def guardian_main(paths):
guardian_corpora = news_cleaner(paths, 'guardian_path')
guardian_corpora = [
norm(i) for i in guardian_corpora if not norm(i) == None
]
guardian_corpora = sum(guardian_corpora, [])
return guardian_corpora
def independent_main(paths):
independent_corpora = news_cleaner(paths, 'independent_path')
independent_corpora = [
norm(i) for i in independent_corpora if not norm(i) == None
]
independent_corpora = sum(independent_corpora, [])
return independent_corpora
def sfchronicle_main(paths):
sfchronicle_corpora = news_cleaner(paths, 'sfchronicle_path')
sfchronicle_corpora = [
norm(i) for i in sfchronicle_corpora if not norm(i) == None
]
sfchronicle_corpora = sum(sfchronicle_corpora, [])
return sfchronicle_corpora
def washington_main(paths):
washington_corpora = news_cleaner(paths, 'washington_path')
washington_corpora = [
norm(i) for i in washington_corpora if not norm(i) == None
]
washington_corpora = sum(washington_corpora, [])
return washington_corpora
def electric_main(paths):
electric_corpora = short_fic_cleaner(paths, 'electric_path')
electric_corpora = [
norm(i) for i in electric_corpora if not norm(i) == None
]
electric_corpora = sum(electric_corpora, [])
return electric_corpora
def waccamaw_main(paths):
waccamaw_corpora = waccamaw_cleaner(paths)
waccamaw_corpora = [
norm(i) for i in waccamaw_corpora if not norm(i) == None
]
waccamaw_corpora = sum(waccamaw_corpora, [])
return waccamaw_corpora
def ny_fiction_main(paths):
ny_fiction_corpora = short_fic_cleaner2(paths, 'ny_fiction_path')
ny_fiction_corpora = [
norm(i) for i in ny_fiction_corpora if not norm(i) == None
]
ny_fiction_corpora = sum(ny_fiction_corpora, [])
return ny_fiction_corpora
def resource_main(paths):
resource_corpora = short_fic_cleaner(paths, 'resource_path')
resource_corpora = [
norm(i) for i in resource_corpora if not norm(i) == None
]
resource_corpora = sum(resource_corpora, [])
return resource_corpora
def bbc_main(paths):
bbc_corpora = news_cleaner(paths, 'bbc_path')
bbc_corpora = [norm(i) for i in bbc_corpora if not norm(i) == None]
bbc_corpora = sum(bbc_corpora, [])
return bbc_corpora
def nytimes_main(paths):
nytimes_corpora = news_cleaner(paths, 'nytimes_path')
nytimes_corpora = [norm(i) for i in nytimes_corpora if not norm(i) == None]
nytimes_corpora = sum(nytimes_corpora, [])
return nytimes_corpora
def oxford_main(paths):
oxford_corpora = ling_blogs_cleaner(paths, 'oxford_path')
oxford_corpora = [norm(i) for i in oxford_corpora if not norm(i) == None]
oxford_corpora = sum(oxford_corpora, [])
return oxford_corpora
def adelaide_main(paths):
adelaide_corpus = adelaide_cleaner(paths)
adelaide_corpus = [norm(i) for i in adelaide_corpus if not norm(i) == None]
adelaide_corpus = sum(adelaide_corpus, [])
return adelaide_corpus
def arab_st_main(paths):
arab_corpora = arab_cleaner(paths)
arab_corpora = [norm(i) for i in arab_corpora if not norm(i) == None]
arab_corpora = sum(arab_corpora, [])
return arab_corpora
def eeuu_main(paths):
eeuu_corpora = eeuu_cleaner(paths)
eeuu_corpora = [norm(i) for i in eeuu_corpora if not norm(i) == None]
eeuu_corpora = sum(eeuu_corpora, [])
return eeuu_corpora
def usa_main(paths):
usa_corpora = short_fic_cleaner2(paths, 'usa_path')
usa_corpora = [norm(i) for i in usa_corpora if not norm(i) == None]
usa_corpora = sum(usa_corpora, [])
return usa_corpora
def latimes_main(paths):
latimes_corpora = news_cleaner(paths, 'latimes_path')
latimes_corpora = [norm(i) for i in latimes_corpora if not norm(i) == None]
latimes_corpora = sum(latimes_corpora, [])
return latimes_corpora
def daily_main(paths):
daily_corpora = news_cleaner(paths, 'daily_path')
daily_corpora = [norm(i) for i in daily_corpora if not norm(i) == None]
daily_corpora = sum(daily_corpora, [])
return daily_corpora
def collins_main(paths):
collins_corpora = ling_blogs_cleaner(paths, 'collins_path')
collins_corpora = [norm(i) for i in collins_corpora if not norm(i) == None]
collins_corpora = sum(collins_corpora, [])
return collins_corpora
def houston_main(paths):
houston_corpora = news_cleaner(paths, 'houston_path')
houston_corpora = [norm(i) for i in houston_corpora if not norm(i) == None]
houston_corpora = sum(houston_corpora, [])
return houston_corpora
def wiki_main(paths):
wiki_corpora = wiki_cleaner(paths)
wiki_corpora = [norm(i) for i in wiki_corpora if not norm(i) == None]
wiki_corpora = sum(wiki_corpora, [])
return wiki_corpora
