|
|
1
|
+#
|
|
|
2
|
+# Copyright (C) 2018 Bloomberg LLC
|
|
|
3
|
+#
|
|
|
4
|
+# This program is free software; you can redistribute it and/or
|
|
|
5
|
+# modify it under the terms of the GNU Lesser General Public
|
|
|
6
|
+# License as published by the Free Software Foundation; either
|
|
|
7
|
+# version 2 of the License, or (at your option) any later version.
|
|
|
8
|
+#
|
|
|
9
|
+# This library is distributed in the hope that it will be useful,
|
|
|
10
|
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
11
|
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
|
|
12
|
+# Lesser General Public License for more details.
|
|
|
13
|
+#
|
|
|
14
|
+# You should have received a copy of the GNU Lesser General Public
|
|
|
15
|
+# License along with this library. If not, see <http://www.gnu.org/licenses/>.
|
|
|
16
|
+#
|
|
|
17
|
+# Authors:
|
|
|
18
|
+# Jim MacArthur <jim macarthur codethink co uk>
|
|
|
19
|
+
|
|
|
20
|
+"""
|
|
|
21
|
+CasBasedDirectory
|
|
|
22
|
+=========
|
|
|
23
|
+
|
|
|
24
|
+Implementation of the Directory class which backs onto a Merkle-tree based content
|
|
|
25
|
+addressable storage system.
|
|
|
26
|
+
|
|
|
27
|
+See also: :ref:`sandboxing`.
|
|
|
28
|
+"""
|
|
|
29
|
+
|
|
|
30
|
+from collections import OrderedDict
|
|
|
31
|
+
|
|
|
32
|
+import os
|
|
|
33
|
+import tempfile
|
|
|
34
|
+import stat
|
|
|
35
|
+
|
|
|
36
|
+from .._protos.build.bazel.remote.execution.v2 import remote_execution_pb2
|
|
|
37
|
+from .._exceptions import BstError
|
|
|
38
|
+from .directory import Directory, VirtualDirectoryError
|
|
|
39
|
+from ._filebaseddirectory import FileBasedDirectory
|
|
|
40
|
+from ..utils import FileListResult, safe_copy, list_relative_paths
|
|
|
41
|
+from .._artifactcache.cascache import CASCache
|
|
|
42
|
+
|
|
|
43
|
+
|
|
|
44
|
+class IndexEntry():
|
|
|
45
|
+ """ Used in our index of names to objects to store the 'modified' flag
|
|
|
46
|
+ for directory entries. Because we need both the remote_execution_pb2 object
|
|
|
47
|
+ and our own Directory object for directory entries, we store both. For files
|
|
|
48
|
+ and symlinks, only pb2_object is used. """
|
|
|
49
|
+ def __init__(self, pb2_object, buildstream_object=None, modified=False):
|
|
|
50
|
+ self.pb2_object = pb2_object
|
|
|
51
|
+ self.buildstream_object = buildstream_object
|
|
|
52
|
+ self.modified = modified
|
|
|
53
|
+
|
|
|
54
|
+
|
|
|
55
|
+# CasBasedDirectory intentionally doesn't call its superclass constuctor,
|
|
|
56
|
+# which is meant to be unimplemented.
|
|
|
57
|
+# pylint: disable=super-init-not-called
|
|
|
58
|
+
|
|
|
59
|
+class CasBasedDirectory(Directory):
|
|
|
60
|
+ """
|
|
|
61
|
+ CAS-based directories can have two names; one is a 'common name' which has no effect
|
|
|
62
|
+ on functionality, and the 'filename'. If a CasBasedDirectory has a parent, then 'filename'
|
|
|
63
|
+ must be the name of an entry in the parent directory's index which points to this object.
|
|
|
64
|
+ This is used to inform a parent directory that it must update the given hash for this
|
|
|
65
|
+ object when this object changes.
|
|
|
66
|
+
|
|
|
67
|
+ Typically a top-level CasBasedDirectory will have a common_name and no filename, and
|
|
|
68
|
+ subdirectories wil have a filename and no common_name. common_name can used to identify
|
|
|
69
|
+ CasBasedDirectory objects in a log file, since they have no unique position in a file
|
|
|
70
|
+ system.
|
|
|
71
|
+ """
|
|
|
72
|
+
|
|
|
73
|
+ # Two constants which define the separators used by the remote execution API.
