Tests for check_handling/regex_constants

This commit is contained in:
Clemens Schwaighofer
2025-10-24 13:45:46 +09:00
parent 1226721bc0
commit 3ee3a0dce0
2 changed files with 337 additions and 0 deletions

View File

@@ -0,0 +1 @@
"""Unit tests for check_handling module."""

View File

@@ -0,0 +1,336 @@
"""
Unit tests for regex_constants module.
Tests all regex patterns defined in the check_handling.regex_constants module.
"""
import re
import pytest
from corelibs.check_handling.regex_constants import (
compile_re,
EMAIL_BASIC_REGEX,
DOMAIN_WITH_LOCALHOST_REGEX,
DOMAIN_WITH_LOCALHOST_PORT_REGEX,
DOMAIN_REGEX,
)
class TestCompileRe:
"""Test cases for the compile_re function."""
def test_compile_re_returns_pattern(self) -> None:
"""Test that compile_re returns a compiled regex Pattern object."""
pattern = compile_re(r"test")
assert isinstance(pattern, re.Pattern)
def test_compile_re_with_verbose_flag(self) -> None:
"""Test that compile_re compiles with VERBOSE flag."""
# Verbose mode allows whitespace and comments in regex
verbose_regex = r"""
\d+ # digits
\s+ # whitespace
"""
pattern = compile_re(verbose_regex)
assert pattern.match("123 ")
assert not pattern.match("abc")
def test_compile_re_simple_pattern(self) -> None:
"""Test compile_re with a simple pattern."""
pattern = compile_re(r"^\d{3}$")
assert pattern.match("123")
assert not pattern.match("12")
assert not pattern.match("1234")
class TestEmailBasicRegex:
"""Test cases for EMAIL_BASIC_REGEX pattern."""
@pytest.fixture
def email_pattern(self) -> re.Pattern[str]:
"""Fixture that returns compiled email regex pattern."""
return compile_re(EMAIL_BASIC_REGEX)
@pytest.mark.parametrize("valid_email", [
"user@example.com",
"test.user@example.com",
"user+tag@example.co.uk",
"first.last@subdomain.example.com",
"user123@test-domain.com",
"a@example.com",
"user_name@example.com",
"user-name@example.com",
"user@sub.domain.example.com",
"test!#$%&'*+-/=?^_`{|}~@example.com",
"1234567890@example.com",
"user@example-domain.com",
"user@domain.co",
# Regex allows these (even if not strictly RFC compliant):
"user.@example.com", # ends with dot before @
"user..name@example.com", # consecutive dots in local part
])
def test_valid_emails(
self, email_pattern: re.Pattern[str], valid_email: str
) -> None:
"""Test that valid email addresses match the pattern."""
assert email_pattern.match(valid_email), (
f"Failed to match valid email: {valid_email}"
)
@pytest.mark.parametrize("invalid_email", [
"", # empty string
"@example.com", # missing local part
"user@", # missing domain
"user", # no @ symbol
"user@.com", # domain starts with dot
"user@domain", # no TLD
"user @example.com", # space in local part
"user@exam ple.com", # space in domain
".user@example.com", # starts with dot
"user@-example.com", # domain starts with hyphen
"user@example-.com", # domain part ends with hyphen
"user@example.c", # TLD too short (1 char)
"user@example.toolong", # TLD too long (>6 chars)
"user@@example.com", # double @
"user@example@com", # multiple @
"user@.example.com", # domain starts with dot
"user@example.com.", # ends with dot
"user@123.456.789.012", # numeric TLD not allowed
])
def test_invalid_emails(
self, email_pattern: re.Pattern[str], invalid_email: str
) -> None:
"""Test that invalid email addresses do not match the pattern."""
assert not email_pattern.match(invalid_email), (
f"Incorrectly matched invalid email: {invalid_email}"
)
def test_email_max_local_part_length(
self, email_pattern: re.Pattern[str]
) -> None:
