cvs2svn_lib/dvcs_common.py

   1 # (Be in -*- python -*- mode.)
   2 #
   3 # ====================================================================
   4 # Copyright (c) 2007-2009 CollabNet.  All rights reserved.
   5 #
   6 # This software is licensed as described in the file COPYING, which
   7 # you should have received as part of this distribution.  The terms
   8 # are also available at http://subversion.tigris.org/license-1.html.
   9 # If newer versions of this license are posted there, you may use a
  10 # newer version instead, at your option.
  11 #
  12 # This software consists of voluntary contributions made by many
  13 # individuals.  For exact contribution history, see the revision
  14 # history and logs, available at http://cvs2svn.tigris.org/.
  15 # ====================================================================
  16
  17 """Miscellaneous utility code common to DVCS backends (like
  18 Git, Mercurial, or Bazaar).
  19 """
  20
  21 import sys
  22
  23 from cvs2svn_lib import config
  24 from cvs2svn_lib.common import FatalError
  25 from cvs2svn_lib.common import InternalError
  26 from cvs2svn_lib.run_options import RunOptions
  27 from cvs2svn_lib.log import logger
  28 from cvs2svn_lib.common import error_prefix
  29 from cvs2svn_lib.context import Ctx
  30 from cvs2svn_lib.artifact_manager import artifact_manager
  31 from cvs2svn_lib.project import Project
  32 from cvs2svn_lib.cvs_item import CVSRevisionAdd
  33 from cvs2svn_lib.cvs_item import CVSRevisionChange
  34 from cvs2svn_lib.cvs_item import CVSRevisionDelete
  35 from cvs2svn_lib.cvs_item import CVSRevisionNoop
  36 from cvs2svn_lib.svn_revision_range import RevisionScores
  37 from cvs2svn_lib.openings_closings import SymbolingsReader
  38 from cvs2svn_lib.repository_mirror import RepositoryMirror
  39 from cvs2svn_lib.output_option import OutputOption
  40 from cvs2svn_lib.property_setters import FilePropertySetter
  41
  42
  43 class KeywordHandlingPropertySetter(FilePropertySetter):
  44   """Set property _keyword_handling to a specified value.
  45
  46   This keyword is used to tell the RevisionReader whether it has to
  47   collapse/expand RCS keywords when generating the fulltext or leave
  48   them alone."""
  49
  50   propname = '_keyword_handling'
  51
  52   def __init__(self, value):
  53     if value not in ['collapsed', 'expanded', 'untouched', None]:
  54       raise FatalError(
  55           'Value for %s must be "collapsed", "expanded", or "untouched"'
  56           % (self.propname,)
  57           )
  58     self.value = value
  59
  60   def set_properties(self, cvs_file):
  61     if self.propname in cvs_file.properties:
  62       return
  63
  64     cvs_file.properties[self.propname] = self.value
  65
  66
  67 class DVCSRunOptions(RunOptions):
  68   """Dumping ground for whatever is common to GitRunOptions and
  69   HgRunOptions."""
  70   def __init__(self, progname, cmd_args, pass_manager):
  71     Ctx().cross_project_commits = False
  72     Ctx().cross_branch_commits = False
  73     RunOptions.__init__(self, progname, cmd_args, pass_manager)
  74
  75   def set_project(
  76         self,
  77         project_cvs_repos_path,
  78         symbol_transforms=None,
  79         symbol_strategy_rules=[],
  80         ):
  81     """Set the project to be converted.
  82
  83     If a project had already been set, overwrite it.
  84
  85     Most arguments are passed straight through to the Project
  86     constructor.  SYMBOL_STRATEGY_RULES is an iterable of
  87     SymbolStrategyRules that will be applied to symbols in this
  88     project."""
