The `SessionOrBasicAuthMixin` and `StandardPagination` classes were
living in the documents app and I needed them in the new `reminders`
app, so this commit breaks them out of `documents` and puts them in the
central `paperless` app instead.
I've broken out the OCR-specific code from the consumers and dumped it
all into its own app, `paperless_tesseract`. This new app should serve
as a sample of how to create one's own consumer for different file
types.
Documentation for how to do this isn't ready yet, but for the impatient:
* Create a new app
* containing a `parsers.py` for your parser modelled after
`paperless_tesseract.parsers.RasterisedDocumentParser`
* containing a `signals.py` with a handler moddelled after
`paperless_tesseract.signals.ConsumerDeclaration`
* connect the signal handler to
`documents.signals.document_consumer_declaration` in
`your_app.apps`
* Install the app into Paperless by declaring
`PAPERLESS_INSTALLED_APPS=your_app`. Additional apps should be
separated with commas.
* Restart the consumer
This was a weird bug to run into. Basically I changed a CharField into
a ForeignKey field and ran `makemigrations` to get the job done.
However, rather than doing a `RemoveField` and an `AddField`, migrations
created a single `AlterField` which worked just fine in SQLite, but blew
up in PostgreSQL with:
psycopg2.ProgrammingError: operator class "varchar_pattern_ops" does
not accept data type integer
The fix was to rewrite the single migration into the two separate steps.
MariaDB/MySQL doesn't handle indexes on TextFields well and for some
reason, Django's migrations opts to blow up rather than handle this in a
more user-friendly way. The fix here isn't ideal, but should be
sufficient should anyone try to use Paperless with MySQL.
This change includes a filthy hack around how Django handles
change_list_results.html -- I'm not thrilled with it, but it's as
elegant as I could come up with. I'm happy to field alternative ideas.
More details can be found in `documents/templatetags/hacks.py`
Specifically, this merge includes a significant facelift to the
documents listing page, moving away from the tabular layout and toward a
tileset look. I tried fiddling with the colours, but I just don't have
any skills in that area, so we're all stuck with Django'd default
colours until someone with an eye for colour can submit a better CSS.
There appears to be quite the mess out there with regard to how DRF
handles filtering. DRF has its own built-in stuff, but recommends
django_filter for the advanced stuff, which has its own overriding
module that explodes with this message when used as per the
documentation:
AttributeError: 'NoneType' object has no attribute 'DjangoFilterBackend'
Then there's djangorestframework-filter, another package that claims to
do the same thing, that does everything just differently enough that
nothing worked while I had it enabled.
I ended up using django_filter, but doing so importing each element
explicitly, rather than just using the recommended (and broken, at least
in this project) method of:
import django_filter.restframework as fitlers
Anyway, this should bring the dependencies up to date, and strips out a
lot of redundant code.
