mirror of
https://github.com/nexus-stc/hyperboria
synced 2024-12-21 17:17:47 +01:00
9ce67ec590
- feat(idm): Rename IDM-2 to IDM - feat(idm): Open IDM 3 internal commit(s) GitOrigin-RevId: e302e9b5cda18cca1adc4ae8a3d906714d222106
35 lines
1.1 KiB
Python
35 lines
1.1 KiB
Python
import re
|
|
from typing import (
|
|
AsyncIterable,
|
|
Callable,
|
|
)
|
|
|
|
from library.logging import error_log
|
|
from nexus.pylon.exceptions import RegexNotFoundError
|
|
from nexus.pylon.sources.base import (
|
|
DoiSource,
|
|
PreparedRequest,
|
|
)
|
|
|
|
|
|
class ResearchSquareSource(DoiSource):
|
|
base_url = 'https://dx.doi.org'
|
|
|
|
async def resolve(self, error_log_func: Callable = error_log) -> AsyncIterable[PreparedRequest]:
|
|
async with self.get_resolve_session() as session:
|
|
url = f'{self.base_url}/{self.doi}'
|
|
async with PreparedRequest(
|
|
method='get',
|
|
url=url,
|
|
timeout=self.resolve_timeout,
|
|
).execute_with(session=session) as resp:
|
|
download_page = await resp.text()
|
|
match = re.search(
|
|
r'\"(https://www\.researchsquare\.com/article/[^\"]+\.pdf)\"',
|
|
download_page,
|
|
re.IGNORECASE,
|
|
)
|
|
if not match:
|
|
raise RegexNotFoundError(url=url)
|
|
yield PreparedRequest(method='get', url=match.group(1), timeout=self.timeout)
|