from corus.sources.meta import METAS
from corus.readme import format_metas, show_html, patch_readme

html = format_metas(METAS, 'https://nbviewer.jupyter.org/github/natasha/corus/blob/master/docs.ipynb')
patch_readme(html, 'README.md')

html = format_metas(METAS)
show_html(html)

from corus import load_lenta

path = 'data/lenta-ru-news.csv.gz'
records = load_lenta(path)
next(records)

from corus import load_lenta2

path = 'data/lenta-ru-news.csv.bz2'
records = load_lenta2(path)
next(records)

from corus import load_factru

dir = 'data/factRuEval-2016-master/'
records = load_factru(dir)
next(records)

from corus import load_ne5

dir = 'data/Collection5/'
records = load_ne5(dir)
next(records)

from corus import load_persons

dir = 'data/Persons-1000.zip'
records = load_persons(dir)
next(records)

from corus import load_gareev

dir = 'data/rus-ner-news-corpus.iob/'
records = load_gareev(dir)
next(records)

from corus import load_wikiner 

path = 'data/aij-wikiner-ru-wp3.bz2'
records = load_wikiner(path)
next(records)

from corus import load_librusec

path = 'data/librusec_fb2.plain.gz'
records = load_librusec(path)
next(records)

from corus import load_taiga_arzamas_metas, load_taiga_arzamas

path = 'data/taiga/Arzamas.tar.gz'
metas = load_taiga_arzamas_metas(path, offset=0, count=1)
records = load_taiga_arzamas(path, metas, offset=0, count=1)
next(records)

from corus import load_taiga_fontanka_metas, load_taiga_fontanka

path = 'data/taiga/Fontanka.tar.gz'
metas = load_taiga_fontanka_metas(path, offset=0, count=1)
records = load_taiga_fontanka(path, metas, offset=0, count=1)
next(records)

from corus import load_taiga_interfax_metas, load_taiga_interfax

path = 'data/taiga/Interfax.tar.gz'
metas = load_taiga_interfax_metas(path, offset=0, count=1)
records = load_taiga_interfax(path, metas, offset=0, count=1)
next(records)

from corus import load_taiga_kp_metas, load_taiga_kp

path = 'data/taiga/KP.tar.gz'
metas = load_taiga_kp_metas(path, offset=0, count=1)
records = load_taiga_kp(path, metas, offset=0, count=1)
next(records)

from corus import load_taiga_lenta_metas, load_taiga_lenta

path = 'data/taiga/Lenta.tar.gz'
metas = load_taiga_lenta_metas(path, offset=0, count=1)
records = load_taiga_lenta(path, metas, offset=0, count=1)
next(records)

from corus import load_taiga_magazines_metas, load_taiga_magazines

path = 'data/taiga/Magazines.tar.gz'
metas = load_taiga_magazines_metas(path, offset=0, count=1)
records = load_taiga_magazines(path, metas, offset=0, count=1)
next(records)

from corus import load_taiga_nplus1_metas, load_taiga_nplus1

path = 'data/taiga/NPlus1.tar.gz'
metas = load_taiga_nplus1_metas(path, offset=0, count=1)
records = load_taiga_nplus1(path, metas, offset=0, count=1)
next(records)

from corus import load_taiga_subtitles_metas, load_taiga_subtitles

path = 'data/taiga/Subtitles.tar.gz'
metas = load_taiga_subtitles_metas(path, offset=0, count=1)
records = load_taiga_subtitles(path, metas, offset=0, count=1)
next(records)

from corus import load_taiga_social

path = 'data/taiga/social.tar.gz'
records = load_taiga_social(path, offset=0, count=4)
list(records)

from corus import load_taiga_proza_metas, load_taiga_proza

path = 'data/taiga/proza_ru.zip'
metas = load_taiga_proza_metas(path, offset=0, count=1)
records = load_taiga_proza(path, metas, offset=0, count=1)
next(records)

from corus import load_taiga_stihi_metas, load_taiga_stihi

path = 'data/taiga/stihi_ru.zip'
metas = load_taiga_stihi_metas(path, offset=0, count=1)
records = load_taiga_stihi(path, metas, offset=0, count=1)
next(records)

