GitLab

When there are no columns white separators and no black
columns separators, then we determine the reading order
by simply looking which line is above which other lines.

Add a workflow diagram for documentation

Unused function select_regions in ocropus-nlbin

Detect the passed number of separators

increased the default value of the parameter --maxcolseps to 3 to be more conform with the "old behaviour"

* see also #137
* the other two lines contained strange symbols coded as FF

Update README.md

This fixes #178 which seems to be a regression from b4381aa7

Travis CI builds

Print summaries to stdout instead of stderr 

[econf] Handle possibility that filtered GT is empty

Cleanup common.py, lstm.py, extract exceptions to Cleanup imports in common.py, lstm.py, extract exceptions

py3k: Use new-style exceptions.

This fixes #45.

py3k: Use print function instead of statement

Signed-off-by: Stefan Weil <sw@weilnetz.de>

The results for the single files were already printed to stdout.
The summary should be printed to stdout, too.
Signed-off-by: Stefan Weil <sw@weilnetz.de>

The results for the single files were already printed to stdout.
The summary should be printed to stdout, too.
Signed-off-by: Stefan Weil <sw@weilnetz.de>

Remove tabs in source

Made output more compatible to the hOCR spec.

- properties "image" and "bbox" for the "ocr_page" element.
- correct orientation of coordinate system

This caching means that fonts don't have to be re-queried on every
build.

This allows the test to ensure it is not accidentally loading the
package from the current directory, but instead from the installed
location.

Search for models relative to installed library.