current status and open questions

• we have a docker container starting languagetool, downloading the libreoffice hunspell dictionaries from an own fork: https://bitbucket.org/mittagqi/hunspell-dictionaries-for-languagetool
  • the repo was forked, the common words were added manually - had to been converted from XML to txt - the config.properties file was created manually
• it contains a fixed config file - manually createad - for using that files, also common word lists from https://github.com/mozilla-b2g/gaia/tree/master/apps/keyboard/js/imes/latin/dictionaries were added
• Open question here:
  1. what happens if a dictionary to a language which is natively provided in languagetool is added? Which data is then used for that language?
  2. In languagetool documentation the usage of morfologik data is suggested instead of using slower hunspell data - https://dev.languagetool.org/hunspell-support.html
  3. How are such dict files then included? Passing them as config as currently the hunspell dict files? Automatically
  4. Are the common wordlists really needed, as far as I understand thats only for language recognition - and we know the language to be used, so...
• Ngrams are automatically downloaded from https://languagetool.org/download/ngram-data/ - are there other sources?

• downloading of the dictionaries and ngrams is done once after start, to update local files they have to be removed, and a container restart / run of  the download scripts makes a download again

Todo:

• Download the dictionaries automatically from the root repo, same for the wordlists, integrate the languages not provided by languagetool (depending on the answer of the above question nr 1).
• Use morfologik files?
• Compare the list of languages with a list of languages in translate5