def test_match_same_uuid(self):
"""Test if there is a match when compares identities with the same UUID"""
uid1 = UniqueIdentity(uuid='John Smith')
uid2 = UniqueIdentity(uuid='John Smith')
matcher = EmailNameMatcher()
result = matcher.match(uid1, uid2)
self.assertEqual(result, True)
result = matcher.match(uid2, uid1)
self.assertEqual(result, True)
# None UUIDs do not produce a positive match
uid1 = UniqueIdentity(uuid=None)
uid2 = UniqueIdentity(uuid=None)
matcher = EmailNameMatcher()
result = matcher.match(uid1, uid2)
self.assertEqual(result, False)
result = matcher.match(uid2, uid1)
self.assertEqual(result, False)
def test_match_with_sources_list(self):
"""Test match when a list of sources to filter is given"""
jsmith = UniqueIdentity(uuid='jsmith')
jsmith.identities = [
Identity(name='John Smith',
email='*****@*****.**',
source='scm'),
Identity(name='John Smith', source='scm'),
Identity(username='******', source='scm'),
Identity(email='', source='scm')
]
jsmith_alt = UniqueIdentity(uuid='J. Smith')
jsmith_alt.identities = [
Identity(name='John Smith JR', username='******',
source='alt'),
Identity(name='John Smith', username='******', source='alt'),
Identity(email='', source='alt'),
Identity(email='jsmith', source='alt')
]
# With these lists there are not matches
matcher = EmailNameMatcher(sources=['github'])
result = matcher.match(jsmith, jsmith_alt)
self.assertEqual(result, False)
matcher = EmailNameMatcher(sources=['scm'])
result = matcher.match(jsmith, jsmith_alt)
self.assertEqual(result, False)
# Only when scm and alt are combined there is a match
matcher = EmailNameMatcher(sources=['scm', 'alt'])
result = matcher.match(jsmith, jsmith_alt)
self.assertEqual(result, True)
def test_match_same_uuid(self):
"""Test if there is a match when compares identities with the same UUID"""
uid1 = UniqueIdentity(uuid='John Smith')
uid2 = UniqueIdentity(uuid='John Smith')
matcher = EmailNameMatcher()
result = matcher.match(uid1, uid2)
self.assertEqual(result, True)
result = matcher.match(uid2, uid1)
self.assertEqual(result, True)
# None UUIDs do not produce a positive match
uid1 = UniqueIdentity(uuid=None)
uid2 = UniqueIdentity(uuid=None)
matcher = EmailNameMatcher()
result = matcher.match(uid1, uid2)
self.assertEqual(result, False)
result = matcher.match(uid2, uid1)
self.assertEqual(result, False)
def test_match_with_sources_list(self):
"""Test match when a list of sources to filter is given"""
jsmith = UniqueIdentity(uuid='jsmith')
jsmith.identities = [Identity(name='John Smith', email='*****@*****.**', source='scm'),
Identity(name='John Smith', source='scm'),
Identity(username='******', source='scm'),
Identity(email='', source='scm')]
jsmith_alt = UniqueIdentity(uuid='J. Smith')
jsmith_alt.identities = [Identity(name='John Smith JR', username='******', source='alt'),
Identity(name='John Smith', username='******', source='alt'),
Identity(email='', source='alt'),
Identity(email='jsmith', source='alt')]
# With these lists there are not matches
matcher = EmailNameMatcher(sources=['github'])
result = matcher.match(jsmith, jsmith_alt)
self.assertEqual(result, False)
matcher = EmailNameMatcher(sources=['scm'])
result = matcher.match(jsmith, jsmith_alt)
self.assertEqual(result, False)
# Only when scm and alt are combined there is a match
matcher = EmailNameMatcher(sources=['scm', 'alt'])
result = matcher.match(jsmith, jsmith_alt)
self.assertEqual(result, True)
def test_match_strict(self):
"""Test strict matching"""
# Let's define some identities first
jsmith_alt = UniqueIdentity(uuid='J. Smith')
jsmith_alt.identities = [
Identity(name='J. Smith', username='******', source='alt'),
Identity(name='John Smith', username='******', source='alt'),
Identity(email='', source='alt'),
Identity(email='jsmith', source='alt')
]
jsmith_not_email = UniqueIdentity(uuid='John Smith')
jsmith_not_email.identities = [Identity(email='jsmith', source='mls')]