"""Test email with maximum local part length (64 characters)."""
# Local part can be up to 64 chars (first char + 63 more)
local_part = "a" * 64
email = f"{local_part}@example.com"
assert email_pattern.match(email)
def test_email_exceeds_local_part_length(
self, email_pattern: re.Pattern[str]
) -> None:
"""Test email exceeding maximum local part length."""
# 65 characters should not match
local_part = "a" * 65
email = f"{local_part}@example.com"
assert not email_pattern.match(email)
class TestDomainWithLocalhostRegex:
"""Test cases for DOMAIN_WITH_LOCALHOST_REGEX pattern."""
@pytest.fixture
def domain_localhost_pattern(self) -> re.Pattern[str]:
"""Fixture that returns compiled domain with localhost regex pattern."""
return compile_re(DOMAIN_WITH_LOCALHOST_REGEX)
@pytest.mark.parametrize("valid_domain", [
"localhost",
"example.com",
"subdomain.example.com",
"sub.domain.example.com",
"test-domain.com",
"example.co.uk",
"a.com",
"test123.example.com",
"my-site.example.org",
"multi.level.subdomain.example.com",
])
def test_valid_domains(
self, domain_localhost_pattern: re.Pattern[str], valid_domain: str
) -> None:
"""Test that valid domains (including localhost) match the pattern."""
assert domain_localhost_pattern.match(valid_domain), (
f"Failed to match valid domain: {valid_domain}"
)
@pytest.mark.parametrize("invalid_domain", [
"", # empty string
"example", # no TLD
"-example.com", # starts with hyphen
"example-.com", # ends with hyphen
".example.com", # starts with dot
"example.com.", # ends with dot
"example..com", # consecutive dots
"exam ple.com", # space in domain
"example.c", # TLD too short
"localhost:8080", # port not allowed in this pattern
"example.com:8080", # port not allowed in this pattern
"@example.com", # invalid character
"example@com", # invalid character
])
def test_invalid_domains(
self, domain_localhost_pattern: re.Pattern[str], invalid_domain: str
) -> None:
"""Test that invalid domains do not match the pattern."""
assert not domain_localhost_pattern.match(invalid_domain), (
f"Incorrectly matched invalid domain: {invalid_domain}"
)
class TestDomainWithLocalhostPortRegex:
"""Test cases for DOMAIN_WITH_LOCALHOST_PORT_REGEX pattern."""
@pytest.fixture
def domain_localhost_port_pattern(self) -> re.Pattern[str]:
"""Fixture that returns compiled domain and localhost with port pattern."""
return compile_re(DOMAIN_WITH_LOCALHOST_PORT_REGEX)
@pytest.mark.parametrize("valid_domain", [
"localhost",
"localhost:8080",
"localhost:3000",
"localhost:80",
"localhost:443",
"localhost:65535",
"example.com",
"example.com:8080",
"subdomain.example.com:3000",
"test-domain.com:443",
"example.co.uk",
"example.co.uk:8000",
"a.com:1",
"multi.level.subdomain.example.com:9999",
])
def test_valid_domains_with_port(
self, domain_localhost_port_pattern: re.Pattern[str], valid_domain: str
) -> None:
"""Test that valid domains with optional ports match the pattern."""
assert domain_localhost_port_pattern.match(valid_domain), (
f"Failed to match valid domain: {valid_domain}"
)
@pytest.mark.parametrize("invalid_domain", [
"", # empty string
"example", # no TLD
"-example.com", # starts with hyphen
"example-.com", # ends with hyphen
".example.com", # starts with dot
"example.com.", # ends with dot
"localhost:", # port without number
"example.com:", # port without number
"example.com:abc", # non-numeric port
"example.com: 8080", # space before port
"example.com:80 80", # space in port
"exam ple.com", # space in domain
"localhost :8080", # space before colon
])
def test_invalid_domains_with_port(
