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