TODO

Thanks, burg113

.github/workflows/build-and-release.yml

@@ -17,9 +17,9 @@ jobs:
         python: ["3.9"]
     steps:
 
-      - uses: actions/checkout@v2
+      - uses: actions/checkout@v3
 
-      - uses: actions/setup-python@v2
+      - uses: actions/setup-python@v4
         with:
           python-version: ${{ matrix.python }}
 
@@ -45,7 +45,7 @@ jobs:
         run: mv dist/pferd* dist/pferd-${{ matrix.os }}
 
       - name: Upload binary
-        uses: actions/upload-artifact@v2
+        uses: actions/upload-artifact@v3
         with:
           name: Binaries
           path: dist/pferd-${{ matrix.os }}
@@ -57,7 +57,7 @@ jobs:
     steps:
 
       - name: Download binaries
-        uses: actions/download-artifact@v2
+        uses: actions/download-artifact@v3
         with:
           name: Binaries
 

.gitignore

@@ -2,7 +2,6 @@
 /.venv/
 /PFERD.egg-info/
 __pycache__/
-/.vscode/
 
 # pyinstaller
 /pferd.spec

.vscode/settings.json

@@ -0,0 +1,8 @@
+{
+    "files.insertFinalNewline": true,
+    "files.trimFinalNewlines": true,
+    "python.formatting.provider": "autopep8",
+    "python.linting.enabled": true,
+    "python.linting.flake8Enabled": true,
+    "python.linting.mypyEnabled": true,
+}

CHANGELOG.md

@@ -23,8 +23,34 @@ ambiguous situations.
 ## Unreleased
 
 ### Fixed
-- Forum crawling crashing when parsing empty (= 0 messages) threads
+- Crawling of courses with the timeline view as the default tab
+- Crawling of file and custom opencast cards
+- Crawling of button cards without descriptions
+
+## 3.4.3 - 2022-11-29
+
+### Added
+- Missing documentation for `forums` option
+
+### Changed
+- Clear up error message shown when multiple paths are found to an element
+
+### Fixed
+- IPD crawler unnecessarily appending trailing slashes
+- Crawling opencast when ILIAS is set to English
+
+## 3.4.2 - 2022-10-26
+
+### Added
+- Recognize and crawl content pages in cards
+- Recognize and ignore surveys
+
+### Fixed
+- Forum crawling crashing when a thread has no messages at all
 - Forum crawling crashing when a forum has no threads at all
+- Ilias login failing in some cases
+- Crawling of paginated future meetings
+- IPD crawler handling of URLs without trailing slash
 
 ## 3.4.1 - 2022-08-17
 

CONFIG.md

@@ -181,6 +181,7 @@ script once per day should be fine.
   redirect to the actual URL. Set to a negative value to disable the automatic
   redirect. (Default: `-1`)
 - `videos`: Whether to download videos. (Default: `no`)
+- `forums`: Whether to download forum threads. (Default: `no`)
 - `http_timeout`: The timeout (in seconds) for all HTTP requests. (Default:
   `20.0`)
 
@@ -289,7 +290,7 @@ path matches `SOURCE`, it is renamed to `TARGET`.
 Example: `foo/bar --> baz`
 - Doesn't match `foo`, `a/foo/bar` or `foo/baz`
 - Converts `foo/bar` into `baz`
-- Converts `foo/bar/wargl` into `bar/wargl`
+- Converts `foo/bar/wargl` into `baz/wargl`
 
 Example: `foo/bar --> !`
 - Doesn't match `foo`, `a/foo/bar` or `foo/baz`

PFERD/crawl/crawler.py

@@ -9,7 +9,6 @@ from typing import Any, Callable, Dict, List, Optional, Sequence, Set, Tuple, Ty
 from ..auth import Authenticator
 from ..config import Config, Section
 from ..deduplicator import Deduplicator
-from ..limiter import Limiter
 from ..logging import ProgressBar, log
 from ..output_dir import FileSink, FileSinkToken, OnConflict, OutputDirectory, OutputDirError, Redownload
 from ..report import MarkConflictError, MarkDuplicateError, Report
@@ -98,10 +97,9 @@ def anoncritical(f: AWrapped) -> AWrapped:
 
 
 class CrawlToken(ReusableAsyncContextManager[ProgressBar]):
-    def __init__(self, limiter: Limiter, path: PurePath):
+    def __init__(self, path: PurePath):
         super().__init__()
 
