def hash(self):
self.size = os.path.getsize(self.path_pdf)
if (self.path_pdf != ''):
self.hash = md5sum_file(self.path_pdf)
else:
self.hash = ""
def __init__(self, url = "", path_pdf = "", path_txt = path_pdf + ".txt"):
self.url = url
self.path_pdf = path_pdf
self.path_txt = path_txt
if (self.path_pdf == ''):
self.hash = ""
self.size = 0
else:
self.hash = md5sum_file(path_pdf)
self.size = os.path.getsize(self.path_pdf)
#If processed = 0, then it has not yet been processed, 1 - successfully processed
#-1 - cannot be processed
self.processed = 0
self.words = []
self.shingles = []
def makehash(self):
if (self.path != None):
self.hash = md5sum_file(self.path)
else:
self.hash = ""