  89
  90     symbol_strategy_rules = list(symbol_strategy_rules)
  91
  92     project = Project(
  93         0,
  94         project_cvs_repos_path,
  95         symbol_transforms=symbol_transforms,
  96         )
  97
  98     self.projects = [project]
  99     self.project_symbol_strategy_rules = [symbol_strategy_rules]
 100
 101   def process_property_setter_options(self):
 102     super(DVCSRunOptions, self).process_property_setter_options()
 103
 104     # Property setters for internal use:
 105     Ctx().file_property_setters.append(
 106         KeywordHandlingPropertySetter('collapsed')
 107         )
 108
 109   def process_options(self):
 110     # Consistency check for options and arguments.
 111     if len(self.args) == 0:
 112       self.usage()
 113       sys.exit(1)
 114
 115     if len(self.args) > 1:
 116       logger.error(error_prefix + ": must pass only one CVS repository.\n")
 117       self.usage()
 118       sys.exit(1)
 119
 120     cvsroot = self.args[0]
 121
 122     self.process_extraction_options()
 123     self.process_output_options()
 124     self.process_symbol_strategy_options()
 125     self.process_property_setter_options()
 126
 127     # Create the project:
 128     self.set_project(
 129         cvsroot,
 130         symbol_transforms=self.options.symbol_transforms,
 131         symbol_strategy_rules=self.options.symbol_strategy_rules,
 132         )
 133
 134
 135 class DVCSOutputOption(OutputOption):
 136   def __init__(self):
 137     self._mirror = RepositoryMirror()
 138     self._symbolings_reader = None
 139
 140   def normalize_author_transforms(self, author_transforms):
 141     """Convert AUTHOR_TRANSFORMS into author strings.
 142
 143     AUTHOR_TRANSFORMS is a dict { CVSAUTHOR : DVCSAUTHOR } where
 144     CVSAUTHOR is the CVS author and DVCSAUTHOR is either:
 145
 146     * a tuple (NAME, EMAIL) where NAME and EMAIL are strings.  Such
 147       entries are converted into a UTF-8 string of the form 'name
 148       <email>'.
 149
 150     * a string already in the form 'name <email>'.
 151
 152     Return a similar dict { CVSAUTHOR : DVCSAUTHOR } where all keys
 153     and values are UTF-8-encoded strings.
 154
 155     Any of the input strings may be Unicode strings (in which case
 156     they are encoded to UTF-8) or 8-bit strings (in which case they
 157     are used as-is).  Also turns None into the empty dict."""
 158
 159     result = {}
 160     if author_transforms is not None:
 161       for (cvsauthor, dvcsauthor) in author_transforms.iteritems():
 162         cvsauthor = to_utf8(cvsauthor)
 163         if isinstance(dvcsauthor, basestring):
 164           dvcsauthor = to_utf8(dvcsauthor)
 165         else:
 166           (name, email,) = dvcsauthor
 167           name = to_utf8(name)
 168           email = to_utf8(email)
 169           dvcsauthor = "%s <%s>" % (name, email,)
 170         result[cvsauthor] = dvcsauthor
 171     return result
 172
 173   def register_artifacts(self, which_pass):
 174     # These artifacts are needed for SymbolingsReader:
 175     artifact_manager.register_temp_file_needed(
 176         config.SYMBOL_OPENINGS_CLOSINGS_SORTED, which_pass
 177         )
 178     artifact_manager.register_temp_file_needed(
 179         config.SYMBOL_OFFSETS_DB, which_pass
 180         )
 181     self._mirror.register_artifacts(which_pass)
 182
 183   def check(self):
 184     if Ctx().cross_project_commits:
 185       raise FatalError(
 186           '%s output is not supported with cross-project commits' % self.name
 187           )
 188     if Ctx().cross_branch_commits:
 189       raise FatalError(
 190           '%s output is not supported with cross-branch commits' % self.name
 191           )
 192     if Ctx().username is None:
 193       raise FatalError(
 194           '%s output requires a default commit username' % self.name
 195           )
 196
 197   def setup(self, svn_rev_count):
 198     self._symbolings_reader = SymbolingsReader()
 199     self._mirror.open()
 200
 201   def cleanup(self):
 202     self._mirror.close()
 203     self._symbolings_reader.close()
 204     del self._symbolings_reader
 205
 206   def _get_source_groups(self, svn_commit):
 207     """Return groups of sources for SVN_COMMIT.
