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