cvs2svn_lib/cvs_path.py

   1 # (Be in -*- python -*- mode.)
   2 #
   3 # ====================================================================
   4 # Copyright (c) 2000-2008 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 """Classes that represent files and directories within CVS repositories."""
  18
  19 import os
  20
  21 from cvs2svn_lib.common import path_join
  22 from cvs2svn_lib.context import Ctx
  23
  24
  25 class CVSPath(object):
  26   """Represent a CVS file or directory.
  27
  28   Members:
  29
  30     id -- (int) unique ID for this CVSPath.  At any moment, there is
  31         at most one CVSPath instance with a particular ID.  (This
  32         means that object identity is the same as object equality, and
  33         objects can be used as map keys even though they don't have a
  34         __hash__() method).
  35
  36     project -- (Project) the project containing this CVSPath.
  37
  38     parent_directory -- (CVSDirectory or None) the CVSDirectory
  39         containing this CVSPath.
  40
  41     basename -- (string) the base name of this CVSPath (no ',v').  The
  42         basename of the root directory of a project is ''.
  43
  44     filename -- (string) the filesystem path to this CVSPath in the
  45         CVS repository.  This is in native format, and already
  46         normalised the way os.path.normpath() normalises paths.  It
  47         starts with the repository path passed to
  48         run_options.add_project() in the options.py file.
  49
  50     ordinal -- (int) the order that this instance should be sorted
  51         relative to other CVSPath instances.  This member is set based
  52         on the ordering imposed by sort_key() by CVSPathDatabase after
  53         all CVSFiles have been processed.  Comparisons of CVSPath
  54         using __cmp__() simply compare the ordinals.
  55
  56   """
  57
  58   __slots__ = [
  59       'id',
  60       'project',
  61       'parent_directory',
  62       'basename',
  63       'ordinal',
  64       'filename',
  65       ]
  66
  67   def __init__(self, id, project, parent_directory, basename):
  68     self.id = id
  69     self.project = project
  70     self.parent_directory = parent_directory
  71     self.basename = basename
  72
  73     # The filename used to be computed on demand, but it turned out to
  74     # be a hot path through the code in some cases.  It's used by
  75     # SubtreeSymbolTransform and similar transforms, so it's called at
  76     # least:
  77     #
  78     #   (num_files * num_symbols_per_file * num_subtree_symbol_transforms)
  79     #
  80     # times.  On a large repository with several subtree symbol
  81     # transforms, that can exceed 100,000,000 calls.  And
  82     # _calculate_filename() is quite complex, so doing that every time
  83     # could add about 10 minutes to the cvs2svn runtime.
  84     #
  85     # So now we precalculate this and just return it.
  86     self.filename = os.path.normpath(self._calculate_filename())
  87
  88   def __getstate__(self):
  89     """This method must only be called after ordinal has been set."""
  90
  91     return (
  92         self.id, self.project.id,
  93         self.parent_directory, self.basename,
  94         self.ordinal,
  95         )
  96
  97   def __setstate__(self, state):
  98     (
  99         self.id, project_id,
 100         self.parent_directory, self.basename,
 101         self.ordinal,
 102         ) = state
 103     self.project = Ctx()._projects[project_id]
 104     self.filename = os.path.normpath(self._calculate_filename())
 105
 106   def get_ancestry(self):
 107     """Return a list of the CVSPaths leading from the root path to SELF.
 108
 109     Return the CVSPaths in a list, starting with
 110     self.project.get_root_cvs_directory() and ending with self."""
 111
 112     ancestry = []
 113     p = self
 114     while p is not None:
 115       ancestry.append(p)
 116       p = p.parent_directory
 117
 118     ancestry.reverse()
 119     return ancestry
 120
 121   def get_cvs_path(self):
 122     """Return the canonical path within the Project.
