add OCR-D/ocrd_fileformat, fix #80

[kba]

No description provided.

[stweil]

I am friend of alphabetic sorting of such lists, but that can be done next time.

[kba]

238f514 core should be before cor-asv-... IMHO

[bertsky]

LGTM.

Can you please add ocrd-pagetopdf as well (this version for now)? I know it should be in ocr-fileformat itself, but that would never be able to replace the full-featured direct conversion.

[kba]

https://app.circleci.com/pipelines/github/OCR-D/ocrd_all/119/workflows/3520cde8-ea4f-4eb8-8e8e-ab229cb8991a/jobs/127 Looks like the submodule is not checked out recursively - @bertsky @stweil what am I doing wrong?

Can you please add ocrd-pagetopdf as well (this version for now)? I know it should be in ocr-fileformat itself, but that would never be able to replace the full-featured direct conversion.

Sure see #84

[bertsky]

https://app.circleci.com/pipelines/github/OCR-D/ocrd_all/119/workflows/3520cde8-ea4f-4eb8-8e8e-ab229cb8991a/jobs/127 Looks like the submodule is not checked out recursively - @bertsky @stweil what am I doing wrong?

You made no mistake: ocrd_fileformat gets cloned recursively:

git submodule sync --recursive ocrd_fileformat
if git submodule status --recursive ocrd_fileformat | grep -qv '^ '; then \
	git submodule update --init --recursive ocrd_fileformat && \
	touch ocrd_fileformat; fi
Submodule 'ocrd_fileformat' (https://github.com/OCR-D/ocrd_fileformat) registered for path 'ocrd_fileformat'
Cloning into '/build/ocrd_fileformat'...
Submodule path 'ocrd_fileformat': checked out 'de5b53282243c0b9696bf1ca9a22c327793ef37d'
Submodule 'ocr-fileformat' (https://github.com/UB-Mannheim/ocr-fileformat) registered for path 'ocrd_fileformat/ocr-fileformat'
Cloning into '/build/ocrd_fileformat/ocr-fileformat'...
Submodule path 'ocrd_fileformat/ocr-fileformat': checked out '86a41edf80ad235a95a612b981ffb68ae49417db'

But then subsequently:

. /usr/local/bin/activate && make -C ocrd_fileformat install-fileformat install
make[1]: Entering directory '/build/ocrd_fileformat'
make -C  ocr-fileformat PREFIX=/usr/local vendor install
make[2]: Entering directory '/build/ocrd_fileformat/ocr-fileformat'
make -C vendor check
make[3]: Entering directory '/build/ocrd_fileformat/ocr-fileformat/vendor'
make[3]: *** [check] Error 1
make[2]: *** [check] Error 2
make[1]: *** [install-fileformat] Error 2

So obviously there's a bug in ocr-fileformat: make vendor does not work, and does not even report a proper error.

[bertsky]

Can you please add ocrd-pagetopdf as well (this version for now)? I know it should be in ocr-fileformat itself, but that would never be able to replace the full-featured direct conversion.

Sure see #84

Thanks @stweil @kba!

[stweil]

So obviously there's a bug in ocr-fileformat: make vendor does not work, and does not even report a proper error.

It checks whether all required executables are available, and the error message is indeed not good. Maybe unzip is missing. I added it to the requirements, so hopefully the Docker build should work now.

[stweil]

@kba, @bertsky, would you prefer a different build fix, or can we merge now?

[bertsky]

would you prefer a different build fix, or can we merge now?

I suppose we could give ocrd_fileformat's Makefile a target deps-ubuntu with all required packages. But we could also do that in a later PR and see how we fare without.