config-fi.json

{
  "LANGUAGE": "finnish",
  "FASTTEXT_PATH": "./fasttext",
  "BIN_FILE": "cc.fi.300.bin",
  "VEC_FILE": "cc.fi.300.vec",
  "OUT_FILE": "cc.fi.300.filtered.vec",
  "VOCAB_FILE": "cc.fi.300.filtered.vocab",
  "_comment_regex_positive": "C, Q, W, X or Z excluded",
  "REGEX_POSITIVE": "([abd-pr-vyABD-PR-VYÄäÖöÅ劚Žž\\-\\']+$)",
  "_comment_regex_negative_1": "Exclude words beginning with dash",
  "_comment_regex_negative_2": "and containing consecutive dashes",
  "REGEX_NEGATIVE": "^-|--",
  "_comment_regex_foreign": "Exclude if matches and doesn't pass spell check",
  "REGEX_FOREIGN": "[Å劚Žž]",
  "_comment_regex_repeat": "Exclude if 3 or more consecutive repeated letters",
  "REGEX_REPEAT": "(\\w)\\1{2,}",
  "RANDOM_CONSONANTS": "bdfghjklmnprstv",
  "RANDOM_MIN_LENGTH": "4",
  "RANDOM_MAX_LENGTH": "20",
  "NN_LOOP_CUTOFF": "100",
  "NN_LOOP_WORDS": "1000",
  "DATABASE": {
    "dbname":  "decrapper",
    "table": "garbwords",
    "user":  "root",
    "password": "",
    "host": "localhost",
    "port": "3306"
  }
}