Bump version to 3.8.0

CHANGELOG.md

@@ -22,22 +22,14 @@ ambiguous situations.
 
 ## Unreleased
 
-## 3.8.1 - 2025-04-17
-
-## Fixed
-- Description html files now specify at UTF-8 encoding
-- Images in descriptions now always have a white background
-
-## 3.8.0 - 2025-04-16
+## 3.8.0 - 2025-04-15
 
 ### Added
 - Support for ILIAS 9
 
 ### Changed
 - Added prettier CSS to forum threads
-- Downloaded forum threads now link to the forum instead of the ILIAS thread
 - Increase minimum supported Python version to 3.11
-- Do not crawl nested courses (courses linked in other courses)
 
 ## Fixed
 - File links in report on Windows

PFERD/crawl/ilias/ilias_html_cleaner.py

@@ -39,10 +39,6 @@ _STYLE_TAG_CONTENT = """
       margin: 0.5rem 0;
     }
 
-    img {
-        background-color: white;
-    }
-
     body {
       padding: 1em;
       grid-template-columns: 1fr min(60rem, 90%) 1fr;
@@ -60,11 +56,12 @@ _ARTICLE_WORTHY_CLASSES = [
 def insert_base_markup(soup: BeautifulSoup) -> BeautifulSoup:
     head = soup.new_tag("head")
     soup.insert(0, head)
-    # Force UTF-8 encoding
-    head.append(soup.new_tag("meta", charset="utf-8"))
 
+    simplecss_link: Tag = soup.new_tag("link")
     # <link rel="stylesheet" href="https://cdn.simplecss.org/simple.css">
-    head.append(soup.new_tag("link", rel="stylesheet", href="https://cdn.simplecss.org/simple.css"))
+    simplecss_link["rel"] = "stylesheet"
+    simplecss_link["href"] = "https://cdn.simplecss.org/simple.css"
+    head.append(simplecss_link)
 
     # Basic style tags for compat
     style: Tag = soup.new_tag("style")

PFERD/crawl/ilias/ilias_web_crawler.py

@@ -216,19 +216,11 @@ instance's greatest bottleneck.
 
     async def _crawl_desktop(self) -> None:
         await self._crawl_url(
-            urljoin(self._base_url, "/ilias.php?baseClass=ilDashboardGUI&cmd=show"),
-            crawl_nested_courses=True
+            urljoin(self._base_url, "/ilias.php?baseClass=ilDashboardGUI&cmd=show")
         )
 
-    async def _crawl_url(
-        self,
-        url: str,
-        expected_id: Optional[int] = None,
-        crawl_nested_courses: bool = False
-    ) -> None:
-        if awaitable := await self._handle_ilias_page(
-            url, None, PurePath("."), expected_id, crawl_nested_courses
-        ):
+    async def _crawl_url(self, url: str, expected_id: Optional[int] = None) -> None:
+        if awaitable := await self._handle_ilias_page(url, None, PurePath("."), expected_id):
             await awaitable
 
     async def _handle_ilias_page(
@@ -237,7 +229,6 @@ instance's greatest bottleneck.
         current_element: Optional[IliasPageElement],
         path: PurePath,
         expected_course_id: Optional[int] = None,
-        crawl_nested_courses: bool = False
     ) -> Optional[Coroutine[Any, Any, None]]:
         maybe_cl = await self.crawl(path)
         if not maybe_cl:
@@ -245,9 +236,7 @@ instance's greatest bottleneck.
         if current_element:
             self._ensure_not_seen(current_element, path)
 
-        return self._crawl_ilias_page(
-            url, current_element, maybe_cl, expected_course_id, crawl_nested_courses
-        )
+        return self._crawl_ilias_page(url, current_element, maybe_cl, expected_course_id)
 
     @anoncritical
     async def _crawl_ilias_page(
@@ -256,7 +245,6 @@ instance's greatest bottleneck.
         current_element: Optional[IliasPageElement],
         cl: CrawlToken,
         expected_course_id: Optional[int] = None,
-        crawl_nested_courses: bool = False,
     ) -> None:
         elements: List[IliasPageElement] = []
         # A list as variable redefinitions are not propagated to outer scopes
@@ -305,7 +293,7 @@ instance's greatest bottleneck.
 
