Anki-Spanish-Creator

I created this project to save time while uploading spanish vocab into anki, a spaced-repetition learning software tool. This will bulk-create spanish works into a learning deck, meant to save a ton of time (the UI would take much longer). Features include:

• automatically creates anki double-sided cards using anki connect
• auto-fetches pronunciation mp3 of spanish word from spanishdict.com
• Web crawls and fetches sample sentences using chrome headless browser, from spanishdict.com

Sample url for pronunciation is something like: https://audio1.spanishdict.com/audio?lang=es&text=hablar&key=494e1f0d93abd4cffaa8d5781d05dd9c

Updates 8/7/19 - can now add a JSON or CSV file, code will handle either case. CSV is kind of unnecessary complications, and then it wasn't possible to parse out comma from part of translation.

Instructions

1. Put words you'd like to learn into a csv or json file under words directory. See csv-parser section below for more info.
2. Make sure anki is running locally, with anki connect plugin installed (https://ankiweb.net/shared/info/2055492159)
3. docker-compose up -d which will spin up a local RMQ, and expose management port to port 8080. Open localhost:8080 in browser to see management UI, with default creds of guest for both username and password.
4. scripts/rmq-init which will ping rmq management API to create 2 queues needed.
5. node sendWords.js <path-to-csv-file> which parses csv and sends to word-pairs queue. This script should be near instantaneous.
6. node fetchSentences.js which parses word pairs and spins up local headless browsers to fetch sentences from spanishdict.com, and publish to word-sentences queue.
7. node createCards will consume from sentences queue and hit local Anki connect api running on port 8765 to create anki cards.

CSV-Parser

• format is english, spanish
• automatically ignores empty lines, and lines starting with // (for commenting)
• you can specify a spanish pronunciation with >>
  • example: ear, oreja (external part) / el oído (internal part) >> oreja
• <> can be used to force spanish pronunciation to include more than one word. otherwise will always be last word (ie, nino instead of el nino)

Notes

This uses forked node processes to have multiple CPUs, for better performance of headless browsers. When running many CPUs though (like > 10) I ran into issues with spanishdict.com rate-limiting by IP (through sophos software looks like). So I looked into proxying, but that led down a path of more problems (proxies not working, SSL issues, slower load times due to international ads, etc). So I bagged the proxying but could be a cool future project.

TODO

• add more error handling
• put constants into config

Words to add:

• mascota (pet)