- setTorrentPath was called only in case DELUGE_DONE_DIRECTORY was set,
forgot about the more general DOWNLOAD_TORRENT_DIR option. Thought about
it in deluge.py, forgot about it in searcher.py. This should fix#2557
Track matching is performed using 'CleanName' which up to now was obtained in
convoluted way, which effectively removed any non-ascii alphanumeric characters
but at the same time left some trash preventing the names to be matched due to
whitespace differences. Current implementation performs most of the
transliteration using Unicode NFD decomposition to remove diacritical marks from
characters in Latin scripts, leaving the others intact. Only alphanumeric chars
are included in resulting string and all the spaces are coalesced. Based on
observations on several-tens GiB library, this allows for much better ratio of
automatic track matches.
- Reading downloaded files now uses response.content instead of
response.text which caused loss of file data due to encoding
shenanigans. This fixes the WHAT issue.
- Fix URL for WFFLS downloads
- Special User-Agent for WHAT downloads, still not sure if this is
necessary
- DLing not working for some users, possibly because of the site - maybe
a different UA will make a difference
- Make links lowercase before testing them for content
- If Deluge is set to use HTTPS but there is no certificate configured
in HP, use HTTPS without verifying the certificate - connection will
still be encrypted but the certificate won't be validated - this will
show up as an ERROR in the logs
- Scrubbing session cookies
This commit adds the possiblity to use variables from tags already present in
the downloaded media files in file, folder and album art renaming routines. The
variable names translate directly to MediaFile field names, so it is now
possible to use variables like $mb_albumid, $genre, $bitrate, $samplerate, etc.
The full list can be read from MediaFile.readable_fields().
