mirrors/pferd
1
0
Fork 0
mirror of https://github.com/Garmelon/PFERD.git synced 2023-12-21 10:23:01 +01:00
Code Issues Packages Projects Releases Wiki Activity

Only enclose get_page in iorepeat in ILIAS crawler

Browse Source 
We previously also gathered in there, which could lead to some more
surprises when the method was retried.
This commit is contained in:
I-Al-Istannen 2021-05-23 17:24:05 +02:00
parent 29d5a40c57
commit 05ad06fbc1
1 changed files with 24 additions and 27 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

51
PFERD/crawlers/ilias/kit_ilias_web_crawler.py
View File

@ -1,6 +1,6 @@
import re
from pathlib import PurePath
from typing import Any, Awaitable, Callable, Dict, Optional, Set, TypeVar, Union
from typing import Any, Awaitable, Callable, Dict, List, Optional, Set, TypeVar, Union
import aiohttp
from aiohttp import hdrs
@ -192,10 +192,11 @@ class KitIliasWebCrawler(HttpCrawler):
return
cl = maybe_cl # Not mypy's fault, but explained here: https://github.com/python/mypy/issues/2608
@_iorepeat(3, "crawling url")
async def impl() -> None:
tasks = []
elements: List[IliasPageElement] = []
@_iorepeat(3, "crawling url")
async def gather_elements() -> None:
elements.clear()
async with cl:
soup = await self._get_page(url)
@ -204,48 +205,44 @@ class KitIliasWebCrawler(HttpCrawler):
if not perma_link_element or "crs_" not in perma_link_element.get("value"):
raise CrawlError("Invalid course id? Didn't find anything looking like a course")
# Duplicated code, but the root page is special - we want to void fetching it twice!
# Duplicated code, but the root page is special - we want to avoid fetching it twice!
page = IliasPage(soup, url, None)
for child in page.get_child_elements():
tasks.append(self._handle_ilias_element(PurePath("."), child))
elements.extend(page.get_child_elements())
# The only point an I/O exception can be thrown is in `get_page`.
# If that happens, no task was spawned yet. Therefore, we only retry
# this method without having spawned a single task. Due to this we do
# not need to cancel anything or worry about this gather call or the forks
# further up.
# Fill up our task list with the found elements
await gather_elements()
tasks = [self._handle_ilias_element(PurePath("."), element) for element in elements]
# And execute them
await self.gather(tasks)
await impl()
async def _handle_ilias_page(self, url: str, parent: IliasPageElement, path: PurePath) -> None:
maybe_cl = await self.crawl(path)
if not maybe_cl:
return
cl = maybe_cl # Not mypy's fault, but explained here: https://github.com/python/mypy/issues/2608
elements: List[IliasPageElement] = []
@_iorepeat(3, "crawling folder")
async def impl() -> None:
tasks = []
async def gather_elements() -> None:
elements.clear()
async with cl:
soup = await self._get_page(url)
page = IliasPage(soup, url, parent)
for child in page.get_child_elements():
tasks.append(self._handle_ilias_element(path, child))
elements.extend(page.get_child_elements())
# The only point an I/O exception can be thrown is in `get_page`.
# If that happens, no task was spawned yet. Therefore, we only retry
# this method without having spawned a single task. Due to this we do
# not need to cancel anything or worry about this gather call or the forks
# further up.
# Fill up our task list with the found elements
await gather_elements()
tasks = [self._handle_ilias_element(PurePath("."), element) for element in elements]
# And execute them
await self.gather(tasks)
await impl()
@anoncritical
# Shouldn't happen but this method must never raise an I/O error as that might interfere with
# handle_ilias_page or crawl_url
# Shouldn't happen but we also really don't want to let I/O errors bubble up to anoncritical.
# If that happens we will be terminated as anoncritical doesn't tream them as non-critical.
@_wrap_io_in_warning("handling ilias element")
async def _handle_ilias_element(self, parent_path: PurePath, element: IliasPageElement) -> None:
element_path = PurePath(parent_path, element.name)