Should url be transcribed?

[ssoto]

I did this tests:

import requests
from urllib.parse import urlencode

content = "Hola esto es una URL:    https://twitter.com/OficinaEspanol  Respetala porfa"
encoded_url_params = urlencode({"spanish": content})
response = requests.get(f"https://api.andaluh.es/epa?{encoded_url_params}")

response.json()
>  {'spanish': 'Hola esto es una URL:    https://twitter.com/OficinaEspanol  Respetala porfa',
>   'andaluh': 'Ola êtto êh una ÛL-L:    ttps://twîttêh.com/OfiçinaÊppanôh  Rêppetala porfa',
>  'rules': {'vaf': 'ç', 'vvf': 'h', 'escapeLinks': False}}

As you can see, transcription algorithm pass over url, appying rules and changing it.
Is this behaviour expected?

[fontanon]

Hi @ssoto please use escapeLinks url parameter to avoid transcribing URL. In other words: add "escapeLinks": True to your encoded_url_params

A curl example below:

curl "https://api.andaluh.es/epa?spanish=Vente%20a%20https://andaluh.es%20a%20ver%20la%20p%C3%A1gina&vaf=s&vvf=j&escapeLinks=true"

Alos, please open issues related with API into andaluh-api repo: https://github.com/andalugeeks/andaluh-api

[ssoto]

Sorry for being harder, did you see the transcription of URL to ÛL-L? Let me know if this behaviour is properly for andaluh-api or for andaluh-py.
regards!

[penyaskito]

Looks like the regex is parsing only the first part of the url

curl https://api.andaluh.es/epa?spanish=Vente%20a%20https://andaluh.es/esto/es/el/resto/de/una/url%20a%20ver%20la%20p%C3%A1gina&vaf=s&vvf=j&escapeLinks=true

returns

{
	  spanish: "Vente a https://andaluh.es/esto/es/el/resto/de/una/url a ver la página",
	  andaluh: "Bente a https://andaluh.es/êtto/êh/el/rêtto/de/una/ûl-l a bêh la pájina",
	  rules: {
		    vaf: "s",
		    vvf: "j",
		    escapeLinks: true
	  }
} 

[fontanon]

Confirmed this behaviour with andaluh-py, reopening this issue.

$ python3
>>> import andaluh
>>> print(andaluh.epa("Hola esto es una URL:    https://twitter.com/OficinaEspanol  Respetala porfa", escape_links=True))
Ola êtto êh una ÛL-L:    https://twitter.com/OfiçinaÊppanôh  Rêppetala porfa
>>> 

[fontanon]

Related: andalugeeks/andaluh-api#3

[fontanon]

Sorry for being harder, did you see the transcription of URL to ÛL-L? Let me know if this behaviour is properly for andaluh-api or for andaluh-py.
regards!

Please preceed with a hashtag those acronyms or foreign language words you do not want to be transcribed. The andaluh-py library just transcribes everything, it cannot be aware those situations.

[ssoto]

Thanks for your analysis guys!
Anyone is working on fix it? I can try fix it, I'm going to asing it to myself!

[fontanon]

I was already, was easy. Thanks.

[fontanon]

@ssoto @penyaskito please use master to test now.

0e9a405