jrae = UniqueIdentity(uuid='jrae')
jrae.identities = [
Identity(name='Jane Rae', source='scm'),
Identity(name='Jane Rae Doe',
email='*****@*****.**',
source='mls')
]
jrae_doe = UniqueIdentity(uuid='jraedoe')
jrae_doe.identities = [
Identity(name='Jane Rae Doe',
email='*****@*****.**',
source='mls'),
Identity(name='jrae', source='scm')
]
jrae_no_name = UniqueIdentity(uuid='Jane Rae')
jrae_no_name.identities = [Identity(name='jrae', source='scm')]
# Tests
strict_matcher = EmailNameMatcher(strict=True)
no_strict_matcher = EmailNameMatcher(strict=False)
# This two unique identities have the same email address
# but due to 'jsmith' is not a valid email address, they
# do not match
result = strict_matcher.match(jsmith_alt, jsmith_not_email)
self.assertEqual(result, False)
# But with strict mode off they do
result = no_strict_matcher.match(jsmith_alt, jsmith_not_email)
self.assertEqual(result, True)
# This two do not match although they share the same name.
# In this case the name is invalid because is not formed
# like "firstname lastname"
result = strict_matcher.match(jrae_doe, jrae_no_name)
self.assertEqual(result, False)
# But with strict mode off they do
result = no_strict_matcher.match(jrae_doe, jrae_no_name)
self.assertEqual(result, True)
def test_match_same_identity(self):
"""Test whether there is a match comparing the same identity"""
uid = UniqueIdentity(uuid='John Smith')
matcher = EmailNameMatcher()
result = matcher.match(uid, uid)
self.assertEqual(result, True)
def test_match_same_identity(self):
"""Test whether there is a match comparing the same identity"""
uid = UniqueIdentity(uuid='John Smith')
matcher = EmailNameMatcher()
result = matcher.match(uid, uid)
self.assertEqual(result, True)
def test_match_with_blacklist(self):
"""Test match when there are entries in the blacklist"""
jsmith = UniqueIdentity(uuid='jsmith')
jsmith.identities = [
Identity(name='John Smith',
email='*****@*****.**',
source='scm'),
Identity(name='John Smith', source='scm'),
Identity(username='******', source='scm'),
Identity(email='', source='scm')
]
john_smith = UniqueIdentity(uuid='js')
john_smith.identities = [
Identity(name='John Smith JR', username='******',
source='scm'),
Identity(username='******', source='scm'),
Identity(name='Smith. J', source='mls'),
Identity(name='Smith. J', email='*****@*****.**', source='mls')
]
jsmith_alt = UniqueIdentity(uuid='J. Smith')
jsmith_alt.identities = [
Identity(name='John Smith JR', username='******',
source='alt'),
Identity(name='John Smith', username='******', source='alt'),
Identity(email='', source='alt'),
Identity(email='jsmith', source='alt')
]
# Tests
bl = [
MatchingBlacklist(excluded='John Smith'),
MatchingBlacklist(excluded='*****@*****.**')
]
matcher = EmailNameMatcher(blacklist=bl)
result = matcher.match(jsmith, john_smith)
self.assertEqual(result, False)
result = matcher.match(john_smith, jsmith)
self.assertEqual(result, False)
# John Smith is blacklisted, so no match
result = matcher.match(jsmith, jsmith_alt)
self.assertEqual(result, False)
result = matcher.match(jsmith_alt, jsmith)
self.assertEqual(result, False)
result = matcher.match(john_smith, jsmith_alt)
self.assertEqual(result, True)
result = matcher.match(jsmith_alt, john_smith)
self.assertEqual(result, True)
def test_match_with_blacklist(self):
"""Test match when there are entries in the blacklist"""
jsmith = UniqueIdentity(uuid='jsmith')
jsmith.identities = [Identity(name='John Smith', email='*****@*****.**', source='scm'),
Identity(name='John Smith', source='scm'),
Identity(username='******', source='scm'),
Identity(email='', source='scm')]
john_smith = UniqueIdentity(uuid='js')
john_smith.identities = [Identity(name='John Smith JR', username='******', source='scm'),
Identity(username='******', source='scm'),
Identity(name='Smith. J', source='mls'),
Identity(name='Smith. J', email='*****@*****.**', source='mls')]
