#Find connections between out links to initial webpages, to add to edges.scala
from avro import schema, datafile, io
import re

def read_avro_file():
	rec_reader = io.DatumReader()
	df_reader = datafile.DataFileReader(open("/Users/vcedeno/Desktop/webpages/social_00000_v2"),rec_reader)
	file2 = open("/Users/vcedeno/Desktop/webpages/edgesID.scala",'w')

	for record in df_reader:
	    id=int(re.search(r'\d+$', record['doc_id']).group())
	    urlN=record['url']
	    #searches in the whole file for an in link
	    file = open("/Users/vcedeno/Desktop/webpages/vertexID.scala",'r')
	    for line in file:
	    	ids,urls=line.split(" ")
	    	if urlN==urls:
	    		file2.write("Edge("+str(id)+"L,"+str(ids)+"L,200),")
	    file.close
	file2.close
	    		
if __name__ == '__main__':
	read_avro_file()