 208
 209     SVN_COMMIT is an instance of SVNSymbolCommit.  Return a list of tuples
 210     (svn_revnum, source_lod, cvs_symbols) where svn_revnum is the revision
 211     that should serve as a source, source_lod is the CVS line of
 212     development, and cvs_symbols is a list of CVSSymbolItems that can be
 213     copied from that source.  The list is in arbitrary order."""
 214
 215     # Get a map {CVSSymbol : SVNRevisionRange}:
 216     range_map = self._symbolings_reader.get_range_map(svn_commit)
 217
 218     # range_map, split up into one map per LOD; i.e., {LOD :
 219     # {CVSSymbol : SVNRevisionRange}}:
 220     lod_range_maps = {}
 221
 222     for (cvs_symbol, range) in range_map.iteritems():
 223       lod_range_map = lod_range_maps.get(range.source_lod)
 224       if lod_range_map is None:
 225         lod_range_map = {}
 226         lod_range_maps[range.source_lod] = lod_range_map
 227       lod_range_map[cvs_symbol] = range
 228
 229     # Sort the sources so that the branch that serves most often as
 230     # parent is processed first:
 231     lod_ranges = lod_range_maps.items()
 232     lod_ranges.sort(
 233         lambda (lod1,lod_range_map1),(lod2,lod_range_map2):
 234         -cmp(len(lod_range_map1), len(lod_range_map2)) or cmp(lod1, lod2)
 235         )
 236
 237     source_groups = []
 238     for (lod, lod_range_map) in lod_ranges:
 239       while lod_range_map:
 240         revision_scores = RevisionScores(lod_range_map.values())
 241         (source_lod, revnum, score) = revision_scores.get_best_revnum()
 242         assert source_lod == lod
 243         cvs_symbols = []
 244         for (cvs_symbol, range) in lod_range_map.items():
 245           if revnum in range:
 246             cvs_symbols.append(cvs_symbol)
 247             del lod_range_map[cvs_symbol]
 248         source_groups.append((revnum, lod, cvs_symbols))
 249
 250     return source_groups
 251
 252   def _is_simple_copy(self, svn_commit, source_groups):
 253     """Return True iff SVN_COMMIT can be created as a simple copy.
 254
 255     SVN_COMMIT is an SVNTagCommit.  Return True iff it can be created
 256     as a simple copy from an existing revision (i.e., if the fixup
 257     branch can be avoided for this tag creation)."""
 258
 259     # The first requirement is that there be exactly one source:
 260     if len(source_groups) != 1:
 261       return False
 262
 263     (svn_revnum, source_lod, cvs_symbols) = source_groups[0]
 264
 265     # The second requirement is that the destination LOD not already
 266     # exist:
 267     try:
 268       self._mirror.get_current_lod_directory(svn_commit.symbol)
 269     except KeyError:
 270       # The LOD doesn't already exist.  This is good.
 271       pass
 272     else:
 273       # The LOD already exists.  It cannot be created by a copy.
 274       return False
 275
 276     # The third requirement is that the source LOD contains exactly
 277     # the same files as we need to add to the symbol:
 278     try:
 279       source_node = self._mirror.get_old_lod_directory(source_lod, svn_revnum)
 280     except KeyError:
 281       raise InternalError('Source %r does not exist' % (source_lod,))
 282     return (
 283         set([cvs_symbol.cvs_file for cvs_symbol in cvs_symbols])
 284         == set(self._get_all_files(source_node))
 285         )
 286
 287   def _get_all_files(self, node):
 288     """Generate all of the CVSFiles under NODE."""