|
|
|
74
|
+ _pb2_path_sep = "/"
|
|
|
75
|
+ _pb2_absolute_path_prefix = "/"
|
|
|
76
|
+
|
|
|
77
|
+ def __init__(self, context, ref=None, parent=None, common_name="untitled", filename=None):
|
|
|
78
|
+ self.context = context
|
|
|
79
|
+ self.cas_directory = os.path.join(context.artifactdir, 'cas')
|
|
|
80
|
+ self.filename = filename
|
|
|
81
|
+ self.common_name = common_name
|
|
|
82
|
+ self.pb2_directory = remote_execution_pb2.Directory()
|
|
|
83
|
+ self.cas_cache = CASCache(context)
|
|
|
84
|
+ if ref:
|
|
|
85
|
+ with open(self.cas_cache.objpath(ref), 'rb') as f:
|
|
|
86
|
+ self.pb2_directory.ParseFromString(f.read())
|
|
|
87
|
+
|
|
|
88
|
+ self.ref = ref
|
|
|
89
|
+ self.index = OrderedDict()
|
|
|
90
|
+ self.parent = parent
|
|
|
91
|
+ self._directory_read = False
|
|
|
92
|
+ self._populate_index()
|
|
|
93
|
+
|
|
|
94
|
+ def _populate_index(self):
|
|
|
95
|
+ if self._directory_read:
|
|
|
96
|
+ return
|
|
|
97
|
+ for entry in self.pb2_directory.directories:
|
|
|
98
|
+ buildStreamDirectory = CasBasedDirectory(self.context, ref=entry.digest,
|
|
|
99
|
+ parent=self, filename=entry.name)
|
|
|
100
|
+ self.index[entry.name] = IndexEntry(entry, buildstream_object=buildStreamDirectory)
|
|
|
101
|
+ for entry in self.pb2_directory.files:
|
|
|
102
|
+ self.index[entry.name] = IndexEntry(entry)
|
|
|
103
|
+ for entry in self.pb2_directory.symlinks:
|
|
|
104
|
+ self.index[entry.name] = IndexEntry(entry)
|
|
|
105
|
+ self._directory_read = True
|
|
|
106
|
+
|
|
|
107
|
+ def _recalculate_recursing_up(self, caller=None):
|
|
|
108
|
+ """Recalcuate the hash for this directory and store the results in
|
|
|
109
|
+ the cache. If this directory has a parent, tell it to
|
|
|
110
|
+ recalculate (since changing this directory changes an entry in
|
|
|
111
|
+ the parent).
|
|
|
112
|
+
|
|
|
113
|
+ """
|
|
|
114
|
+ self.ref = self.cas_cache.add_object(buffer=self.pb2_directory.SerializeToString())
|
|
|
115
|
+ if caller:
|
|
|
116
|
+ old_dir = self._find_pb2_entry(caller.filename)
|
|
|
117
|
+ self.cas_cache.add_object(digest=old_dir.digest, buffer=caller.pb2_directory.SerializeToString())
|
|
|
118
|
+ if self.parent:
|
|
|
119
|
+ self.parent._recalculate_recursing_up(self)
|
|
|
120
|
+
|
|
|
121
|
+ def _recalculate_recursing_down(self, parent=None):
|
|
|
122
|
+ """Recalcuate the hash for this directory and any
|
|
|
123
|
+ subdirectories. Hashes for subdirectories should be calculated
|
|
|
124
|
+ and stored after a significant operation (e.g. an
|
|
|
125
|
+ import_files() call) but not after adding each file, as that
|
|
|
126
|
+ is extremely wasteful.
|
|
|
127
|
+
|
|
|
128
|
+ """
|
|
|
129
|
+ for entry in self.pb2_directory.directories:
|
|
|
130
|
+ self.index[entry.name].buildstream_object._recalculate_recursing_down(entry)
|
|
|
131
|
+
|
|
|
132
|
+ if parent:
|
|
|
133
|
+ self.ref = self.cas_cache.add_object(digest=parent.digest, buffer=self.pb2_directory.SerializeToString())
|
|
|
134
|
+ else:
|
|
|
135
|
+ self.ref = self.cas_cache.add_object(buffer=self.pb2_directory.SerializeToString())
|
|
|
136
|
+ # We don't need to do anything more than that; files were already added ealier, and symlinks are
|
|
|
137
|
+ # part of the directory structure.
