Add Pferd class

PFERD/__init__.py

@@ -7,6 +7,8 @@ more complex configuration, you need to import the other submodules manually.
 
 import logging
 
+from .pferd import Pferd
+
 STYLE = "{"
 FORMAT = "[{levelname:<7}] {message}"
 DATE_FORMAT = "%F %T"

PFERD/downloaders.py

@@ -3,7 +3,6 @@ General downloaders useful in many situations
 """
 
 from dataclasses import dataclass, field
-from pathlib import Path
 from typing import Any, Dict, List, Optional
 
 import requests
@@ -67,7 +66,7 @@ class HttpDownloader:
 
         with self._session.get(info.url, params=info.parameters, stream=True) as response:
             if response.status_code == 200:
-                tmp_file = self._tmp_dir.new_file()
+                tmp_file = self._tmp_dir.new_path()
                 stream_to_path(response, tmp_file)
                 self._organizer.accept_file(tmp_file, info.path)
             else:

PFERD/ilias/__init__.py

@@ -2,5 +2,6 @@
 Synchronizing files from ILIAS instances (https://www.ilias.de/).
 """
 
-from .authenticators import *
-from .downloader import *
+from .authenticators import IliasAuthenticator, KitShibbolethAuthenticator
+from .crawler import IliasCrawler, IliasFilter
+from .downloader import IliasDownloader

PFERD/ilias/authenticators.py

@@ -15,6 +15,9 @@ from ..utils import soupify
 LOGGER = logging.getLogger(__name__)
 
 
+# TODO save cookies whenever we know they're good
+
+
 class IliasAuthenticator(abc.ABC):
     # pylint: disable=too-few-public-methods
 

PFERD/ilias/crawler.py

@@ -7,11 +7,12 @@ import json
 import logging
 import re
 from pathlib import Path
-from typing import Any, Dict, List, Optional
+from typing import Any, Callable, Dict, List, Optional
 from urllib.parse import (parse_qs, urlencode, urljoin, urlparse, urlsplit,
                           urlunsplit)
 
 import bs4
+import requests
 
 from ..cookie_jar import CookieJar
 from ..utils import soupify
@@ -22,23 +23,36 @@ from .downloader import IliasDownloadInfo
 LOGGER = logging.getLogger(__name__)
 
 
+IliasFilter = Callable[[Path], bool]
+
+
 class IliasCrawler:
     # pylint: disable=too-few-public-methods
+
+    # TODO use the filter as appropriate
+    # TODO log the things that were discovered to the console on INFO
+
     """
     A crawler for ILIAS.
     """
 
-    def __init__(self, authenticator: IliasAuthenticator, base_url: str, course_id: str):
+    def __init__(
+            self,
+            base_url: str,
+            course_id: str,
+            session: requests.Session,
+            authenticator: IliasAuthenticator,
+            filter_: IliasFilter
+    ):
         """
         Create a new ILIAS crawler.
         """
-        self._cookie_jar = CookieJar(Path("/tmp/test/cookies"))
-        self._cookie_jar.load_cookies()
 
         self._base_url = base_url
         self._course_id = course_id
-        self._session = self._cookie_jar.create_session()
+        self._session = session
         self._authenticator = authenticator
+        self._filter = filter_
 
     def _abs_url_from_link(self, link_tag: bs4.Tag) -> str:
         """
@@ -342,8 +356,6 @@ class IliasCrawler:
 
         self._authenticator.authenticate(self._session)
 
-        self._cookie_jar.save_cookies("Authed")
-
         return self._get_page(url, params)
 
     @staticmethod
@@ -369,11 +381,3 @@ class IliasCrawler:
             LOGGER.debug("Auth: Found #playerContainer")
             return True
         return False
-
-
-def run_as_test(ilias_url: str, course_id: int) -> List[IliasDownloadInfo]:
-    from ..organizer import Organizer
-    from .authenticators import KitShibbolethAuthenticator
-
-    crawler = IliasCrawler(KitShibbolethAuthenticator(), ilias_url, str(course_id))
-    return crawler.crawl()

PFERD/ilias/downloader.py

@@ -33,12 +33,21 @@ class IliasDownloadInfo(Transformable):
 class IliasDownloader:
     """A downloader for ILIAS."""
 