-        self._limiter = limiter
         self._path = path
 
     @property
@@ -110,17 +108,15 @@ class CrawlToken(ReusableAsyncContextManager[ProgressBar]):
 
     async def _on_aenter(self) -> ProgressBar:
         self._stack.callback(lambda: log.status("[bold cyan]", "Crawled", fmt_path(self._path)))
-        await self._stack.enter_async_context(self._limiter.limit_crawl())
         bar = self._stack.enter_context(log.crawl_bar("[bold bright_cyan]", "Crawling", fmt_path(self._path)))
 
         return bar
 
 
 class DownloadToken(ReusableAsyncContextManager[Tuple[ProgressBar, FileSink]]):
-    def __init__(self, limiter: Limiter, fs_token: FileSinkToken, path: PurePath):
+    def __init__(self, fs_token: FileSinkToken, path: PurePath):
         super().__init__()
 
-        self._limiter = limiter
         self._fs_token = fs_token
         self._path = path
 
@@ -129,7 +125,6 @@ class DownloadToken(ReusableAsyncContextManager[Tuple[ProgressBar, FileSink]]):
         return self._path
 
     async def _on_aenter(self) -> Tuple[ProgressBar, FileSink]:
-        await self._stack.enter_async_context(self._limiter.limit_download())
         sink = await self._stack.enter_async_context(self._fs_token)
         # The "Downloaded ..." message is printed in the output dir, not here
         bar = self._stack.enter_context(log.download_bar("[bold bright_cyan]", "Downloading",
@@ -235,12 +230,6 @@ class Crawler(ABC):
         self.name = name
         self.error_free = True
 
-        self._limiter = Limiter(
-            task_limit=section.tasks(),
-            download_limit=section.downloads(),
-            task_delay=section.task_delay(),
-        )
-
         self._deduplicator = Deduplicator(section.windows_paths())
         self._transformer = Transformer(section.transform())
 
@@ -288,7 +277,7 @@ class Crawler(ABC):
             return None
 
         log.explain("Answer: Yes")
-        return CrawlToken(self._limiter, path)
+        return CrawlToken(path)
 
     async def download(
             self,
@@ -313,7 +302,7 @@ class Crawler(ABC):
             return None
 
         log.explain("Answer: Yes")
-        return DownloadToken(self._limiter, fs_token, path)
+        return DownloadToken(fs_token, path)
 
     async def _cleanup(self) -> None:
         log.explain_topic("Decision: Clean up files")

PFERD/crawl/http_crawler.py

@@ -1,12 +1,9 @@
 import asyncio
-import http.cookies
-import ssl
+from http.cookiejar import LWPCookieJar
 from pathlib import Path, PurePath
-from typing import Any, Dict, List, Optional
+from typing import Dict, List, Optional
 
-import aiohttp
-import certifi
-from aiohttp.client import ClientTimeout
+import requests
 
 from ..auth import Authenticator
 from ..config import Config
@@ -35,9 +32,9 @@ class HttpCrawler(Crawler):
 
         self._authentication_id = 0
         self._authentication_lock = asyncio.Lock()
-        self._request_count = 0
-        self._http_timeout = section.http_timeout()
+        self._http_timeout = section.http_timeout()  # TODO Use or remove
 
+        self._cookie_jar = LWPCookieJar()
         self._cookie_jar_path = self._output_dir.resolve(self.COOKIE_FILE)
         self._shared_cookie_jar_paths: Optional[List[Path]] = None
         self._shared_auth = shared_auth
@@ -57,7 +54,6 @@ class HttpCrawler(Crawler):
         # This should reduce the amount of requests we make: If an authentication is in progress
         # all future requests wait for authentication to complete.
         async with self._authentication_lock:
-            self._request_count += 1
             return self._authentication_id
 
     async def authenticate(self, caller_auth_id: int) -> None:
@@ -106,32 +102,13 @@ class HttpCrawler(Crawler):
 
         self._shared_cookie_jar_paths.append(self._cookie_jar_path)
 
-    def _load_cookies_from_file(self, path: Path) -> None:
-        jar: Any = http.cookies.SimpleCookie()
-        with open(path, encoding="utf-8") as f:
-            for i, line in enumerate(f):
-                # Names of headers are case insensitive
-                if line[:11].lower() == "set-cookie:":
-                    jar.load(line[11:])
-                else:
-                    log.explain(f"Line {i} doesn't start with 'Set-Cookie:', ignoring it")
-        self._cookie_jar.update_cookies(jar)
-
-    def _save_cookies_to_file(self, path: Path) -> None:
-        jar: Any = http.cookies.SimpleCookie()
-        for morsel in self._cookie_jar:
-            jar[morsel.key] = morsel
-        with open(path, "w", encoding="utf-8") as f:
-            f.write(jar.output(sep="\n"))
-            f.write("\n")  # A trailing newline is just common courtesy
-
     def _load_cookies(self) -> None:
         log.explain_topic("Loading cookies")
 