 123
 124     The canonical path:
 125
 126     - Uses forward slashes
 127
 128     - Doesn't include ',v' for files
 129
 130     - This doesn't include the 'Attic' segment of the path unless the
 131       file is to be left in an Attic directory in the SVN repository;
 132       i.e., if a filename exists in and out of Attic and the
 133       --retain-conflicting-attic-files option was specified.
 134
 135     """
 136
 137     return path_join(*[p.basename for p in self.get_ancestry()[1:]])
 138
 139   cvs_path = property(get_cvs_path)
 140
 141   def _get_dir_components(self):
 142     """Return a list containing the components of the path leading to SELF.
 143
 144     The return value contains the base names of all of the parent
 145     directories (except for the root directory) and SELF."""
 146
 147     return [p.basename for p in self.get_ancestry()[1:]]
 148
 149   def __eq__(a, b):
 150     """Compare two CVSPath instances for equality.
 151
 152     This method is supplied to avoid using __cmp__() for comparing for
 153     equality."""
 154
 155     return a is b
 156
 157   def sort_key(self):
 158     """Return the key that should be used for sorting CVSPath instances.
 159
 160     This is a relatively expensive computation, so it is only used
 161     once, the the results are used to set the ordinal member."""
 162
 163     return (
 164         # Sort first by project:
 165         self.project,
 166         # Then by directory components:
 167         self._get_dir_components(),
 168         )
 169
 170   def __cmp__(a, b):
 171     """This method must only be called after ordinal has been set."""
 172
 173     return cmp(a.ordinal, b.ordinal)
 174
 175
 176 class CVSDirectory(CVSPath):
 177   """Represent a CVS directory.
 178
 179   Members:
 180
 181     id -- (int or None) unique id for this file.  If None, a new id is
 182         generated.
 183
 184     project -- (Project) the project containing this file.
 185
 186     parent_directory -- (CVSDirectory or None) the CVSDirectory
 187         containing this CVSDirectory.
 188
 189     basename -- (string) the base name of this CVSDirectory (no ',v').
 190
 191     ordinal -- (int) the order that this instance should be sorted
 192         relative to other CVSPath instances.  See CVSPath.ordinal.
 193
 194     empty_subdirectory_ids -- (list of int) a list of the ids of any
 195         direct subdirectories that are empty.  (An empty directory is
 196         defined to be a directory that doesn't contain any RCS files
 197         or non-empty subdirectories.
 198
 199   """
 200
 201   __slots__ = ['empty_subdirectory_ids']
 202
 203   def __init__(self, id, project, parent_directory, basename):
 204     """Initialize a new CVSDirectory object."""
 205
 206     CVSPath.__init__(self, id, project, parent_directory, basename)
 207
 208     # This member is filled in by CollectData.close():
 209     self.empty_subdirectory_ids = []
 210
 211   def _calculate_filename(self):
 212     """Return the filesystem path to this CVSPath in the CVS repository."""
 213
 214     if self.parent_directory is None:
 215       return self.project.project_cvs_repos_path
 216     else:
 217       return os.path.join(
 218           self.parent_directory.filename, self.basename
 219           )
 220
 221   def __getstate__(self):
 222     return (
 223         CVSPath.__getstate__(self),
 224         self.empty_subdirectory_ids,
 225         )
 226
 227   def __setstate__(self, state):
 228     (
 229         cvs_path_state,
 230         self.empty_subdirectory_ids,
 231         ) = state
 232     CVSPath.__setstate__(self, cvs_path_state)
 233
 234   def __str__(self):
 235     """For convenience only.  The format is subject to change at any time."""
 236
 237     return self.cvs_path + '/'
 238
 239   def __repr__(self):
 240     return 'CVSDirectory<%x>(%r)' % (self.id, str(self),)
 241
 242
 243 class CVSFile(CVSPath):
 244   """Represent a CVS file.
 245
 246   Members:
 247
 248     id -- (int) unique id for this file.
 249
 250     project -- (Project) the project containing this file.
 251
 252     parent_directory -- (CVSDirectory) the CVSDirectory containing
 253         this CVSFile.
