phunspell 0.1.6

Pure Python spell checker, utilizing Spylls a port of Hunspell

Phunspell

A pure Python spell checker utilizing spylls a port of Hunspell.

NOTE: If you are only supporting languages: English, Russian or Swedish then use spylls directly: (pip install spylls)

This library includes dictionaries for all languages supported by LibreOffice.

Just a note giving credit where it's due, spylls is a fantastic project which deserves all the credit. There is a corresponding blog entry which is a good read. (and of course Hunspell itself)

Usage

import phunspell

pspell = phunspell.Phunspell('en_US')
print(pspell.lookup("phunspell")) # False
print(pspell.lookup("about")) # True

mispelled = pspell.lookup_list("Bill's TV is borken".split(" "))
print(mispelled) # ["borken"]

for suggestion in pspell.suggest('phunspell'):
    print(suggestion) # Hunspell

Installation

pip install phunspell

Supported Languages

Language	Language Code
Afrikaans	af_ZA
Aragonese	an_ES
Arabic	ar
Belarusian	be_BY
Bulgarian	bg_BG
Breton	br_FR
Catalan	ca_ES
Czech	cs_CZ
Danish	da_DK
German	de_AT
German	de_CH
German	de_DE
Greek	el_GR
English (Australian)	en_AU
English (Canada)	en_CA
English (Great Britain)	en_GB
English (US)	en_US
English (South African)	en_ZA
Spanish (all variants)	es
Spanish	es_AR
Spanish	es_BO
Spanish	es_CL
Spanish	es_CO
Spanish	es_CR
Spanish	es_CU
Spanish	es_DO
Spanish	es_EC
Spanish	es_ES
Spanish	es_GQ
Spanish	es_GT
Spanish	es_HN
Spanish	es_MX
Spanish	es_NI
Spanish	es_PA
Spanish	es_PE
Spanish	es_PH
Spanish	es_PR
Spanish	es_PY
Spanish	es_SV
Spanish	es_US
Spanish	es_UY
Spanish	es_VE
Estonian	et_EE
French	fr_FR
Scottish Gaelic	gd_GB
Gujarati	gu_IN
Guarani	gug_PY
Hebrew	he_IL
Hindi	hi_IN
Croatian	hr_HR
Hungarian	hu_HU (TODO)
Icelandic	is
Indonesian	id_ID
Italian	it_IT
Kurdish (Turkey)	ku_TR
Lithuanian	lt_LT
Latvian	lv_LV
Mapudüngun	md (arn) (TODO)
Netherlands	nl_NL
Norwegian	nb_NO
Norwegian	nn_NO
Occitan	oc_FR
Polish	pl_PL
Brazilian Portuguese	pt_BR
Portuguese	pt_PT
Romanian	ro_RO
Sinhala	si_LK
Slovak	sk_SK
Slovenian	sl_SI
Serbian (Cyrillic)	sr
Serbian (Latin)	sr-Latn
Swedish	sv_SE
Swahili	sw_TZ
Tamil	Ta (TODO)
Thai	th_TH
Turkish	tr_TR
Ukrainian	uk_UA
Vietnamese	vi_VN

Tests

python -m unittest discover -s phunspell/tests -p "test_*.py"

Experimental

    # Extended Optional:

    # First time usage:
    # create a directory of dictionaries stored as object
    # makes loading/access much faster

    storage_path = "/home/dvwright/data/phunspell/dictionary_objects"
    # run once only:
    pspell_object_create = PhunspellObjectStore(path=storage_path)


    # Then, typical usage:
    pspell = Phunspell(object_storage=storage_path)

    dicts_words = {
        "an_ES": "vengar",
        "be_BY": "ідалапаклонніцкі",
        "bg_BG": "удържехме",
    }

    for loc in dicts_words.keys():
        print(pspell.lookup(dicts_words[loc], locs=loc))

There is an option to build/store all the dictionaries as pickled data. Since there are security risks associated with pickled data we will not include that data in the distrubution.

To create your own local pickled dictionaries:

enter a python shell:

$ python
storage_path = "/home/dvwright/data/phunspell/dictionary_objects"
pspell = PhunspellObjectStore(path=storage_path)

NOTE: You only have to do this once before using the library and it's optional (this will consume a lot of resources!)

Once completed you should have a picked object for every dictionary supported by this lib.

$ ls /home/dwright/python/phunspell/pickled_data/
af_ZA
an_ES
be_BY
bg_BG
bn_BD
br_FR
bs_BA
cs_CZ
da_DK
de_AT
de_CH
...
...
...

NOTE: will take up almost 2 GB of space

$ du -sh .
1.4G

For all future uses of the library just pass the directory as an argument.

storage_path = "/home/dvwright/data/phunspell/dictionary_objects"
pspell = Phunspell(object_storage=storage_path)

# load the specific locale on lookups
pspell.lookup_list(['us-word1', 'us-word2'], locs='en_US')
pspell.lookup('german-word', locs='de_DE')

NOTE: If you ever update dictionary data, you will need to create a new pickle store for it.

and it should find the dictionaries and load them quickly

Misc

python, python3, hunspell, libreoffice, spell, spell checking