Responsible for reading the Linux provided dictionaries from '/usr/share/dict/' and creating dictionaries out of it (or any other file that contains a list of words with each line containing one word). Those dictionaries contain a collection of words and their near neighbours, defined through the Levenshtein distance (which can be specified through the additional parameter --levenshtein).

Through that we can create complex replacement dictionaries for the generation of our error detection framework.

make lang_dict

./lang_dict 
  --input=<PATH/TO/INPUT_WORDS>
  --output=<PATH/TO/OUTPUT_FILE>
  --archaic=<PATH/TO/ARCHAIC_WORDS>
  --levenshtein=1

For single language:

./lang_dict --output=<PATH/TO/FILE.pkl> [--archaic=FILE] --input=<LIST OF FILES>

Example:

./lang_dict --output=./langs/en/en_US.json --input=/usr/share/dict/american-english-insane

The generated json has the following structure:

[
 "real": [
  "<WORD_0>": {"id":xxx, "type": "REAL_WORD", "neighbor": [], "archaic": []},
  ...
  "<WORD_N>": {"id":xxx, "type": "REAL_WORD", "neighbor": [], "archaic": []}
 ],
 "archaic": [
   "<WORD_0>": {"id":xxx, "type": "ARCHAIC", "neighbor": [], "archaic": []},
   ...
   "<WORD_M>": {"id":xxx, "type": "ARCHAIC", "neighbor": [], "archaic": []},
 ]
]

Where the neighbor list is a collection of IDs that are real word neighbors of the current word, according to the levenshtein edit distance. The IDs within the archaic list are archaic word neighbors of the current one, according to the levenshtein edit distance.