 254
 255     basename -- (string) the base name of this CVSFile (no ',v').
 256
 257     ordinal -- (int) the order that this instance should be sorted
 258         relative to other CVSPath instances.  See CVSPath.ordinal.
 259
 260     _in_attic -- (bool) True if RCS file is in an Attic subdirectory
 261         that is not considered the parent directory.  (If a file is
 262         in-and-out-of-attic and one copy is to be left in Attic after
 263         the conversion, then the Attic directory is that file's
 264         PARENT_DIRECTORY and _IN_ATTIC is False.)
 265
 266     executable -- (bool) True iff RCS file has executable bit set.
 267
 268     file_size -- (long) size of the RCS file in bytes.
 269
 270     mode -- (string or None) 'kkv', 'kb', etc., as read from the CVS
 271         file.
 272
 273     description -- (string or None) the file description as read from
 274         the RCS file.
 275
 276     properties -- (dict) file properties that are preserved across
 277         this history of this file.  Keys are strings; values are
 278         strings (indicating the property value) or None (indicating
 279         that the property should be left unset).  These properties can
 280         be overridden by CVSRevision.properties.  Different backends
 281         can use these properties for different purposes; for cvs2svn
 282         they become SVN versioned properties.  Properties whose names
 283         start with underscore are reserved for internal cvs2svn
 284         purposes.
 285
 286   PARENT_DIRECTORY might contain an 'Attic' component if it should be
 287   retained in the SVN repository; i.e., if the same filename exists out
 288   of Attic and the --retain-conflicting-attic-files option was specified.
 289
 290   """
 291
 292   __slots__ = [
 293       '_in_attic',
 294       'executable',
 295       'file_size',
 296       'mode',
 297       'description',
 298       'properties',
 299       ]
 300
 301   def __init__(
 302         self, id, project, parent_directory, basename, in_attic,
 303         executable, file_size, mode, description
 304         ):
 305     """Initialize a new CVSFile object."""
 306
 307     assert parent_directory is not None
 308
 309     # This member is needed by _calculate_filename(), which is called
 310     # by CVSPath.__init__().  So initialize it before calling
 311     # CVSPath.__init__().
 312     self._in_attic = in_attic
 313     CVSPath.__init__(self, id, project, parent_directory, basename)
 314
 315     self.executable = executable
 316     self.file_size = file_size
 317     self.mode = mode
 318     self.description = description
 319     self.properties = None
 320
 321   def determine_file_properties(self, file_property_setters):
 322     """Determine the properties for this file from FILE_PROPERTY_SETTERS.
 323
 324     This must only be called after SELF.mode and SELF.description have
 325     been set by CollectData."""
 326
 327     self.properties = {}
 328
 329     for file_property_setter in file_property_setters:
 330       file_property_setter.set_properties(self)
 331
 332   def _calculate_filename(self):
 333     """Return the filesystem path to this CVSPath in the CVS repository."""
 334
 335     if self._in_attic:
 336       return os.path.join(
 337           self.parent_directory.filename, 'Attic', self.basename + ',v'
 338           )
 339     else:
 340       return os.path.join(
 341           self.parent_directory.filename, self.basename + ',v'
 342           )
 343
 344   def __getstate__(self):
 345     return (
 346         CVSPath.__getstate__(self),
 347         self._in_attic, self.executable, self.file_size, self.mode,
 348         self.description, self.properties,
 349         )
 350
 351   def __setstate__(self, state):
 352     (
 353         cvs_path_state,
 354         self._in_attic, self.executable, self.file_size, self.mode,
 355         self.description, self.properties,
 356         ) = state
 357     CVSPath.__setstate__(self, cvs_path_state)
 358
 359   def __str__(self):
 360     """For convenience only.  The format is subject to change at any time."""
 361
 362     return self.cvs_path
 363
 364   def __repr__(self):
 365     return 'CVSFile<%x>(%r)' % (self.id, str(self),)
 366
 367