+ # Build the upload file list, excluding requested files and counting the
+ # bytes expected to be uploaded.
+ self._build_upload_list()
+
+ def _build_upload_list(self):
+ """
+ Scan the requested paths to count file sizes, excluding files & dirs if requested
+ and building the upload file list.
+ """
+ # If there aren't special files to be read, reset total bytes count to zero
+ # to start counting.
+ if not any(filter(lambda p: not (os.path.isfile(p) or os.path.isdir(p)),
+ self.paths)):
+ self.bytes_expected = 0
+
+ for path in self.paths:
+ # Test for stdin first, in case some file named '-' exist
+ if path == '-':
+ if self.dry_run:
+ raise ArvPutUploadIsPending()
+ self._write_stdin(self.filename or 'stdin')
+ elif not os.path.exists(path):
+ raise PathDoesNotExistError("file or directory '{}' does not exist.".format(path))
+ elif os.path.isdir(path):
+ # Use absolute paths on cache index so CWD doesn't interfere
+ # with the caching logic.
+ orig_path = path
+ path = os.path.abspath(path)
+ if orig_path[-1:] == os.sep:
+ # When passing a directory reference with a trailing slash,
+ # its contents should be uploaded directly to the
+ # collection's root.
+ prefixdir = path
+ else:
+ # When passing a directory reference with no trailing slash,
+ # upload the directory to the collection's root.
+ prefixdir = os.path.dirname(path)
+ prefixdir += os.sep
+ for root, dirs, files in os.walk(path,
+ followlinks=self.follow_links):
+ root_relpath = os.path.relpath(root, path)
+ if root_relpath == '.':
+ root_relpath = ''
+ # Exclude files/dirs by full path matching pattern
+ if self.exclude_paths:
+ dirs[:] = list(filter(
+ lambda d: not any(
+ [pathname_match(os.path.join(root_relpath, d),
+ pat)
+ for pat in self.exclude_paths]),
+ dirs))
+ files = list(filter(
+ lambda f: not any(
+ [pathname_match(os.path.join(root_relpath, f),
+ pat)
+ for pat in self.exclude_paths]),
+ files))
+ # Exclude files/dirs by name matching pattern
+ if self.exclude_names is not None:
+ dirs[:] = list(filter(lambda d: not self.exclude_names.match(d), dirs))
+ files = list(filter(lambda f: not self.exclude_names.match(f), files))
+ # Make os.walk()'s dir traversing order deterministic
+ dirs.sort()
+ files.sort()
+ for f in files:
+ filepath = os.path.join(root, f)
+ # Add its size to the total bytes count (if applicable)
+ if self.follow_links or (not os.path.islink(filepath)):
+ if self.bytes_expected is not None:
+ self.bytes_expected += os.path.getsize(filepath)
+ self._check_file(filepath,
+ os.path.join(root[len(prefixdir):], f))
+ else:
+ filepath = os.path.abspath(path)
+ # Add its size to the total bytes count (if applicable)
+ if self.follow_links or (not os.path.islink(filepath)):
+ if self.bytes_expected is not None:
+ self.bytes_expected += os.path.getsize(filepath)
+ self._check_file(filepath,
+ self.filename or os.path.basename(path))
+ # If dry-mode is on, and got up to this point, then we should notify that
+ # there aren't any file to upload.
+ if self.dry_run:
+ raise ArvPutUploadNotPending()
+ # Remove local_collection's files that don't exist locally anymore, so the
+ # bytes_written count is correct.
+ for f in self.collection_file_paths(self._local_collection,
+ path_prefix=""):
+ if f != 'stdin' and f != self.filename and not f in self._file_paths:
+ self._local_collection.remove(f)
+
projects / arvados.git / blobdiff