mirror of
https://github.com/LonamiWebs/Telethon.git
synced 2024-11-24 02:13:45 +03:00
149 lines
4.5 KiB
Python
149 lines
4.5 KiB
Python
import collections
|
|
import re
|
|
|
|
from .tlarg import TLArg
|
|
from .tlobject import TLObject
|
|
from ..methods import Usability
|
|
|
|
|
|
CORE_TYPES = {
|
|
0xbc799737, # boolFalse#bc799737 = Bool;
|
|
0x997275b5, # boolTrue#997275b5 = Bool;
|
|
0x3fedd339, # true#3fedd339 = True;
|
|
0xc4b9f9bb, # error#c4b9f9bb code:int text:string = Error;
|
|
0x56730bcc # null#56730bcc = Null;
|
|
}
|
|
|
|
# Telegram Desktop (C++) doesn't care about string/bytes, and the .tl files
|
|
# don't either. However in Python we *do*, and we want to deal with bytes
|
|
# for the authorization key process, not UTF-8 strings (they won't be).
|
|
#
|
|
# Every type with an ID that's in here should get their attribute types
|
|
# with string being replaced with bytes.
|
|
AUTH_KEY_TYPES = {
|
|
0x05162463, # resPQ,
|
|
0x83c95aec, # p_q_inner_data
|
|
0xa9f55f95, # p_q_inner_data_dc
|
|
0x3c6a84d4, # p_q_inner_data_temp
|
|
0x56fddf88, # p_q_inner_data_temp_dc
|
|
0xd0e8075c, # server_DH_params_ok
|
|
0xb5890dba, # server_DH_inner_data
|
|
0x6643b654, # client_DH_inner_data
|
|
0xd712e4be, # req_DH_params
|
|
0xf5045f1f, # set_client_DH_params
|
|
0x3072cfa1 # gzip_packed
|
|
}
|
|
|
|
|
|
def _from_line(line, is_function, method_info, layer):
|
|
match = re.match(
|
|
r'^([\w.]+)' # 'name'
|
|
r'(?:#([0-9a-fA-F]+))?' # '#optionalcode'
|
|
r'(?:\s{?\w+:[\w\d<>#.?!]+}?)*' # '{args:.0?type}'
|
|
r'\s=\s' # ' = '
|
|
r'([\w\d<>#.?]+);$', # '<result.type>;'
|
|
line
|
|
)
|
|
if match is None:
|
|
# Probably "vector#1cb5c415 {t:Type} # [ t ] = Vector t;"
|
|
raise ValueError('Cannot parse TLObject {}'.format(line))
|
|
|
|
args_match = re.findall(
|
|
r'({)?'
|
|
r'(\w+)'
|
|
r':'
|
|
r'([\w\d<>#.?!]+)'
|
|
r'}?',
|
|
line
|
|
)
|
|
|
|
name = match.group(1)
|
|
method_info = method_info.get(name)
|
|
if method_info:
|
|
usability = method_info.usability
|
|
friendly = method_info.friendly
|
|
else:
|
|
usability = Usability.UNKNOWN
|
|
friendly = None
|
|
|
|
return TLObject(
|
|
fullname=name,
|
|
object_id=match.group(2),
|
|
result=match.group(3),
|
|
is_function=is_function,
|
|
layer=layer,
|
|
usability=usability,
|
|
friendly=friendly,
|
|
args=[TLArg(name, arg_type, brace != '')
|
|
for brace, name, arg_type in args_match]
|
|
)
|
|
|
|
|
|
def parse_tl(file_path, layer, methods=None, ignored_ids=CORE_TYPES):
|
|
"""
|
|
This method yields TLObjects from a given .tl file.
|
|
|
|
Note that the file is parsed completely before the function yields
|
|
because references to other objects may appear later in the file.
|
|
"""
|
|
method_info = {m.name: m for m in (methods or [])}
|
|
obj_all = []
|
|
obj_by_name = {}
|
|
obj_by_type = collections.defaultdict(list)
|
|
with file_path.open() as file:
|
|
is_function = False
|
|
for line in file:
|
|
comment_index = line.find('//')
|
|
if comment_index != -1:
|
|
line = line[:comment_index]
|
|
|
|
line = line.strip()
|
|
if not line:
|
|
continue
|
|
|
|
match = re.match(r'---(\w+)---', line)
|
|
if match:
|
|
following_types = match.group(1)
|
|
is_function = following_types == 'functions'
|
|
continue
|
|
|
|
try:
|
|
result = _from_line(
|
|
line, is_function, method_info, layer=layer)
|
|
|
|
if result.id in ignored_ids:
|
|
continue
|
|
|
|
obj_all.append(result)
|
|
if not result.is_function:
|
|
obj_by_name[result.fullname] = result
|
|
obj_by_type[result.result].append(result)
|
|
except ValueError as e:
|
|
if 'vector#1cb5c415' not in str(e):
|
|
raise
|
|
|
|
# Once all objects have been parsed, replace the
|
|
# string type from the arguments with references
|
|
for obj in obj_all:
|
|
if obj.id in AUTH_KEY_TYPES:
|
|
for arg in obj.args:
|
|
if arg.type == 'string':
|
|
arg.type = 'bytes'
|
|
|
|
for arg in obj.args:
|
|
arg.cls = obj_by_type.get(arg.type) or (
|
|
[obj_by_name[arg.type]] if arg.type in obj_by_name else []
|
|
)
|
|
|
|
yield from obj_all
|
|
|
|
|
|
def find_layer(file_path):
|
|
"""Finds the layer used on the specified scheme.tl file."""
|
|
layer_regex = re.compile(r'^//\s*LAYER\s*(\d+)$')
|
|
with file_path.open('r') as file:
|
|
for line in file:
|
|
match = layer_regex.match(line)
|
|
if match:
|
|
return int(match.group(1))
|