|
|
|
138
|
+
|
|
|
139
|
+ def _find_pb2_entry(self, name):
|
|
|
140
|
+ if name in self.index:
|
|
|
141
|
+ return self.index[name].pb2_object
|
|
|
142
|
+ return None
|
|
|
143
|
+
|
|
|
144
|
+ def _add_directory(self, name):
|
|
|
145
|
+ if name in self.index:
|
|
|
146
|
+ newdir = self.index[name].buildstream_object
|
|
|
147
|
+ if not isinstance(newdir, CasBasedDirectory):
|
|
|
148
|
+ # TODO: This may not be an actual error; it may actually overwrite it
|
|
|
149
|
+ raise VirtualDirectoryError("New directory {} in {} would overwrite existing non-directory of type {}"
|
|
|
150
|
+ .format(name, str(self), type(newdir)))
|
|
|
151
|
+ dirnode = self._find_pb2_entry(name)
|
|
|
152
|
+ else:
|
|
|
153
|
+ newdir = CasBasedDirectory(self.context, parent=self, filename=name)
|
|
|
154
|
+ dirnode = self.pb2_directory.directories.add()
|
|
|
155
|
+
|
|
|
156
|
+ dirnode.name = name
|
|
|
157
|
+
|
|
|
158
|
+ # Calculate the hash for an empty directory
|
|
|
159
|
+ new_directory = remote_execution_pb2.Directory()
|
|
|
160
|
+ self.cas_cache.add_object(digest=dirnode.digest, buffer=new_directory.SerializeToString())
|
|
|
161
|
+ self.index[name] = IndexEntry(dirnode, buildstream_object=newdir)
|
|
|
162
|
+ return newdir
|
|
|
163
|
+
|
|
|
164
|
+ def _add_new_file(self, basename, filename):
|
|
|
165
|
+ filenode = self.pb2_directory.files.add()
|
|
|
166
|
+ filenode.name = filename
|
|
|
167
|
+ self.cas_cache.add_object(digest=filenode.digest, path=os.path.join(basename, filename))
|
|
|
168
|
+ is_executable = os.access(os.path.join(basename, filename), os.X_OK)
|
|
|
169
|
+ filenode.is_executable = is_executable
|
|
|
170
|
+ self.index[filename] = IndexEntry(filenode, modified=(filename in self.index))
|
|
|
171
|
+
|
|
|
172
|
+ def _add_new_link(self, basename, filename):
|
|
|
173
|
+ existing_link = self._find_pb2_entry(filename)
|
|
|
174
|
+ if existing_link:
|
|
|
175
|
+ symlinknode = existing_link
|
|
|
176
|
+ else:
|
|
|
177
|
+ symlinknode = self.pb2_directory.symlinks.add()
|
|
|
178
|
+ symlinknode.name = filename
|
|
|
179
|
+ # A symlink node has no digest.
|
|
|
180
|
+ symlinknode.target = os.readlink(os.path.join(basename, filename))
|
|
|
181
|
+ self.index[filename] = IndexEntry(symlinknode, modified=(existing_link is not None))
|
|
|
182
|
+
|
|
|
183
|
+ def delete_entry(self, name):
|
|
|
184
|
+ for collection in [self.pb2_directory.files, self.pb2_directory.symlinks, self.pb2_directory.directories]:
|
|
|
185
|
+ if name in collection:
|
|
|
186
|
+ collection.remove(name)
|
|
|
187
|
+ if name in self.index:
|
|
|
188
|
+ del self.index[name]
|
|
|
189
|
+
|
|
|
190
|
+ def descend(self, subdirectory_spec, create=False):
|
|
|
191
|
+ """Descend one or more levels of directory hierarchy and return a new
|
|
|
192
|
+ Directory object for that directory.
|
|
|
193
|
+
|
|
|
194
|
+ Arguments:
|
|
|
195
|
+ * subdirectory_spec (list of strings): A list of strings which are all directory
|
|
|
196
|
+ names.
|
|
|
197
|
+ * create (boolean): If this is true, the directories will be created if
|
|
|
198
|
+ they don't already exist.
|
|
|
199
|
+
|
|
|
200
|
+ Note: At the moment, creating a directory by descending does
|
|
|
201
|
+ not update this object in the CAS cache. However, performing
|
|
|
202
|
+ an import_files() into a subdirectory of any depth obtained by
|
|
|
203
|
+ descending from this object *will* cause this directory to be
|
|
|
204
|
+ updated and stored.