         cookie_jar_path: Optional[Path] = None
 
         if self._shared_cookie_jar_paths is None:
-            log.explain("Not sharing any cookies")
+            log.explain("Not sharing cookies")
             cookie_jar_path = self._cookie_jar_path
         else:
             log.explain("Sharing cookies")
@@ -154,46 +131,38 @@ class HttpCrawler(Crawler):
 
         log.explain(f"Loading cookies from {fmt_real_path(cookie_jar_path)}")
         try:
-            self._load_cookies_from_file(cookie_jar_path)
+            self._cookie_jar.load(filename=str(cookie_jar_path))
         except Exception as e:
-            log.explain("Failed to load cookies")
-            log.explain(str(e))
+            log.explain(f"Failed to load cookies: {e}")
+            log.explain("Proceeding without cookies")
 
     def _save_cookies(self) -> None:
         log.explain_topic("Saving cookies")
 
         try:
             log.explain(f"Saving cookies to {fmt_real_path(self._cookie_jar_path)}")
-            self._save_cookies_to_file(self._cookie_jar_path)
+            self._cookie_jar.save(filename=str(self._cookie_jar_path))
         except Exception as e:
-            log.warn(f"Failed to save cookies to {fmt_real_path(self._cookie_jar_path)}")
-            log.warn(str(e))
+            log.warn(f"Failed to save cookies: {e}")
 
     async def run(self) -> None:
         self._request_count = 0
-        self._cookie_jar = aiohttp.CookieJar()
         self._load_cookies()
 
-        async with aiohttp.ClientSession(
-                headers={"User-Agent": f"{NAME}/{VERSION}"},
-                cookie_jar=self._cookie_jar,
-                connector=aiohttp.TCPConnector(ssl=ssl.create_default_context(cafile=certifi.where())),
-                timeout=ClientTimeout(
-                    # 30 minutes. No download in the history of downloads was longer than 30 minutes.
-                    # This is enough to transfer a 600 MB file over a 3 Mib/s connection.
-                    # Allowing an arbitrary value could be annoying for overnight batch jobs
-                    total=15 * 60,
-                    connect=self._http_timeout,
-                    sock_connect=self._http_timeout,
-                    sock_read=self._http_timeout,
-                )
-        ) as session:
-            self.session = session
+        self.session = requests.Session()
+        self.session.headers["User-Agent"] = f"{NAME}/{VERSION}"
+
+        # From the request docs: "All requests code should work out of the box
+        # with externally provided instances of CookieJar, e.g. LWPCookieJar and
+        # FileCookieJar."
+        # https://requests.readthedocs.io/en/latest/api/#requests.cookies.RequestsCookieJar
+        self.session.cookies = self._cookie_jar  # type: ignore
+
+        with self.session:
             try:
                 await super().run()
             finally:
                 del self.session
-        log.explain_topic(f"Total amount of HTTP requests: {self._request_count}")
 
         # They are saved in authenticate, but a final save won't hurt
         self._save_cookies()

PFERD/crawl/ilias/kit_ilias_html.py

@@ -24,6 +24,7 @@ class IliasElementType(Enum):
     LINK = "link"
     BOOKING = "booking"
     MEETING = "meeting"
+    SURVEY = "survey"
     VIDEO = "video"
     VIDEO_PLAYER = "video_player"
     VIDEO_FOLDER = "video_folder"
@@ -133,7 +134,7 @@ class IliasPage:
 
         thread_ids = [f["value"] for f in form.find_all(attrs={"name": "thread_ids[]"})]
 
-        form_data: Dict[str, Union[str, List[ſtr]]] = {
+        form_data: Dict[str, Union[str, List[str]]] = {
             "thread_ids[]": thread_ids,
             "selected_cmd2": "html",
             "select_cmd2": "Ausführen",
@@ -157,6 +158,8 @@ class IliasPage:
         if self._contains_collapsed_future_meetings():
             log.explain("Requesting *all* future meetings")
             return self._uncollapse_future_meetings_url()
+        if not self._is_content_tab_selected():
+            return self._select_content_page_url()
         return None
 
     def _is_forum_page(self) -> bool:
@@ -219,6 +222,27 @@ class IliasPage:
         link = self._abs_url_from_link(element)
         return IliasPageElement(IliasElementType.FOLDER, link, "show all meetings")
 
