Check for new versions at startup

CHANGELOG.md

@@ -22,6 +22,10 @@ ambiguous situations.
 
 ## Unreleased
 
+### Fixed
+- Forum crawling crashing when parsing empty (= 0 messages) threads
+- Forum crawling crashing when a forum has no threads at all
+
 ## 3.4.1 - 2022-08-17
 
 ### Added

PFERD/__main__.py

@@ -5,6 +5,8 @@ import os
 import sys
 from pathlib import Path
 
+from PFERD.update import check_for_updates
+
 from .auth import AuthLoadError
 from .cli import PARSER, ParserLoadError, load_default_section
 from .config import Config, ConfigDumpError, ConfigLoadError, ConfigOptionError
@@ -134,6 +136,11 @@ def main() -> None:
             loop.run_until_complete(asyncio.sleep(1))
             loop.close()
         else:
+            log.explain_topic("Checking for updates")
+            if not args.skip_update_check:
+                asyncio.run(check_for_updates())
+            else:
+                log.explain("Update check skipped due to configuration option")
             asyncio.run(pferd.run(args.debug_transforms))
     except (ConfigOptionError, AuthLoadError) as e:
         log.unlock()

PFERD/cli/parser.py

@@ -151,6 +151,11 @@ PARSER.add_argument(
     action="version",
     version=f"{NAME} {VERSION} (https://github.com/Garmelon/PFERD)",
 )
+PARSER.add_argument(
+    "--skip-update-check",
+    action="store_true",
+    help="disable automatic update checks at startup"
+)
 PARSER.add_argument(
     "--config", "-c",
     type=Path,

PFERD/crawl/ilias/kit_ilias_html.py

@@ -59,6 +59,7 @@ class IliasPageElement:
 class IliasDownloadForumData:
     url: str
     form_data: Dict[str, Union[str, List[str]]]
+    empty: bool
 
 
 @dataclass
@@ -130,24 +131,32 @@ class IliasPage:
             return None
         post_url = self._abs_url_from_relative(form["action"])
 
+        thread_ids = [f["value"] for f in form.find_all(attrs={"name": "thread_ids[]"})]
+
         form_data: Dict[str, Union[str, List[ſtr]]] = {
-            "thread_ids[]": [f["value"] for f in form.find_all(attrs={"name": "thread_ids[]"})],
+            "thread_ids[]": thread_ids,
             "selected_cmd2": "html",
             "select_cmd2": "Ausführen",
             "selected_cmd": "",
         }
 
-        return IliasDownloadForumData(post_url, form_data)
+        return IliasDownloadForumData(url=post_url, form_data=form_data, empty=len(thread_ids) == 0)
 
     def get_next_stage_element(self) -> Optional[IliasPageElement]:
         if self._is_forum_page():
             if "trows=800" in self._page_url:
                 return None
+            log.explain("Requesting *all* forum threads")
             return self._get_show_max_forum_entries_per_page_url()
         if self._is_ilias_opencast_embedding():
+            log.explain("Unwrapping opencast embedding")
             return self.get_child_elements()[0]
         if self._page_type == IliasElementType.VIDEO_FOLDER_MAYBE_PAGINATED:
+            log.explain("Unwrapping video pagination")
             return self._find_video_entries_paginated()[0]
+        if self._contains_collapsed_future_meetings():
+            log.explain("Requesting *all* future meetings")
+            return self._uncollapse_future_meetings_url()
         return None
 
     def _is_forum_page(self) -> bool:
@@ -200,6 +209,16 @@ class IliasPage:
             return False
         return "target=copa_" in link.get("value")
 
+    def _contains_collapsed_future_meetings(self) -> bool:
+        return self._uncollapse_future_meetings_url() is not None
+
+    def _uncollapse_future_meetings_url(self) -> Optional[IliasPageElement]:
+        element = self._soup.find("a", attrs={"href": lambda x: x and "crs_next_sess=1" in x})
+        if not element:
+            return None
+        link = self._abs_url_from_link(element)
+        return IliasPageElement(IliasElementType.FOLDER, link, "show all meetings")
+
     def _player_to_video(self) -> List[IliasPageElement]:
         # Fetch the actual video page. This is a small wrapper page initializing a javscript
         # player. Sadly we can not execute that JS. The actual video stream url is nowhere
@@ -790,6 +809,10 @@ class IliasPage:
         if img_tag is None:
             img_tag = found_parent.select_one("img.icon")
 