self,
domain_localhost_port_pattern: re.Pattern[str],
invalid_domain: str,
) -> None:
"""Test that invalid domains do not match the pattern."""
assert not domain_localhost_port_pattern.match(invalid_domain), (
f"Incorrectly matched invalid domain: {invalid_domain}"
)
def test_large_port_number(
self, domain_localhost_port_pattern: re.Pattern[str]
) -> None:
"""Test domain with large port numbers."""
assert domain_localhost_port_pattern.match("example.com:65535")
# Regex doesn't validate port range
assert domain_localhost_port_pattern.match("example.com:99999")
class TestDomainRegex:
"""Test cases for DOMAIN_REGEX pattern (no localhost)."""
@pytest.fixture
def domain_pattern(self) -> re.Pattern[str]:
"""Fixture that returns compiled domain regex pattern."""
return compile_re(DOMAIN_REGEX)
@pytest.mark.parametrize("valid_domain", [
"example.com",
"subdomain.example.com",
"sub.domain.example.com",
"test-domain.com",
"example.co.uk",
"a.com",
"test123.example.com",
"my-site.example.org",
"multi.level.subdomain.example.com",
"example.co",
])
def test_valid_domains_no_localhost(
self, domain_pattern: re.Pattern[str], valid_domain: str
) -> None:
"""Test that valid domains match the pattern."""
assert domain_pattern.match(valid_domain), (
f"Failed to match valid domain: {valid_domain}"
)
@pytest.mark.parametrize("invalid_domain", [
"", # empty string
"localhost", # localhost not allowed
"example", # no TLD
"-example.com", # starts with hyphen
"example-.com", # ends with hyphen
".example.com", # starts with dot
"example.com.", # ends with dot
"example..com", # consecutive dots
"exam ple.com", # space in domain
"example.c", # TLD too short
"example.com:8080", # port not allowed
"@example.com", # invalid character
"example@com", # invalid character
])
def test_invalid_domains_no_localhost(
self, domain_pattern: re.Pattern[str], invalid_domain: str
) -> None:
"""Test that invalid domains do not match the pattern."""
assert not domain_pattern.match(invalid_domain), (
f"Incorrectly matched invalid domain: {invalid_domain}"
)
def test_localhost_not_allowed(
self, domain_pattern: re.Pattern[str]
) -> None:
"""Test that localhost is explicitly not allowed in DOMAIN_REGEX."""
assert not domain_pattern.match("localhost")
class TestRegexPatternConsistency:
"""Test cases for consistency across regex patterns."""
def test_all_patterns_compile(self) -> None:
"""Test that all regex patterns can be compiled without errors."""
patterns = [
EMAIL_BASIC_REGEX,
DOMAIN_WITH_LOCALHOST_REGEX,
DOMAIN_WITH_LOCALHOST_PORT_REGEX,
DOMAIN_REGEX,
]
for pattern in patterns:
compiled = compile_re(pattern)
assert isinstance(compiled, re.Pattern)
def test_domain_patterns_are_strings(self) -> None:
"""Test that all regex constants are strings."""
assert isinstance(EMAIL_BASIC_REGEX, str)
assert isinstance(DOMAIN_WITH_LOCALHOST_REGEX, str)
assert isinstance(DOMAIN_WITH_LOCALHOST_PORT_REGEX, str)
assert isinstance(DOMAIN_REGEX, str)
def test_domain_patterns_hierarchy(self) -> None:
"""Test that domain patterns follow expected hierarchy."""
# DOMAIN_WITH_LOCALHOST_PORT_REGEX should accept everything
# DOMAIN_WITH_LOCALHOST_REGEX accepts
domain_localhost = compile_re(DOMAIN_WITH_LOCALHOST_REGEX)
domain_localhost_port = compile_re(DOMAIN_WITH_LOCALHOST_PORT_REGEX)
test_cases = ["example.com", "subdomain.example.com", "localhost"]
for test_case in test_cases:
if domain_localhost.match(test_case):
assert domain_localhost_port.match(test_case), (
f"{test_case} should match both patterns"
)