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     rcs_basename -- (string) the base name of the filename path in the
  42         CVS repository corresponding to this CVSPath (but with ',v'
  43         removed for CVSFiles).  The rcs_basename of the root directory
  44         of a project is ''.
  45
  46     rcs_path -- (string) the filesystem path to this CVSPath in the
  47         CVS repository.  This is in native format, and already
  48         normalised the way os.path.normpath() normalises paths.  It
  49         starts with the repository path passed to
  50         run_options.add_project() in the options.py file.
  51
  52     ordinal -- (int) the order that this instance should be sorted
  53         relative to other CVSPath instances.  This member is set based
  54         on the ordering imposed by sort_key() by CVSPathDatabase after
  55         all CVSFiles have been processed.  Comparisons of CVSPath
  56         using __cmp__() simply compare the ordinals.
  57
  58   """
  59
  60   __slots__ = [
  61       'id',
  62       'project',
  63       'parent_directory',
  64       'rcs_basename',
  65       'ordinal',
  66       'rcs_path',
  67       ]
  68
  69   def __init__(self, id, project, parent_directory, rcs_basename):
  70     self.id = id
  71     self.project = project
  72     self.parent_directory = parent_directory
  73     self.rcs_basename = rcs_basename
  74
  75     # The rcs_path used to be computed on demand, but it turned out to
  76     # be a hot path through the code in some cases.  It's used by
  77     # SubtreeSymbolTransform and similar transforms, so it's called at
  78     # least:
  79     #
  80     #   (num_files * num_symbols_per_file * num_subtree_symbol_transforms)
  81     #
  82     # times.  On a large repository with several subtree symbol
  83     # transforms, that can exceed 100,000,000 calls.  And
  84     # _calculate_rcs_path() is quite complex, so doing that every time
  85     # could add about 10 minutes to the cvs2svn runtime.
  86     #
  87     # So now we precalculate this and just return it.
  88     self.rcs_path = os.path.normpath(self._calculate_rcs_path())
  89
  90   def __getstate__(self):
  91     """This method must only be called after ordinal has been set."""
  92
  93     return (
  94         self.id, self.project.id,
  95         self.parent_directory, self.rcs_basename,
  96         self.ordinal,
  97         )
  98
  99   def __setstate__(self, state):
 100     (
 101         self.id, project_id,
 102         self.parent_directory, self.rcs_basename,
 103         self.ordinal,
 104         ) = state
 105     self.project = Ctx()._projects[project_id]
 106     self.rcs_path = os.path.normpath(self._calculate_rcs_path())
 107
 108   def get_ancestry(self):
 109     """Return a list of the CVSPaths leading from the root path to SELF.
 110
 111     Return the CVSPaths in a list, starting with
 112     self.project.get_root_cvs_directory() and ending with self."""
 113
 114     ancestry = []
 115     p = self
 116     while p is not None:
 117       ancestry.append(p)
 118       p = p.parent_directory
 119
 120     ancestry.reverse()
 121     return ancestry
 122
 123   def get_cvs_path(self):
 124     """Return the canonical path within the Project.
 125
 126     The canonical path:
 127
 128     - Uses forward slashes
 129
 130     - Doesn't include ',v' for files
 131
 132     - This doesn't include the 'Attic' segment of the path unless the
 133       file is to be left in an Attic directory in the SVN repository;
 134       i.e., if a filename exists in and out of Attic and the
 135       --retain-conflicting-attic-files option was specified.
 136
 137     """
 138
 139     return path_join(*[p.rcs_basename for p in self.get_ancestry()[1:]])
 140
 141   cvs_path = property(get_cvs_path)
 142
 143   def _get_dir_components(self):
 144     """Return a list containing the components of the path leading to SELF.
 145
 146     The return value contains the base names of all of the parent
 147     directories (except for the root directory) and SELF."""
 148
 149     return [p.rcs_basename for p in self.get_ancestry()[1:]]
 150
 151   def __eq__(a, b):
 152     """Compare two CVSPath instances for equality.
 153
 154     This method is supplied to avoid using __cmp__() for comparing for
 155     equality."""
 156
 157     return a is b
 158
 159   def sort_key(self):
 160     """Return the key that should be used for sorting CVSPath instances.
 161
 162     This is a relatively expensive computation, so it is only used
 163     once, the the results are used to set the ordinal member."""
 164
 165     return (
 166         # Sort first by project:
 167         self.project,
 168         # Then by directory components:
 169         self._get_dir_components(),
 170         )
 171
 172   def __cmp__(a, b):
 173     """This method must only be called after ordinal has been set."""
 174
 175     return cmp(a.ordinal, b.ordinal)
 176
 177
 178 class CVSDirectory(CVSPath):
 179   """Represent a CVS directory.
 180
 181   Members:
 182
 183     id -- (int or None) unique id for this file.  If None, a new id is
 184         generated.
 185
 186     project -- (Project) the project containing this file.
 187
 188     parent_directory -- (CVSDirectory or None) the CVSDirectory
 189         containing this CVSDirectory.
 190
 191     rcs_basename -- (string) the base name of the filename path in the
 192         CVS repository corresponding to this CVSDirectory.  The
 193         rcs_basename of the root directory of a project is ''.
 194
 195     ordinal -- (int) the order that this instance should be sorted
 196         relative to other CVSPath instances.  See CVSPath.ordinal.
 197
 198     empty_subdirectory_ids -- (list of int) a list of the ids of any
 199         direct subdirectories that are empty.  (An empty directory is
 200         defined to be a directory that doesn't contain any RCS files
 201         or non-empty subdirectories.
