | return None |
|---|
| 133 | |
|---|
| 134 | def rev_older_than(self, rev1, rev2): |
|---|
| 135 | return self.normalize_rev(rev1) < self.normalize_rev(rev2) |
|---|
| 136 | |
|---|
| 137 | def get_path_history(self, path, rev=None, limit=None): |
|---|
| 138 | # FIXME: this is not correct |
|---|
| 139 | return self.get_node(path, rev).get_history(limit) |
|---|
| 140 | |
|---|
| 141 | def normalize_path(self, path): |
|---|
| 142 | return path and path.strip('/') or '/' |
|---|
| 143 | |
|---|
| 144 | def normalize_rev(self, rev): |
|---|
| 145 | if isinstance(rev, basestring) and len(rev) in (61,64): |
|---|
| 146 | if rev.endswith('.gz'): |
|---|
| 147 | # We don't store ending .gz in the db |
|---|
| 148 | rev = rev[:-3] |
|---|
| 149 | c = self.db.cursor() |
|---|
| 150 | c.execute('SELECT rev FROM darcs_changesets ' |
|---|
| 151 | 'WHERE repo_id = %s AND hash = %s', (self.id, rev)) |
|---|
| 152 | row = c.fetchone() |
|---|
| 153 | if row is None: |
|---|
| 154 | raise NoSuchChangeset(rev) |
|---|
| 155 | rev = int(row[0]) |
|---|
| 156 | else: |
|---|
| 157 | youngest = self.get_youngest_rev() |
|---|
| 158 | if rev is None or rev == "": |
|---|
| 159 | return youngest |
|---|
| 160 | try: |
|---|
| 161 | rev = int(rev) |
|---|
| 162 | except ValueError, le: |
|---|
| 163 | raise TracError('Ill-formed revision: %s, error: %s' % (rev, le)) |
|---|
| 164 | if rev > youngest: |
|---|
| 165 | rev = youngest |
|---|
| 166 | return rev |
|---|
| 167 | |
|---|
| 168 | def get_changes(self, old_path, old_rev, new_path, new_rev, ignore_ancestry=1): |
|---|
| 169 | old_path = self.normalize_path(old_path) |
|---|
| 170 | old_rev = self.normalize_rev(old_rev) |
|---|
| 171 | new_path = self.normalize_path(new_path) |
|---|
| 172 | new_rev = self.normalize_rev(new_rev) |
|---|
| 173 | old_node = self.get_node(old_path, old_rev) |
|---|
| 174 | new_node = self.get_node(new_path, new_rev) |
|---|
| 175 | |
|---|
| 176 | node_id = old_node._get_node_id() |
|---|
| 177 | if node_id != new_node._get_node_id(): |
|---|
| 178 | raise TracError('Node mismatch: base is %s in rev %d ' |
|---|
| 179 | 'and target is %s in rev %d' % (old_path, old_rev, |
|---|
| 180 | new_path, new_rev)) |
|---|
| 181 | |
|---|
| 182 | if old_node.kind == Node.FILE: |
|---|
| 183 | if old_node.rev != new_node.rev: |
|---|
| 184 | yield (old_node, new_node, Node.FILE, Changeset.EDIT) |
|---|
| 185 | return |
|---|
| 186 | |
|---|
| 187 | c = self.db.cursor() |
|---|
| 188 | if node_id is not None: |
|---|
| 189 | c.execute('SELECT rev,path FROM darcs_node_changes ' |
|---|
| 190 | 'WHERE repo_id = %s AND node_id = %s AND rev >= %s AND rev <= %s', |
|---|
| 191 | (self.id, node_id, old_rev, new_rev)) |
|---|
| 192 | else: |
|---|
| 193 | c.execute('SELECT rev,path FROM darcs_node_changes ' |
|---|
| 194 | 'WHERE repo_id = %s AND rev >= %s AND rev <= %s', |
|---|
| 195 | (self.id, old_rev, new_rev)) |
|---|
| 196 | node_set = dict() |
|---|
| 197 | node_list = [] |
|---|
| 198 | c1 = self.db.cursor() |
|---|
| 199 | for rev,path in c: |
|---|
| 200 | c1.execute('SELECT node_id FROM darcs_node_changes ' |
|---|
| 201 | 'WHERE repo_id = %s AND rev = %s AND path LIKE %s', |
|---|
| 202 | (self.id, rev, path+'/%')) |
