def traverse_inner(list, pattern, depth, count):
for url in list:
if re.match(pattern, url) and (not UrlDbHelper.inCheckedTable(url)):
try:
UrlDbHelper.addToCheckedTable(url)
count += 1
print " "*depth + url + " depth:" + str(depth)
except:
print "error adding url:" + url
if count % 100 == 0:
UrlDbHelper.commit()
if depth < MAX_DEPTH:
try:
result = UrlFetcher.fetch_all_links(url)
traverse_inner(result, pattern, depth+1, count)
except:
continue
return
def traverse(root, pattern):
UrlDbHelper.initDB()
count = 0
traverse_inner([root,], pattern, count, 0)
UrlDbHelper.commit()
print str(count) + " tuples have been added"