pirtleshell / lwt Goto Github PK

I didn't find a docker image for this project so I set up my own. The setup in a modern environment is a bit tricky with MariaDB (requires some additional steps setting up a root password which are not well documented), so I think this should be quite helpful.

https://hub.docker.com/r/csalg/lwt_fork

Unlike the original, it incorrectly deals with a sentence including decimal fractions.
For example, when you add a multiword term (say "goes there") for

5.5 learning curves, 6.66 here and there, 7.0 goes there .8

the sentence becomes

5.5 learning curves, 6.66 here and ²there, 7.0 goes there .8

with Show All on, which means that "there, 7." is selected instead of "goes there."

The Non-Word List is also different from the original's list. For example, the original takes "5.5 " as a single element not as two elements, "5." and "5 ".

Like you I don't like that LWT doesn´t allow contributions even if it has a lot of flaws. (but a very good basic contents to be fair).
I forked it myself 4 years ago, then work on it from time to time. But if I knew you worked on it, I would have worked with you...
I had to rewrite completely LWT in Python (https://github.com/gustavklopp/LingL) to get what I wanted (LWT has a terrible way to store words for example).

Anyway, if you are still interested I would be very happy for a collaboration to make the LWT that we wanted to.

I am thinking of pulling and adding a sentence-by-sentence reader feature. I'm opening this issue for discussion, maybe you guys have a better idea? If you think the idea is good I would be happy to contribute this.

I have recently been making flashcards from subtitles using some subtitle translation websites and SubtitleMemorize. This approach works quite well and I prefer to review sentences than individual words. The problem is that I prefer reading books to watching TV, so I was thinking of adding a similar feature to LWT.
The feature would work like this:

• Parse the text and break it up into sentences.
• Use Google Translate API to translate each individual sentence, similar to how some websites do it.
• Store both sentences in a table, and also an array with newline locations in the original texts so that the text can be reassembled from the table.
• Now, when reading, one clicks on the sentence and the flashcard editor appears. Front is target language, back is native language. One can select some words, for example by using curly braces as LWT does right now, in both languages, to highlight the terms when reviewing in Anki.

This feature would also really improve user experience IMO as it is much faster to read the cached translation from local database than call external dictionary API's every time.

I was thinking of having a separate page for this, or perhaps extending the current text reader using some shortcut (for example, click to display term edit options, ctrl+click for sentence edit options).

Opinions?