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_path_components(self, rcs=False):
 124     """Return the path components to this CVSPath.
 125
 126     Return the components of this CVSPath's path, relative to the
 127     project's project_cvs_repos_path, as a list of strings.  If rcs is
 128     True, return the components of the filesystem path to the RCS file
 129     corresponding to this CVSPath (i.e., including any 'Attic'
 130     component and trailing ',v'.  If rcs is False, return the
 131     components of the logical CVS path name (i.e., including 'Attic'
 132     only if the file is to be left in an Attic directory in the SVN
 133     repository and without trailing ',v')."""
 134
 135     raise NotImplementedError()
 136
 137   def get_cvs_path(self):
 138     """Return the canonical path within the Project.
 139
 140     The canonical path:
 141
 142     - Uses forward slashes
 143
 144     - Doesn't include ',v' for files
 145
 146     - This doesn't include the 'Attic' segment of the path unless the
 147       file is to be left in an Attic directory in the SVN repository;
 148       i.e., if a filename exists in and out of Attic and the
 149       --retain-conflicting-attic-files option was specified.
 150
 151     """
 152
 153     return path_join(*[p.rcs_basename for p in self.get_ancestry()[1:]])
 154
 155   cvs_path = property(get_cvs_path)
 156
 157   def _get_dir_components(self):
 158     """Return a list containing the components of the path leading to SELF.
 159
 160     The return value contains the base names of all of the parent
 161     directories (except for the root directory) and SELF."""
 162
 163     return [p.rcs_basename for p in self.get_ancestry()[1:]]
 164
 165   def _calculate_rcs_path(self):
 166     """Return the filesystem path in the CVS repo corresponding to SELF."""
 167
 168     return os.path.join(
 169         self.project.project_cvs_repos_path,
 170         *self.get_path_components(rcs=True)
 171         )
 172
 173   def __eq__(a, b):
 174     """Compare two CVSPath instances for equality.
 175
 176     This method is supplied to avoid using __cmp__() for comparing for
 177     equality."""
 178
 179     return a is b
 180
 181   def sort_key(self):
 182     """Return the key that should be used for sorting CVSPath instances.
 183
 184     This is a relatively expensive computation, so it is only used
 185     once, the the results are used to set the ordinal member."""
 186
 187     return (
 188         # Sort first by project:
 189         self.project,
 190         # Then by directory components:
 191         self._get_dir_components(),
 192         )
 193
 194   def __cmp__(a, b):
 195     """This method must only be called after ordinal has been set."""
 196
 197     return cmp(a.ordinal, b.ordinal)
 198
 199
 200 class CVSDirectory(CVSPath):
 201   """Represent a CVS directory.
 202
 203   Members:
 204
 205     id -- (int or None) unique id for this file.  If None, a new id is
 206         generated.
 207
 208     project -- (Project) the project containing this file.
 209
 210     parent_directory -- (CVSDirectory or None) the CVSDirectory
 211         containing this CVSDirectory.
 212
 213     rcs_basename -- (string) the base name of the filename path in the
 214         CVS repository corresponding to this CVSDirectory.  The
 215         rcs_basename of the root directory of a project is ''.
 216
 217     ordinal -- (int) the order that this instance should be sorted
 218         relative to other CVSPath instances.  See CVSPath.ordinal.
 219
 220     empty_subdirectory_ids -- (list of int) a list of the ids of any
 221         direct subdirectories that are empty.  (An empty directory is
 222         defined to be a directory that doesn't contain any RCS files
 223         or non-empty subdirectories.
 224
 225   """
 226
 227   __slots__ = ['empty_subdirectory_ids']
 228
 229   def __init__(self, id, project, parent_directory, rcs_basename):
 230     """Initialize a new CVSDirectory object."""
 231
 232     CVSPath.__init__(self, id, project, parent_directory, rcs_basename)
 233
 234     # This member is filled in by CollectData.close():
 235     self.empty_subdirectory_ids = []
 236
 237   def get_path_components(self, rcs=False):
 238     components = []
 239     p = self
 240     while p.parent_directory is not None:
 241       components.append(p.rcs_basename)
 242       p = p.parent_directory
 243
 244     components.reverse()
 245     return components
 246
 247   def __getstate__(self):
 248     return (
 249         CVSPath.__getstate__(self),
 250         self.empty_subdirectory_ids,
 251         )
 252
 253   def __setstate__(self, state):
 254     (
 255         cvs_path_state,
 256         self.empty_subdirectory_ids,
 257         ) = state
 258     CVSPath.__setstate__(self, cvs_path_state)
 259
 260   def __str__(self):
 261     """For convenience only.  The format is subject to change at any time."""
