Pyfranc

Text language detection basic on trigrams. Support 414 language from franc-all

Install

This package is tested in Python 3.8, but should work on the whole 3rd revision of Python.

pip:

pip install pyfranc

Use

How library

from pyfranc import franc

franc.lang_detect('Alle menslike wesens word vry')[0][0] # 'afr'
franc.lang_detect('এটি একটি ভাষা একক IBM স্ক্রিপ্ট')[0][0]  # 'ben'
franc.lang_detect('Alle menneske er fødde til fridom')[0][0] # 'nno'
franc.lang_detect('')[0][0] # 'und'

# You can change what’s too short (default: 10):
franc.lang_detect('the')[0][0] # 'und'
franc.lang_detect('the', minlength=3)[0][0] # 'sco'

[0][0] has taken first value (iso code lang) in first element in output array.

whitelist

franc.lang_detect('Considerando ser essencial que os direitos humanos', whitelist = ['por', 'spa'])
# [['por', 1], ['spa', 0.6034146900423971]]

blacklist

franc.lang_detect('Considerando ser essencial que os direitos humanos', blacklist = ['src', 'glg'])
#[['por', 1],
# ['ina', 0.6211756617394293], 
# ['spa', 0.6034146900423971], 
# ['ast', 0.5628509224246592], 
# ['oci', 0.5583820327718574],
# ... 317 more items]

How CLI

usage: pyfranc_cli [-h] [-v] [-s STRING] [-t TOP] [-m MINLENGTH] [-w [WHITELIST [WHITELIST ...]]]
                   [-b [BLACKLIST [BLACKLIST ...]]] [-a] [-f] [-p]

CLI to detect the language of text.

optional arguments:
  -h, --help            show this help message and exit
  -v, --version         Print version number.
  -s STRING, --string STRING
                        Input string.
  -t TOP, --top TOP     Print top results.
  -m MINLENGTH, --minlength MINLENGTH
                        Minimum string length to accept.
  -w [WHITELIST [WHITELIST ...]], --whitelist [WHITELIST [WHITELIST ...]], 
  -o [WHITELIST [WHITELIST ...]], --only [WHITELIST [WHITELIST ...]]
                        Allow languages.
  -b [BLACKLIST [BLACKLIST ...]], --blacklist [BLACKLIST [BLACKLIST ...]], 
  -i [BLACKLIST [BLACKLIST ...]], --ignore [BLACKLIST [BLACKLIST ...]]
                        Disallow languages.
  -a, --all             Output all raw results.
  -f, --full            Print full name of language (with lang code).
  -p, --percentage      Print relative match value (in percent).

usage:

# output language
$ pyfranc_cli -t 1 -s "Alle menslike wesens word vry"
# 'afr' : 1.0

# output language from stdin (expects utf8)
$ echo "এটি একটি ভাষা একক IBM স্ক্রিপ্ট" | pyfranc_cli -t 1 -s $0
# 'ben' : 1.0

# ignore certain languages
$ pyfranc_cli --blacklist por glg -s "O Brasil caiu 26 posições"
# 'vec' : 1.0

# output language from stdin with only
$ echo "Alle mennesker er født frie og" | pyfranc_cli -t 1 --whitelist nob dan -s $0
# 'nob' : 1.0'

# output all results in raw-list format
$ pyfranc_cli --all -s "Considerando ser essencial que os direitos humanos"
# [['por', 1.0], ['glg', 0.771284519307895], ... 320 more items]

# display the result language name
$ pyfranc_cli --full -t 1 -s "Alle menslike wesens word vry"
# Afrikaans (afr) : 1.0

# output result with relative percentage of value
$ pyfranc_cli -t 5 --percentage -s "Considerando ser essencial que os direitos humanos"
# por : 28%
# glg : 22%
# ina : 17%
# spa : 17%
# ast : 16%

Derivation

Pyfranc is a outright port from Franc (JavaScript, MIT), trigram-utils (JavaScript, MIT), collapse-white-space (JavaScript, MIT), and n-gram (JavaScript, MIT). All this by Titus Wormer.

License

MIT © cyb3rk0tik