-    def __init__(self, tmp_dir: TmpDir, organizer: Organizer, authenticator: IliasAuthenticator):
-        """Create a new IliasDownloader."""
-        self._authenticator = authenticator
-        self._session = requests.Session()
+    def __init__(
+            self,
+            tmp_dir: TmpDir,
+            organizer: Organizer,
+            session: requests.Session,
+            authenticator: IliasAuthenticator,
+    ):
+        """
+        Create a new IliasDownloader.
+        """
+
         self._tmp_dir = tmp_dir
         self._organizer = organizer
+        self._session = session
+        self._authenticator = authenticator
 
     def download_all(self, infos: List[IliasDownloadInfo]) -> None:
         """
@@ -55,7 +64,7 @@ class IliasDownloader:
         Retries authentication until eternity if it could not fetch the file.
         """
 
-        tmp_file = self._tmp_dir.new_file()
+        tmp_file = self._tmp_dir.new_path()
 
         while not self._try_download(info, tmp_file):
             self._authenticator.authenticate(self._session)

PFERD/pferd.py

@@ -0,0 +1,65 @@
+from pathlib import Path
+from typing import Optional
+
+from .cookie_jar import CookieJar
+from .ilias import (IliasAuthenticator, IliasCrawler, IliasDownloader,
+                    IliasFilter, KitShibbolethAuthenticator)
+from .organizer import Organizer
+from .tmp_dir import TmpDir
+from .transform import Transform, apply_transform
+from .utils import Location
+
+
+class Pferd(Location):
+    # pylint: disable=too-many-arguments
+
+    def __init__(self, base_dir: Path, tmp_dir: Path = Path(".tmp")):
+        super().__init__(Path(base_dir))
+
+        self._tmp_dir = TmpDir(self.resolve(tmp_dir))
+
+    def _ilias(
+            self,
+            target: Path,
+            base_url: str,
+            course_id: str,
+            authenticator: IliasAuthenticator,
+            cookies: Optional[Path],
+            filter_: IliasFilter,
+            transform: Transform,
+    ) -> None:
+        cookie_jar = CookieJar(cookies)
+        session = cookie_jar.create_session()
+        tmp_dir = self._tmp_dir.new_subdir()
+        organizer = Organizer(self.resolve(target))
+
+        crawler = IliasCrawler(base_url, course_id, session, authenticator, filter_)
+        downloader = IliasDownloader(tmp_dir, organizer, session, authenticator)
+
+        cookie_jar.load_cookies()
+        info = crawler.crawl()
+        cookie_jar.save_cookies()
+        downloader.download_all(apply_transform(transform, info))
+        cookie_jar.save_cookies()
+
+    def ilias_kit(
+            self,
+            target: Path,
+            course_id: str,
+            filter_: IliasFilter = lambda x: True,
+            transform: Transform = lambda x: x,
+            cookies: Optional[Path] = None,
+            username: Optional[str] = None,
+            password: Optional[str] = None,
+    ) -> None:
+        # This authenticator only works with the KIT ilias instance.
+        authenticator = KitShibbolethAuthenticator(username=username, password=password)
+        self._ilias(
+            target=target,
+            base_url="https://ilias.studium.kit.edu/",
+            course_id=course_id,
+            authenticator=authenticator,
+            cookies=cookies,
+            filter_=filter_,
+            transform=transform,
+        )

PFERD/tmp_dir.py

@@ -39,18 +39,24 @@ class TmpDir(Location):
         self.cleanup()
         return None
 
-    def new_file(self, prefix: Optional[str] = None) -> Path:
-        """Return a unique path inside the folder, but don't create a file."""
+    def new_path(self, prefix: Optional[str] = None) -> Path:
+        """
+        Return a unique path inside the directory. Doesn't create a file or
+        directory.
+        """
+
         name = f"{prefix if prefix else 'tmp'}-{self._inc_and_get_counter():03}"
 
         LOGGER.debug("Creating temp file %s", name)
 
         return self.resolve(Path(name))
 
-    def new_folder(self, prefix: Optional[str] = None) -> 'TmpDir':
-        """Create a new nested temporary folder and return its path."""
-        name = f"{prefix if prefix else 'tmp'}-{self._inc_and_get_counter():03}"
+    def new_subdir(self, prefix: Optional[str] = None) -> 'TmpDir':
+        """
+        Create a new nested temporary folder and return it.
+        """
 
+        name = f"{prefix if prefix else 'tmp'}-{self._inc_and_get_counter():03}"
         sub_path = self.resolve(Path(name))
         sub_path.mkdir(parents=True)