|
|
|
205
|
+
|
|
|
206
|
+ """
|
|
|
207
|
+
|
|
|
208
|
+ # It's very common to send a directory name instead of a list and this causes
|
|
|
209
|
+ # bizarre errors, so check for it here
|
|
|
210
|
+ if not isinstance(subdirectory_spec, list):
|
|
|
211
|
+ subdirectory_spec = [subdirectory_spec]
|
|
|
212
|
+
|
|
|
213
|
+ # Because of the way split works, it's common to get a list which begins with
|
|
|
214
|
+ # an empty string. Detect these and remove them.
|
|
|
215
|
+ while subdirectory_spec and subdirectory_spec[0] == "":
|
|
|
216
|
+ subdirectory_spec.pop(0)
|
|
|
217
|
+
|
|
|
218
|
+ # Descending into [] returns the same directory.
|
|
|
219
|
+ if not subdirectory_spec:
|
|
|
220
|
+ return self
|
|
|
221
|
+
|
|
|
222
|
+ if subdirectory_spec[0] in self.index:
|
|
|
223
|
+ entry = self.index[subdirectory_spec[0]].buildstream_object
|
|
|
224
|
+ if isinstance(entry, CasBasedDirectory):
|
|
|
225
|
+ return entry.descend(subdirectory_spec[1:], create)
|
|
|
226
|
+ else:
|
|
|
227
|
+ error = "Cannot descend into {}, which is a '{}' in the directory {}"
|
|
|
228
|
+ raise VirtualDirectoryError(error.format(subdirectory_spec[0],
|
|
|
229
|
+ type(entry).__name__,
|
|
|
230
|
+ self))
|
|
|
231
|
+ else:
|
|
|
232
|
+ if create:
|
|
|
233
|
+ newdir = self._add_directory(subdirectory_spec[0])
|
|
|
234
|
+ return newdir.descend(subdirectory_spec[1:], create)
|
|
|
235
|
+ else:
|
|
|
236
|
+ error = "No entry called '{}' found in {}. There are directories called {}."
|
|
|
237
|
+ directory_list = ",".join([entry.name for entry in self.pb2_directory.directories])
|
|
|
238
|
+ raise VirtualDirectoryError(error.format(subdirectory_spec[0], str(self),
|
|
|
239
|
+ directory_list))
|
|
|
240
|
+ return None
|
|
|
241
|
+
|
|
|
242
|
+ def find_root(self):
|
|
|
243
|
+ """ Finds the root of this directory tree by following 'parent' until there is
|
|
|
244
|
+ no parent. """
|
|
|
245
|
+ if self.parent:
|
|
|
246
|
+ return self.parent.find_root()
|
|
|
247
|
+ else:
|
|
|
248
|
+ return self
|
|
|
249
|
+
|
|
|
250
|
+ def _resolve_symlink_or_directory(self, name):
|
|
|
251
|
+ """Used only by _import_files_from_directory. Tries to resolve a
|
|
|
252
|
+ directory name or symlink name. 'name' must be an entry in this
|
|
|
253
|
+ directory. It must be a single symlink or directory name, not a path
|
|
|
254
|
+ separated by path separators. If it's an existing directory name, it
|
|
|
255
|
+ just returns the Directory object for that. If it's a symlink, it will
|
|
|
256
|
+ attempt to find the target of the symlink and return that as a
|
|
|
257
|
+ Directory object.
|
|
|
258
|
+
|
|
|
259
|
+ If a symlink target doesn't exist, it will attempt to create it
|
|
|
260
|
+ as a directory as long as it's within this directory tree.
|
|
|
261
|
+ """
|
|
|
262
|
+
|
|
|
263
|
+ if isinstance(self.index[name].buildstream_object, Directory):
|
|
|
264
|
+ return self.index[name].buildstream_object
|
|
|
265
|
+ # OK then, it's a symlink
|
|
|
266
|
+ symlink = self._find_pb2_entry(name)
|
|
|
267
|
+ absolute = symlink.target.startswith(CasBasedDirectory._pb2_absolute_path_prefix)
|
|
|
268
|
+ if absolute:
|
|
|
269
|
+ root = self.find_root()
|
|
|
270
|
+ else:
|
|
|
271
|
+ root = self
|
|
|
272
|
+ directory = root
|
|
|
273
|
+ components = symlink.target.split(CasBasedDirectory._pb2_path_sep)
|
|
|
274
|
+ for c in components:
|
|
|
275
|
+ if c == "..":
|
|
|
276
|
+ directory = directory.parent
|
|
|
277
|
+ else:
|
|
|
278
|
+ directory = directory.descend(c, create=True)
|
|
|
279
|
+ return directory
|
|
|
280
|
+
|
|
|
281
|
+ def _check_replacement(self, name, path_prefix, fileListResult):
|
|
|
282
|
+ """ Checks whether 'name' exists, and if so, whether we can overwrite it.
