def _create_file_info_dict(self, file_path, piece_size, include_md5_sum=True):
"""
Returns a dictionary with the following keys:
- pieces: concatenated 20-byte SHA-1 hashes
- name: basename of the file
- length: size of the file in bytes
- md5sum: md5sum of the file (if include_md5_sum is True)
@rtype: dict
"""
if not os.path.isfile(file_path):
msg = '"{path}" is not a file'
raise TorrentError(msg.format(path=file_path))
if os.path.getsize(file_path) == 0:
msg = '"{path}" is a zero byte file!'
raise TorrentError(msg.format(path=file_path))
file_extension = os.path.splitext(file_path)[1].lower()
if (self.extension_whitelist is not None) and (file_extension not in self.extension_whitelist):
msg = '"{path}" is not a valid file type for upload to this tracker'
raise TorrentError(msg.format(path=file_path))
# Concatenated 20-byte SHA-1 hashes of all the file's pieces
pieces = bytearray()
# Aggregate MD5 sum
md5 = hashlib.md5() if include_md5_sum else None
msg = 'Hashing file "{path}"... '
logging.info(msg.format(path=file_path))
file_pieces = create_piece_generator(file_path, piece_size)
for piece in file_pieces:
piece_hash = files.sha1(piece)
pieces.extend(piece_hash)
if include_md5_sum:
md5.update(piece)
info = {
'pieces': pieces,
'piece length': piece_size,
'name': os.path.basename(file_path),
'length': os.path.getsize(file_path),
}
if include_md5_sum:
info['md5sum'] = md5.hexdigest()
assert len(info['pieces']) % 20 == 0, 'len(pieces) is not a multiple of 20 bytes!'
return info
def _create_directory_info_dict(self, root_dir_path, piece_size, include_md5_sum=True):
"""
Returns a dictionary with the following keys:
- pieces: concatenated 20-byte SHA-1 hashes
- name: basename of the directory (default name of all torrents)
- files: a list of dictionaries with the following keys:
- length: size of the file in bytes
- md5sum: md5 sum of the file (unless disabled via include_md5)
- path: list of the file's path components, relative to the directory
@rtype: dict
"""
if not os.path.isdir(root_dir_path):
msg = '"{path}" is not a directory'
raise TorrentError(msg.format(path=root_dir_path))
if piece_size < (16 * KB):
msg = 'Piece size {size} is less than 16 KiB!'
raise TorrentError(msg.format(size=piece_size))
# Concatenated 20-byte SHA-1 hashes of all the torrent's pieces.
info_pieces = bytearray()
# This bytearray will be used for the calculation of info_pieces.
# Consecutive files will be written into data_buffer as a continuous
# stream, as required by the BitTorrent specification.
data_buffer = bytearray()
file_dicts = []
for (dir_path, dir_names, file_names) in os.walk(root_dir_path):
for file_name in file_names:
# If the file's extension isn't in the whitelist, ignore it
file_extension = os.path.splitext(file_name)[1].lower()
if (self.extension_whitelist is not None) and (file_extension not in self.extension_whitelist):
continue
file_path = os.path.join(dir_path, file_name)
# Build the current file's dictionary.
file_dict = {
'length': os.path.getsize(file_path),
'path': files.split_path(os.path.relpath(file_path, root_dir_path))
}
# Keep track of the file's MD5 sum
md5 = hashlib.md5() if include_md5_sum else None
logging.info(
'Hashing file "{path}"... '.format(
path=os.path.relpath(file_path, root_dir_path)
)
)
file_pieces = create_piece_generator(file_path, piece_size)
for piece in file_pieces:
data_buffer.extend(piece)
if len(data_buffer) >= piece_size:
piece_hash = files.sha1(data_buffer[:piece_size])
info_pieces.extend(piece_hash)
data_buffer[:] = data_buffer[piece_size:]
if include_md5_sum:
md5.update(piece)
if include_md5_sum:
file_dict['md5sum'] = md5.hexdigest()
file_dicts.append(file_dict)
# Hash any remaining data that is fewer than piece_size bytes
if len(data_buffer) > 0:
piece_hash = files.sha1(data_buffer)
info_pieces.extend(piece_hash)
info = {
'pieces': info_pieces,
'piece length': piece_size,
'name': os.path.basename(root_dir_path.strip(os.path.sep)),
'files': file_dicts,
}
assert len(info['pieces']) % 20 == 0, 'len(pieces) is not a multiple of 20 bytes!'
return info