|---|
| 203 | for nid, in c1: |
|---|
| 204 | if nid not in node_set: |
|---|
| 205 | node_set[nid] = 1 |
|---|
| 206 | node_list.append(nid) |
|---|
| 207 | for nid in node_list: |
|---|
| 208 | old_node = new_node = None |
|---|
| 209 | c1.execute('SELECT rev,path FROM darcs_node_changes ' |
|---|
| 210 | 'WHERE repo_id = %s AND node_id = %s AND rev < %s ' |
|---|
| 211 | 'ORDER BY rev DESC LIMIT 1', |
|---|
| 212 | (self.id, nid, old_rev)) |
|---|
| 213 | row = c1.fetchone() |
|---|
| 214 | if row is not None: |
|---|
| 215 | rev,path = row |
|---|
| 216 | old_node = self.get_node(path, rev) |
|---|
| 217 | c1.execute('SELECT rev,path,change_kind FROM darcs_node_changes ' |
|---|
| 218 | 'WHERE repo_id = %s AND node_id = %s AND rev >= %s AND rev <= %s ' |
|---|
| 219 | 'ORDER BY rev DESC LIMIT 1', |
|---|
| 220 | (self.id, nid, old_rev, new_rev)) |
|---|
| 221 | rev,path,change = c1.fetchone() |
|---|
| 222 | if change != CHANGE_REMOVED: |
|---|
| 223 | new_node = self.get_node(path, rev) |
|---|
| 224 | assert (old_node is not None) or (new_node is not None) |
|---|
| 225 | kind = old_node and old_node.kind or new_node.kind |
|---|
| 226 | if old_node is None: |
|---|
| 227 | change = Changeset.ADD |
|---|
| 228 | elif new_node is None: |
|---|
| 229 | change = Changeset.DELETE |
|---|
| 230 | elif old_node.path != new_node.path: |
|---|
| 231 | change = Changeset.MOVE |
|---|
| 232 | else: |
|---|
| 233 | change = Changeset.EDIT |
|---|
| 234 | yield (old_node,new_node,kind,change) |
|---|
| 235 | |
|---|
| 236 | def sync(self, rev_callback=None, clean=False): |
|---|
| 237 | from dbutil import format_elapsed_time as trepr |
|---|
| 238 | |
|---|
| 239 | # Import any new changesets, if any |
|---|
| 240 | |
|---|
| 241 | newrevs = update_darcsdb(self.db, self.__cmd, self.log, self.id, |
|---|
| 242 | rev_callback=rev_callback, clean=clean) |
|---|
| 243 | |
|---|
| 244 | # In eager mode, precompute the content and the annotations for |
|---|
| 245 | # each file modified or added by latest changesets |
|---|
| 246 | |
|---|
| 247 | if self.eager_annotations and newrevs: |
|---|
| 248 | i = 1 |
|---|
| 249 | count = len(newrevs) |
|---|
| 250 | for rev in newrevs: |
|---|
| 251 | t0 = time.time() |
|---|
| 252 | |
|---|
| 253 | c = self.get_changeset(rev) |
|---|
| 254 | for path,kind,change,prev_path,prev_rev in c.get_changes(): |
|---|
| 255 | if kind==Node.FILE and (change==Changeset.EDIT or change==Changeset.ADD): |
|---|
| 256 | node = self.get_node(path, rev) |
|---|
| 257 | node.get_content() |
|---|
| 258 | node.get_annotations() |
|---|
| 259 | |
|---|
| 260 | t1 = time.time() |
|---|
| 261 | usec = (t1-t0) * 1e6 |
|---|
| 262 | self.log.info('Preannotated changeset %d/%d: %s', i, count, trepr(usec)) |
|---|
| 263 | i += 1 |
|---|
| 264 | |
|---|
| 265 | class DarcsNode(Node): |
|---|
| 266 | def __init__(self, node_id, node_type, path, rev, |
|---|
| 267 | repos, cmd, log=None): |
|---|
| 268 | kind = _node_type_map[node_type] |
|---|
| 269 | Node.__init__(self, repos, path, rev, kind) |
|---|
| 270 | self.__node_id = node_id |
|---|
| 271 | self.__node_type = node_type |
|---|
| 272 | self.__cmd = cmd |
|---|
| 273 | self.__log = log |
|---|
