def readURLFile(filename):
# Read in list of urls from file
url_list = open(filename).readlines()
# Build a list of canonicalized versions of the urls
canon_url_list = buildCanonicalList(url_list)
# Print out url information
for url in url_list:
print "Source:\n" + url
print "Valid: " + str(url_validate(url))
print "Canonical: " + url_canonicalize(url)
print "Source unique: " + str((url_list.count(url) == 1))
print "Canonicalized URL unique: " + str((canon_url_list.count(url_canonicalize(url)) == 1))
def readURLFile(filename):
# Read in list of urls from file
url_list = open(filename).readlines()
# Build a list of canonicalized versions of the urls
canon_url_list = buildCanonicalList(url_list)
# Print out url information
for url in url_list:
print "Source:\n" + url
print "Valid: " + str(url_validate(url))
print "Canonical: " + url_canonicalize(url)
print "Source unique: " + str((url_list.count(url) == 1))
print "Canonicalized URL unique: " + str(
(canon_url_list.count(url_canonicalize(url)) == 1))
def buildCanonicalList(url_list):
canon_url_list = []
for url in url_list:
canon_url_list.append(url_canonicalize(url))
return canon_url_list
def buildCanonicalList(url_list):
canon_url_list = []
for url in url_list:
canon_url_list.append(url_canonicalize(url))
return canon_url_list
def runTest(self):
assert url_canonicalize(original) == normalized, (original, normalized, url_canonicalize(original))
def runTest(self):
assert (url_canonicalize(value) == value) == expected, (expected, value, url_canonicalize(value))
