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 Log
  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       Log().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   # name of output format (for error messages); must be set by
 137   # subclasses
 138   name = None
 139
 140   def __init__(self):
 141     self._mirror = RepositoryMirror()
 142     self._symbolings_reader = None
 143
 144   def normalize_author_transforms(self, author_transforms):
 145     """Convert AUTHOR_TRANSFORMS into author strings.
 146
 147     AUTHOR_TRANSFORMS is a dict { CVSAUTHOR : DVCSAUTHOR } where
 148     CVSAUTHOR is the CVS author and DVCSAUTHOR is either:
 149
 150     * a tuple (NAME, EMAIL) where NAME and EMAIL are strings.  Such
 151       entries are converted into a UTF-8 string of the form 'name
 152       <email>'.
 153
 154     * a string already in the form 'name <email>'.
 155
 156     Return a similar dict { CVSAUTHOR : DVCSAUTHOR } where all keys
 157     and values are UTF-8-encoded strings.
 158
 159     Any of the input strings may be Unicode strings (in which case
 160     they are encoded to UTF-8) or 8-bit strings (in which case they
 161     are used as-is).  Also turns None into the empty dict."""
 162
 163     result = {}
 164     if author_transforms is not None:
 165       for (cvsauthor, dvcsauthor) in author_transforms.iteritems():
 166         cvsauthor = to_utf8(cvsauthor)
 167         if isinstance(dvcsauthor, basestring):
 168           dvcsauthor = to_utf8(dvcsauthor)
 169         else:
 170           (name, email,) = dvcsauthor
 171           name = to_utf8(name)
 172           email = to_utf8(email)
 173           dvcsauthor = "%s <%s>" % (name, email,)
 174         result[cvsauthor] = dvcsauthor
 175     return result
 176
 177   def register_artifacts(self, which_pass):
 178     # These artifacts are needed for SymbolingsReader:
 179     artifact_manager.register_temp_file_needed(
 180         config.SYMBOL_OPENINGS_CLOSINGS_SORTED, which_pass
 181         )
 182     artifact_manager.register_temp_file_needed(
 183         config.SYMBOL_OFFSETS_DB, which_pass
 184         )
 185     self._mirror.register_artifacts(which_pass)
 186
 187   def check(self):
 188     if Ctx().cross_project_commits:
 189       raise FatalError(
 190           '%s output is not supported with cross-project commits' % self.name
 191           )
 192     if Ctx().cross_branch_commits:
 193       raise FatalError(
 194           '%s output is not supported with cross-branch commits' % self.name
 195           )
 196     if Ctx().username is None:
 197       raise FatalError(
 198           '%s output requires a default commit username' % self.name
 199           )
 200
 201   def setup(self, svn_rev_count):
 202     self._symbolings_reader = SymbolingsReader()
 203     self._mirror.open()
 204
 205   def cleanup(self):
 206     self._mirror.close()
 207     self._symbolings_reader.close()
 208     del self._symbolings_reader
 209
 210   def _get_source_groups(self, svn_commit):
 211     """Return groups of sources for SVN_COMMIT.
 212
 213     SVN_COMMIT is an instance of SVNSymbolCommit.  Yield tuples
 214     (source_lod, svn_revnum, cvs_symbols) where source_lod is the line
 215     of development and svn_revnum is the revision that should serve as
 216     a source, and cvs_symbols is a list of CVSSymbolItems that can be
 217     copied from that source.  The groups are returned in arbitrary
 218     order."""
 219
 220     # Get a map {CVSSymbol : SVNRevisionRange}:
 221     range_map = self._symbolings_reader.get_range_map(svn_commit)
 222
 223     # range_map, split up into one map per LOD; i.e., {LOD :
 224     # {CVSSymbol : SVNRevisionRange}}:
 225     lod_range_maps = {}
 226
 227     for (cvs_symbol, range) in range_map.iteritems():
 228       lod_range_map = lod_range_maps.get(range.source_lod)
 229       if lod_range_map is None:
 230         lod_range_map = {}
 231         lod_range_maps[range.source_lod] = lod_range_map
 232       lod_range_map[cvs_symbol] = range
 233
 234     # Sort the sources so that the branch that serves most often as
 235     # parent is processed first:
 236     lod_ranges = lod_range_maps.items()
 237     lod_ranges.sort(
 238         lambda (lod1,lod_range_map1),(lod2,lod_range_map2):
 239         -cmp(len(lod_range_map1), len(lod_range_map2)) or cmp(lod1, lod2)
 240         )
 241
 242     for (lod, lod_range_map) in lod_ranges:
 243       while lod_range_map:
 244         revision_scores = RevisionScores(lod_range_map.values())
 245         (source_lod, revnum, score) = revision_scores.get_best_revnum()
 246         assert source_lod == lod
 247         cvs_symbols = []
 248         for (cvs_symbol, range) in lod_range_map.items():
 249           if revnum in range:
 250             cvs_symbols.append(cvs_symbol)
 251             del lod_range_map[cvs_symbol]
 252         yield (lod, revnum, cvs_symbols)
 253
 254   def _is_simple_copy(self, svn_commit, source_groups):
 255     """Return True iff SVN_COMMIT can be created as a simple copy.
