Add new docker image to CL
Add simple test to begin replacing weird binaries

Use our new docker container to generate the thumbnail
and return the contents

Refactor subprocess into our new BTE container

Not sure its necessary or used in any way
The method only extracts page count for PDFs
prior to this refactor.

Add volumes to new containers to task server
Remove unneeded installs to task server

Update tests

Currently tests.yml is not running
this is an attempt to get it working in
a new syntax

…ypes

Built and pushed beta containers for Django
and the celery task-server.

Task server is imported from django image and
neither has the 'weird binaries'.

The new images are 36% of the old ones

Add JUDGE_PICS_PATH
add SEAL_ROOKERY_PATH

Add BTE URL to settings and point to it in functions

Remove extension from get_page_count

Remove error catching from generate thumnail method.
Move exceptions to the calling code.

…ect file type to bte

…tern

Check to validate successful response from BTE

Update test to reflect BTE changes

Remove `u` prefix from unicode strings - all strings are unicode.

Change `unicode(...)` to `str(...)`.

Change `isinstance(x, basestring)` to `isinstance(x, str)`.

Some functions, particularly those on dicts like `d.items()` and
`d.values()` return iterators, not lists. In some circumstances (i.e.
when you're not immediately iterating through the results) you need to
construct the list with `list(...)` to work with py3.

httplib => http.client
HTMLParser => html.parser
cStringIO, StringIO => io.StringIO
urlparse => urllib.parse
urllib2 => urllib.error
izip => builtin zip

I didn't think this was a Python 2 vs. 3 thing but hey.

__nonzero__ => __bool__
__unicode__ => __str__

Update method to call for audio conversion.
Includes a serializer for passing the necessary information to BTE.

Using dashes in submodule names isn't valid Python (2 or 3, I think).

Seal-rookery, and eyed3 are not needed anymore

add test_BTE_audio_conversion for audio conversion and metadata addition

No longer used in CL

# Conflicts:
#	cl/scrapers/tasks.py
#	cl/search/management/commands/cl_count_recap_pages.py

Building on @drewsilcock I merged his branch
with the dockerize weird binaries modified juriscraper slightly and voila.

I havent tested it out but it will load the homepage and search solr.

Include changes to migration files help text encoding

Include changes to migration files help text encoding

…ener into python38-merger

…ener into python38-merger

…ener into python38-merger

…ener into python38-merger

…ener into python38-merger

…ener into python38-merger

use // instead of / for integer division

…ison

In cleanup main query and get related clusters with cache
replaced sunburnt with scorched

Upgrade scorched, ipython, cryptography and comment out psycopg2

Remove old magic code with python magic, add to req.

Where pip needs `==a.b.*,>=a.b.c`, poetry can just do `^a.b.c`. Strictly
speaking this isn't exactly the same as `^a.b.c` will satisfy any
`a.d.e` where d > b and e > c. This makes the versioning more reliable,
because the pip version could install major package updates without you
realising.

…date