Ability to convert between simple tense, past tense and past participle?

[MehdiK]

Not sure how involved this is going to be; but this would be a nice addition if there is a relatively simple light solution.

Thoughts?

[hazzik]

Hi @MehdiK, please check my old repository https://github.com/hazzik/fluent-english probably it is what you are looking for.

[MehdiK]

Thanks a lot for your response.

COOL. That does a whole lot more than what I am looking for 👍

Have you used it in production? The reason I ask is that one of my main fears with this feature is that there might be unforeseen edge cases. If it's used in production for a while then that fear is largely mitigated and the few potential edge cases can be dealt with.

Also would you want a new home for it?! If yes, then would you be happy to restyle the code and add some tests to make it consistent with the rest of Humanizer? I am happy to do it too - that way I'll learn a lot of awesome stuff you've done in there. I'm happy either way.

[hazzik]

Have you used it in production?

No it is just a concept

The current big issue here that the lib does not have irregular verbs dictionary.

Also it accepts only the full sentence in present simple tense.

[MehdiK]

Ah, cool. I didn't read through the code - it was 3am. Thanks for clarification. This feature does need a dictionary of irregular verbs.

We might add the whole English NLP thing over time but as the first step I would like to add the conversion in rather soon. I will try to get started on it in the next few days (unless you send me a PR before that :p)

[MehdiK]

I coded this in PR #64 but there is a problem: The coded solution works perfectly for all cases except when the verb is not irregular, is more than one syllable, the stress is not on the final syllable and the last letter is not x and w. I know of a few words like "happened", "listened" and "opened" that fit this profile and I can take care of them one by one; but I am sure there are more and I don't know all of them.

I hopelessly did some research to see if there is any algorithm that helps find where the stress is on a word; but there is none. So the only way to achieve this is to import a COMPLETE English dictionary and get rid of the rules in the code; but I don't personally consider this a good solution. If one wants to use a complete dictionary there are other libraries out there that implement this feature and a lot more; e.g. NetSpell.

So I need your input. Do you think I should ship #64 and put a warning on readme, ask people to report the exceptions and build up a list over time or I should just forget about this altogether?

[MehdiK]

Closing this issue for now.