+    def _is_content_tab_selected(self) -> bool:
+        return self._select_content_page_url() is None
+
+    def _select_content_page_url(self) -> Optional[IliasPageElement]:
+        tab = self._soup.find(
+            id="tab_view_content",
+            attrs={"class": lambda x: x is not None and "active" not in x}
+        )
+        # Already selected (or not found)
+        if not tab:
+            return None
+        link = tab.find("a")
+        if link:
+            link = self._abs_url_from_link(link)
+            return IliasPageElement(IliasElementType.FOLDER, link, "select content page")
+
+        _unexpected_html_warning()
+        log.warn_contd(f"Could not find content tab URL on {self._page_url!r}.")
+        log.warn_contd("PFERD might not find content on the course's main page.")
+        return None
+
     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
@@ -365,7 +389,7 @@ class IliasPage:
         """
         # Video start links are marked with an "Abspielen" link
         video_links: List[Tag] = self._soup.findAll(
-            name="a", text=re.compile(r"\s*Abspielen\s*")
+            name="a", text=re.compile(r"\s*(Abspielen|Play)\s*")
         )
 
         results: List[IliasPageElement] = []
@@ -684,7 +708,11 @@ class IliasPage:
                 "div",
                 attrs={"class": lambda x: x and "caption" in x},
             )
-            description = caption_parent.find_next_sibling("div").getText().strip()
+            caption_container = caption_parent.find_next_sibling("div")
+            if caption_container:
+                description = caption_container.getText().strip()
+            else:
+                description = None
 
             if not type:
                 _unexpected_html_warning()
@@ -714,7 +742,7 @@ class IliasPage:
 
         icon: Tag = card_root.select_one(".il-card-repository-head .icon")
 
-        if "opencast" in icon["class"]:
+        if "opencast" in icon["class"] or "xoct" in icon["class"]:
             return IliasElementType.VIDEO_FOLDER_MAYBE_PAGINATED
         if "exc" in icon["class"]:
             return IliasElementType.EXERCISE
@@ -730,6 +758,12 @@ class IliasPage:
             return IliasElementType.TEST
         if "fold" in icon["class"]:
             return IliasElementType.FOLDER
+        if "copa" in icon["class"]:
+            return IliasElementType.FOLDER
+        if "svy" in icon["class"]:
+            return IliasElementType.SURVEY
+        if "file" in icon["class"]:
+            return IliasElementType.FILE
 
         _unexpected_html_warning()
         log.warn_contd(f"Could not extract type from {icon} for card title {card_title}")

PFERD/crawl/ilias/kit_ilias_web_crawler.py

@@ -126,13 +126,6 @@ def _iorepeat(attempts: int, name: str, failure_is_error: bool = False) -> Calla
     return decorator
 
 
-def _wrap_io_in_warning(name: str) -> Callable[[AWrapped], AWrapped]:
-    """
-    Wraps any I/O exception in a CrawlWarning.
-    """
-    return _iorepeat(1, name)
-
-
 # Crawler control flow:
 #
 #     crawl_desktop -+
@@ -194,7 +187,7 @@ instance's greatest bottleneck.
         self._links = section.links()
         self._videos = section.videos()
         self._forums = section.forums()
-        self._visited_urls: Set[str] = set()
+        self._visited_urls: Dict[str, PurePath] = dict()
 
     async def _run(self) -> None:
         if isinstance(self._target, int):
@@ -226,114 +219,45 @@ instance's greatest bottleneck.
             return
         cl = maybe_cl  # Not mypy's fault, but explained here: https://github.com/python/mypy/issues/2608
 
-        elements: List[IliasPageElement] = []
-        # A list as variable redefinitions are not propagated to outer scopes
-        description: List[BeautifulSoup] = []
+        def ensure_is_valid_course_id(parent: Optional[IliasPageElement], soup: BeautifulSoup) -> None:
+            if 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")
 
-        @_iorepeat(3, "crawling url")
-        async def gather_elements() -> None:
-            elements.clear()
-            async with cl:
-                next_stage_url: Optional[str] = url
-                current_parent = None
-
-                # Duplicated code, but the root page is special - we want to avoid fetching it twice!
-                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)
-
-        # Fill up our task list with the found elements
-        await gather_elements()
-
-        if description:
-            await self._download_description(PurePath("."), description[0])
-
-        elements.sort(key=lambda e: e.id())
-
-        tasks: List[Awaitable[None]] = []
-        for element in elements:
-            if handle := await self._handle_ilias_element(PurePath("."), element):
-                tasks.append(asyncio.create_task(handle))
-
-        # And execute them
-        await self.gather(tasks)
-
-    async def _handle_ilias_page(
-        self,
-        url: str,
-        parent: IliasPageElement,
-        path: PurePath,
-    ) -> Optional[Coroutine[Any, Any, None]]:
-        maybe_cl = await self.crawl(path)
-        if not maybe_cl:
-            return None
-        return self._crawl_ilias_page(url, parent, maybe_cl)
+        await self._crawl_ilias_page(url, None, cl, ensure_is_valid_course_id)
 