+        if img_tag is None and found_parent.find("a", attrs={"href": lambda x: x and "crs_next_sess=" in x}):
+            log.explain("Found session expansion button, skipping it as it has no content")
+            return None
+
         if img_tag is None:
             _unexpected_html_warning()
             log.warn_contd(f"Tried to figure out element type, but did not find an image for {url}")
@@ -937,6 +960,13 @@ def parse_ilias_forum_export(forum_export: BeautifulSoup) -> List[IliasForumThre
     for p in forum_export.select("body > p"):
         title_tag = p
         content_tag = p.find_next_sibling("ul")
+
+        if not content_tag:
+            # ILIAS allows users to delete the initial post while keeping the thread open
+            # This produces empty threads without *any* content.
+            # I am not sure why you would want this, but ILIAS makes it easy to do.
+            continue
+
         title = p.find("b").text
         if ":" in title:
             title = title[title.find(":") + 1:]

PFERD/crawl/ilias/kit_ilias_web_crawler.py

@@ -23,6 +23,12 @@ from .kit_ilias_html import (IliasElementType, IliasForumThread, IliasPage, Ilia
 
 TargetType = Union[str, int]
 
+_ILIAS_URL = "https://ilias.studium.kit.edu"
+
+
+class KitShibbolethBackgroundLoginSuccessful():
+    pass
+
 
 class KitIliasWebCrawlerSection(HttpCrawlerSection):
     def target(self) -> TargetType:
@@ -36,7 +42,7 @@ class KitIliasWebCrawlerSection(HttpCrawlerSection):
         if target == "desktop":
             # Full personal desktop
             return target
-        if target.startswith("https://ilias.studium.kit.edu"):
+        if target.startswith(_ILIAS_URL):
             # ILIAS URL
             return target
 
@@ -181,7 +187,7 @@ instance's greatest bottleneck.
             section.tfa_auth(authenticators),
         )
 
-        self._base_url = "https://ilias.studium.kit.edu"
+        self._base_url = _ILIAS_URL
 
         self._target = section.target()
         self._link_file_redirect_delay = section.link_redirect_delay()
@@ -228,19 +234,28 @@ instance's greatest bottleneck.
         async def gather_elements() -> None:
             elements.clear()
             async with cl:
-                soup = await self._get_page(url)
-
-                if expected_id is not None:
-                    perma_link_element: Tag = soup.find(id="current_perma_link")
-                    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")
+                next_stage_url: Optional[str] = url
+                current_parent = None
 
                 # Duplicated code, but the root page is special - we want to avoid fetching it twice!
-                log.explain_topic("Parsing root HTML page")
-                log.explain(f"URL: {url}")
-                page = IliasPage(soup, url, None)
-                elements.extend(page.get_child_elements())
+                while next_stage_url:
+                    soup = await self._get_page(next_stage_url)
 
+                    if current_parent is None and expected_id is not None:
+                        perma_link_element: Tag = soup.find(id="current_perma_link")
+                        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")
+
+                    log.explain_topic(f"Parsing HTML page for {fmt_path(cl.path)}")
+                    log.explain(f"URL: {next_stage_url}")
+                    page = IliasPage(soup, next_stage_url, current_parent)
+                    if next_element := page.get_next_stage_element():
+                        current_parent = next_element
+                        next_stage_url = next_element.url
+                    else:
+                        next_stage_url = None
+
+                elements.extend(page.get_child_elements())
                 if description_string := page.get_description():
                     description.append(description_string)
 
@@ -658,7 +673,7 @@ instance's greatest bottleneck.
     @_iorepeat(3, "crawling forum")
     @anoncritical
     async def _crawl_forum(self, element: IliasPageElement, cl: CrawlToken) -> None:
-        elements = []
+        elements: List[IliasForumThread] = []
 
         async with cl:
             next_stage_url = element.url
@@ -677,6 +692,10 @@ instance's greatest bottleneck.
             download_data = page.get_download_forum_data()
             if not download_data:
                 raise CrawlWarning("Failed to extract forum data")
+            if download_data.empty:
+                log.explain("Forum had no threads")
+                elements = []
+                return
             html = await self._post_authenticated(download_data.url, download_data.form_data)
             elements = parse_ilias_forum_export(soupify(html))
 
@@ -804,14 +823,17 @@ class KitShibbolethLogin:
 
         # Equivalent: Click on "Mit KIT-Account anmelden" button in
         # https://ilias.studium.kit.edu/login.php
-        url = "https://ilias.studium.kit.edu/shib_login.php"
+        url = f"{_ILIAS_URL}/shib_login.php"
         data = {
             "sendLogin": "1",
             "idp_selection": "https://idp.scc.kit.edu/idp/shibboleth",
             "il_target": "",
             "home_organization_selection": "Weiter",
         }
-        soup: BeautifulSoup = await _shib_post(sess, url, data)
+        soup: Union[BeautifulSoup, KitShibbolethBackgroundLoginSuccessful] = await _shib_post(sess, url, data)
+
+        if isinstance(soup, KitShibbolethBackgroundLoginSuccessful):
+            return
 
