feat(backend): block tld (#12240)

Co-authored-by: sp.wack <83104063+amanape@users.noreply.github.com>
This commit is contained in:
Hiep Le
2026-01-03 00:42:22 +07:00
committed by GitHub
parent 1bae1fc4e6
commit 63d5ceada6
2 changed files with 338 additions and 7 deletions

View File

@@ -31,7 +31,13 @@ class DomainBlocker:
return None
def is_domain_blocked(self, email: str) -> bool:
"""Check if email domain is blocked"""
"""Check if email domain is blocked
Supports blocking:
- Exact domains: 'example.com' blocks 'user@example.com'
- Subdomains: 'example.com' blocks 'user@subdomain.example.com'
- TLDs: '.us' blocks 'user@company.us' and 'user@subdomain.company.us'
"""
if not self.is_active():
return False
@@ -44,13 +50,26 @@ class DomainBlocker:
logger.debug(f'Could not extract domain from email: {email}')
return False
is_blocked = domain in self.blocked_domains
if is_blocked:
logger.warning(f'Email domain {domain} is blocked for email: {email}')
else:
logger.debug(f'Email domain {domain} is not blocked')
# Check if domain matches any blocked pattern
for blocked_pattern in self.blocked_domains:
if blocked_pattern.startswith('.'):
# TLD pattern (e.g., '.us') - check if domain ends with it
if domain.endswith(blocked_pattern):
logger.warning(
f'Email domain {domain} is blocked by TLD pattern {blocked_pattern} for email: {email}'
)
return True
else:
# Full domain pattern (e.g., 'example.com')
# Block exact match or subdomains
if domain == blocked_pattern or domain.endswith(f'.{blocked_pattern}'):
logger.warning(
f'Email domain {domain} is blocked by domain pattern {blocked_pattern} for email: {email}'
)
return True
return is_blocked
logger.debug(f'Email domain {domain} is not blocked')
return False
domain_blocker = DomainBlocker()

