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.context import Ctx
  29 from cvs2svn_lib.artifact_manager import artifact_manager
  30 from cvs2svn_lib.project import Project
  31 from cvs2svn_lib.svn_revision_range import RevisionScores
  32 from cvs2svn_lib.openings_closings import SymbolingsReader
  33 from cvs2svn_lib.repository_mirror import RepositoryMirror
  34 from cvs2svn_lib.output_option import OutputOption
  35
  36
  37 class DVCSRunOptions(RunOptions):
  38   """Dumping ground for whatever is common to GitRunOptions and
  39   HgRunOptions."""
  40   def __init__(self, progname, cmd_args, pass_manager):
  41     Ctx().cross_project_commits = False
  42     Ctx().cross_branch_commits = False
  43     RunOptions.__init__(self, progname, cmd_args, pass_manager)
  44
  45   def set_project(
  46         self,
  47         project_cvs_repos_path,
  48         symbol_transforms=None,
  49         symbol_strategy_rules=[],
  50         ):
  51     """Set the project to be converted.
  52
  53     If a project had already been set, overwrite it.
  54
  55     Most arguments are passed straight through to the Project
  56     constructor.  SYMBOL_STRATEGY_RULES is an iterable of
  57     SymbolStrategyRules that will be applied to symbols in this
  58     project."""
  59
  60     symbol_strategy_rules = list(symbol_strategy_rules)
  61
  62     project = Project(
  63         0,
  64         project_cvs_repos_path,
  65         symbol_transforms=symbol_transforms,
  66         )
  67
  68     self.projects = [project]
  69     self.project_symbol_strategy_rules = [symbol_strategy_rules]
  70
  71   def process_options(self):
  72     # Consistency check for options and arguments.
  73     if len(self.args) == 0:
  74       self.usage()
  75       sys.exit(1)
  76
  77     if len(self.args) > 1:
  78       Log().error(error_prefix + ": must pass only one CVS repository.\n")
  79       self.usage()
  80       sys.exit(1)
  81
  82     cvsroot = self.args[0]
  83
  84     self.process_extraction_options()
  85     self.process_output_options()
  86     self.process_symbol_strategy_options()
  87     self.process_property_setter_options()
  88
  89     # Create the project:
  90     self.set_project(
  91         cvsroot,
  92         symbol_transforms=self.options.symbol_transforms,
  93         symbol_strategy_rules=self.options.symbol_strategy_rules,
  94         )
  95
  96
  97 class DVCSOutputOption(OutputOption):
  98   # name of output format (for error messages); must be set by
  99   # subclasses
 100   name = None
 101
 102   def __init__(self):
 103     self._mirror = RepositoryMirror()
 104     self._symbolings_reader = None
 105
 106   def normalize_author_transforms(self, author_transforms):
 107     """Return a new dict with the same content as author_transforms, but all
 108     strings encoded to UTF-8.  Also turns None into the empty dict."""
 109     result = {}
 110     if author_transforms is not None:
 111       for (cvsauthor, (name, email,)) in author_transforms.iteritems():
 112         cvsauthor = to_utf8(cvsauthor)
 113         name = to_utf8(name)
 114         email = to_utf8(email)
 115         result[cvsauthor] = (name, email,)
 116     return result
 117
 118   def register_artifacts(self, which_pass):
 119     # These artifacts are needed for SymbolingsReader:
 120     artifact_manager.register_temp_file_needed(
 121         config.SYMBOL_OPENINGS_CLOSINGS_SORTED, which_pass
 122         )
 123     artifact_manager.register_temp_file_needed(
 124         config.SYMBOL_OFFSETS_DB, which_pass
 125         )
 126     self._mirror.register_artifacts(which_pass)
 127
 128   def check(self):
 129     if Ctx().cross_project_commits:
 130       raise FatalError(
 131           '%s output is not supported with cross-project commits' % self.name
 132           )
 133     if Ctx().cross_branch_commits:
 134       raise FatalError(
 135           '%s output is not supported with cross-branch commits' % self.name
 136           )
 137     if Ctx().username is None:
 138       raise FatalError(
 139           '%s output requires a default commit username' % self.name
 140           )
 141
 142   def setup(self, svn_rev_count):
 143     self._symbolings_reader = SymbolingsReader()
 144     self._mirror.open()
 145
 146   def cleanup(self):
 147     self._mirror.close()
 148     self._symbolings_reader.close()
 149     del self._symbolings_reader
 150
 151   def _get_source_groups(self, svn_commit):
 152     """Return groups of sources for SVN_COMMIT.