         # Attempt to login using credentials, if necessary
         while not self._login_successful(soup):
@@ -850,7 +872,7 @@ class KitShibbolethLogin:
         # (or clicking "Continue" if you have JS disabled)
         relay_state = soup.find("input", {"name": "RelayState"})
         saml_response = soup.find("input", {"name": "SAMLResponse"})
-        url = "https://ilias.studium.kit.edu/Shibboleth.sso/SAML2/POST"
+        url = f"{_ILIAS_URL}/Shibboleth.sso/SAML2/POST"
         data = {  # using the info obtained in the while loop above
             "RelayState": relay_state["value"],
             "SAMLResponse": saml_response["value"],
@@ -899,22 +921,35 @@ async def _post(session: aiohttp.ClientSession, url: str, data: Any) -> Beautifu
         return soupify(await response.read())
 
 
-async def _shib_post(session: aiohttp.ClientSession, url: str, data: Any) -> BeautifulSoup:
+async def _shib_post(
+    session: aiohttp.ClientSession,
+    url: str,
+    data: Any
+) -> Union[BeautifulSoup, KitShibbolethBackgroundLoginSuccessful]:
     """
     aiohttp unescapes '/' and ':' in URL query parameters which is not RFC compliant and rejected
     by Shibboleth. Thanks a lot. So now we unroll the requests manually, parse location headers and
     build encoded URL objects ourselves... Who thought mangling location header was a good idea??
     """
+    log.explain_topic("Shib login POST")
     async with session.post(url, data=data, allow_redirects=False) as response:
         location = response.headers.get("location")
+        log.explain(f"Got location {location!r}")
         if not location:
             raise CrawlWarning(f"Login failed (1), no location header present at {url}")
         correct_url = yarl.URL(location, encoded=True)
+        log.explain(f"Corrected location to {correct_url!r}")
+
+        if str(correct_url).startswith(_ILIAS_URL):
+            log.explain("ILIAS recognized our shib token and logged us in in the background, returning")
+            return KitShibbolethBackgroundLoginSuccessful()
 
         async with session.get(correct_url, allow_redirects=False) as response:
             location = response.headers.get("location")
+            log.explain(f"Redirected to {location!r} with status {response.status}")
             # If shib still still has a valid session, it will directly respond to the request
             if location is None:
+                log.explain("Shib recognized us, returning its response directly")
                 return soupify(await response.read())
 
             as_yarl = yarl.URL(response.url)
@@ -928,6 +963,7 @@ async def _shib_post(session: aiohttp.ClientSession, url: str, data: Any) -> Bea
                 path=location,
                 encoded=True
             )
+            log.explain(f"Corrected location to {correct_url!r}")
 
             async with session.get(correct_url, allow_redirects=False) as response:
                 return soupify(await response.read())

PFERD/update.py

@@ -0,0 +1,53 @@
+from dataclasses import dataclass
+import ssl
+from typing import Optional
+import aiohttp
+import certifi
+
+from .version import NAME, VERSION
+from .logging import log
+
+
+@dataclass
+class PferdUpdate:
+    release_url: str
+    version: str
+
+
+def _build_session() -> aiohttp.ClientSession:
+    return aiohttp.ClientSession(
+        headers={"User-Agent": f"{NAME}/{VERSION}"},
+        connector=aiohttp.TCPConnector(ssl=ssl.create_default_context(cafile=certifi.where())),
+        timeout=aiohttp.ClientTimeout(
+            total=15 * 60,
+            connect=10,
+            sock_connect=10,
+            sock_read=10,
+        )
+    )
+
+
+async def check_for_updates() -> None:
+    if new_version := await get_newer_version():
+        log.warn(
+            f"{NAME} version out of date. "
+            + f"You are running version {VERSION!r} but {new_version.version!r} was found on GitHub."
+        )
+        log.warn_contd(f"You can download it on GitHub: {new_version.release_url}")
+    else:
+        log.explain("No update found")
+
+
+async def get_newer_version() -> Optional[PferdUpdate]:
+    async with _build_session() as session:
+        async with session.get(
+            "https://api.github.com/repos/Garmelon/Pferd/releases/latest",
+            headers={"Accept": "application/vnd.github+json"}
+        ) as response:
+            release_information = await response.json()
+            tag_name: str = release_information["tag_name"]
+            tag_name = tag_name.removeprefix("v")
+            if VERSION == tag_name:
+                return None
+
+            return PferdUpdate(release_url=release_information["html_url"], version=tag_name)