     @anoncritical
     async def _crawl_ilias_page(
         self,
         url: str,
-        parent: IliasPageElement,
+        parent: Optional[IliasPageElement],
         cl: CrawlToken,
+        next_stage_hook: Callable[[Optional[IliasPageElement], BeautifulSoup], None] = lambda a, b: None
     ) -> None:
-        elements: List[IliasPageElement] = []
-        # A list as variable redefinitions are not propagated to outer scopes
-        description: List[BeautifulSoup] = []
+        async with cl:
+            next_stage_url: Optional[str] = url
+            current_parent = parent
 
-        @_iorepeat(3, "crawling folder")
-        async def gather_elements() -> None:
-            elements.clear()
-            async with cl:
-                next_stage_url: Optional[str] = url
-                current_parent = parent
+            while next_stage_url:
+                soup = await self._get_page(next_stage_url)
+                log.explain_topic(f"Parsing HTML page for {fmt_path(cl.path)}")
+                log.explain(f"URL: {next_stage_url}")
 
-                while next_stage_url:
-                    soup = await self._get_page(next_stage_url)
-                    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
+                next_stage_hook(current_parent, soup)
 
-                elements.extend(page.get_child_elements())
-                if description_string := page.get_description():
-                    description.append(description_string)
+                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
 
-        # Fill up our task list with the found elements
-        await gather_elements()
+        for element in sorted(page.get_child_elements(), key=lambda e: e.id()):
+            await self._handle_ilias_element(cl.path, element)
 
-        if description:
-            await self._download_description(cl.path, description[0])
-
-        elements.sort(key=lambda e: e.id())
-
-        tasks: List[Awaitable[None]] = []
-        for element in elements:
-            if handle := await self._handle_ilias_element(cl.path, element):
-                tasks.append(asyncio.create_task(handle))
-
-        # And execute them
-        await self.gather(tasks)
+        if description_string := page.get_description():
+            await self._download_description(cl.path, description_string)
 
     # These decorators only apply *to this method* and *NOT* to the returned
     # awaitables!
@@ -345,12 +269,14 @@ instance's greatest bottleneck.
         self,
         parent_path: PurePath,
         element: IliasPageElement,
-    ) -> Optional[Coroutine[Any, Any, None]]:
+    ) -> None:
         if element.url in self._visited_urls:
             raise CrawlWarning(
-                f"Found second path to element {element.name!r} at {element.url!r}. Aborting subpath"
+                f"Found second path to element {element.name!r} at {element.url!r}. "
+                + f"First path: {fmt_path(self._visited_urls[element.url])}. "
+                + f"Second path: {fmt_path(parent_path)}."
             )
-        self._visited_urls.add(element.url)
+        self._visited_urls[element.url] = parent_path
 
         element_path = PurePath(parent_path, element.name)
 
@@ -365,7 +291,7 @@ instance's greatest bottleneck.
                 return None
 
         if element.type == IliasElementType.FILE:
-            return await self._handle_file(element, element_path)
+            await self._handle_file(element, element_path)
         elif element.type == IliasElementType.FORUM:
             if not self._forums:
                 log.status(
@@ -375,22 +301,36 @@ instance's greatest bottleneck.
                     "[bright_black](enable with option 'forums')"
                 )
                 return None
-            return await self._handle_forum(element, element_path)
+            await self._handle_forum(element, element_path)
         elif element.type == IliasElementType.TEST:
-            log.explain_topic(f"Decision: Crawl {fmt_path(element_path)}")
-            log.explain("Tests contain no relevant files")
-            log.explain("Answer: No")
+            log.status(
+                "[bold bright_black]",
+                "Ignored",
+                fmt_path(element_path),
+                "[bright_black](tests contain no relevant data)"
+            )
+            return None
+        elif element.type == IliasElementType.SURVEY:
+            log.status(
+                "[bold bright_black]",
+                "Ignored",
+                fmt_path(element_path),
+                "[bright_black](surveys contain no relevant data)"
+            )
             return None
         elif element.type == IliasElementType.LINK:
-            return await self._handle_link(element, element_path)
+            await self._handle_link(element, element_path)
         elif element.type == IliasElementType.BOOKING:
-            return await self._handle_booking(element, element_path)
+            await self._handle_booking(element, element_path)
         elif element.type == IliasElementType.VIDEO:
-            return await self._handle_file(element, element_path)
+            await self._handle_file(element, element_path)
         elif element.type == IliasElementType.VIDEO_PLAYER:
-            return await self._handle_video(element, element_path)
+            await self._handle_video(element, element_path)
         elif element.type in _DIRECTORY_PAGES:
-            return await self._handle_ilias_page(element.url, element, element_path)
+            maybe_cl = await self.crawl(element_path)
+            if not maybe_cl:
+                return None
+            await self._crawl_ilias_page(element.url, element, maybe_cl)
         else:
             # This will retry it a few times, failing everytime. It doesn't make any network
             # requests, so that's fine.
@@ -400,7 +340,7 @@ instance's greatest bottleneck.
         self,
         element: IliasPageElement,
         element_path: PurePath,
-    ) -> Optional[Coroutine[Any, Any, None]]:
+    ) -> None:
         log.explain_topic(f"Decision: Crawl Link {fmt_path(element_path)}")
         log.explain(f"Links type is {self._links}")
 
