command not found

[bablf]

after using pip install I cant call cdliconll2conllu. It is installed which I can check with pip freeze | grep cdli and in my PATH variable.

I get an command not found error.

[bablf]

Managed to get it to work with python3.9 and only if I use python3.9 -m cdliconll2conllu.cli --input_path "./train/cnn_0001.v4_gold_conll" -s "./train/conllu/" But I still get an error saying TypeError: stat: path should be string, bytes, os.PathLike or integer, not tuple

[megamattc]

I currently cannot get the program to run. I execute cdliconll2conllu -i input_file.conll but it says the command cannot be found.

[larsgw]

I think 359b1f2 should fix it, could you try again?

[larsgw]

(by reinstalling)

[megamattc]

I reinstalled the package and it does execute, but when I run the command

cdliconll2conllu -i inscriptions.conll

where inscriptions.conll is in the same directory, I get a prompt

Input path with sytnax: 

I don't understand what to do with this.

[larsgw]

I only checked whether it was installed, sorry. It looks like some (relatively) recent changes are causing problems, but I don't know what to do about those at the moment. Reverting to 8c72487 worked for me:

pip3 install git+https://github.com/cdli-gh/CDLI-CoNLL-to-CoNLLU-Converter.git@8c7248734beb9b165c08c4e8dfbb3c58c82f0389

[megamattc]

Ok. It now does convert.

However, I did not realize it can only process one text at a time. I fed it a file with many cdliconll texts in it, each prefixed by a comment line with P-number and line numbers, but that all gets collapsed into one text in the output. Since I am trying to convert search results from cdli (which puts many atf transliterations in one file), this is problematic.

[epageperron]

I don't think that in our pipeline we have had needed to process multiple texts from one file for this specific conversion; this is why the app cannot handle this. It should be a trivial change to implement. I'll open a separate issue about this.

[megamattc]

Hmm. I think this would be an issue for people searching for a group of texts in cdli because the export feature puts all the found texts in one file.

[epageperron]

I would presume that people would want to download annotated texts in those formats, not unannotated ones. As for the ATF files, we need the atf texts in one file so editors can easily update multiple texts using grep in a simple text processing software and then re-upload the changes. Uploading atf and cdli-CoNLL annotations in cdli is done with one file. The cdli-CoNLL 2 CoNLL-U converter is used in the backend, with texts sent one by one to the app.

In all cases, we have very few consumers, so I am happy that we adapt to the needs that arise; however, we are understaffed. In the email I just sent you I outlined a potential solution if you are interested.

[megamattc]

Ok. Thanks.