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
  41
  42 class DVCSRunOptions(RunOptions):
  43   """Dumping ground for whatever is common to GitRunOptions and
  44   HgRunOptions."""
  45   def __init__(self, progname, cmd_args, pass_manager):
  46     Ctx().cross_project_commits = False
  47     Ctx().cross_branch_commits = False
  48     RunOptions.__init__(self, progname, cmd_args, pass_manager)
  49
  50   def set_project(
  51         self,
  52         project_cvs_repos_path,
  53         symbol_transforms=None,
  54         symbol_strategy_rules=[],
  55         ):
  56     """Set the project to be converted.
  57
  58     If a project had already been set, overwrite it.
  59
  60     Most arguments are passed straight through to the Project
  61     constructor.  SYMBOL_STRATEGY_RULES is an iterable of
  62     SymbolStrategyRules that will be applied to symbols in this
  63     project."""
  64
  65     symbol_strategy_rules = list(symbol_strategy_rules)
  66
  67     project = Project(
  68         0,
  69         project_cvs_repos_path,
  70         symbol_transforms=symbol_transforms,
  71         )
  72
  73     self.projects = [project]
  74     self.project_symbol_strategy_rules = [symbol_strategy_rules]
  75
  76   def process_options(self):
  77     # Consistency check for options and arguments.
  78     if len(self.args) == 0:
  79       self.usage()
  80       sys.exit(1)
  81
  82     if len(self.args) > 1:
  83       Log().error(error_prefix + ": must pass only one CVS repository.\n")
  84       self.usage()
  85       sys.exit(1)
  86
  87     cvsroot = self.args[0]
  88
  89     self.process_extraction_options()
  90     self.process_output_options()
  91     self.process_symbol_strategy_options()
  92     self.process_property_setter_options()
  93
  94     # Create the project:
  95     self.set_project(
  96         cvsroot,
  97         symbol_transforms=self.options.symbol_transforms,
  98         symbol_strategy_rules=self.options.symbol_strategy_rules,
  99         )
 100
 101
 102 class DVCSOutputOption(OutputOption):
 103   # name of output format (for error messages); must be set by
 104   # subclasses
 105   name = None
 106
 107   def __init__(self):
 108     self._mirror = RepositoryMirror()
 109     self._symbolings_reader = None
 110
 111   def normalize_author_transforms(self, author_transforms):
 112     """Convert AUTHOR_TRANSFORMS into author strings.
 113
 114     AUTHOR_TRANSFORMS is a dict { CVSAUTHOR : DVCSAUTHOR } where
 115     CVSAUTHOR is the CVS author and DVCSAUTHOR is either:
 116
 117     * a tuple (NAME, EMAIL) where NAME and EMAIL are strings.  Such
 118       entries are converted into a UTF-8 string of the form 'name
 119       <email>'.
 120
 121     * a string already in the form 'name <email>'.
 122
 123     Return a similar dict { CVSAUTHOR : DVCSAUTHOR } where all keys
 124     and values are UTF-8-encoded strings.
 125
 126     Any of the input strings may be Unicode strings (in which case
 127     they are encoded to UTF-8) or 8-bit strings (in which case they
 128     are used as-is).  Also turns None into the empty dict."""