 202
 203   """
 204
 205   __slots__ = ['empty_subdirectory_ids']
 206
 207   def __init__(self, id, project, parent_directory, rcs_basename):
 208     """Initialize a new CVSDirectory object."""
 209
 210     CVSPath.__init__(self, id, project, parent_directory, rcs_basename)
 211
 212     # This member is filled in by CollectData.close():
 213     self.empty_subdirectory_ids = []
 214
 215   def _calculate_rcs_path(self):
 216     """Return the filesystem path in the CVS repo corresponding to SELF."""
 217
 218     if self.parent_directory is None:
 219       return self.project.project_cvs_repos_path
 220     else:
 221       return os.path.join(
 222           self.parent_directory.rcs_path, self.rcs_basename
 223           )
 224
 225   def __getstate__(self):
 226     return (
 227         CVSPath.__getstate__(self),
 228         self.empty_subdirectory_ids,
 229         )
 230
 231   def __setstate__(self, state):
 232     (
 233         cvs_path_state,
 234         self.empty_subdirectory_ids,
 235         ) = state
 236     CVSPath.__setstate__(self, cvs_path_state)
 237
 238   def __str__(self):
 239     """For convenience only.  The format is subject to change at any time."""
 240
 241     return self.cvs_path + '/'
 242
 243   def __repr__(self):
 244     return 'CVSDirectory<%x>(%r)' % (self.id, str(self),)
 245
 246
 247 class CVSFile(CVSPath):
 248   """Represent a CVS file.
 249
 250   Members:
 251
 252     id -- (int) unique id for this file.
 253
 254     project -- (Project) the project containing this file.
 255
 256     parent_directory -- (CVSDirectory) the CVSDirectory containing
 257         this CVSFile.
 258
 259     rcs_basename -- (string) the base name of the RCS file in the CVS
 260         repository corresponding to this CVSPath (but with the ',v'
 261         removed).
 262
 263     ordinal -- (int) the order that this instance should be sorted
 264         relative to other CVSPath instances.  See CVSPath.ordinal.
 265
 266     _in_attic -- (bool) True if RCS file is in an Attic subdirectory
 267         that is not considered the parent directory.  (If a file is
 268         in-and-out-of-attic and one copy is to be left in Attic after
 269         the conversion, then the Attic directory is that file's
 270         PARENT_DIRECTORY and _IN_ATTIC is False.)
 271
 272     executable -- (bool) True iff RCS file has executable bit set.
 273
 274     file_size -- (long) size of the RCS file in bytes.
 275
 276     mode -- (string or None) 'kv', 'b', etc., as read from the CVS
 277         file.
 278
 279     description -- (string or None) the file description as read from
 280         the RCS file.
 281
 282     properties -- (dict) file properties that are preserved across
 283         this history of this file.  Keys are strings; values are
 284         strings (indicating the property value) or None (indicating
 285         that the property should be left unset).  These properties can
 286         be overridden by CVSRevision.properties.  Different backends
 287         can use these properties for different purposes; for cvs2svn
 288         they become SVN versioned properties.  Properties whose names
 289         start with underscore are reserved for internal cvs2svn
 290         purposes.
 291
 292   PARENT_DIRECTORY might contain an 'Attic' component if it should be
 293   retained in the SVN repository; i.e., if the same filename exists
 294   out of Attic and the --retain-conflicting-attic-files option was
 295   specified.
 296
 297   """
 298
 299   __slots__ = [
 300       '_in_attic',
 301       'executable',
 302       'file_size',
 303       'mode',
 304       'description',
 305       'properties',
 306       ]
 307
 308   def __init__(
 309         self, id, project, parent_directory, rcs_basename, in_attic,
 310         executable, file_size, mode, description
 311         ):
 312     """Initialize a new CVSFile object."""
 313
 314     assert parent_directory is not None
 315
 316     # This member is needed by _calculate_rcs_path(), which is
 317     # called by CVSPath.__init__().  So initialize it before calling
 318     # CVSPath.__init__().
 319     self._in_attic = in_attic
 320     CVSPath.__init__(self, id, project, parent_directory, rcs_basename)
 321
 322     self.executable = executable
 323     self.file_size = file_size
 324     self.mode = mode
 325     self.description = description
 326     self.properties = None
 327
 328   def determine_file_properties(self, file_property_setters):
 329     """Determine the properties for this file from FILE_PROPERTY_SETTERS.
 330
 331     This must only be called after SELF.mode and SELF.description have
 332     been set by CollectData."""
 333
 334     self.properties = {}
 335
 336     for file_property_setter in file_property_setters:
 337       file_property_setter.set_properties(self)
 338
 339   def _calculate_rcs_path(self):
 340     """Return the filesystem path to the RCS file for this CVSFile."""
 341
 342     if self._in_attic:
 343       return os.path.join(
 344           self.parent_directory.rcs_path, 'Attic', self.rcs_basename + ',v'
 345           )
 346     else:
 347       return os.path.join(
 348           self.parent_directory.rcs_path, self.rcs_basename + ',v'
 349           )
 350
 351   def __getstate__(self):
 352     return (
 353         CVSPath.__getstate__(self),
 354         self._in_attic, self.executable, self.file_size, self.mode,
 355         self.description, self.properties,
 356         )
 357
 358   def __setstate__(self, state):
 359     (
 360         cvs_path_state,
 361         self._in_attic, self.executable, self.file_size, self.mode,
 362         self.description, self.properties,
 363         ) = state
 364     CVSPath.__setstate__(self, cvs_path_state)
 365
 366   def __str__(self):
 367     """For convenience only.  The format is subject to change at any time."""
 368
 369     return self.cvs_path
 370
 371   def __repr__(self):
 372     return 'CVSFile<%x>(%r)' % (self.id, str(self),)
 373
 374