         tasks: List[Awaitable[None]] = []
         for element in elements:
-            if handle := await self._handle_ilias_element(cl.path, element, crawl_nested_courses):
+            if handle := await self._handle_ilias_element(cl.path, element):
                 tasks.append(asyncio.create_task(handle))
 
         # And execute them
@@ -321,7 +309,6 @@ instance's greatest bottleneck.
         self,
         parent_path: PurePath,
         element: IliasPageElement,
-        crawl_nested_courses: bool = False
     ) -> Optional[Coroutine[Any, Any, None]]:
         # element.name might contain `/` if the crawler created nested elements,
         # so we can not sanitize it here. We trust in the output dir to thwart worst-case
@@ -415,13 +402,11 @@ instance's greatest bottleneck.
             )
             return None
         elif element.type == IliasElementType.COURSE:
-            if crawl_nested_courses:
-                return await self._handle_ilias_page(element.url, element, element_path)
             log.status(
                 "[bold bright_black]",
                 "Ignored",
                 fmt_path(element_path),
-                "[bright_black](not descending into linked course)"
+                "[bright_black](not descending into linked course, download it separately)"
             )
             return None
         elif element.type == IliasElementType.LEARNING_MODULE:
@@ -774,23 +759,70 @@ instance's greatest bottleneck.
     @_iorepeat(3, "crawling forum")
     @anoncritical
     async def _crawl_forum(self, element: IliasPageElement, cl: CrawlToken) -> None:
+        elements: List[IliasForumThread] = []
+
         async with cl:
-            inner = IliasPage(await self._get_page(element.url), element)
-            export_url = inner.get_forum_export_url()
-            if not export_url:
-                log.warn("Could not extract forum export url")
+            next_stage_url = element.url
+            page = None
+
+            while next_stage_url:
+                log.explain_topic(f"Parsing HTML page for {fmt_path(cl.path)}")
+                log.explain(f"URL: {next_stage_url}")
+
+                soup = await self._get_page(next_stage_url)
+                page = IliasPage(soup, element)
+
+                if next := page.get_next_stage_element():
+                    next_stage_url = next.url
+                else:
+                    break
+
+            forum_threads: list[tuple[IliasPageElement, bool]] = []
+            for entry in cast(IliasPage, page).get_forum_entries():
+                path = cl.path / (_sanitize_path_name(entry.name) + ".html")
+                forum_threads.append((entry, self.should_try_download(path, mtime=entry.mtime)))
+
+            # Sort the ids. The forum download will *preserve* this ordering
+            forum_threads.sort(key=lambda elem: elem[0].id())
+
+            if not forum_threads:
+                log.explain("Forum had no threads")
                 return
 
-            export = await self._post(export_url, {
-                "format": "html",
-                "cmd[createExportFile]": ""
-            })
+            download_data = cast(IliasPage, page).get_download_forum_data(
+                [thread.id() for thread, download in forum_threads if download]
+            )
+            if not download_data:
+                raise CrawlWarning("Failed to extract forum data")
 
-            elements = parse_ilias_forum_export(soupify(export))
+            if not download_data.empty:
+                html = await self._post_authenticated(download_data.url, download_data.form_data)
+                elements = parse_ilias_forum_export(soupify(html))
+            else:
+                elements = []
+
+        # Verify that ILIAS does not change the order, as we depend on it later. Otherwise, we could not call
+        # download in the correct order, potentially messing up duplication handling.
+        expected_element_titles = [thread.name for thread, download in forum_threads if download]
+        actual_element_titles = [_sanitize_path_name(thread.name) for thread in elements]
+        if expected_element_titles != actual_element_titles:
+            raise CrawlWarning(
+                f"Forum thread order mismatch: {expected_element_titles} != {actual_element_titles}"
+            )
 