 129
 130     result = {}
 131     if author_transforms is not None:
 132       for (cvsauthor, dvcsauthor) in author_transforms.iteritems():
 133         cvsauthor = to_utf8(cvsauthor)
 134         if isinstance(dvcsauthor, basestring):
 135           dvcsauthor = to_utf8(dvcsauthor)
 136         else:
 137           (name, email,) = dvcsauthor
 138           name = to_utf8(name)
 139           email = to_utf8(email)
 140           dvcsauthor = "%s <%s>" % (name, email,)
 141         result[cvsauthor] = dvcsauthor
 142     return result
 143
 144   def register_artifacts(self, which_pass):
 145     # These artifacts are needed for SymbolingsReader:
 146     artifact_manager.register_temp_file_needed(
 147         config.SYMBOL_OPENINGS_CLOSINGS_SORTED, which_pass
 148         )
 149     artifact_manager.register_temp_file_needed(
 150         config.SYMBOL_OFFSETS_DB, which_pass
 151         )
 152     self._mirror.register_artifacts(which_pass)
 153
 154   def check(self):
 155     if Ctx().cross_project_commits:
 156       raise FatalError(
 157           '%s output is not supported with cross-project commits' % self.name
 158           )
 159     if Ctx().cross_branch_commits:
 160       raise FatalError(
 161           '%s output is not supported with cross-branch commits' % self.name
 162           )
 163     if Ctx().username is None:
 164       raise FatalError(
 165           '%s output requires a default commit username' % self.name
 166           )
 167
 168   def setup(self, svn_rev_count):
 169     self._symbolings_reader = SymbolingsReader()
 170     self._mirror.open()
 171
 172   def cleanup(self):
 173     self._mirror.close()
 174     self._symbolings_reader.close()
 175     del self._symbolings_reader
 176
 177   def _get_source_groups(self, svn_commit):
 178     """Return groups of sources for SVN_COMMIT.
 179
 180     SVN_COMMIT is an instance of SVNSymbolCommit.  Yield tuples
 181     (source_lod, svn_revnum, cvs_symbols) where source_lod is the line
 182     of development and svn_revnum is the revision that should serve as
 183     a source, and cvs_symbols is a list of CVSSymbolItems that can be
 184     copied from that source.  The groups are returned in arbitrary
 185     order."""
 186
 187     # Get a map {CVSSymbol : SVNRevisionRange}:
 188     range_map = self._symbolings_reader.get_range_map(svn_commit)
 189
 190     # range_map, split up into one map per LOD; i.e., {LOD :
 191     # {CVSSymbol : SVNRevisionRange}}:
 192     lod_range_maps = {}
 193
 194     for (cvs_symbol, range) in range_map.iteritems():
 195       lod_range_map = lod_range_maps.get(range.source_lod)
 196       if lod_range_map is None:
 197         lod_range_map = {}
 198         lod_range_maps[range.source_lod] = lod_range_map
 199       lod_range_map[cvs_symbol] = range
 200
 201     # Sort the sources so that the branch that serves most often as
 202     # parent is processed first:
 203     lod_ranges = lod_range_maps.items()
 204     lod_ranges.sort(
 205         lambda (lod1,lod_range_map1),(lod2,lod_range_map2):
 206         -cmp(len(lod_range_map1), len(lod_range_map2)) or cmp(lod1, lod2)
 207         )
 208
 209     for (lod, lod_range_map) in lod_ranges:
 210       while lod_range_map:
 211         revision_scores = RevisionScores(lod_range_map.values())
 212         (source_lod, revnum, score) = revision_scores.get_best_revnum()
 213         assert source_lod == lod
 214         cvs_symbols = []
 215         for (cvs_symbol, range) in lod_range_map.items():
 216           if revnum in range:
 217             cvs_symbols.append(cvs_symbol)
 218             del lod_range_map[cvs_symbol]
 219         yield (lod, revnum, cvs_symbols)
 220
 221   def _is_simple_copy(self, svn_commit, source_groups):
 222     """Return True iff SVN_COMMIT can be created as a simple copy.
 223
 224     SVN_COMMIT is an SVNTagCommit.  Return True iff it can be created
 225     as a simple copy from an existing revision (i.e., if the fixup
 226     branch can be avoided for this tag creation)."""
 227
 228     # The first requirement is that there be exactly one source:
 229     if len(source_groups) != 1:
 230       return False
 231
 232     (source_lod, svn_revnum, cvs_symbols) = source_groups[0]
 233
 234     # The second requirement is that the destination LOD not already
 235     # exist:
 236     try:
 237       self._mirror.get_current_lod_directory(svn_commit.symbol)
 238     except KeyError:
 239       # The LOD doesn't already exist.  This is good.
 240       pass
 241     else:
 242       # The LOD already exists.  It cannot be created by a copy.