| 274 | self.created_path = path |
|---|
| 275 | self.created_rev = rev |
|---|
| 276 | |
|---|
| 277 | def _get_node_id(self): |
|---|
| 278 | return self.__node_id |
|---|
| 279 | |
|---|
| 280 | def _get_cached_rev(self): |
|---|
| 281 | # if there are no future versions, use the HEAD |
|---|
| 282 | nrev = self._get_next() |
|---|
| 283 | if nrev is None: |
|---|
| 284 | return None |
|---|
| 285 | |
|---|
| 286 | maxrange = nrev[1]-1 |
|---|
| 287 | |
|---|
| 288 | # if it's just one hop from node's revision, we're done |
|---|
| 289 | if maxrange == self.rev: |
|---|
| 290 | return maxrange |
|---|
| 291 | |
|---|
| 292 | # ok, let's see if there is already a cache in the range |
|---|
| 293 | c = self.repos.db.cursor() |
|---|
| 294 | c.execute('SELECT max(rev) FROM darcs_cache ' |
|---|
| 295 | 'WHERE repo_id = %s AND node_id = %s AND rev >= %s AND rev <= %s AND content IS NOT NULL', |
|---|
| 296 | (self.repos.id, self.__node_id, self.rev, maxrange)) |
|---|
| 297 | row = c.fetchone() |
|---|
| 298 | if row[0] is not None: |
|---|
| 299 | return row[0] |
|---|
| 300 | |
|---|
| 301 | # No luck, return the most recent revision before the next |
|---|
| 302 | return maxrange |
|---|
| 303 | |
|---|
| 304 | def get_content(self): |
|---|
| 305 | if self.__node_type == NODE_DIR_TYPE: |
|---|
| 306 | return None |
|---|
| 307 | c = self.repos.db.cursor() |
|---|
| 308 | |
|---|
| 309 | # Since darcs is faster and faster in building the content |
|---|
| 310 | # of a file for more and more recent changes, compute the |
|---|
| 311 | # optimal revision to build the cache of |
|---|
| 312 | crev = self._get_cached_rev() |
|---|
| 313 | |
|---|
| 314 | if crev is not None: |
|---|
| 315 | # check if the file content is there in the cache |
|---|
| 316 | c.execute('SELECT content FROM darcs_cache ' |
|---|
| 317 | 'WHERE repo_id = %s AND node_id = %s AND rev = %s', |
|---|
| 318 | (self.repos.id, self.__node_id, crev)) |
|---|
| 319 | row = c.fetchone() |
|---|
| 320 | if row is not None: |
|---|
| 321 | self.__log.debug('Cache hit %s at rev %s', self.path, crev) |
|---|
| 322 | # if present just return it |
|---|
| 323 | data = str(buffer(row[0])) |
|---|
| 324 | else: |
|---|
| 325 | self.__log.debug('Building content cache for %s at rev %s', self.path, crev) |
|---|
| 326 | |
|---|
| 327 | # load the file content from the repo |
|---|
| 328 | c.execute('SELECT hash FROM darcs_changesets ' |
|---|
| 329 | 'WHERE repo_id = %s AND rev = %s', (self.repos.id, crev,)) |
|---|
| 330 | hash = c.fetchone()[0] |
|---|
| 331 | data = self.__cmd.cat(hash, self.path) |
|---|
| 332 | |
|---|
| 333 | # save the file content in the cache |
|---|
| 334 | c = self.repos.db.cursor() |
|---|
| 335 | try: |
|---|
| 336 | c.execute('INSERT INTO darcs_cache (repo_id,node_id,rev,content,size) ' |
|---|
| 337 | 'VALUES (%s,%s,%s,%s,%s)', |
|---|
| 338 | (self.repos.id, self.__node_id, crev, buffer(data), len(data))) |
|---|
| 339 | except: |
|---|
| 340 | self.repos.db.rollback() |
|---|
| 341 | c = self.repos.db.cursor() |
|---|
| 342 | # Maybe some other thread computed the same content |
|---|
| 343 | c.execute('SELECT content FROM darcs_cache ' |
|---|
| 344 | 'WHERE repo_id = %s AND node_id = %s AND rev = %s', |