@@ -417,7 +357,7 @@ instance's greatest bottleneck.
         if not maybe_dl:
             return None
 
-        return self._download_link(element, link_template_maybe, maybe_dl)
+        await self._download_link(element, link_template_maybe, maybe_dl)
 
     @anoncritical
     @_iorepeat(3, "resolving link")
@@ -509,7 +449,7 @@ instance's greatest bottleneck.
         self,
         element: IliasPageElement,
         element_path: PurePath,
-    ) -> Optional[Coroutine[Any, Any, None]]:
+    ) -> None:
         # Copy old mapping as it is likely still relevant
         if self.prev_report:
             self.report.add_custom_value(
@@ -535,7 +475,7 @@ instance's greatest bottleneck.
 
             return None
 
-        return self._download_video(element_path, element, maybe_dl)
+        await self._download_video(element_path, element, maybe_dl)
 
     def _previous_contained_videos(self, video_path: PurePath) -> List[PurePath]:
         if not self.prev_report:
@@ -617,11 +557,11 @@ instance's greatest bottleneck.
         self,
         element: IliasPageElement,
         element_path: PurePath,
-    ) -> Optional[Coroutine[Any, Any, None]]:
+    ) -> None:
         maybe_dl = await self.download(element_path, mtime=element.mtime)
         if not maybe_dl:
             return None
-        return self._download_file(element, maybe_dl)
+        await self._download_file(element, maybe_dl)
 
     @anoncritical
     @_iorepeat(3, "downloading file")
@@ -664,11 +604,11 @@ instance's greatest bottleneck.
         self,
         element: IliasPageElement,
         element_path: PurePath,
-    ) -> Optional[Coroutine[Any, Any, None]]:
+    ) -> None:
         maybe_cl = await self.crawl(element_path)
         if not maybe_cl:
             return None
-        return self._crawl_forum(element, maybe_cl)
+        await self._crawl_forum(element, maybe_cl)
 
     @_iorepeat(3, "crawling forum")
     @anoncritical

PFERD/crawl/kit_ipd_crawler.py

@@ -2,7 +2,7 @@ import os
 import re
 from dataclasses import dataclass
 from pathlib import PurePath
-from typing import Awaitable, List, Optional, Pattern, Set, Union
+from typing import List, Optional, Pattern, Set, Tuple, Union
 from urllib.parse import urljoin
 
 from bs4 import BeautifulSoup, Tag
@@ -64,67 +64,62 @@ class KitIpdCrawler(HttpCrawler):
         self._file_regex = section.link_regex()
 
     async def _run(self) -> None:
-        maybe_cl = await self.crawl(PurePath("."))
-        if not maybe_cl:
+        cl = await self.crawl(PurePath("."))
+        if not cl:
             return
 
-        tasks: List[Awaitable[None]] = []
-
-        async with maybe_cl:
+        async with cl:
             for item in await self._fetch_items():
                 if isinstance(item, KitIpdFolder):
-                    tasks.append(self._crawl_folder(item))
+                    await self._crawl_folder(item)
                 else:
                     # Orphan files are placed in the root folder
-                    tasks.append(self._download_file(PurePath("."), item))
-
-        await self.gather(tasks)
+                    await self._download_file(PurePath("."), item)
 
     async def _crawl_folder(self, folder: KitIpdFolder) -> None:
         path = PurePath(folder.name)
         if not await self.crawl(path):
             return
 
-        tasks = [self._download_file(path, file) for file in folder.files]
-
-        await self.gather(tasks)
+        for file in folder.files:
+            await self._download_file(path, file)
 
     async def _download_file(self, parent: PurePath, file: KitIpdFile) -> None:
         element_path = parent / file.name
-        maybe_dl = await self.download(element_path)
-        if not maybe_dl:
+        dl = await self.download(element_path)
+        if not dl:
             return
 
