def prepare(self):
Const.log("URLID " + str(self.url_id))
Const.log("BESTTERM " + self.bestterm)
if(self.snippet == "" and not self.bestterm == ""):
Const.log("in if")
filename = Const.PGFOLDER + str(self.url_id)
ucfile = None
with open(filename, 'r') as myfile:
Const.log("File open")
entirefile = myfile.read()
ucfile = unicode(entirefile, 'utf-8').lower()
anIndex=0
look=True
while(look):
anIndex = ucfile.find(self.bestterm, anIndex+1)
Const.log("Find ind " + str(anIndex))
if(not anIndex == -1):
start = max(anIndex-RANGE, 0)
Const.log("START : " + str(start))
end = min(len(ucfile), anIndex + RANGE)
Const.log("END : " + str(end))
rng = end - start
Const.log("RNG : " + str(rng))
self.snippet = ucfile[start:end]
Const.log("SNIPPET : " + self.snippet)
look=check(self.snippet)
else:
look=False
self.snippet=''