         tasks: List[Awaitable[None]] = []
-        for thread in elements:
-            tasks.append(asyncio.create_task(self._download_forum_thread(cl.path, thread, element.url)))
+        for thread, download in forum_threads:
+            if download:
+                # This only works because ILIAS keeps the order in the export
+                elem = elements.pop(0)
+                tasks.append(asyncio.create_task(self._download_forum_thread(cl.path, elem, thread)))
+            else:
+                # We only downloaded the threads we "should_try_download"ed. This can be an
+                # over-approximation and all will be fine.
+                # If we selected too few, e.g. because there was a duplicate title and the mtime of the
+                # original is newer than the update of the duplicate.
+                # This causes stale data locally, but I consider this problem acceptable right now.
+                tasks.append(asyncio.create_task(self._download_forum_thread(cl.path, thread, thread)))
 
         # And execute them
         await self.gather(tasks)
@@ -801,7 +833,7 @@ instance's greatest bottleneck.
         self,
         parent_path: PurePath,
         thread: Union[IliasForumThread, IliasPageElement],
-        forum_url: str
+        element: IliasPageElement
     ) -> None:
         path = parent_path / (_sanitize_path_name(thread.name) + ".html")
         maybe_dl = await self.download(path, mtime=thread.mtime)
@@ -811,7 +843,7 @@ instance's greatest bottleneck.
         async with maybe_dl as (bar, sink):
             rendered = forum_thread_template(
                 thread.name,
-                forum_url,
+                element.url,
                 thread.name_tag,
                 await self.internalize_images(thread.content_tag)
             )
@@ -989,19 +1021,29 @@ instance's greatest bottleneck.
             )
         return soup
 
-    async def _post(
+    async def _post_authenticated(
         self,
         url: str,
         data: dict[str, Union[str, List[str]]]
     ) -> bytes:
+        auth_id = await self._current_auth_id()
+
         form_data = aiohttp.FormData()
         for key, val in data.items():
             form_data.add_field(key, val)
 
-        async with self.session.post(url, data=form_data()) as request:
+        async with self.session.post(url, data=form_data(), allow_redirects=False) as request:
             if request.status == 200:
                 return await request.read()
-            raise CrawlError(f"post failed with status {request.status}")
+
+        # We weren't authenticated, so try to do that
+        await self.authenticate(auth_id)
+
+        # Retry once after authenticating. If this fails, we will die.
+        async with self.session.post(url, data=data, allow_redirects=False) as request:
+            if request.status == 200:
+                return await request.read()
+        raise CrawlError("post_authenticated failed even after authenticating")
 
     async def _get_authenticated(self, url: str) -> bytes:
         auth_id = await self._current_auth_id()

PFERD/crawl/ilias/kit_ilias_html.py

@@ -271,7 +271,6 @@ class IliasPageElement:
             r"mcst/(?P<id>\d+)",  # mediacast
             r"pg/(?P<id>(\d|_)+)",  # page?
             r"svy/(?P<id>\d+)",  # survey
-            r"sess/(?P<id>\d+)",  # session
             r"webr/(?P<id>\d+)",  # web referene (link)
             r"thr_pk=(?P<id>\d+)",  # forums
             r"ref_id=(?P<id>\d+)",
@@ -490,31 +489,79 @@ class IliasPage:
             return url
         return None
 
-    def get_forum_export_url(self) -> Optional[str]:
-        forum_link = self._soup.select_one("#tab_forums_threads > a")
-        if not forum_link:
-            log.explain("Found no forum link")
+    def get_forum_entries(self) -> list[IliasPageElement]:
+        form = self._get_forum_form()
+        if not form:
+            return []
+        threads = []
+
+        for row in cast(list[Tag], form.select("table > tbody > tr")):
+            url_tag = cast(
+                Optional[Tag],
+                row.find(name="a", attrs={"href": lambda x: x is not None and "cmd=viewthread" in x.lower()})
+            )
+            if url_tag is None:
+                log.explain(f"Skipping row without URL: {row}")
+                continue
+            name = url_tag.get_text().strip()
+            columns = [td.get_text().strip() for td in cast(list[Tag], row.find_all(name="td"))]
+            potential_dates_opt = [IliasPage._find_date_in_text(column) for column in columns]
+            potential_dates = [x for x in potential_dates_opt if x is not None]
+            mtime = max(potential_dates) if potential_dates else None
+
+            threads.append(IliasPageElement.create_new(
+                IliasElementType.FORUM_THREAD,
+                self._abs_url_from_link(url_tag),
+                name,
+                mtime=mtime
+            ))
+
+        return threads
+
+    def get_download_forum_data(self, thread_ids: list[str]) -> Optional[IliasDownloadForumData]:
+        form = cast(Optional[Tag], self._soup.find(
+            "form",
+            attrs={"action": lambda x: x is not None and "fallbackCmd=showThreads" in x}
+        ))
+        if not form:
             return None
+        post_url = self._abs_url_from_relative(cast(str, form["action"]))
 
