The previous version of `tests_date_3.png` had too much spacing
between the `0` and the `8` glyphs, which resulted in the year getting
parsed as `200 8` in Tesseract 3.05.00 (+ tessdata 3.04.00).
This caused the date parsing test to fail.
It turns out that the Lorem ipsum text in the sample files was confuing the language guesser, causing it to think the file was in Catalan and not English or German.
It is not obvious that the PAPERLESS_FORGIVING_OCR allows to let
document consumption happen even if no language can be detected.
Mentioning it in the actual error message in the log seems like the best
way to make it clear.
I'm not sure what the circumstances are, but it looks like unpaper can
attempt to write a temporary file that already exists [0]. This then
fails the consumption. As per daedadu's comment simply letting unpaper
overwrite files fixes this.
[0]
unpaper: error: output file '/tmp/paperless/paperless-pjkrcr4l/convert-0000.unpaper.pnm' already present.
See https://web.archive.org/web/20181008081515/https://github.com/danielquinn/paperless/issues/406#issue-360651630
Now when you export a document, the `storage_type` value is always
`unencrypted` (since that's what it is when it's exported anyway), and
the flag is set by the importing script instead, based on the existence
of a `PAPERLESS_PASSPHRASE` environment variable, indicating that
encryption is enabled.
- Debug mode is now configurable in the configuration file. This way, we don't have to edit versioned files to disable it on production systems.
- Recent correspondents filter (enable in configuration file)
- Document actions: Edit tags and correspondents on multiple documents at once
- Replaced month list filter with date drilldown
- Sortable document count columns on Tag and Correspondent admin
- Last correspondence column on Correspondent admin
- Save and edit next functionality for document editing