jsmith_alt = UniqueIdentity(uuid='J. Smith')
jsmith_alt.identities = [Identity(name='John Smith JR', username='******', source='alt'),
Identity(name='John Smith', username='******', source='alt'),
Identity(email='', source='alt'),
Identity(email='jsmith', source='alt')]
# Tests
bl = [MatchingBlacklist(excluded='John Smith'),
MatchingBlacklist(excluded='*****@*****.**')]
matcher = EmailNameMatcher(blacklist=bl)
result = matcher.match(jsmith, john_smith)
self.assertEqual(result, False)
result = matcher.match(john_smith, jsmith)
self.assertEqual(result, False)
# John Smith is blacklisted, so no match
result = matcher.match(jsmith, jsmith_alt)
self.assertEqual(result, False)
result = matcher.match(jsmith_alt, jsmith)
self.assertEqual(result, False)
result = matcher.match(john_smith, jsmith_alt)
self.assertEqual(result, True)
result = matcher.match(jsmith_alt, john_smith)
self.assertEqual(result, True)
def test_match(self):
"""Test match method"""
# Let's define some identities first
jsmith = UniqueIdentity(uuid='jsmith')
jsmith.identities = [
Identity(name='John Smith',
email='*****@*****.**',
source='scm'),
Identity(name='John Smith', source='scm'),
Identity(username='******', source='scm'),
Identity(email='', source='scm')
]
john_smith = UniqueIdentity(uuid='js')
john_smith.identities = [
Identity(name='J. Smith', username='******', source='scm'),
Identity(username='******', source='scm'),
Identity(name='Smith. J', source='mls'),
Identity(name='Smith. J', email='*****@*****.**', source='mls')
]
jsmith_alt = UniqueIdentity(uuid='J. Smith')
jsmith_alt.identities = [
Identity(name='J. Smith', username='******', source='alt'),
Identity(name='John Smith', username='******', source='alt'),
Identity(email='', source='alt'),
Identity(email='jsmith', source='alt')
]
jsmith_not_email = UniqueIdentity(uuid='John Smith')
jsmith_not_email.identities = [Identity(email='jsmith', source='mls')]
jrae = UniqueIdentity(uuid='jrae')
jrae.identities = [
Identity(name='Jane Rae', source='scm'),
Identity(name='Jane Rae Doe',
email='*****@*****.**',
source='mls')
]
jrae_doe = UniqueIdentity(uuid='jraedoe')
jrae_doe.identities = [
Identity(name='Jane Rae Doe',
email='*****@*****.**',
source='mls'),
Identity(name='jrae', source='scm')
]
jrae_no_name = UniqueIdentity(uuid='Jane Rae')
jrae_no_name.identities = [Identity(name='jrae', source='scm')]
# Tests
matcher = EmailNameMatcher()
# First two unique identities must match
result = matcher.match(jsmith, john_smith)
self.assertEqual(result, True)
result = matcher.match(john_smith, jsmith)
self.assertEqual(result, True)
# Comparing with the third only the first one
# produces a match because of "John Smith" name
result = matcher.match(jsmith, jsmith_alt)
self.assertEqual(result, True)
result = matcher.match(jsmith_alt, jsmith)
self.assertEqual(result, True)
result = matcher.match(john_smith, jsmith_alt)
self.assertEqual(result, False)
result = matcher.match(jsmith_alt, john_smith)
self.assertEqual(result, False)
# Jane Rae matches Jane Rae Doe because they share
# the same name "Jane Rae Doe"
result = matcher.match(jrae, jrae_doe)
self.assertEqual(result, True)
result = matcher.match(jrae, jrae_doe)
self.assertEqual(result, True)
# No match with Jane Rae
result = matcher.match(jsmith, jrae)
self.assertEqual(result, False)
result = matcher.match(jsmith, jrae_doe)
self.assertEqual(result, False)
result = matcher.match(john_smith, jrae)
self.assertEqual(result, False)
result = matcher.match(john_smith, jrae_doe)
self.assertEqual(result, False)
result = matcher.match(jsmith_alt, jrae)
self.assertEqual(result, False)
result = matcher.match(jsmith_alt, jrae_doe)
self.assertEqual(result, False)
# This two unique identities have the same email address
# but due to 'jsmith' is not a valid email address, they
# do not match
result = matcher.match(jsmith_alt, jsmith_not_email)
self.assertEqual(result, False)
