import requests
import json
import sys

#parse the parameter indicating the number of sites to create the statics for
for  arg in sys.argv:
	try:
		numItems = int(arg)
	except:
		pass

#function for reading the raw results
def getState():
	statefile = open("crawl-state.json","r")
	state = json.loads(statefile.read())
	statefile.close()
	return state

#read the list of sites
domains = []
fileHandle = open("top-1m.csv","r")
line = fileHandle.readline()
while (line):
        splited = line.split(",")

        domains.append([int(splited[0]),splited[1][:-1]])
        line = fileHandle.readline()
fileHandle.close()

#read the raw results
results = getState()

#the containervariables for the statistics
scriptDomains = 0
noScriptDomains = 0
errorDomains = 0

#calculate the statistics 
for i in range(1,numItems+1):
	try:
		results[str(i)]
	except:
		print("skipped "+str(i))	
		errorDomains += 1
		continue

	if results[str(i)][1] == "Sucess":
		if results[str(i)][2] == "script":
			scriptDomains += 1
		else:
			noScriptDomains += 1
	else:
		errorDomains += 1

#print the statistics
print("scriptDomains: "+str(scriptDomains))
print("noScriptDomains: "+str(noScriptDomains))
print("errorDomains: "+str(errorDomains))