 256
 257     SVN_COMMIT is an SVNTagCommit.  Return True iff it can be created
 258     as a simple copy from an existing revision (i.e., if the fixup
 259     branch can be avoided for this tag creation)."""
 260
 261     # The first requirement is that there be exactly one source:
 262     if len(source_groups) != 1:
 263       return False
 264
 265     (source_lod, svn_revnum, cvs_symbols) = source_groups[0]
 266
 267     # The second requirement is that the destination LOD not already
 268     # exist:
 269     try:
 270       self._mirror.get_current_lod_directory(svn_commit.symbol)
 271     except KeyError:
 272       # The LOD doesn't already exist.  This is good.
 273       pass
 274     else:
 275       # The LOD already exists.  It cannot be created by a copy.
 276       return False
 277
 278     # The third requirement is that the source LOD contains exactly
 279     # the same files as we need to add to the symbol:
 280     try:
 281       source_node = self._mirror.get_old_lod_directory(source_lod, svn_revnum)
 282     except KeyError:
 283       raise InternalError('Source %r does not exist' % (source_lod,))
 284     return (
 285         set([cvs_symbol.cvs_file for cvs_symbol in cvs_symbols])
 286         == set(self._get_all_files(source_node))
 287         )
 288
 289   def _get_all_files(self, node):
 290     """Generate all of the CVSFiles under NODE."""
 291
 292     for cvs_path in node:
 293       subnode = node[cvs_path]
 294       if subnode is None:
 295         yield cvs_path
 296       else:
 297         for sub_cvs_path in self._get_all_files(subnode):
 298           yield sub_cvs_path
 299
 300
 301 class MirrorUpdater(object):
 302   def register_artifacts(self, which_pass):
 303     pass
 304
 305   def start(self, mirror):
 306     self._mirror = mirror
 307
 308   def _mkdir_p(self, cvs_directory, lod):
 309     """Make sure that CVS_DIRECTORY exists in LOD.
 310
 311     If not, create it.  Return the node for CVS_DIRECTORY."""
 312
 313     try:
 314       node = self._mirror.get_current_lod_directory(lod)
 315     except KeyError:
 316       node = self._mirror.add_lod(lod)
 317
 318     for sub_path in cvs_directory.get_ancestry()[1:]:
 319       try:
 320         node = node[sub_path]
 321       except KeyError:
 322         node = node.mkdir(sub_path)
 323       if node is None:
 324         raise ExpectedDirectoryError(
 325             'File found at \'%s\' where directory was expected.' % (sub_path,)
 326             )
 327
 328     return node
 329
 330   def add_file(self, cvs_rev, post_commit):
 331     cvs_file = cvs_rev.cvs_file
 332     if post_commit:
 333       lod = cvs_file.project.get_trunk()
 334     else:
 335       lod = cvs_rev.lod
 336     parent_node = self._mkdir_p(cvs_file.parent_directory, lod)
 337     parent_node.add_file(cvs_file)
 338
 339   def modify_file(self, cvs_rev, post_commit):
 340     cvs_file = cvs_rev.cvs_file
 341     if post_commit:
 342       lod = cvs_file.project.get_trunk()
 343     else:
 344       lod = cvs_rev.lod
 345     if self._mirror.get_current_path(cvs_file, lod) is not None:
 346       raise ExpectedFileError(
 347           'Directory found at \'%s\' where file was expected.' % (cvs_file,)
 348           )
 349
 350   def delete_file(self, cvs_rev, post_commit):
 351     cvs_file = cvs_rev.cvs_file
 352     if post_commit:
 353       lod = cvs_file.project.get_trunk()
 354     else:
 355       lod = cvs_rev.lod
 356     parent_node = self._mirror.get_current_path(
 357         cvs_file.parent_directory, lod
 358         )
 359     if parent_node[cvs_file] is not None:
 360       raise ExpectedFileError(
 361           'Directory found at \'%s\' where file was expected.' % (cvs_file,)
 362           )
 363     del parent_node[cvs_file]
 364
 365   def process_revision(self, cvs_rev, post_commit):
 366     if isinstance(cvs_rev, CVSRevisionAdd):
 367       self.add_file(cvs_rev, post_commit)
 368     elif isinstance(cvs_rev, CVSRevisionChange):
 369       self.modify_file(cvs_rev, post_commit)
 370     elif isinstance(cvs_rev, CVSRevisionDelete):
 371       self.delete_file(cvs_rev, post_commit)
 372     elif isinstance(cvs_rev, CVSRevisionNoop):
 373       pass
 374     else:
 375       raise InternalError('Unexpected CVSRevision type: %s' % (cvs_rev,))
 376
 377   def branch_file(self, cvs_symbol):
 378     cvs_file = cvs_symbol.cvs_file
 379     parent_node = self._mkdir_p(cvs_file.parent_directory, cvs_symbol.symbol)
 380     parent_node.add_file(cvs_file)
 381
 382   def finish(self):
 383     del self._mirror
 384
 385
 386 def to_utf8(s):
 387   if isinstance(s, unicode):
 388     return s.encode('utf8')
 389   else:
 390     return s
 391
 392