-        async with maybe_dl as (bar, sink):
+        async with dl as (bar, sink):
             await self._stream_from_url(file.url, sink, bar)
 
     async def _fetch_items(self) -> Set[Union[KitIpdFile, KitIpdFolder]]:
-        page = await self.get_page()
+        page, url = await self._get_page()
         elements: List[Tag] = self._find_file_links(page)
         items: Set[Union[KitIpdFile, KitIpdFolder]] = set()
 
         for element in elements:
             folder_label = self._find_folder_label(element)
             if folder_label:
-                folder = self._extract_folder(folder_label)
+                folder = self._extract_folder(folder_label, url)
                 if folder not in items:
                     items.add(folder)
                     folder.explain()
             else:
-                file = self._extract_file(element)
+                file = self._extract_file(element, url)
                 items.add(file)
                 log.explain_topic(f"Orphan file {file.name!r} (href={file.url!r})")
                 log.explain("Attributing it to root folder")
 
         return items
 
-    def _extract_folder(self, folder_tag: Tag) -> KitIpdFolder:
+    def _extract_folder(self, folder_tag: Tag, url: str) -> KitIpdFolder:
         files: List[KitIpdFile] = []
         name = folder_tag.getText().strip()
 
         container: Tag = folder_tag.findNextSibling(name="table")
         for link in self._find_file_links(container):
-            files.append(self._extract_file(link))
+            files.append(self._extract_file(link, url))
 
         return KitIpdFolder(name, files)
 
@@ -135,16 +130,16 @@ class KitIpdCrawler(HttpCrawler):
             return None
         return enclosing_table.findPreviousSibling(name=re.compile("^h[1-6]$"))
 
-    def _extract_file(self, link: Tag) -> KitIpdFile:
-        url = self._abs_url_from_link(link)
+    def _extract_file(self, link: Tag, url: str) -> KitIpdFile:
+        url = self._abs_url_from_link(url, link)
         name = os.path.basename(url)
         return KitIpdFile(name, url)
 
     def _find_file_links(self, tag: Union[Tag, BeautifulSoup]) -> List[Tag]:
         return tag.findAll(name="a", attrs={"href": self._file_regex})
 
-    def _abs_url_from_link(self, link_tag: Tag) -> str:
-        return urljoin(self._url, link_tag.get("href"))
+    def _abs_url_from_link(self, url: str, link_tag: Tag) -> str:
+        return urljoin(url, link_tag.get("href"))
 
     async def _stream_from_url(self, url: str, sink: FileSink, bar: ProgressBar) -> None:
         async with self.session.get(url, allow_redirects=False) as resp:
@@ -159,12 +154,12 @@ class KitIpdCrawler(HttpCrawler):
 
             sink.done()
 
-    async def get_page(self) -> BeautifulSoup:
-        async with self.session.get(self._url) as request:
-            # The web page for Algorithmen für Routenplanung contains some
-            # weird comments that beautifulsoup doesn't parse correctly. This
-            # hack enables those pages to be crawled, and should hopefully not
-            # cause issues on other pages.
-            content = (await request.read()).decode("utf-8")
-            content = re.sub(r"<!--.*?-->", "", content)
-            return soupify(content.encode("utf-8"))
+    async def _get_page(self) -> Tuple[BeautifulSoup, str]:
+        response = self.session.get(self._url)
+
+        # The web page for Algorithmen für Routenplanung contains some
+        # weird comments that beautifulsoup doesn't parse correctly. This
+        # hack enables those pages to be crawled, and should hopefully not
+        # cause issues on other pages.
+        content = re.sub(r"<!--.*?-->", "", response.text)
+        return soupify(content.encode("utf-8")), str(request.url)

PFERD/crawl/local_crawler.py

@@ -71,8 +71,6 @@ class LocalCrawler(Crawler):
         if not cl:
             return
 
-        tasks = []
-
         async with cl:
             await asyncio.sleep(random.uniform(
                 0.5 * self._crawl_delay,
@@ -81,9 +79,7 @@ class LocalCrawler(Crawler):
 
             for child in path.iterdir():
                 pure_child = cl.path / child.name
-                tasks.append(self._crawl_path(child, pure_child))
-
-        await self.gather(tasks)
+                await self._crawl_path(child, pure_child)
 
     async def _crawl_file(self, path: Path, pure: PurePath) -> None:
         stat = path.stat()

