hyperboria/nexus/pylon/sources/base.py
the-superpirate 7389f85ef2 - Send Pylon to golden
-  Move fancy naming to Hub
-  Adopt translations
-  Fix long authors list in bot
-  Support CJK in scitech
-  Fixed pylon bug with decoding improper unicode ch...

GitOrigin-RevId: 74f73c44f749a71cb65dd5ddd3416f32a83d329f
2021-01-08 23:09:19 +03:00

203 lines
6.0 KiB
Python

import hashlib
import random
from typing import (
AsyncIterable,
Optional,
)
import aiohttp
import aiohttp.client_exceptions
from aiohttp_socks import (
ProxyConnector,
ProxyError,
)
from aiokit import AioThing
from nexus.pylon.exceptions import (
BadResponseError,
IncorrectMD5Error,
NotFoundError,
)
from nexus.pylon.pdftools import is_pdf
from nexus.pylon.proto.file_pb2 import Chunk as ChunkPb
from nexus.pylon.proto.file_pb2 import FileResponse as FileResponsePb
from python_socks import ProxyTimeoutError
from tenacity import (
retry,
retry_if_exception_type,
stop_after_attempt,
)
DEFAULT_USER_AGENT = 'PylonBot/1.0 (Linux x86_64) PylonBot/1.0.0'
class PreparedRequest:
def __init__(
self,
method: str,
url: str,
headers: dict = None,
params: dict = None,
cookies: dict = None,
ssl: bool = True,
):
self.method = method
self.url = url
self.headers = {
'User-Agent': DEFAULT_USER_AGENT,
}
if headers:
self.headers.update(headers)
self.params = params
self.cookies = cookies
self.ssl = ssl
def __repr__(self):
return f'{self.method} {self.url} {self.headers} {self.params}'
def __str__(self):
return repr(self)
class BaseValidator:
def update(self, chunk: bytes):
pass
def validate(self):
pass
class Md5Validator(BaseValidator):
def __init__(self, md5: str):
self.md5 = md5
self.v = hashlib.md5()
def update(self, chunk: bytes):
self.v.update(chunk)
def validate(self):
digest = self.v.hexdigest()
if self.md5.lower() != self.v.hexdigest().lower():
raise IncorrectMD5Error(requested_md5=self.md5, downloaded_md5=digest)
class DoiValidator(BaseValidator):
def __init__(self, doi: str):
self.doi = doi
self.file = bytes()
def update(self, chunk):
self.file += chunk
def validate(self):
return is_pdf(self.file)
class BaseSource(AioThing):
allowed_content_type = None
base_url = None
is_enabled = True
resolve_timeout = None
ssl = True
timeout = None
use_proxy = None
def __init__(self, proxy: str = None, resolve_proxy: str = None):
super().__init__()
self.proxy = proxy
self.resolve_proxy = resolve_proxy
def get_proxy(self):
if self.proxy and self.use_proxy is not False:
return ProxyConnector.from_url(self.proxy, verify_ssl=self.ssl)
return aiohttp.TCPConnector(verify_ssl=self.ssl)
def get_resolve_proxy(self):
if self.resolve_proxy and self.use_proxy is not False:
return ProxyConnector.from_url(self.resolve_proxy, verify_ssl=self.ssl)
return aiohttp.TCPConnector(verify_ssl=self.ssl)
def get_session(self):
return aiohttp.ClientSession(connector=self.get_proxy())
def get_resolve_session(self):
return aiohttp.ClientSession(connector=self.get_resolve_proxy())
def resolve(self) -> AsyncIterable[PreparedRequest]:
raise NotImplementedError("`resolve` for BaseSource is not implemented")
def get_validator(self):
return BaseValidator()
@retry(
reraise=True,
stop=stop_after_attempt(3),
retry=retry_if_exception_type((ProxyError, aiohttp.client_exceptions.ClientPayloadError, ProxyTimeoutError)),
)
async def execute_prepared_request(self, prepared_request: PreparedRequest):
async with self.get_session() as session:
async with session.request(
method=prepared_request.method,
url=prepared_request.url,
timeout=self.timeout,
headers=prepared_request.headers,
cookies=prepared_request.cookies,
params=prepared_request.params,
ssl=prepared_request.ssl,
) as resp:
if resp.status == 404:
raise NotFoundError(url=prepared_request.url)
elif (
resp.status != 200
or (
self.allowed_content_type
and resp.headers.get('Content-Type', '').lower() not in self.allowed_content_type
)
):
raise BadResponseError(
request_headers=prepared_request.headers,
url=prepared_request.url,
status=resp.status,
headers=str(resp.headers),
)
file_validator = self.get_validator()
# Randomness is required due to annoying bug of when separators
# (\r\n) are splitted to different chunks
# https://github.com/aio-libs/aiohttp/issues/4677
yield FileResponsePb(status=FileResponsePb.Status.BEGIN_TRANSMISSION, source=prepared_request.url)
async for content in resp.content.iter_chunked(1024 * 100 + random.randint(-1024, 1024)):
file_validator.update(content)
yield FileResponsePb(chunk=ChunkPb(content=content), source=prepared_request.url)
file_validator.validate()
class Md5Source(BaseSource):
def __init__(
self,
md5: str,
proxy: Optional[str] = None,
resolve_proxy: Optional[str] = None,
):
super().__init__(proxy=proxy, resolve_proxy=resolve_proxy)
self.md5 = md5
def get_validator(self):
return Md5Validator(self.md5)
class DoiSource(BaseSource):
def __init__(
self,
doi: str,
md5: Optional[str] = None,
proxy: Optional[str] = None,
resolve_proxy: Optional[str] = None,
):
super().__init__(proxy=proxy, resolve_proxy=resolve_proxy)
self.doi = doi
self.md5 = md5
def get_validator(self):
if self.md5:
return Md5Validator(self.md5)
return DoiValidator(self.doi)