|
|
|
283
|
+ If we can, add the name to 'overwritten_files' and delete the existing entry.
|
|
|
284
|
+ Returns 'True' if the import should go ahead.
|
|
|
285
|
+ fileListResult.overwritten and fileListResult.ignore are updated depending
|
|
|
286
|
+ on the result. """
|
|
|
287
|
+ existing_entry = self._find_pb2_entry(name)
|
|
|
288
|
+ relative_pathname = os.path.join(path_prefix, name)
|
|
|
289
|
+ if existing_entry is None:
|
|
|
290
|
+ return True
|
|
|
291
|
+ if (isinstance(existing_entry,
|
|
|
292
|
+ (remote_execution_pb2.FileNode, remote_execution_pb2.SymlinkNode))):
|
|
|
293
|
+ self.delete_entry(name)
|
|
|
294
|
+ fileListResult.overwritten.append(relative_pathname)
|
|
|
295
|
+ return True
|
|
|
296
|
+ elif isinstance(existing_entry, remote_execution_pb2.DirectoryNode):
|
|
|
297
|
+ # If 'name' maps to a DirectoryNode, then there must be an entry in index
|
|
|
298
|
+ # pointing to another Directory.
|
|
|
299
|
+ if self.index[name].buildstream_object.is_empty():
|
|
|
300
|
+ self.delete_entry(name)
|
|
|
301
|
+ fileListResult.overwritten.append(relative_pathname)
|
|
|
302
|
+ return True
|
|
|
303
|
+ else:
|
|
|
304
|
+ # We can't overwrite a non-empty directory, so we just ignore it.
|
|
|
305
|
+ fileListResult.ignored.append(relative_pathname)
|
|
|
306
|
+ return False
|
|
|
307
|
+ raise VirtualDirectoryError("Entry '{}' is not a recognised file/link/directory and not None; it is {}"
|
|
|
308
|
+ .format(name, type(existing_entry)))
|
|
|
309
|
+
|
|
|
310
|
+ def _import_directory_recursively(self, directory_name, source_directory, remaining_path, path_prefix):
|
|
|
311
|
+ """ _import_directory_recursively and _import_files_from_directory will be called alternately
|
|
|
312
|
+ as a directory tree is descended. """
|
|
|
313
|
+ if directory_name in self.index:
|
|
|
314
|
+ subdir = self._resolve_symlink_or_directory(directory_name)
|
|
|
315
|
+ else:
|
|
|
316
|
+ subdir = self._add_directory(directory_name)
|
|
|
317
|
+ new_path_prefix = os.path.join(path_prefix, directory_name)
|
|
|
318
|
+ subdir_result = subdir._import_files_from_directory(os.path.join(source_directory, directory_name),
|
|
|
319
|
+ [os.path.sep.join(remaining_path)],
|
|
|
320
|
+ path_prefix=new_path_prefix)
|
|
|
321
|
+ return subdir_result
|
|
|
322
|
+
|
|
|
323
|
+ def _import_files_from_directory(self, source_directory, files, path_prefix=""):
|
|
|
324
|
+ """ Imports files from a traditional directory """
|
|
|
325
|
+ result = FileListResult()
|
|
|
326
|
+ for entry in sorted(files):
|
|
|
327
|
+ split_path = entry.split(os.path.sep)
|
|
|
328
|
+ # The actual file on the FS we're importing
|
|
|
329
|
+ import_file = os.path.join(source_directory, entry)
|
|
|
330
|
+ # The destination filename, relative to the root where the import started
|
|
|
331
|
+ relative_pathname = os.path.join(path_prefix, entry)
|
|
|
332
|
+ if len(split_path) > 1:
|
|
|
333
|
+ directory_name = split_path[0]
|
|
|
334
|
+ # Hand this off to the importer for that subdir. This will only do one file -
|
|
|
335
|
+ # a better way would be to hand off all the files in this subdir at once.