from corus import load_buriy_news

paths = [
    'data/buriy/news-articles-2014.tar.bz2',
    'data/buriy/news-articles-2015-part1.tar.bz2',
    'data/buriy/news-articles-2015-part2.tar.bz2'
]
records = (
    record
    for path in paths
    for record in load_buriy_news(path)
)
next(records)

from corus import load_buriy_webhose

path = 'data/buriy/webhose-2016.tar.bz2'
records = load_buriy_webhose(path)
next(records)

from corus import load_mokoron

path = 'data/mokoron/db.sql'
records = load_mokoron(path)
list(records)

from corus import load_wiki

path = 'data/ruwiki-latest-pages-articles.xml.bz2'
records = load_wiki(path)
list(records)

from corus import load_bsnlp

path = 'data/bsnlp'
records = load_bsnlp(path)
record = next(records)
record

from corus import load_ods_interfax

path = 'data/ods/interfax.csv.gz'
records = load_ods_interfax(path)
next(records)

from corus import load_ods_gazeta

path = 'data/ods/gazeta.csv.gz'
records = load_ods_gazeta(path)
next(records)

from corus import load_ods_izvestia

path = 'data/ods/iz.csv.gz'
records = load_ods_izvestia(path)
next(records)

from corus import load_ods_meduza

path = 'data/ods/meduza.csv.gz'
records = load_ods_meduza(path)
next(records)

from corus import load_ods_ria

path = 'data/ods/ria.csv.gz'
records = load_ods_ria(path)
next(records)

from corus import load_ods_rt

path = 'data/ods/rt.csv.gz'
records = load_ods_rt(path)
next(records)

from corus import load_ods_tass

path = 'data/ods/tass-001.csv.gz'
records = load_ods_tass(path)
next(records)

from corus import load_ria_raw

path = 'data/ria.json.gz'
records = load_ria_raw(path)
next(records)

from corus import load_ria

path = 'data/ria.json.gz'
records = load_ria(path)
next(records)

from corus import load_ud_gsd

path = 'data/ud/ru_gsd-ud-dev.conllu'
records = load_ud_gsd(path)
next(records)

from corus import load_ud_taiga

path = 'data/ud/ru_taiga-ud-dev.conllu'
records = load_ud_taiga(path)
next(records)

from corus import load_ud_pud

path = 'data/ud/ru_pud-ud-test.conllu'
records = load_ud_pud(path)
next(records)

from corus import load_ud_syntag

path = 'data/ud/ru_syntagrus-ud-dev.conllu'
records = load_ud_syntag(path)
next(records)

from corus import load_morphoru_gicrya

path = 'data/morphoru/gikrya_new_test.out'
records = load_morphoru_gicrya(path)
next(records)

from corus import load_morphoru_rnc

path = 'data/morphoru/RNCgoldInUD_Morpho.conll'
records = load_morphoru_rnc(path)
next(records)

from corus import load_morphoru_corpora

path = 'data/morphoru/unamb_sent_14_6.conllu'
records = load_morphoru_corpora(path)
next(records)

from corus import load_gramru

path = 'data/gramru/GramEval_private_test.conllu'
records = load_gramru(path)
next(records)

from corus import load_corpora

path = 'data/annot.opcorpora.xml.byfile.zip'
records = load_corpora(path)
next(records)

from corus import load_russe_hj

path = 'data/russe/sem/hj.csv'
records = load_russe_hj(path)
list(records)

from corus import load_russe_rt

path = 'data/russe/sem/rt.csv'
records = load_russe_rt(path)
list(records)

from corus import load_russe_ae

path = 'data/russe/sem/ae2.csv'
records = load_russe_ae(path)
list(records)

from corus import load_toloka_lrwc

path = 'data/toloka/lrwc-1.1-aggregated.tsv'
records = load_toloka_lrwc(path)
list(records)

from corus import load_simlex

path = 'data/simlex/ru_simlex965_tagged.tsv'
records = load_simlex(path)
list(records)

from corus import load_omnia

path = 'data/ru_om1000a.x1_.xz'
docs = load_omnia(path)
next(docs)

from corus import load_ruadrect

path = 'data/toloka/ruadrect/task2_ru_test.tsv'
records = load_ruadrect(path)
next(records)

from corus import load_rudrec

path = 'data/rudrec/rudrec_annotated.json'
records = load_rudrec(path)
next(records)