|---|
| 345 | (self.repos.id, self.__node_id, crev)) |
|---|
| 346 | row = c.fetchone() |
|---|
| 347 | if row is not None: |
|---|
| 348 | self.__log.debug('Late cache hit %s at rev %s', self.path, crev) |
|---|
| 349 | data = str(buffer(row[0])) |
|---|
| 350 | else: |
|---|
| 351 | raise |
|---|
| 352 | else: |
|---|
| 353 | self.repos.db.commit() |
|---|
| 354 | else: |
|---|
| 355 | # Use the HEAD |
|---|
| 356 | self.__log.debug('Serving pristine file %s, no changes since rev %s', self.path, self.rev) |
|---|
| 357 | data = self.__cmd.cat(None, self.path) |
|---|
| 358 | |
|---|
| 359 | return StringIO.StringIO(data) |
|---|
| 360 | |
|---|
| 361 | def get_entries(self): |
|---|
| 362 | if self.__node_type == NODE_FILE_TYPE: |
|---|
| 363 | return |
|---|
| 364 | if self.__node_id is None: |
|---|
| 365 | cond = 'dnc.parent_id IS NULL' |
|---|
| 366 | else: |
|---|
| 367 | cond = 'dnc.parent_id = %d' % self.__node_id |
|---|
| 368 | q,args = query_nodes_for_revision(self.repos.id, self.rev, cond) |
|---|
| 369 | c = self.repos.db.cursor() |
|---|
| 370 | c.execute(q, args) |
|---|
| 371 | for node_id,rev,path,_ in c: |
|---|
| 372 | node_type = get_node_type(self.repos.db, self.repos.id, node_id) |
|---|
| 373 | yield DarcsNode(node_id, node_type, path, rev, |
|---|
| 374 | self.repos, self.__cmd, self.__log) |
|---|
| 375 | |
|---|
| 376 | def get_history(self, limit=None): |
|---|
| 377 | if self.path == '/': |
|---|
| 378 | for i in range(self.rev,0,-1): |
|---|
| 379 | yield (self.path, i, Changeset.EDIT) |
|---|
| 380 | return |
|---|
| 381 | c = self.repos.db.cursor() |
|---|
| 382 | q = ('SELECT path,rev,change_kind FROM darcs_node_changes ' |
|---|
| 383 | 'WHERE repo_id = %s AND node_id = %s AND rev <= %s ' |
|---|
| 384 | 'ORDER BY rev DESC') |
|---|
| 385 | if limit is not None: |
|---|
| 386 | q += ' LIMIT %d' % limit |
|---|
| 387 | c.execute(q, (self.repos.id, self.__node_id, self.rev)) |
|---|
| 388 | for path,rev,change in c: |
|---|
| 389 | yield (path, rev, _change_map[change]) |
|---|
| 390 | |
|---|
| 391 | def _get_next(self): |
|---|
| 392 | try: |
|---|
| 393 | return self._get_future(1).next() |
|---|
| 394 | except StopIteration: |
|---|
| 395 | return None |
|---|
| 396 | |
|---|
| 397 | def _get_future(self, limit=None): |
|---|
| 398 | if self.path == '/': |
|---|
| 399 | youngest = self.get_youngest_rev() |
|---|
| 400 | for i in range(youngest, self.rev, -1): |
|---|
| 401 | yield (self.path, i, Changeset.EDIT) |
|---|
| 402 | return |
|---|
| 403 | c = self.repos.db.cursor() |
|---|
| 404 | q = ('SELECT path,rev,change_kind FROM darcs_node_changes ' |
|---|
| 405 | 'WHERE repo_id = %s AND node_id = %s AND rev > %s ' |
|---|
| 406 | 'ORDER BY rev') |
|---|
| 407 | if limit is not None: |
|---|
| 408 | q += ' LIMIT %d' % limit |
|---|
| 409 | c.execute(q, (self.repos.id, self.__node_id, self.rev)) |
|---|
| 410 | for path,rev,change in c: |
|---|
| 411 | yield (path, rev, _change_map[change]) |
|---|
| 412 | |
|---|
| 413 | def get_annotations(self): |
|---|
| 414 | """Provide detailed backward history for the content of this Node. |
|---|
| 415 | |
|---|
| 416 | Retrieve an array of revisions parsing `darcs annotate`. Since |