|
|
|
336
|
+ subdir_result = self._import_directory_recursively(directory_name, source_directory,
|
|
|
337
|
+ split_path[1:], path_prefix)
|
|
|
338
|
+ result.combine(subdir_result)
|
|
|
339
|
+ elif os.path.islink(import_file):
|
|
|
340
|
+ if self._check_replacement(entry, path_prefix, result):
|
|
|
341
|
+ self._add_new_link(source_directory, entry)
|
|
|
342
|
+ result.files_written.append(relative_pathname)
|
|
|
343
|
+ elif os.path.isdir(import_file):
|
|
|
344
|
+ # A plain directory which already exists isn't a problem; just ignore it.
|
|
|
345
|
+ if entry not in self.index:
|
|
|
346
|
+ self._add_directory(entry)
|
|
|
347
|
+ elif os.path.isfile(import_file):
|
|
|
348
|
+ if self._check_replacement(entry, path_prefix, result):
|
|
|
349
|
+ self._add_new_file(source_directory, entry)
|
|
|
350
|
+ result.files_written.append(relative_pathname)
|
|
|
351
|
+ return result
|
|
|
352
|
+
|
|
|
353
|
+ def import_files(self, external_pathspec, *, files=None,
|
|
|
354
|
+ report_written=True, update_utimes=False,
|
|
|
355
|
+ can_link=False):
|
|
|
356
|
+ """Imports some or all files from external_path into this directory.
|
|
|
357
|
+
|
|
|
358
|
+ Keyword arguments: external_pathspec: Either a string
|
|
|
359
|
+ containing a pathname, or a Directory object, to use as the
|
|
|
360
|
+ source.
|
|
|
361
|
+
|
|
|
362
|
+ files (list of strings): A list of all the files relative to
|
|
|
363
|
+ the external_pathspec to copy. If 'None' is supplied, all
|
|
|
364
|
+ files are copied.
|
|
|
365
|
+
|
|
|
366
|
+ report_written (bool): Return the full list of files
|
|
|
367
|
+ written. Defaults to true. If false, only a list of
|
|
|
368
|
+ overwritten files is returned.
|
|
|
369
|
+
|
|
|
370
|
+ update_utimes (bool): Currently ignored, since CAS does not store utimes.
|
|
|
371
|
+
|
|
|
372
|
+ can_link (bool): Ignored, since hard links do not have any meaning within CAS.
|
|
|
373
|
+ """
|
|
|
374
|
+ if isinstance(external_pathspec, FileBasedDirectory):
|
|
|
375
|
+ source_directory = external_pathspec.get_underlying_directory()
|
|
|
376
|
+ elif isinstance(external_pathspec, CasBasedDirectory):
|
|
|
377
|
+ # TODO: This transfers from one CAS to another via the
|
|
|
378
|
+ # filesystem, which is very inefficient. Alter this so it
|
|
|
379
|
+ # transfers refs across directly.
|
|
|
380
|
+ with tempfile.TemporaryDirectory(prefix="roundtrip") as tmpdir:
|
|
|
381
|
+ external_pathspec.export_files(tmpdir)
|
|
|
382
|
+ if files is None:
|
|
|
383
|
+ files = list_relative_paths(tmpdir)
|
|
|
384
|
+ result = self._import_files_from_directory(tmpdir, files=files)
|
|
|
385
|
+ return result
|
|
|
386
|
+ else:
|
|
|
387
|
+ source_directory = external_pathspec
|
|
|
388
|
+
|
|
|
389
|
+ if files is None:
|
|
|
390
|
+ files = list_relative_paths(source_directory)
|
|
|
391
|
+
|
|
|
392
|
+ # TODO: No notice is taken of report_written, update_utimes or can_link.
|
|
|
393
|
+ # Current behaviour is to fully populate the report, which is inefficient,
|
|
|
394
|
+ # but still correct.
|
|
|
395
|
+ result = self._import_files_from_directory(source_directory, files=files)
|
|
|
396
|
+
|
|
|
397
|
+ # We need to recalculate and store the hashes of all directories both
|
|
|
398
|
+ # up and down the tree; we have changed our directory by importing files
|
|
|
399
|
+ # which changes our hash and all our parents' hashes of us. The trees
|
|
|
400
|
+ # lower down need to be stored in the CAS as they are not automatically
|
|
|
401
|
+ # added during construction.
