Rename images with line id

Two main changes:

• Images are renamed {page_id}_{line_number}_{line_id} (previously {page_id}_{line_number}) as it is easier to visualize on Arkindex
• Line number (e.g line_032) should come from the element's name and not from enumerate to avoid offset on pages containing lines without transcription

Note: I'm assuming that there are at most 999 lines/image (line_number has 3 digits)

assigned to @starride

added 1 commit

• f83d7ac9 - rename images to {page_id}_{line_number}_{line_id} and read line number from element's name

Compare with previous version

added 1 commit

• 7cf359e3 - Remove useless print

Compare with previous version

added 1 commit

• e541cf49 - Renaming transcription files

Compare with previous version

added 1 commit

• 33172e7f - Fix linting

Compare with previous version

marked this merge request as ready

changed title from Rename images with line to Rename images with line id

changed the description

requested review from @babadie

Is this project used a lot ? I'm totally discovering it (cc @kermorvant )

If it's used a lot, it would be better to have at least one unit test that trigger the main workflow (I only see utils & init being covered right now).

approved this merge request

Yes, it's used to format datasets for HTR. I use it every time I need to train PyLaia on a new dataset, and I think @Chaza_Abdelwahab and @melodie.boillet also use it frequently.

I can try to implement unit tests, but this would be a first for me, so I would probably need some guidance. Could you share a reference project and/or other resources on how to write tests properly ?

added 5 commits

• dcbea608 - 1 commit from branch master
• 0607db20 - rename images to {page_id}_{line_number}_{line_id} and read line number from element's name
• bf3d0296 - Remove useless print
• 65e5467c - Renaming transcription files
• 7a6d75db - Fix linting

Compare with previous version

added 6 commits

• 7a6d75db...fd06f077 - 2 commits from branch master
• 5b7929b6 - rename images to {page_id}_{line_number}_{line_id} and read line number from element's name
• 6fd680d9 - Remove useless print
• dabceec7 - Renaming transcription files
• 7e589180 - Fix linting

Compare with previous version

I was able to test with

kaldi-data-generator -f kaldi --dataset_name demo --out_dir xxx --pages d29439d3-5078-44ac-a30d-568eff2e483a --transcription_type transcription_line

It's a nice project, @martin_teklia put a lot of effort into it.

I'll add some issues regarding unit tests

@starride You can already look at these slides https://notes.teklia.com/p/IOhDTSui5#/

Once I have written issues, we can discuss

merged