-        base_url = self._abs_url_from_link(forum_link)
-        base_url = re.sub(r"cmd=\w+", "cmd=post", base_url)
-        base_url = re.sub(r"cmdClass=\w+", "cmdClass=ilExportGUI", base_url)
+        log.explain(f"Fetching forum threads {thread_ids}")
 
-        rtoken_form = cast(
-            Optional[Tag],
-            self._soup.find("form", attrs={"action": lambda x: x is not None and "rtoken=" in x})
-        )
-        if not rtoken_form:
-            log.explain("Found no rtoken anywhere")
-            return None
-        match = cast(re.Match[str], re.search(r"rtoken=(\w+)", str(rtoken_form.attrs["action"])))
-        rtoken = match.group(1)
+        form_data: Dict[str, Union[str, list[str]]] = {
+            "thread_ids[]": cast(list[str], thread_ids),
+            "selected_cmd2": "html",
+            "select_cmd2": "Ausführen",
+            "selected_cmd": "",
+        }
 
-        base_url = base_url + "&rtoken=" + rtoken
+        return IliasDownloadForumData(url=post_url, form_data=form_data, empty=len(thread_ids) == 0)
 
-        return base_url
+    def _get_forum_form(self) -> Optional[Tag]:
+        return cast(Optional[Tag], self._soup.find(
+            "form",
+            attrs={"action": lambda x: x is not None and "fallbackCmd=showThreads" in x}
+        ))
 
     def get_next_stage_element(self) -> Optional[IliasPageElement]:
+        if self._is_forum_page():
+            if "trows=" in self._page_url:
+                log.explain("Manual row override detected, accepting it as good")
+                return None
+            log.explain("Requesting *all* forum threads")
+            thread_count = self._get_forum_thread_count()
+            if thread_count is not None and thread_count > 400:
+                log.warn(
+                    "Forum has more than 400 threads, fetching all threads will take a while. "
+                    "You might need to adjust your http_timeout config option."
+                )
+
+            # Fetch at least 400 in case we detect it wrong
+            if thread_count is not None and thread_count < 400:
+                thread_count = 400
+
+            return self._get_show_max_forum_entries_per_page_url(thread_count)
         if self._is_ilias_opencast_embedding():
             log.explain("Unwrapping opencast embedding")
             return self.get_child_elements()[0]
@@ -532,6 +579,11 @@ class IliasPage:
                 log.explain("Crawling info tab, skipping content select")
         return None
 
+    def _is_forum_page(self) -> bool:
+        if perma_link := self.get_permalink():
+            return "/frm/" in perma_link
+        return False
+
     def _is_video_player(self) -> bool:
         return "paella_config_file" in str(self._soup)
 
@@ -1075,7 +1127,7 @@ class IliasPage:
 
             videos.append(IliasPageElement.create_new(
                 typ=IliasElementType.MOB_VIDEO,
-                url=url,
+                url=self._abs_url_from_relative(url),
                 name=_sanitize_path_name(title),
                 mtime=None
             ))
@@ -1101,9 +1153,6 @@ class IliasPage:
         else:
             title = f"unknown video {figure}"
 
-        if url:
-            url = self._abs_url_from_relative(url)
-
         return url, title
 
     def _is_in_expanded_meeting(self, tag: Tag) -> bool:

PFERD/version.py

@@ -1,2 +1,2 @@
 NAME = "PFERD"
-VERSION = "3.8.1"
+VERSION = "3.8.0"