View File

@@ -179,3 +179,315 @@ def test_is_domain_blocked_with_whitespace(domain_blocker):
# Assert
assert result is True
# ============================================================================
# TLD Blocking Tests (patterns starting with '.')
# ============================================================================
def test_is_domain_blocked_tld_pattern_blocks_matching_domain(domain_blocker):
"""Test that TLD pattern blocks domains ending with that TLD."""
# Arrange
domain_blocker.blocked_domains = ['.us']
# Act
result = domain_blocker.is_domain_blocked('user@company.us')
# Assert
assert result is True
def test_is_domain_blocked_tld_pattern_blocks_subdomain_with_tld(domain_blocker):
"""Test that TLD pattern blocks subdomains with that TLD."""
# Arrange
domain_blocker.blocked_domains = ['.us']
# Act
result = domain_blocker.is_domain_blocked('user@subdomain.company.us')
# Assert
assert result is True
def test_is_domain_blocked_tld_pattern_does_not_block_different_tld(domain_blocker):
"""Test that TLD pattern does not block domains with different TLD."""
# Arrange
domain_blocker.blocked_domains = ['.us']
# Act
result = domain_blocker.is_domain_blocked('user@company.com')
# Assert
assert result is False
def test_is_domain_blocked_tld_pattern_does_not_block_substring_match(
domain_blocker,
):
"""Test that TLD pattern does not block domains that contain but don't end with the TLD."""
# Arrange
domain_blocker.blocked_domains = ['.us']
# Act
result = domain_blocker.is_domain_blocked('user@focus.com')
# Assert
assert result is False
def test_is_domain_blocked_tld_pattern_case_insensitive(domain_blocker):
"""Test that TLD pattern matching is case-insensitive."""
# Arrange
domain_blocker.blocked_domains = ['.us']
# Act
result = domain_blocker.is_domain_blocked('user@COMPANY.US')
# Assert
assert result is True
def test_is_domain_blocked_multiple_tld_patterns(domain_blocker):
"""Test blocking with multiple TLD patterns."""
# Arrange
domain_blocker.blocked_domains = ['.us', '.vn', '.com']
# Act
result_us = domain_blocker.is_domain_blocked('user@test.us')
result_vn = domain_blocker.is_domain_blocked('user@test.vn')
result_com = domain_blocker.is_domain_blocked('user@test.com')
result_org = domain_blocker.is_domain_blocked('user@test.org')
# Assert
assert result_us is True
assert result_vn is True
assert result_com is True
assert result_org is False
def test_is_domain_blocked_tld_pattern_with_multi_level_tld(domain_blocker):
"""Test that TLD pattern works with multi-level TLDs like .co.uk."""
# Arrange
domain_blocker.blocked_domains = ['.co.uk']
# Act
result_match = domain_blocker.is_domain_blocked('user@example.co.uk')
result_subdomain = domain_blocker.is_domain_blocked('user@api.example.co.uk')
result_no_match = domain_blocker.is_domain_blocked('user@example.uk')
# Assert
assert result_match is True
assert result_subdomain is True
assert result_no_match is False
# ============================================================================
# Subdomain Blocking Tests (domain patterns now block subdomains)
# ============================================================================
def test_is_domain_blocked_domain_pattern_blocks_exact_match(domain_blocker):
"""Test that domain pattern blocks exact domain match."""
# Arrange
domain_blocker.blocked_domains = ['example.com']
# Act
result = domain_blocker.is_domain_blocked('user@example.com')
# Assert
assert result is True
def test_is_domain_blocked_domain_pattern_blocks_subdomain(domain_blocker):
"""Test that domain pattern blocks subdomains of that domain."""
# Arrange
domain_blocker.blocked_domains = ['example.com']
# Act
result = domain_blocker.is_domain_blocked('user@subdomain.example.com')
# Assert
assert result is True
def test_is_domain_blocked_domain_pattern_blocks_multi_level_subdomain(
domain_blocker,
):
"""Test that domain pattern blocks multi-level subdomains."""
# Arrange
domain_blocker.blocked_domains = ['example.com']
# Act
result = domain_blocker.is_domain_blocked('user@api.v2.example.com')
# Assert
assert result is True
def test_is_domain_blocked_domain_pattern_does_not_block_similar_domain(
domain_blocker,
):
"""Test that domain pattern does not block domains that contain but don't match the pattern."""
# Arrange
domain_blocker.blocked_domains = ['example.com']
# Act
result = domain_blocker.is_domain_blocked('user@notexample.com')
# Assert
assert result is False
def test_is_domain_blocked_domain_pattern_does_not_block_different_tld(
domain_blocker,
):
"""Test that domain pattern does not block same domain with different TLD."""
# Arrange
domain_blocker.blocked_domains = ['example.com']
# Act
result = domain_blocker.is_domain_blocked('user@example.org')
# Assert
assert result is False
def test_is_domain_blocked_subdomain_pattern_blocks_exact_and_nested(domain_blocker):
"""Test that blocking a subdomain also blocks its nested subdomains."""
# Arrange
domain_blocker.blocked_domains = ['api.example.com']
# Act
result_exact = domain_blocker.is_domain_blocked('user@api.example.com')
result_nested = domain_blocker.is_domain_blocked('user@v1.api.example.com')
result_parent = domain_blocker.is_domain_blocked('user@example.com')
# Assert
assert result_exact is True
assert result_nested is True
assert result_parent is False
# ============================================================================
# Mixed Pattern Tests (TLD + domain patterns together)
# ============================================================================
def test_is_domain_blocked_mixed_patterns_tld_and_domain(domain_blocker):
"""Test blocking with both TLD and domain patterns."""
# Arrange
domain_blocker.blocked_domains = ['.us', 'openhands.dev']
# Act
result_tld = domain_blocker.is_domain_blocked('user@company.us')
result_domain = domain_blocker.is_domain_blocked('user@openhands.dev')
result_subdomain = domain_blocker.is_domain_blocked('user@api.openhands.dev')
result_allowed = domain_blocker.is_domain_blocked('user@example.com')
# Assert
assert result_tld is True
assert result_domain is True
assert result_subdomain is True
assert result_allowed is False
def test_is_domain_blocked_overlapping_patterns(domain_blocker):
"""Test that overlapping patterns (TLD and specific domain) both work."""
# Arrange
domain_blocker.blocked_domains = ['.us', 'test.us']
# Act
result_specific = domain_blocker.is_domain_blocked('user@test.us')
result_other_us = domain_blocker.is_domain_blocked('user@other.us')
# Assert
assert result_specific is True
assert result_other_us is True
def test_is_domain_blocked_complex_multi_pattern_scenario(domain_blocker):
"""Test complex scenario with multiple TLD and domain patterns."""
# Arrange
domain_blocker.blocked_domains = [
'.us',
'.vn',
'test.com',
'openhands.dev',
]
# Act & Assert
# TLD patterns
assert domain_blocker.is_domain_blocked('user@anything.us') is True
assert domain_blocker.is_domain_blocked('user@company.vn') is True
# Domain patterns (exact)
assert domain_blocker.is_domain_blocked('user@test.com') is True
assert domain_blocker.is_domain_blocked('user@openhands.dev') is True
# Domain patterns (subdomains)
assert domain_blocker.is_domain_blocked('user@api.test.com') is True
assert domain_blocker.is_domain_blocked('user@staging.openhands.dev') is True
# Not blocked
assert domain_blocker.is_domain_blocked('user@allowed.com') is False
assert domain_blocker.is_domain_blocked('user@example.org') is False
# ============================================================================
# Edge Case Tests
# ============================================================================
def test_is_domain_blocked_domain_with_hyphens(domain_blocker):
"""Test that domain patterns work with hyphenated domains."""
# Arrange
domain_blocker.blocked_domains = ['my-company.com']
# Act
result_exact = domain_blocker.is_domain_blocked('user@my-company.com')
result_subdomain = domain_blocker.is_domain_blocked('user@api.my-company.com')
# Assert
assert result_exact is True
assert result_subdomain is True
def test_is_domain_blocked_domain_with_numbers(domain_blocker):
"""Test that domain patterns work with numeric domains."""
# Arrange
domain_blocker.blocked_domains = ['test123.com']
# Act
result_exact = domain_blocker.is_domain_blocked('user@test123.com')
result_subdomain = domain_blocker.is_domain_blocked('user@api.test123.com')
# Assert
assert result_exact is True
assert result_subdomain is True
def test_is_domain_blocked_short_tld(domain_blocker):
"""Test that short TLD patterns work correctly."""
# Arrange
domain_blocker.blocked_domains = ['.io']
# Act
result = domain_blocker.is_domain_blocked('user@company.io')
# Assert
assert result is True
def test_is_domain_blocked_very_long_subdomain_chain(domain_blocker):
"""Test that blocking works with very long subdomain chains."""
# Arrange
domain_blocker.blocked_domains = ['example.com']
# Act
result = domain_blocker.is_domain_blocked(
'user@level4.level3.level2.level1.example.com'
)
# Assert
assert result is True