def main():
""" main script function """
def extr_mini_maxi(titl):
""" extract bounds from title"""
res = re.split(u"[()-]", titl)
return res[1], res[2]
items = [ item_by_title("fr", title) for title in MAINS ]
ranges = [ extr_mini_maxi(title) for title in MAINS ]
# items : the main articles, 7 main ranges, separated into subranges each
all_items = items
all_ranges = ranges
make_sequence(items)
for (item, rang_) in zip(items, ranges):
(min_, max_) = rang_
prefix = min_[0:-4]
print ("====================='{}'========================".format(prefix))
def gen_title(lrange):
""" title gen"""
mi_ = ('{}{}'.format(prefix, lrange.split(" ")[0]))
ma_ = ('{}{}'.format(prefix, lrange.split(" ")[1]))
# import pdb ; pdb.set_trace()
return frtitle(mi_, ma_)
titles = [ gen_title(lrange) for lrange in LESSER.split("\n") ]
items = [ item_by_title("fr", title) for title in titles ]
ranges = [ extr_mini_maxi(title) for title in titles ]
make_sequence(items)
# suboptimal
all_items = all_items + items
all_ranges = all_ranges + ranges
for (item, (min_, max_)) in zip(all_items, all_ranges):
set_for_lang( item, u"Table des caractères Unicode", "fr", label(min_, max_), "ambiguity and label correction")
set_for_lang( item, u"", "en", enlabel(min_, max_), "ambiguity and label correction")
# correction of previous bug as it seems
set_for_lang( item, u"Unicode characters from 100000 to 10FFFF codepoints",\
"en", enlabel(min_, max_), "ambiguity and label correction")
def treat_serie(serie_name, site_name = 'en', \
main_page_name = None, num = None, start_year = None, title_pattern = "{}_{}"):
""" main """
if not main_page_name:
main_page_name = serie_name
site = pywikibot.getSite(site_name)
print("Serie : {}, Page: {}".format(serie_name, main_page_name) )
serie_item = wd_lib.item_by_title(site, main_page_name)
# Patterns of the titles of the series
has_previous = True
current = 1
items = {}
if not num:
num = 1000
year = start_year
while has_previous and current < num:
title = title_pattern.format(serie_name, year)
pywikibot.output("searching article : {}".format(title))
page = pywikibot.Page(site, title)
print(title)
if page.exists():
datapage = pywikibot.DataPage(page)
if datapage.exists():
datapage.get()
items[current] = datapage
else:
raise Exception()
current += 1
year = year + 1
else:
has_previous = False
num_season = current - 1
print("Number of seasons : {}".format(num_season))
year = start_year
for i in range(1, len(items) + 1):
print("season {}, item: {}". format(i, items[i]))
set_season_labels(items[i], serie_name, i, year)
year = year + 1
# part of (P361): this item is a part of that item
items[i] = wd_lib.reloaditempage(items[i])
wd_lib.maybe_set_claim(items[i], 361, serie_item)
# wd_lib.instance_of(items[i], wd_lib.item_by_title("fr", u"Saison (télévision)"))
type_item = wd_lib.item_by_title("fr", u"Saison (télévision)")
print(type_item)
wd_lib.make_sequence(items.itervalues(), type_item)