 262
 263     return self.cvs_path + '/'
 264
 265   def __repr__(self):
 266     return 'CVSDirectory<%x>(%r)' % (self.id, str(self),)
 267
 268
 269 class CVSFile(CVSPath):
 270   """Represent a CVS file.
 271
 272   Members:
 273
 274     id -- (int) unique id for this file.
 275
 276     project -- (Project) the project containing this file.
 277
 278     parent_directory -- (CVSDirectory) the CVSDirectory containing
 279         this CVSFile.
 280
 281     rcs_basename -- (string) the base name of the RCS file in the CVS
 282         repository corresponding to this CVSPath (but with the ',v'
 283         removed).
 284
 285     ordinal -- (int) the order that this instance should be sorted
 286         relative to other CVSPath instances.  See CVSPath.ordinal.
 287
 288     _in_attic -- (bool) True if RCS file is in an Attic subdirectory
 289         that is not considered the parent directory.  (If a file is
 290         in-and-out-of-attic and one copy is to be left in Attic after
 291         the conversion, then the Attic directory is that file's
 292         PARENT_DIRECTORY and _IN_ATTIC is False.)
 293
 294     executable -- (bool) True iff RCS file has executable bit set.
 295
 296     file_size -- (long) size of the RCS file in bytes.
 297
 298     mode -- (string or None) 'kv', 'b', etc., as read from the CVS
 299         file.
 300
 301     description -- (string or None) the file description as read from
 302         the RCS file.
 303
 304     properties -- (dict) file properties that are preserved across
 305         this history of this file.  Keys are strings; values are
 306         strings (indicating the property value) or None (indicating
 307         that the property should be left unset).  These properties can
 308         be overridden by CVSRevision.properties.  Different backends
 309         can use these properties for different purposes; for cvs2svn
 310         they become SVN versioned properties.  Properties whose names
 311         start with underscore are reserved for internal cvs2svn
 312         purposes.
 313
 314   PARENT_DIRECTORY might contain an 'Attic' component if it should be
 315   retained in the SVN repository; i.e., if the same filename exists
 316   out of Attic and the --retain-conflicting-attic-files option was
 317   specified.
 318
 319   """
 320
 321   __slots__ = [
 322       '_in_attic',
 323       'executable',
 324       'file_size',
 325       'mode',
 326       'description',
 327       'properties',
 328       ]
 329
 330   def __init__(
 331         self, id, project, parent_directory, rcs_basename, in_attic,
 332         executable, file_size, mode, description
 333         ):
 334     """Initialize a new CVSFile object."""
 335
 336     assert parent_directory is not None
 337
 338     # This member is needed by _calculate_rcs_path(), which is
 339     # called by CVSPath.__init__().  So initialize it before calling
 340     # CVSPath.__init__().
 341     self._in_attic = in_attic
 342     CVSPath.__init__(self, id, project, parent_directory, rcs_basename)
 343
 344     self.executable = executable
 345     self.file_size = file_size
 346     self.mode = mode
 347     self.description = description
 348     self.properties = None
 349
 350   def determine_file_properties(self, file_property_setters):
 351     """Determine the properties for this file from FILE_PROPERTY_SETTERS.
 352
 353     This must only be called after SELF.mode and SELF.description have
 354     been set by CollectData."""
 355
 356     self.properties = {}
 357
 358     for file_property_setter in file_property_setters:
 359       file_property_setter.set_properties(self)
 360
 361   def get_path_components(self, rcs=False):
 362     components = self.parent_directory.get_path_components(rcs=rcs)
 363     if rcs:
 364       if self._in_attic:
 365         components.append('Attic')
 366       components.append(self.rcs_basename + ',v')
 367     else:
 368       components.append(self.rcs_basename)
 369     return components
 370
 371   def __getstate__(self):
 372     return (
 373         CVSPath.__getstate__(self),
 374         self._in_attic, self.executable, self.file_size, self.mode,
 375         self.description, self.properties,
 376         )
 377
 378   def __setstate__(self, state):
 379     (
 380         cvs_path_state,
 381         self._in_attic, self.executable, self.file_size, self.mode,
 382         self.description, self.properties,
 383         ) = state
 384     CVSPath.__setstate__(self, cvs_path_state)
 385
 386   def __str__(self):
 387     """For convenience only.  The format is subject to change at any time."""
 388
 389     return self.cvs_path
 390
 391   def __repr__(self):
 392     return 'CVSFile<%x>(%r)' % (self.id, str(self),)
 393
 394