|---|
| 417 | that is (still) not fast enough for some repository, we write |
|---|
| 418 | a cache of the information: a future annotate on the same file |
|---|
| 419 | at the same revision won't reexecute `darcs annotate`. |
|---|
| 420 | """ |
|---|
| 421 | |
|---|
| 422 | from xml.sax import make_parser |
|---|
| 423 | from xml.sax.handler import ContentHandler, ErrorHandler |
|---|
| 424 | from datetime import datetime |
|---|
| 425 | |
|---|
| 426 | c = self.repos.db.cursor() |
|---|
| 427 | |
|---|
| 428 | # Since darcs is faster and faster in building the content |
|---|
| 429 | # of a file for more and more recent changes, compute the |
|---|
| 430 | # optimal revision to build the cache of |
|---|
| 431 | crev = self._get_cached_rev() |
|---|
| 432 | if crev is None: |
|---|
| 433 | crev = self.rev |
|---|
| 434 | |
|---|
| 435 | # Check if the annotate cache is already present |
|---|
| 436 | c.execute('SELECT up_to_line,blame_rev FROM darcs_annotate_cache ' |
|---|
| 437 | 'WHERE repo_id = %s AND node_id = %s AND rev = %s ' |
|---|
| 438 | 'ORDER BY up_to_line', (self.repos.id, self.__node_id, crev)) |
|---|
| 439 | row = c.fetchone() |
|---|
| 440 | if row is not None: |
|---|
| 441 | self.__log.debug('Annotate cache hit for %s at rev %s', self.path, crev) |
|---|
| 442 | revs = [] |
|---|
| 443 | line = 0 |
|---|
| 444 | # Expand the cache, producing a list of revisions, one per line |
|---|
| 445 | while row is not None: |
|---|
| 446 | while line<row[0]: |
|---|
| 447 | revs.append(row[1]) |
|---|
| 448 | line += 1 |
|---|
| 449 | row = c.fetchone() |
|---|
| 450 | return revs |
|---|
| 451 | |
|---|
| 452 | # No cache, build it |
|---|
| 453 | |
|---|
| 454 | self.__log.debug('Building annotate cache for %s at rev %s', self.path, crev) |
|---|
| 455 | |
|---|
| 456 | class DarcsXMLAnnotateHandler(ContentHandler): |
|---|
| 457 | def __init__(self): |
|---|
| 458 | self.revisions = [] |
|---|
| 459 | self.known_hashes = {} |
|---|
| 460 | |
|---|
| 461 | def startElement(self, name, attributes): |
|---|
| 462 | if name == 'patch': |
|---|
| 463 | self.current_hash = attributes['hash'][:-3] |
|---|
| 464 | |
|---|
| 465 | def endElement(self, name): |
|---|
| 466 | if name == 'normal_line': |
|---|
| 467 | self.revisions.append(self.findRevision(self.current_hash)) |
|---|
| 468 | elif name == 'added_line': |
|---|
| 469 | self.revisions.append(self.findRevision(self.last_changed_hash)) |
|---|
| 470 | elif name == 'modified': |
|---|
| 471 | self.last_changed_hash = self.current_hash |
|---|
| 472 | |
|---|
| 473 | def findRevision(self, hash): |
|---|
| 474 | # Return the trac revision for the given patch hash |
|---|
| 475 | try: |
|---|
| 476 | return self.known_hashes[hash] |
|---|
| 477 | except KeyError: |
|---|
| 478 | c.execute('SELECT rev FROM darcs_changesets ' |
|---|
| 479 | 'WHERE hash = %s', (hash,)) |
|---|
| 480 | rev = self.known_hashes[hash] = c.fetchone()[0] |
|---|
| 481 | return rev |
|---|
| 482 | |
|---|
| 483 | # Get the hash of the patch |
|---|
| 484 | c.execute('SELECT hash FROM darcs_changesets ' |
|---|
| 485 | 'WHERE rev = %s', (self.rev,)) |
|---|
| 486 | hash = c.fetchone()[0] |
|---|
| 487 | |
|---|
