flag to dynamically consider publication limits, such as, GEMA-licenced
content available only +/-7 days around an broadcast

radio program now with teaser and/or descrption as well as references to
internal/external websites/email

generated with cmd:
python3 manage.py dumpdata --settings=lohrothek.local_settings
--settings=lohrothek.settings --indent 4 program.RadioProgram
program.RadioProgramReference program.BroadcastTemplate
program.BroadcastRepeatTemplate > scripts/program_fixture.json

apply with cmd:
python3 manage.py loaddata scripts/program_fixture.json

erlaubt zu jeder sendereihe verweise zu setzen auf webseiten, emails,
telefon

API-auto-documentation

docstrings should be quoted with double quotes following the pep257 and
pep8 standard.

redirect on nonset get parameter to current date param page. usage of
session if page was visited before

gapfill items are sliced and rendered in gaps of broadcasts, if
applicable

some migrations use features only available in more recent
versions of sqlite.

debian stretch ships a rather old version of sqlite by default, but
a newer one is available from backports. as stretch will be
superseded in the coming 2-3 months, there is not point in
maintaining explicit compatibility with the old sqlite version.

so this change and 0f8947fa
effectively makes sqlite 3.26+ a required version for anyone
who wants to use sqlite during development or production.

This was caused by renaming the `AudioFile` model to `Audiofile` and
only got worse with each new ForeignKey relation to `Audiofile`.

This is a bug in Django:
https://code.djangoproject.com/ticket/23916

* move transcode logic into processing module
  to make it reusable
* use sql query instead of python code for filtering audio files
  to improve performance
* show progress during transcoding when tqdm library is available
  and has not been supressed by a non-tty stdout or by explicitly
  passing the --no-progress option
* add option to just show audio files that are not transcoded
* add optional positional file ids to only transcode one or
  more, but not all audio files in database
* transcode formats are now configurable in settings
* fix: don’t attach transcoded audio files to report.
       transcoded files do not represent new content but just
       the same content in a different format.
* fix: container can be null for raw-data-streams like mp3
* fix: sample_rate may change after encoding and it may
       be null for formats unsupported by mutagen (opus).
* fix: add container/codec data for existing mp3 files.
* fix: increase audio file name max length.
       audio file names may exceed django’s default limit
* fix: use project setting ffmpeg binary

rendering issue known and fixed by contributor, but not yet merged in
package repository. overwrote relevant methods of TagSelect2 in
django-autocomplete-light package

This development-middleware will handle HttpResponseSendfile response
types with X-Sendfile/X-Accel-Redirects headers and convert them into
a simple FileResponse.

Should only be used during development to circumvent the need for a
dedicated web- or application server that brings native support for
these kind of headers.

working...

Webservers (in our case nginx) will pass the Content-Type header
provided by our application back to the client. But its the webserver
that should figure out the proper Content-Type via its static file
handler.

The test target now creates its own python environment with the help of
the venv module (provided by the python3-venv package). Installing
python dependencies directly into the system is therefore no longer
necessary.

With the inclusion of makefilet in commit
38e3798d wget is a first-class
dependency of our ci make targets.