Compare commits

...

13 Commits

Author SHA1 Message Date
52852d11a6 Bump version to 1.1.8 2019-09-22 11:56:41 +00:00
f94629a7fa Fix exceptions with weird content types
(hopefully)
2019-09-22 11:55:47 +00:00
c8ee456d33 Bump version to 1.1.7 2019-07-26 08:14:55 +00:00
2752e98621 Fix relative url joining in ti downloader 2019-07-26 10:06:01 +02:00
1572e11da8 Bump version to 1.1.6 2019-07-05 08:49:26 +00:00
ea01dc7cb2 Allow even more types of files 2019-07-05 08:48:43 +00:00
aba8d46d26 Bump version to 1.1.5 2019-07-04 12:17:33 +00:00
77056e6f8d Allow more types of files 2019-07-04 12:16:42 +00:00
064f12c14c Ignore mypy files 2019-07-04 12:16:26 +00:00
2eb834afc3 Bump version to 1.1.4 2019-06-11 12:46:40 +00:00
d468a45662 Allow wolfram files 2019-06-11 12:42:55 +00:00
50e25346e5 Bump version to 1.1.3 2019-06-07 11:36:41 +00:00
67da4e69fa Add colorful log output
Highlight the important operations (new, modified) in different colours.
2019-06-07 13:28:55 +02:00
10 changed files with 59 additions and 35 deletions

1
.gitignore vendored
View File

@ -10,3 +10,4 @@ pyvenv.cfg
.tmp/
pip-selfcheck.json
.mypy_cache/

View File

@ -8,10 +8,11 @@ import bs4
import requests
from .organizer import Organizer
from .utils import stream_to_path
from .utils import stream_to_path, PrettyLogger
__all__ = ["FfM"]
logger = logging.getLogger(__name__)
pretty = PrettyLogger(logger)
class FfM:
BASE_URL = "http://www.math.kit.edu/"
@ -23,7 +24,7 @@ class FfM:
self._session = requests.Session()
def synchronize(self, urlpart, to_dir, transform=lambda x: x):
logger.info(f" Synchronizing {urlpart} to {to_dir} using the FfM synchronizer.")
pretty.starting_synchronizer(to_dir, "FfM", urlpart)
sync_path = pathlib.Path(self.base_path, to_dir)

View File

@ -4,13 +4,13 @@ import logging
import pathlib
import re
import bs4
from .ilias_authenticators import ShibbolethAuthenticator
from .organizer import Organizer
from .utils import PrettyLogger
__all__ = ["Ilias"]
logger = logging.getLogger(__name__)
pretty = PrettyLogger(logger)
class Ilias:
FILE_RE = re.compile(r"goto\.php\?target=(file_\d+_download)")
@ -22,7 +22,7 @@ class Ilias:
self._auth = ShibbolethAuthenticator(base_path / cookie_file)
def synchronize(self, ref_id, to_dir, transform=lambda x: x, filter=lambda x: True):
logger.info(f" Synchronizing ref_id {ref_id} to {to_dir} using the Ilias synchronizer.")
pretty.starting_synchronizer(to_dir, "ILIAS", f"ref_id {ref_id}")
sync_path = pathlib.Path(self.base_path, to_dir)
orga = Organizer(self.base_path, sync_path)

View File

@ -23,16 +23,6 @@ logger = logging.getLogger(__name__)
class ShibbolethAuthenticator:
ILIAS_GOTO = "https://ilias.studium.kit.edu/goto.php"
ALLOWED_CONTENT_TYPES = [
"application/pdf",
"application/zip",
"application/msword",
"text/xml",
"text/plain",
"image/jpeg",
"image/png",
]
def __init__(self, cookie_file) -> None:
# Because LWPCookieJar insists on the path being str-like instead of
# Path-like.
@ -147,11 +137,7 @@ class ShibbolethAuthenticator:
with self._session.get(url, params=params, stream=True) as r:
content_type = r.headers["content-type"]
if content_type in self.ALLOWED_CONTENT_TYPES:
# Yay, we got the file :)
stream_to_path(r, to_path)
return True
elif content_type == "text/html":
if content_type.startswith("text/html"):
# Dangit, we're probably not logged in.
soup = bs4.BeautifulSoup(r.text, "html.parser")
if self._is_logged_in(soup):
@ -159,9 +145,9 @@ class ShibbolethAuthenticator:
"Attempting to download a web page, not a file")
return False
else:
# What *did* we get?
raise ContentTypeException(
f"Unknown file of type {content_type}")
# Yay, we got the file :)
stream_to_path(r, to_path)
return True
def download_file(self, file_id, to_path):
params = {"target": file_id}

View File

@ -9,10 +9,11 @@ import bs4
import requests
from .organizer import Organizer
from .utils import rename, stream_to_path
from .utils import rename, stream_to_path, PrettyLogger
__all__ = ["Norbert"]
logger = logging.getLogger(__name__)
pretty = PrettyLogger(logger)
class Norbert:
BASE_URL = "https://studwww.informatik.kit.edu/~s_blueml/"
@ -24,7 +25,7 @@ class Norbert:
self._session = requests.Session()
def synchronize(self, to_dir, transform=lambda x: x, unzip=lambda _: True):
logger.info(f" Synchronizing to {to_dir} using the Norbert synchronizer.")
pretty.starting_synchronizer(to_dir, "Norbert")
sync_path = pathlib.Path(self.base_path, to_dir)
orga = Organizer(self.base_path, sync_path)