 289
 290     for cvs_path in node:
 291       subnode = node[cvs_path]
 292       if subnode is None:
 293         yield cvs_path
 294       else:
 295         for sub_cvs_path in self._get_all_files(subnode):
 296           yield sub_cvs_path
 297
 298
 299 class MirrorUpdater(object):
 300   def register_artifacts(self, which_pass):
 301     pass
 302
 303   def start(self, mirror):
 304     self._mirror = mirror
 305
 306   def _mkdir_p(self, cvs_directory, lod):
 307     """Make sure that CVS_DIRECTORY exists in LOD.
 308
 309     If not, create it.  Return the node for CVS_DIRECTORY."""
 310
 311     try:
 312       node = self._mirror.get_current_lod_directory(lod)
 313     except KeyError:
 314       node = self._mirror.add_lod(lod)
 315
 316     for sub_path in cvs_directory.get_ancestry()[1:]:
 317       try:
 318         node = node[sub_path]
 319       except KeyError:
 320         node = node.mkdir(sub_path)
 321       if node is None:
 322         raise ExpectedDirectoryError(
 323             'File found at \'%s\' where directory was expected.' % (sub_path,)
 324             )
 325
 326     return node
 327
 328   def add_file(self, cvs_rev, post_commit):
 329     cvs_file = cvs_rev.cvs_file
 330     if post_commit:
 331       lod = cvs_file.project.get_trunk()
 332     else:
 333       lod = cvs_rev.lod
 334     parent_node = self._mkdir_p(cvs_file.parent_directory, lod)
 335     parent_node.add_file(cvs_file)
 336
 337   def modify_file(self, cvs_rev, post_commit):
 338     cvs_file = cvs_rev.cvs_file
 339     if post_commit:
 340       lod = cvs_file.project.get_trunk()
 341     else:
 342       lod = cvs_rev.lod
 343     if self._mirror.get_current_path(cvs_file, lod) is not None:
 344       raise ExpectedFileError(
 345           'Directory found at \'%s\' where file was expected.' % (cvs_file,)
 346           )
 347
 348   def delete_file(self, cvs_rev, post_commit):
 349     cvs_file = cvs_rev.cvs_file
 350     if post_commit:
 351       lod = cvs_file.project.get_trunk()
 352     else:
 353       lod = cvs_rev.lod
 354     parent_node = self._mirror.get_current_path(
 355         cvs_file.parent_directory, lod
 356         )
 357     if parent_node[cvs_file] is not None:
 358       raise ExpectedFileError(
 359           'Directory found at \'%s\' where file was expected.' % (cvs_file,)
 360           )
 361     del parent_node[cvs_file]
 362
 363   def process_revision(self, cvs_rev, post_commit):
 364     if isinstance(cvs_rev, CVSRevisionAdd):
 365       self.add_file(cvs_rev, post_commit)
 366     elif isinstance(cvs_rev, CVSRevisionChange):
 367       self.modify_file(cvs_rev, post_commit)
 368     elif isinstance(cvs_rev, CVSRevisionDelete):
 369       self.delete_file(cvs_rev, post_commit)
 370     elif isinstance(cvs_rev, CVSRevisionNoop):
 371       pass
 372     else:
 373       raise InternalError('Unexpected CVSRevision type: %s' % (cvs_rev,))
 374
 375   def branch_file(self, cvs_symbol):
 376     cvs_file = cvs_symbol.cvs_file
 377     parent_node = self._mkdir_p(cvs_file.parent_directory, cvs_symbol.symbol)
 378     parent_node.add_file(cvs_file)
 379
 380   def finish(self):
 381     del self._mirror
 382
 383
 384 def to_utf8(s):
 385   if isinstance(s, unicode):
 386     return s.encode('utf8')
 387   else:
 388     return s
 389
 390