PFERD/limiter.py

@@ -1,97 +0,0 @@
-import asyncio
-import time
-from contextlib import asynccontextmanager
-from dataclasses import dataclass
-from typing import AsyncIterator, Optional
-
-
-@dataclass
-class Slot:
-    active: bool = False
-    last_left: Optional[float] = None
-
-
-class Limiter:
-    def __init__(
-            self,
-            task_limit: int,
-            download_limit: int,
-            task_delay: float
-    ):
-        if task_limit <= 0:
-            raise ValueError("task limit must be at least 1")
-        if download_limit <= 0:
-            raise ValueError("download limit must be at least 1")
-        if download_limit > task_limit:
-            raise ValueError("download limit can't be greater than task limit")
-        if task_delay < 0:
-            raise ValueError("Task delay must not be negative")
-
-        self._slots = [Slot() for _ in range(task_limit)]
-        self._downloads = download_limit
-        self._delay = task_delay
-
-        self._condition = asyncio.Condition()
-
-    def _acquire_slot(self) -> Optional[Slot]:
-        for slot in self._slots:
-            if not slot.active:
-                slot.active = True
-                return slot
-
-        return None
-
-    async def _wait_for_slot_delay(self, slot: Slot) -> None:
-        if slot.last_left is not None:
-            delay = slot.last_left + self._delay - time.time()
-            if delay > 0:
-                await asyncio.sleep(delay)
-
-    def _release_slot(self, slot: Slot) -> None:
-        slot.last_left = time.time()
-        slot.active = False
-
-    @asynccontextmanager
-    async def limit_crawl(self) -> AsyncIterator[None]:
-        slot: Slot
-        async with self._condition:
-            while True:
-                if found_slot := self._acquire_slot():
-                    slot = found_slot
-                    break
-                await self._condition.wait()
-
-        await self._wait_for_slot_delay(slot)
-
-        try:
-            yield
-        finally:
-            async with self._condition:
-                self._release_slot(slot)
-                self._condition.notify_all()
-
-    @asynccontextmanager
-    async def limit_download(self) -> AsyncIterator[None]:
-        slot: Slot
-        async with self._condition:
-            while True:
-                if self._downloads <= 0:
-                    await self._condition.wait()
-                    continue
-
-                if found_slot := self._acquire_slot():
-                    slot = found_slot
-                    self._downloads -= 1
-                    break
-
-                await self._condition.wait()
-
-        await self._wait_for_slot_delay(slot)
-
-        try:
-            yield
-        finally:
-            async with self._condition:
-                self._release_slot(slot)
-                self._downloads += 1
-                self._condition.notify_all()

PFERD/utils.py

@@ -92,17 +92,32 @@ def url_set_query_params(url: str, params: Dict[str, str]) -> str:
 
 
 def str_path(path: PurePath) -> str:
+    """
+    Turn a path into a string, in a platform-independent way.
+
+    This function always uses "/" as path separator, even on Windows.
+    """
     if not path.parts:
         return "."
     return "/".join(path.parts)
 
 
 def fmt_path(path: PurePath) -> str:
+    """
+    Turn a path into a delimited string.
+
+    This is useful if file or directory names contain weird characters like
+    newlines, leading/trailing whitespace or unprintable characters. This way,
+    they are escaped and visible to the user.
+    """
     return repr(str_path(path))
 
 
 def fmt_real_path(path: Path) -> str:
-    return repr(str(path.absolute()))
+    """
+    Like fmt_path, but resolves the path before converting it to a string.
+    """
+    return fmt_path(path.absolute())
 
 
 class ReusableAsyncContextManager(ABC, Generic[T]):

PFERD/version.py

@@ -1,2 +1,2 @@
 NAME = "PFERD"
-VERSION = "3.4.1"
+VERSION = "3.4.3"

README.md

@@ -30,7 +30,10 @@ The use of [venv](https://docs.python.org/3/library/venv.html) is recommended.
 
 Unofficial packages are available for:
 - [AUR](https://aur.archlinux.org/packages/pferd)
+- [brew](https://formulae.brew.sh/formula/pferd)
+- [conda-forge](https://github.com/conda-forge/pferd-feedstock)
 - [nixpkgs](https://github.com/NixOS/nixpkgs/blob/master/pkgs/tools/misc/pferd/default.nix)
+- [PyPi](https://pypi.org/project/pferd)
 
 See also PFERD's [repology page](https://repology.org/project/pferd/versions).
 

scripts/setup

@@ -14,4 +14,4 @@ pip install --editable .
 
 # Installing tools and type hints
 pip install --upgrade mypy flake8 autopep8 isort pyinstaller
-pip install --upgrade types-chardet types-certifi
+mypy PFERD --install-types --non-interactive

setup.cfg

@@ -11,6 +11,7 @@ install_requires =
   rich>=11.0.0
   keyring>=23.5.0
   certifi>=2021.10.8
+  requests>=2.28.1
 
 [options.entry_points]
 console_scripts =