View File

@ -7,6 +7,7 @@ from . import utils
__all__ = ["Organizer"]
logger = logging.getLogger(__name__)
pretty = utils.PrettyLogger(logger)
class Organizer:
def __init__(self, base_dir, sync_dir):
@ -64,7 +65,7 @@ class Organizer:
if to_path.exists():
if filecmp.cmp(from_path, to_path, shallow=False):
logger.info(f"Ignored {to_path}")
pretty.ignored_file(to_path)
# remember path for later reference
self._added_files.add(to_path.resolve())
@ -73,9 +74,9 @@ class Organizer:
# No further action needed, especially not overwriting symlinks...
return
else:
logger.info(f"Different file at {to_path}")
pretty.modified_file(to_path)
else:
logger.info(f"New file at {to_path}")
pretty.new_file(to_path)
# copy the file from from_path to sync_dir/to_path
# If the file being replaced was a symlink, the link itself is overwritten,

View File

@ -4,15 +4,17 @@ import getpass
import logging
import pathlib
import re
from urllib.parse import urljoin
import bs4
import requests
from .organizer import Organizer
from .utils import stream_to_path
from .utils import stream_to_path, PrettyLogger
__all__ = ["Ti"]
logger = logging.getLogger(__name__)
pretty = PrettyLogger(logger)
class Ti:
BASE_URL = "http://ti.ira.uka.de/"
@ -26,7 +28,7 @@ class Ti:
def synchronize(self, urlpart, to_dir, transform=lambda x: x,
filter=lambda x: True):
logger.info(f" Synchronizing {urlpart} to {to_dir} using the Ti synchronizer.")
pretty.starting_synchronizer(to_dir, "Ti", urlpart)
sync_path = pathlib.Path(self.base_path, to_dir)
@ -42,7 +44,7 @@ class Ti:
if filter(path):
self._crawl(urlpart + address, path, orga, transform)
else:
loggwe.info(f"Skipping {name}/")
logger.info(f"Skipping {name}/")
orga.clean_sync_dir()
orga.clean_temp_dir()
@ -75,7 +77,7 @@ class Ti:
for filelink in soup.find_all("a", href=self.FILE_RE):
filepath = path / filelink["href"]
fileurl = url + "/" + filelink["href"]
fileurl = urljoin(url, filelink["href"])
new_path = transform(filepath)
if new_path is None:

View File

@ -1,5 +1,8 @@
import os
import sys
import pathlib
from colorama import Style
from colorama import Fore
__all__ = [
"get_base_dir",
@ -8,6 +11,7 @@ __all__ = [
"stream_to_path",
"ContentTypeException",
"FileNotFoundException",
"PrettyLogger",
]
def get_base_dir(script_file):
@ -26,8 +30,35 @@ def stream_to_path(response, to_path, chunk_size=1024**2):
for chunk in response.iter_content(chunk_size=chunk_size):
fd.write(chunk)
def isOutputPipe():
"""Returns whether this program's output is attached to a pipe.
"""
return sys.stdout.isatty
class ContentTypeException(Exception):
pass
class FileNotFoundException(Exception):
pass
class PrettyLogger:
def __init__(self, logger):
self.logger = logger
def modified_file(self, file_name):
self.logger.info(f"{Fore.MAGENTA}{Style.BRIGHT}Modified {file_name}.{Style.RESET_ALL}")
def new_file(self, file_name):
self.logger.info(f"{Fore.GREEN}{Style.BRIGHT}Created {file_name}.{Style.RESET_ALL}")
def ignored_file(self, file_name):
self.logger.info(f"{Style.DIM}Ignored {file_name}.{Style.RESET_ALL}")
def starting_synchronizer(self, target_directory, synchronizer_name, subject=None):
subject_str = f"{subject} " if subject else ""
self.logger.info("")
self.logger.info((
f"{Fore.CYAN}{Style.BRIGHT}Synchronizing {subject_str}to {target_directory}"
f" using the {synchronizer_name} synchronizer.{Style.RESET_ALL}"
))

View File

@ -10,7 +10,7 @@ test it though).
To install PFERD or update your installation to the latest version, run this
wherever you want to install/have installed PFERD:
```
$ pip install git+https://github.com/Garmelon/PFERD@v1.1.2
$ pip install git+https://github.com/Garmelon/PFERD@v1.1.8
```
The use of [venv](https://docs.python.org/3/library/venv.html) is recommended.
@ -25,7 +25,7 @@ $ mkdir Vorlesungen
$ cd Vorlesungen
$ python3 -m venv .
$ . bin/activate
$ pip install git+https://github.com/Garmelon/PFERD@v1.1.2
$ pip install git+https://github.com/Garmelon/PFERD@v1.1.8
$ curl -O https://raw.githubusercontent.com/Garmelon/PFERD/master/example_config.py
$ python3 example_config.py
$ deactivate

View File

@ -2,11 +2,12 @@ from setuptools import setup
setup(
name="PFERD",
version="1.1.2",
version="1.1.8",
packages=["PFERD"],
install_requires=[
"requests>=2.21.0",
"beautifulsoup4>=4.7.1",
"colorama>=0.4.1"
],
)