|
|
|
402
|
+ self._recalculate_recursing_down()
|
|
|
403
|
+ if self.parent:
|
|
|
404
|
+ self.parent._recalculate_recursing_up(self)
|
|
|
405
|
+ return result
|
|
|
406
|
+
|
|
|
407
|
+ def set_deterministic_mtime(self):
|
|
|
408
|
+ """ Sets a static modification time for all regular files in this directory.
|
|
|
409
|
+ Since we don't store any modification time, we don't need to do anything.
|
|
|
410
|
+ """
|
|
|
411
|
+ pass
|
|
|
412
|
+
|
|
|
413
|
+ def set_deterministic_user(self):
|
|
|
414
|
+ """ Sets all files in this directory to the current user's euid/egid.
|
|
|
415
|
+ We also don't store user data, so this can be ignored.
|
|
|
416
|
+ """
|
|
|
417
|
+ pass
|
|
|
418
|
+
|
|
|
419
|
+ def export_files(self, to_directory, *, can_link=False, can_destroy=False):
|
|
|
420
|
+ """Copies everything from this into to_directory, which must be the name
|
|
|
421
|
+ of a traditional filesystem directory.
|
|
|
422
|
+
|
|
|
423
|
+ Arguments:
|
|
|
424
|
+
|
|
|
425
|
+ to_directory (string): a path outside this directory object
|
|
|
426
|
+ where the contents will be copied to.
|
|
|
427
|
+
|
|
|
428
|
+ can_link (bool): Whether we can create hard links in to_directory
|
|
|
429
|
+ instead of copying.
|
|
|
430
|
+
|
|
|
431
|
+ can_destroy (bool): Whether we can destroy elements in this
|
|
|
432
|
+ directory to export them (e.g. by renaming them as the
|
|
|
433
|
+ target).
|
|
|
434
|
+
|
|
|
435
|
+ """
|
|
|
436
|
+
|
|
|
437
|
+ if not os.path.exists(to_directory):
|
|
|
438
|
+ os.mkdir(to_directory)
|
|
|
439
|
+
|
|
|
440
|
+ for entry in self.pb2_directory.directories:
|
|
|
441
|
+ if entry.name not in self.index:
|
|
|
442
|
+ raise VirtualDirectoryError("CasDir {} contained {} in directories but not in the index"
|
|
|
443
|
+ .format(str(self), entry.name))
|
|
|
444
|
+ if not self._directory_read:
|
|
|
445
|
+ raise VirtualDirectoryError("CasDir {} has not been indexed yet".format(str(self)))
|
|
|
446
|
+ dest_dir = os.path.join(to_directory, entry.name)
|
|
|
447
|
+ if not os.path.exists(dest_dir):
|
|
|
448
|
+ os.mkdir(dest_dir)
|
|
|
449
|
+ target = self.descend([entry.name])
|
|
|
450
|
+ target.export_files(dest_dir)
|
|
|
451
|
+ for entry in self.pb2_directory.files:
|
|
|
452
|
+ # Extract the entry to a single file
|
|
|
453
|
+ dest_name = os.path.join(to_directory, entry.name)
|
|
|
454
|
+ src_name = self.cas_cache.objpath(entry.digest)
|
|
|
455
|
+ safe_copy(src_name, dest_name)
|
|
|
456
|
+ if entry.is_executable:
|
|
|
457
|
+ os.chmod(dest_name, stat.S_IRUSR | stat.S_IWUSR | stat.S_IXUSR |
|
|
|
458
|
+ stat.S_IRGRP | stat.S_IXGRP |
|
|
|
459
|
+ stat.S_IROTH | stat.S_IXOTH)
|
|
|
460
|
+ for entry in self.pb2_directory.symlinks:
|
|
|
461
|
+ src_name = os.path.join(to_directory, entry.name)
|
|
|
462
|
+ target_name = entry.target
|
|
|
463
|
+ try:
|
|
|
464
|
+ os.symlink(target_name, src_name)
|
|
|
465
|
+ except FileExistsError as e:
|
|
|
466
|
+ raise BstError(("Cannot create a symlink named {} pointing to {}." +
|
|
|
467
|
+ " The original error was: {}").