# This two do not match although they share the same name.
# In this case the name is invalid because is not formed
# like "firstname lastname"
result = matcher.match(jrae_doe, jrae_no_name)
self.assertEqual(result, False)
def test_match(self):
"""Test match method"""
# Let's define some identities first
jsmith = UniqueIdentity(uuid='jsmith')
jsmith.identities = [Identity(name='John Smith', email='*****@*****.**', source='scm'),
Identity(name='John Smith', source='scm'),
Identity(username='******', source='scm'),
Identity(email='', source='scm')]
john_smith = UniqueIdentity(uuid='js')
john_smith.identities = [Identity(name='J. Smith', username='******', source='scm'),
Identity(username='******', source='scm'),
Identity(name='Smith. J', source='mls'),
Identity(name='Smith. J', email='*****@*****.**', source='mls')]
jsmith_alt = UniqueIdentity(uuid='J. Smith')
jsmith_alt.identities = [Identity(name='J. Smith', username='******', source='alt'),
Identity(name='John Smith', username='******', source='alt'),
Identity(email='', source='alt'),
Identity(email='jsmith', source='alt')]
jsmith_not_email = UniqueIdentity(uuid='John Smith')
jsmith_not_email.identities = [Identity(email='jsmith', source='mls')]
jrae = UniqueIdentity(uuid='jrae')
jrae.identities = [Identity(name='Jane Rae', source='scm'),
Identity(name='Jane Rae Doe', email='*****@*****.**', source='mls')]
jrae_doe = UniqueIdentity(uuid='jraedoe')
jrae_doe.identities = [Identity(name='Jane Rae Doe', email='*****@*****.**', source='mls'),
Identity(name='jrae', source='scm')]
jrae_no_name = UniqueIdentity(uuid='Jane Rae')
jrae_no_name.identities = [Identity(name='jrae', source='scm')]
# Tests
matcher = EmailNameMatcher()
# First two unique identities must match
result = matcher.match(jsmith, john_smith)
self.assertEqual(result, True)
result = matcher.match(john_smith, jsmith)
self.assertEqual(result, True)
# Comparing with the third only the first one
# produces a match because of "John Smith" name
result = matcher.match(jsmith, jsmith_alt)
self.assertEqual(result, True)
result = matcher.match(jsmith_alt, jsmith)
self.assertEqual(result, True)
result = matcher.match(john_smith, jsmith_alt)
self.assertEqual(result, False)
result = matcher.match(jsmith_alt, john_smith)
self.assertEqual(result, False)
# Jane Rae matches Jane Rae Doe because they share
# the same name "Jane Rae Doe"
result = matcher.match(jrae, jrae_doe)
self.assertEqual(result, True)
result = matcher.match(jrae, jrae_doe)
self.assertEqual(result, True)
# No match with Jane Rae
result = matcher.match(jsmith, jrae)
self.assertEqual(result, False)
result = matcher.match(jsmith, jrae_doe)
self.assertEqual(result, False)
result = matcher.match(john_smith, jrae)
self.assertEqual(result, False)
result = matcher.match(john_smith, jrae_doe)
self.assertEqual(result, False)
result = matcher.match(jsmith_alt, jrae)
self.assertEqual(result, False)
result = matcher.match(jsmith_alt, jrae_doe)
self.assertEqual(result, False)
# This two unique identities have the same email address
# but due to 'jsmith' is not a valid email address, they
# do not match
result = matcher.match(jsmith_alt, jsmith_not_email)
self.assertEqual(result, False)
# This two do not match although they share the same name.
# In this case the name is invalid because is not formed
# like "firstname lastname"
result = matcher.match(jrae_doe, jrae_no_name)
self.assertEqual(result, False)