 153
 154     SVN_COMMIT is an instance of SVNSymbolCommit.  Yield tuples
 155     (source_lod, svn_revnum, cvs_symbols) where source_lod is the line
 156     of development and svn_revnum is the revision that should serve as
 157     a source, and cvs_symbols is a list of CVSSymbolItems that can be
 158     copied from that source.  The groups are returned in arbitrary
 159     order."""
 160
 161     # Get a map {CVSSymbol : SVNRevisionRange}:
 162     range_map = self._symbolings_reader.get_range_map(svn_commit)
 163
 164     # range_map, split up into one map per LOD; i.e., {LOD :
 165     # {CVSSymbol : SVNRevisionRange}}:
 166     lod_range_maps = {}
 167
 168     for (cvs_symbol, range) in range_map.iteritems():
 169       lod_range_map = lod_range_maps.get(range.source_lod)
 170       if lod_range_map is None:
 171         lod_range_map = {}
 172         lod_range_maps[range.source_lod] = lod_range_map
 173       lod_range_map[cvs_symbol] = range
 174
 175     # Sort the sources so that the branch that serves most often as
 176     # parent is processed first:
 177     lod_ranges = lod_range_maps.items()
 178     lod_ranges.sort(
 179         lambda (lod1,lod_range_map1),(lod2,lod_range_map2):
 180         -cmp(len(lod_range_map1), len(lod_range_map2)) or cmp(lod1, lod2)
 181         )
 182
 183     for (lod, lod_range_map) in lod_ranges:
 184       while lod_range_map:
 185         revision_scores = RevisionScores(lod_range_map.values())
 186         (source_lod, revnum, score) = revision_scores.get_best_revnum()
 187         assert source_lod == lod
 188         cvs_symbols = []
 189         for (cvs_symbol, range) in lod_range_map.items():
 190           if revnum in range:
 191             cvs_symbols.append(cvs_symbol)
 192             del lod_range_map[cvs_symbol]
 193         yield (lod, revnum, cvs_symbols)
 194
 195   def _is_simple_copy(self, svn_commit, source_groups):
 196     """Return True iff SVN_COMMIT can be created as a simple copy.
 197
 198     SVN_COMMIT is an SVNTagCommit.  Return True iff it can be created
 199     as a simple copy from an existing revision (i.e., if the fixup
 200     branch can be avoided for this tag creation)."""
 201
 202     # The first requirement is that there be exactly one source:
 203     if len(source_groups) != 1:
 204       return False
 205
 206     (source_lod, svn_revnum, cvs_symbols) = source_groups[0]
 207
 208     # The second requirement is that the destination LOD not already
 209     # exist:
 210     try:
 211       self._mirror.get_current_lod_directory(svn_commit.symbol)
 212     except KeyError:
 213       # The LOD doesn't already exist.  This is good.
 214       pass
 215     else:
 216       # The LOD already exists.  It cannot be created by a copy.
 217       return False
 218
 219     # The third requirement is that the source LOD contains exactly
 220     # the same files as we need to add to the symbol:
 221     try:
 222       source_node = self._mirror.get_old_lod_directory(source_lod, svn_revnum)
 223     except KeyError:
 224       raise InternalError('Source %r does not exist' % (source_lod,))
 225     return (
 226         set([cvs_symbol.cvs_file for cvs_symbol in cvs_symbols])
 227         == set(self._get_all_files(source_node))
 228         )
 229
 230   def _get_all_files(self, node):
 231     """Generate all of the CVSFiles under NODE."""
 232
 233     for cvs_path in node:
 234       subnode = node[cvs_path]
 235       if subnode is None:
 236         yield cvs_path
 237       else:
 238         for sub_cvs_path in self._get_all_files(subnode):
 239           yield sub_cvs_path
 240
 241
 242 def to_utf8(s):
 243   if isinstance(s, unicode):
 244     return s.encode('utf8')
 245   else:
 246     return s
 247
 248