def run(self):
import readWikipedia
pattern = ''
first = True
for line in open(self.input, 'rU'):
line = line.strip()
if first:
if len(line) > 0:
pattern = line
first = False
else:
if len(line) > 0:
pattern = pattern + '|' + line
if len(self.section_name.strip()) > 0:
pattern = pattern + ':' + self.section_name.strip()
print pattern
readWikipedia.ReadWikipedia(self.wiki_directory, pattern,
self.categories, self.output,
'temp/pickfromwikipediatitles',
self.break_sentences)
def run(self):
import readWikipedia
readWikipedia.ReadWikipedia(self.wiki_directory, self.pattern,
self.categories, self.out_file,
self.titles_file, self.break_sentences)