|
|
|
@ -23,16 +23,16 @@ from random import shuffle
|
|
|
|
|
################
|
|
|
|
|
|
|
|
|
|
######
|
|
|
|
|
# Args
|
|
|
|
|
# user_args
|
|
|
|
|
####
|
|
|
|
|
p = ArgumentParser()
|
|
|
|
|
p.add_argument("--host", default="pzwiki.wdka.nl")
|
|
|
|
|
p.add_argument("--path", default="/mw-mediadesign/", help="nb: should end with /")
|
|
|
|
|
p.add_argument("--category", "-c", nargs="*", help="category to query, use -c foo bar to intersect multiple categories")
|
|
|
|
|
p.add_argument("--preview", help='Preview page. Will override category querying. Use: --page "Name Of Wiki Page"')
|
|
|
|
|
parser = ArgumentParser()
|
|
|
|
|
parser.add_argument("--host", default="pzwiki.wdka.nl")
|
|
|
|
|
parser.add_argument("--path", default="/mw-mediadesign/", help="nb: should end with /")
|
|
|
|
|
parser.add_argument("--category", "-c", nargs="*", help="category to query, use -c foo bar to intersect multiple categories")
|
|
|
|
|
parser.add_argument("--preview", help='Preview page. Will override category querying. Use: --page "Name Of Wiki Page"')
|
|
|
|
|
|
|
|
|
|
args = p.parse_args()
|
|
|
|
|
print 'args', args
|
|
|
|
|
user_args = parser.parse_args()
|
|
|
|
|
print 'user_args', user_args
|
|
|
|
|
|
|
|
|
|
######
|
|
|
|
|
# DEFS: create_pages create_index
|
|
|
|
@ -46,6 +46,7 @@ def create_pages(memberpages, mode):
|
|
|
|
|
for member in memberpages:
|
|
|
|
|
page = mw_page(site, member)
|
|
|
|
|
page_text = mw_page_text(site, page)
|
|
|
|
|
# import pdb; pdb.set_trace()
|
|
|
|
|
articledict = parse_work(site, member, page_text) # create dictionary
|
|
|
|
|
# Title, Creator, Date, Website, Thumbnail, Bio, Description, Extra
|
|
|
|
|
if len(articledict['Creator'])>0 and len(articledict['Title'])>0 and len(articledict['Thumbnail'])>0:
|
|
|
|
@ -176,10 +177,14 @@ def create_index(indexdict, year):
|
|
|
|
|
#####
|
|
|
|
|
# ACTION
|
|
|
|
|
#####
|
|
|
|
|
site = mwsite(args.host, args.path)
|
|
|
|
|
|
|
|
|
|
memberpages=mw_cats(site, args)
|
|
|
|
|
site = mwsite(user_args.host, user_args.path)
|
|
|
|
|
# import pdb; pdb.set_trace()
|
|
|
|
|
if user_args.preview:
|
|
|
|
|
memberpages=[unicode(user_args.preview)]
|
|
|
|
|
else:
|
|
|
|
|
memberpages=mw_cats(site, user_args)
|
|
|
|
|
# print '\n\nmemberpages:\n\n', memberpages, '\n\n********\n\n' # memberpages include years, from current to 2015
|
|
|
|
|
|
|
|
|
|
indexdict = create_pages(memberpages, 'index')
|
|
|
|
|
indexdict_byyear={year:{} for year in years } # index of all page organized according to year
|
|
|
|
|
# indexdict_byyear structure:
|
|
|
|
|