CLI commands should be verbs/actions

[bact]

Current pythainlp commands are ['corpus', 'tokenization', 'soundex', 'tagging'], which are

• not verbs
• not consistent with module names (.corpus, .tokenize, .transliterate, .tag)
  (see: https://github.com/PyThaiNLP/pythainlp/blob/dev/pythainlp/cli/__init__.py )

Verb commands are more intuitive (and many time also shorter).

Example of verb commands:

• git add
• git commit
• git push
• git pull

We should try to use verb commands when possible.
Work in progress in https://github.com/PyThaiNLP/pythainlp/tree/verb-commands

Todo:

• thainlp data purge to remove all downloaded datasets
• Add benchmark command: https://github.com/PyThaiNLP/pythainlp/blob/dev/bin/word-tokenization-benchmark
• Write testcases (maybe using subprocess)
• Update document at https://github.com/PyThaiNLP/pythainlp/blob/verb-commands/docs/notes/command_line.rst
• Record usage demo with https://asciinema.org

[bact]

DRAFT DESIGN for soundex

• thainlp soundex -h - help
• thainlp soundex TEXT - get soundex of TEXT, using default algorithm (udom83)
• thainlp soundex -a ALGO TEXT - get soundex of TEXT, using ALGO algorithm

[bact]

DRAFT DESIGN for data (formerly corpus)

• thainlp data catalog - print dataset/corpus available for download (download the list from internet)
  • thainlp data ls ? show dataset downloaded/available locally? NOT IMPLEMENTED YET
• thainlp data info DATASET_NAME - show information about DATASET
• thainlp data get DATASET_NAME - download DATASET
• thainlp data rm DATASET_NAME - remove DATASET
  • thainlp data del ?
• thainlp data path - print path for local dataset

[bact]

DRAFT DESIGN for tokenize

(NOT YET IMPLEMENTED, comments are very welcome)

• thainlp tokenize TEXT - tokenize TEXT at word level (default)
• thainlp tokenize -kw TEXT - tokenize TEXT at word level and keep whitespaces
• thainlp tokenize -sub -word -sent TEXT - tokenize TEXT at subword, word, and sentence levels
  • should be very straightforward to implement, but the issue is about the output representation like which character to use for separator of each level?
• thainlp tokenize -worda ALGO -s : TEXT - tokenize TEXT at word level using ALGO as algorithm, and using : as seperator
  • -word argument is assumed if -worda exists
• thainlp tokenize --infile INPUT_FILENAME - tokenize text from INPUT_FILE and print output to stdout
• thainlp tokenize -i INPUT_FILE -o OUTPUT_FILE

[bact]

DRAFT DESIGN for tag

(NOT YET IMPLEMENTED, comments are very welcome)

• thainlp tag -pos TEXT
• thainlp tag -pos -ne TEXT
• thainlp tag -pos --infile INPUT_FILE --outfile OUTPUT_FILE

IMPLEMENTED

• thainlp tag pos -s . การ.เดินทาง.ไกล

การ / FIXN
เดินทาง / VACT
ไกล / ADVN

[p16i]

We also have a benchmark command here:
https://github.com/PyThaiNLP/pythainlp/blob/dev/bin/word-tokenization-benchmark.

For data command, we might also have purge to remove all downloaded datasets.

[p16i]

One idea. We might use https://asciinema.org to advertise + show how to leverage this functionality.