Migrate document_ocr v10

[cmsalmeida]

Migration of module document_ocr to 10.0
With some enhancements:

1. System parameters to control DPI and Quality of conversion from PDF to image
2. System parameter to define default language for OCR
3. OnChange() language in attachment form, apply OCR again
4. OCR of PDF one page at a time, this allow one to process big PDF files
5. All PDFs will pass through OCR

[cmsalmeida]

It's failing in tesseract dependency.
It must be installed like: apt-get install tesseract-ocr tesseract-ocr-eng
Can someone help me how to configure repo to pass this requirement?
Thanks in advance

[pedrobaeza]

Just put a requirements.txt.

[cmsalmeida]

I did, it's there

[cmsalmeida]

It's here!

[cmsalmeida]

and here

[pedrobaeza]

Then everything is OK in that part. The problem should be another.

[pedrobaeza]

Maybe an outdated .travis.yml? Check latest template on https://github.com/OCA/maintainer-quality-tools/blob/master/sample_files/.travis.yml and compare with this.

[cmsalmeida]

This module needs tesseract.
In Ubuntu one can install it issuing:
sudo apt-get install tesseract-ocr tesseract-ocr-eng
How to make it work in travis?

[pedrobaeza]

How is this install in previous versions? Use the same technique.

[cmsalmeida]

I didn't did the previous versions, I got the code from 8.0, did the migration and enhancements, it's working in my servers.
I did this PR to OCA to share, because now it's really good, and it works for big PDFs (many pages).
It's shared, if someone can merge it would be great.

[cmsalmeida]

I can't spend more time on this module, sorry....
It's demotivator trying to contribute, and not being able to :(
Any way you have the code there, or if someone interested in use it it's available for all at:
https://github.com/thinkopensolutions/oca-knowledge
Thanks any way.

[pedrobaeza]

OK, as you consider. I think knowing all the internals of the processes, like Travis and so on, make you better later facing other developments, but if you don't want to expense more time, we can't force you.

[cmsalmeida]

Can you please point me to documentation on that?
Or any code/configuration that works?

[pedrobaeza]

In 8.0, the dependency is added via Travis packages: https://github.com/OCA/knowledge/blob/8.0/.travis.yml#L10.

You can make the same here.

[ovnicraft]

👍

[cmsalmeida]

Hi,
For the tests to run I need to force the import of PdfImagePlugin and PalmImagePlugin from PIL.
But there is no statement where to use that import. Flake8 is giving import not used error.
Can one tell me how to solve this?

[cmsalmeida]

I found that, if one use pylint and flake8 one can ignore unused import warning in both tools in this way:
import IMPORT # noqa # pylint: disable=unused-import
This way you explicitly tell pylint and flake8 that this unused import in intended.

[pedrobaeza]

Yeah, that's it. I don't understand why don't you use it if you import it, but you can bypass it that way.

[hbrunn]

Unfortunately, you didn't follow the guidelines for migrations, so previous commits are not preserved and it's quite inefficient to review this way (there's no way to tell what you added and what existed before), that's why I'm not going to go into details right now. I'd suggest to follow the above and squash all your commits to one, then reviewing this commit is sufficient.
I understand newcomer's frustration, but most of this can be avoided by reading the new project's contribution guidelines before contributing.

[pedrobaeza]

I also add the link to the specific migration guide to 10.0. As this module is not in 9.0, you should replace 9.0 by 8.0: https://github.com/OCA/maintainer-tools/wiki/Migration-to-version-10.0

[cmsalmeida]

Thank you for your help.
Unfortunately I really can't spend more time in this process.
This module is now publicly available at: https://github.com/thinkopensolutions/tko-addons/tree/10.0/tko_document_ocr
It is available on Odoo apps also.

[cmsalmeida]

I will try this again soon...