|
|
|
468
|
+ format(src_name, entry.target, e))
|
|
|
469
|
+
|
|
|
470
|
+ def export_to_tar(self, tarfile, destination_dir, mtime=0):
|
|
|
471
|
+ raise NotImplementedError()
|
|
|
472
|
+
|
|
|
473
|
+ def mark_changed(self):
|
|
|
474
|
+ """ It should not be possible to externally modify a CAS-based
|
|
|
475
|
+ directory at the moment."""
|
|
|
476
|
+ raise NotImplementedError()
|
|
|
477
|
+
|
|
|
478
|
+ def is_empty(self):
|
|
|
479
|
+ """ Return true if this directory has no files, subdirectories or links in it.
|
|
|
480
|
+ """
|
|
|
481
|
+ return len(self.index) == 0
|
|
|
482
|
+
|
|
|
483
|
+ def mark_unmodified(self):
|
|
|
484
|
+ """ Marks all files in this directory (recursively) as unmodified.
|
|
|
485
|
+ """
|
|
|
486
|
+ # TODO: We don't actually mark our own directory unmodified
|
|
|
487
|
+ # here, because we can't get to the containing IndexEntry -
|
|
|
488
|
+ # just the objects we contain.
|
|
|
489
|
+ for i in self.index.values():
|
|
|
490
|
+ i.modified = False
|
|
|
491
|
+ if isinstance(i.buildstream_object, CasBasedDirectory):
|
|
|
492
|
+ i.buildstream_object.mark_unmodified()
|
|
|
493
|
+
|
|
|
494
|
+ def list_modified_paths(self):
|
|
|
495
|
+ """Provide a list of relative paths which have been modified since the
|
|
|
496
|
+ last call to mark_unmodified.
|
|
|
497
|
+
|
|
|
498
|
+ Return value: List(str) - list of modified paths
|
|
|
499
|
+ """
|
|
|
500
|
+
|
|
|
501
|
+ filelist = []
|
|
|
502
|
+ for (k, v) in self.index.items():
|
|
|
503
|
+ if isinstance(v.buildstream_object, CasBasedDirectory):
|
|
|
504
|
+ filelist.extend([k + os.path.sep + x for x in v.buildstream_object.list_modified_paths()])
|
|
|
505
|
+ elif isinstance(v.pb2_object, remote_execution_pb2.FileNode) and v.modified:
|
|
|
506
|
+ filelist.append(k)
|
|
|
507
|
+ return filelist
|
|
|
508
|
+
|
|
|
509
|
+ def list_relative_paths(self):
|
|
|
510
|
+ """Provide a list of all relative paths.
|
|
|
511
|
+
|
|
|
512
|
+ NOTE: This list is not in the same order as utils.list_relative_paths.
|
|
|
513
|
+
|
|
|
514
|
+ Return value: List(str) - list of all paths
|
|
|
515
|
+ """
|
|
|
516
|
+
|
|
|
517
|
+ filelist = []
|
|
|
518
|
+ for (k, v) in self.index.items():
|
|
|
519
|
+ if isinstance(v.buildstream_object, CasBasedDirectory):
|
|
|
520
|
+ filelist.extend([k + os.path.sep + x for x in v.buildstream_object.list_relative_paths()])
|
|
|
521
|
+ elif isinstance(v.pb2_object, remote_execution_pb2.FileNode):
|
|
|
522
|
+ filelist.append(k)
|
|
|
523
|
+ return filelist
|
|
|
524
|
+
|
|
|
525
|
+ def _get_identifier(self):
|
|
|
526
|
+ path = ""
|
|
|
527
|
+ if self.parent:
|
|
|
528
|
+ path = self.parent._get_identifier()
|
|
|
529
|
+ if self.filename:
|
|
|
530
|
+ path += "/" + self.filename
|
|
|
531
|
+ else:
|
|
|
532
|
+ path += "/" + self.common_name
|
|
|
533
|
+ return path
|
|
|
534
|
+
|
|
|
535
|
+ def __str__(self):
|
|
|
536
|
+ return "[CAS:{}]".format(self._get_identifier())
|
|
|
537
|
+
|
|
|
538
|
+ def get_underlying_directory(self):
|
|
|
539
|
+ """ There is no underlying directory for a CAS-backed directory, so
|
|
|
540
|
+ throw an exception. """
|
|
|
541
|
+ raise VirtualDirectoryError("get_underlying_directory was called on a CAS-backed directory," +
|
|
|
542
|
+ " which has no underlying directory.")
|