 243       return False
 244
 245     # The third requirement is that the source LOD contains exactly
 246     # the same files as we need to add to the symbol:
 247     try:
 248       source_node = self._mirror.get_old_lod_directory(source_lod, svn_revnum)
 249     except KeyError:
 250       raise InternalError('Source %r does not exist' % (source_lod,))
 251     return (
 252         set([cvs_symbol.cvs_file for cvs_symbol in cvs_symbols])
 253         == set(self._get_all_files(source_node))
 254         )
 255
 256   def _get_all_files(self, node):
 257     """Generate all of the CVSFiles under NODE."""
 258
 259     for cvs_path in node:
 260       subnode = node[cvs_path]
 261       if subnode is None:
 262         yield cvs_path
 263       else:
 264         for sub_cvs_path in self._get_all_files(subnode):
 265           yield sub_cvs_path
 266
 267
 268 class MirrorUpdater(object):
 269   def register_artifacts(self, which_pass):
 270     pass
 271
 272   def start(self, mirror):
 273     self._mirror = mirror
 274
 275   def _mkdir_p(self, cvs_directory, lod):
 276     """Make sure that CVS_DIRECTORY exists in LOD.
 277
 278     If not, create it.  Return the node for CVS_DIRECTORY."""
 279
 280     try:
 281       node = self._mirror.get_current_lod_directory(lod)
 282     except KeyError:
 283       node = self._mirror.add_lod(lod)
 284
 285     for sub_path in cvs_directory.get_ancestry()[1:]:
 286       try:
 287         node = node[sub_path]
 288       except KeyError:
 289         node = node.mkdir(sub_path)
 290       if node is None:
 291         raise ExpectedDirectoryError(
 292             'File found at \'%s\' where directory was expected.' % (sub_path,)
 293             )
 294
 295     return node
 296
 297   def add_file(self, cvs_rev, post_commit):
 298     cvs_file = cvs_rev.cvs_file
 299     if post_commit:
 300       lod = cvs_file.project.get_trunk()
 301     else:
 302       lod = cvs_rev.lod
 303     parent_node = self._mkdir_p(cvs_file.parent_directory, lod)
 304     parent_node.add_file(cvs_file)
 305
 306   def modify_file(self, cvs_rev, post_commit):
 307     cvs_file = cvs_rev.cvs_file
 308     if post_commit:
 309       lod = cvs_file.project.get_trunk()
 310     else:
 311       lod = cvs_rev.lod
 312     if self._mirror.get_current_path(cvs_file, lod) is not None:
 313       raise ExpectedFileError(
 314           'Directory found at \'%s\' where file was expected.' % (cvs_file,)
 315           )
 316
 317   def delete_file(self, cvs_rev, post_commit):
 318     cvs_file = cvs_rev.cvs_file
 319     if post_commit:
 320       lod = cvs_file.project.get_trunk()
 321     else:
 322       lod = cvs_rev.lod
 323     parent_node = self._mirror.get_current_path(
 324         cvs_file.parent_directory, lod
 325         )
 326     if parent_node[cvs_file] is not None:
 327       raise ExpectedFileError(
 328           'Directory found at \'%s\' where file was expected.' % (cvs_file,)
 329           )
 330     del parent_node[cvs_file]
 331
 332   def process_revision(self, cvs_rev, post_commit):
 333     if isinstance(cvs_rev, CVSRevisionAdd):
 334       self.add_file(cvs_rev, post_commit)
 335     elif isinstance(cvs_rev, CVSRevisionChange):
 336       self.modify_file(cvs_rev, post_commit)
 337     elif isinstance(cvs_rev, CVSRevisionDelete):
 338       self.delete_file(cvs_rev, post_commit)
 339     elif isinstance(cvs_rev, CVSRevisionNoop):
 340       pass
 341     else:
 342       raise InternalError('Unexpected CVSRevision type: %s' % (cvs_rev,))
 343
 344   def branch_file(self, cvs_symbol):
 345     cvs_file = cvs_symbol.cvs_file
 346     parent_node = self._mkdir_p(cvs_file.parent_directory, cvs_symbol.symbol)
 347     parent_node.add_file(cvs_file)
 348
 349   def finish(self):
 350     del self._mirror
 351
 352
 353 def to_utf8(s):
 354   if isinstance(s, unicode):
 355     return s.encode('utf8')
 356   else:
 357     return s
 358
 359