| 488 | # Get darcs annotate output for the given entry and patch hash |
|---|
| 489 | annotate = self.__cmd.annotate(hash, self.path) |
|---|
| 490 | |
|---|
| 491 | parser = make_parser() |
|---|
| 492 | handler = DarcsXMLAnnotateHandler() |
|---|
| 493 | parser.setContentHandler(handler) |
|---|
| 494 | parser.setErrorHandler(ErrorHandler()) |
|---|
| 495 | |
|---|
| 496 | parser.feed(annotate) |
|---|
| 497 | parser.close() |
|---|
| 498 | |
|---|
| 499 | revs = handler.revisions |
|---|
| 500 | if not revs: |
|---|
| 501 | self.__log.debug('Empty file, no annotations for %s at rev %s', self.path, crev) |
|---|
| 502 | return revs |
|---|
| 503 | |
|---|
| 504 | # Write a compressed representation |
|---|
| 505 | |
|---|
| 506 | prev = None |
|---|
| 507 | for i,rev in enumerate(revs): |
|---|
| 508 | if prev is not None: |
|---|
| 509 | if prev != rev: |
|---|
| 510 | c.execute('INSERT INTO darcs_annotate_cache (repo_id,node_id,rev,up_to_line,blame_rev) ' |
|---|
| 511 | 'VALUES (%s,%s,%s,%s,%s)', |
|---|
| 512 | (self.repos.id, self.__node_id, crev, i, prev)) |
|---|
| 513 | prev = rev |
|---|
| 514 | lastline = i |
|---|
| 515 | else: |
|---|
| 516 | prev = rev |
|---|
| 517 | lastline = 0 |
|---|
| 518 | if lastline != len(revs): |
|---|
| 519 | c.execute('INSERT INTO darcs_annotate_cache (repo_id,node_id,rev,up_to_line,blame_rev) ' |
|---|
| 520 | 'VALUES (%s,%s,%s,%s,%s)', |
|---|
| 521 | (self.repos.id, self.__node_id, crev, len(revs), revs[-1])) |
|---|
| 522 | |
|---|
| 523 | self.repos.db.commit() |
|---|
| 524 | |
|---|
| 525 | return revs |
|---|
| 526 | |
|---|
| 527 | def get_properties(self): |
|---|
| 528 | return {} |
|---|
| 529 | |
|---|
| 530 | def get_content_length(self): |
|---|
| 531 | if self.isdir: |
|---|
| 532 | return None |
|---|
| 533 | |
|---|
| 534 | # first check if the file is already in the cache |
|---|
| 535 | c = self.repos.db.cursor() |
|---|
| 536 | c.execute('SELECT size FROM darcs_cache ' |
|---|
| 537 | 'WHERE repo_id = %s AND node_id = %s AND rev = %s', |
|---|
| 538 | (self.repos.id, self.__node_id, self.rev)) |
|---|
| 539 | row = c.fetchone() |
|---|
| 540 | if row is not None: |
|---|
| 541 | return row[0] |
|---|
| 542 | |
|---|
| 543 | # if it's not, get the whole content and count... |
|---|
| 544 | # next time you'll be luckier, promise! |
|---|
| 545 | return len(self.get_content().read()) |
|---|
| 546 | |
|---|
| 547 | def get_content_type(self): |
|---|
| 548 | if self.isdir: |
|---|
| 549 | return None |
|---|
| 550 | return mimetypes.guess_type(self.path)[0] |
|---|
| 551 | |
|---|
| 552 | def get_name(self): |
|---|
| 553 | return os.path.split(self.path)[1] |
|---|
| 554 | |
|---|
| 555 | def get_last_modified(self): |
|---|
| 556 | if self.__node_id is None: |
|---|
| 557 | return 0 |
|---|
| 558 | c = self.repos.db.cursor() |
|---|
| 559 | c.execute('SELECT rev FROM darcs_node_changes ' |
|---|
| 560 | 'WHERE repo_id = %s AND node_id = %s AND rev = %s', |
|---|
| 561 | (self.repos.id, self.__node_id, self.rev)) |
|---|
| 562 | rev = c.fetchone()[0] |
|---|
| 563 | if IS_TRAC_0_12_OR_BETTER: |
|---|
| 564 | c.execute('SELECT time FROM revision ' |
|---|
| 565 | 'WHERE repos = %s AND rev = %s', (self.repos.id, rev)) |
|---|
