OSError: Invalid data format: can't load _dawg.Dictionary

[mobileunit]

Hi Duygu,

thanks for the tool DEMorphy. I was trying to install the tool using your Readme

$ git clone https://github.com/DuyguA/DEMorphy $ cd DEMorphy $ python setup.py install

All seems to work out properly. When starting your code example in Python I get the error message

OSError: Invalid data format: can't load _dawg.Dictionary

Do I somehow have to connetc the installation to the dictionary sources available here?

https://github.com/DuyguA/german-morph-dictionaries

I tried to put them simply in the demorphy/morph_dict/ folder but that doesn't seem to work.

Do you have a clue why it is not working for me?

Many thanks and best regards, Andy

[DuyguA]

Hello,

Yes, it seems that I uploaded the old version of README file. Due to Github restrictions, I changed downloading dictionary a bit. You need to download demorphy/data/words.dg manually from this repo and then launch the installation script.

I'll change the README soon. Thanx for detecting.

[mobileunit]

Wooohooo :D It is working now! Thanks a lot!!! Best, Andy

[DuyguA]

Thanks a lot for noticing also, I uploaded the wrong version😯

[AlessandroVol23]

Hi, I've got the same error with this example script:

from demorphy import Analyzer
print("after import")
lemmatizer = Analyzer()
# Same with lemmatizer = Analyzer(char_subs_allowed=True)
print("test")

I've installed the package within a virtual env in python 3.6.8 with the command python setup.py install

[DuyguA]

Hello @AlessandroVol23 ,

Did you follow the this instruction in the README:

Download the dictionary file demorphy/data/words.dg and replace it under the corresponding directory again. Then you're ready to launch the setup script:

Before launching the install script, one needs to download the words graph manually.

[AlessandroVol23]

Hi, yes that was my mistake at the beginning, I downloaded the file via git lfs pull and now I can initialize the Analyzer. Thanks

[DuyguA]

I'll close the issue then , would that be OK?