| 566 | else: |
|---|
| 567 | c.execute('SELECT time FROM revision ' |
|---|
| 568 | 'WHERE rev = %s', (rev,)) |
|---|
| 569 | return datetime.fromtimestamp(c.fetchone()[0], utc) |
|---|
| 570 | |
|---|
| 571 | class DarcsChangeset(Changeset): |
|---|
| 572 | def __init__(self, repos, rev): |
|---|
| 573 | repo_id = repos.id |
|---|
| 574 | c = repos.db.cursor() |
|---|
| 575 | if IS_TRAC_0_12_OR_BETTER: |
|---|
| 576 | c.execute('SELECT r.author,r.time,c.name,r.message,c.hash ' |
|---|
| 577 | 'FROM revision as r, darcs_changesets as c ' |
|---|
| 578 | 'WHERE r.repos = %s AND c.repo_id = r.repos ' |
|---|
| 579 | ' AND r.rev = %s AND c.rev = r.rev', |
|---|
| 580 | (repo_id, rev)) |
|---|
| 581 | else: |
|---|
| 582 | c.execute('SELECT r.author,r.time,c.name,r.message,c.hash ' |
|---|
| 583 | 'FROM revision as r, darcs_changesets as c ' |
|---|
| 584 | 'WHERE r.rev = %s ' |
|---|
| 585 | ' AND c.rev = r.rev AND c.repo_id = %s', (rev, 0)) |
|---|
| 586 | row = c.fetchone() |
|---|
| 587 | if row is None: |
|---|
| 588 | raise NoSuchChangeset(rev) |
|---|
| 589 | author,date,name,comment,hash = row |
|---|
| 590 | date = datetime.fromtimestamp(date, utc) |
|---|
| 591 | # Trac 0.10.x hack |
|---|
| 592 | if IS_TRAC_0_10_X: |
|---|
| 593 | date = time.mktime(date.timetuple()) |
|---|
| 594 | msg = name |
|---|
| 595 | if comment: |
|---|
| 596 | msg += '\n' + comment |
|---|
| 597 | Changeset.__init__(self, repos, rev, msg, author, date) |
|---|
| 598 | self.__hash = hash |
|---|
| 599 | |
|---|
| 600 | def get_changes(self): |
|---|
| 601 | c = self.repos.db.cursor() |
|---|
| 602 | repo_id = self.repos.id |
|---|
| 603 | c.execute('SELECT node_id,path,change_kind FROM darcs_node_changes ' |
|---|
| 604 | 'WHERE repo_id = %s AND rev = %s', (repo_id, self.rev,)) |
|---|
| 605 | for node_id,path,change in c: |
|---|
| 606 | node_type = get_node_type(self.repos.db, repo_id, node_id) |
|---|
| 607 | kind = _node_type_map[node_type] |
|---|
| 608 | if change == CHANGE_ADDED: |
|---|
| 609 | prev_path = prev_rev = None |
|---|
| 610 | else: |
|---|
| 611 | prev_path,prev_rev = get_prev_path_rev(self.repos.db, repo_id, |
|---|
| 612 | node_id, self.rev) |
|---|
| 613 | change = _change_map[change] |
|---|
| 614 | yield (path,kind,change,prev_path,prev_rev) |
|---|
| 615 | |
|---|
| 616 | def get_properties(self): |
|---|
| 617 | props = dict(Hashname=self.__hash) |
|---|
| 618 | |
|---|
| 619 | # Compute a list of "equivalent changesets", when the same |
|---|
| 620 | # changeset is present in other repositories. |
|---|
| 621 | |
|---|
| 622 | c = self.repos.db.cursor() |
|---|
| 623 | c.execute('SELECT rep.value, dcs.rev ' |
|---|
| 624 | 'FROM darcs_changesets dcs, darcs_changesets dcs2, repository rep ' |
|---|
| 625 | 'WHERE dcs2.repo_id = %s AND dcs2.rev = %s ' |
|---|
| 626 | ' AND dcs.hash = dcs2.hash ' |
|---|
| 627 | ' AND dcs.repo_id <> dcs2.repo_id' |
|---|
| 628 | ' AND rep.id = dcs.repo_id AND rep.name = \'name\'' |
|---|
| 629 | 'ORDER BY rep.value', (self.repos.id, self.rev)) |
|---|
| 630 | eqcsets = [(repo, rev) for repo,rev in c.fetchall()] |
|---|
| 631 | if eqcsets: |
|---|
| 632 | props['EqChangesets'] = eqcsets |
|---|
| 633 | |
|---|
| 634 | return props |
|---|