resolutionflow/backend/tests/test_redaction_service.py

"""Tests for sensitive data redaction service."""

import pytest
from app.services.redaction_service import (
    apply_redaction_to_text,
    format_redaction_footer,
    RedactionSummary,
)


class TestIPv4Redaction:
    def test_single_ipv4(self):
        text = "Server at 192.168.1.100 is down"
        result, summary = apply_redaction_to_text(text)
        assert result == "Server at [IP REDACTED] is down"
        assert summary.ips == 1

    def test_multiple_ipv4(self):
        text = "Route from 10.0.0.1 to 172.16.0.5"
        result, summary = apply_redaction_to_text(text)
        assert "[IP REDACTED]" in result
        assert "10.0.0.1" not in result
        assert "172.16.0.5" not in result
        assert summary.ips == 2

    def test_ipv4_at_boundaries(self):
        text = "10.0.0.1\n192.168.1.1"
        result, summary = apply_redaction_to_text(text)
        assert summary.ips == 2
        assert "10.0.0.1" not in result


class TestIPv6Redaction:
    def test_full_ipv6(self):
        text = "Address: 2001:0db8:85a3:0000:0000:8a2e:0370:7334"
        result, summary = apply_redaction_to_text(text)
        assert result == "Address: [IP REDACTED]"
        assert summary.ips == 1

    def test_abbreviated_ipv6(self):
        text = "fe80:1234:abcd:5678:9abc"
        result, summary = apply_redaction_to_text(text)
        assert "[IP REDACTED]" in result
        assert summary.ips == 1


class TestEmailRedaction:
    def test_simple_email(self):
        text = "Contact admin@company.com for help"
        result, summary = apply_redaction_to_text(text)
        assert result == "Contact [EMAIL REDACTED] for help"
        assert summary.emails == 1

    def test_complex_email(self):
        text = "Send to john.doe+tag@sub.domain.co.uk"
        result, summary = apply_redaction_to_text(text)
        assert "[EMAIL REDACTED]" in result
        assert summary.emails == 1

    def test_multiple_emails(self):
        text = "From user@a.com to admin@b.org"
        result, summary = apply_redaction_to_text(text)
        assert summary.emails == 2
        assert "user@a.com" not in result


class TestTokenRedaction:
    def test_bearer_token(self):
        text = "Authorization: Bearer eyJhbGciOiJIUzI1NiJ9.payload.sig"
        result, summary = apply_redaction_to_text(text)
        assert "Bearer" not in result or "[TOKEN REDACTED]" in result
        assert summary.tokens >= 1

    def test_long_api_key(self):
        text = "API key: a1b2c3d4e5f6a1b2c3d4e5f6a1b2c3d4"
        result, summary = apply_redaction_to_text(text)
        assert "[TOKEN REDACTED]" in result
        assert summary.tokens >= 1

    def test_short_string_not_matched(self):
        text = "Short code: abc123"
        result, summary = apply_redaction_to_text(text)
        assert "abc123" in result  # Too short to be a token


class TestUNCPathRedaction:
    def test_simple_unc(self):
        text = r"Map drive to \\fileserver\shared"
        result, summary = apply_redaction_to_text(text)
        assert result == "Map drive to [UNC PATH REDACTED]"
        assert summary.unc_paths == 1

    def test_unc_with_dollar_share(self):
        text = r"Access \\server01\C$"
        result, summary = apply_redaction_to_text(text)
        assert "[UNC PATH REDACTED]" in result
        assert summary.unc_paths == 1


class TestMixedContent:
    def test_multiple_pattern_types(self):
        text = (
            "Server 192.168.1.1 has user admin@corp.com "
            r"and share \\filesvr\data"
        )
        result, summary = apply_redaction_to_text(text)
        assert "192.168.1.1" not in result
        assert "admin@corp.com" not in result
        assert r"\\filesvr\data" not in result
        assert summary.ips == 1
        assert summary.emails == 1
        assert summary.unc_paths == 1

    def test_no_sensitive_data(self):
        text = "Everything is working fine. No issues found."
        result, summary = apply_redaction_to_text(text)
        assert result == text
        assert summary.total == 0


class TestEdgeCases:
    def test_empty_string(self):
        result, summary = apply_redaction_to_text("")
        assert result == ""
        assert summary.total == 0

    def test_idempotency(self):
        """Already-redacted content should not produce extra matches."""
        text = "Server at [IP REDACTED] and [EMAIL REDACTED]"
        result, summary = apply_redaction_to_text(text)
        assert result == text
        assert summary.total == 0

    def test_redaction_then_re_redaction(self):
        """Running redaction twice produces the same output."""
        text = "Contact admin@test.com at 10.0.0.1"
        first_pass, _ = apply_redaction_to_text(text)
        second_pass, summary2 = apply_redaction_to_text(first_pass)
        assert first_pass == second_pass
        assert summary2.total == 0


class TestRedactionSummary:
    def test_total_calculation(self):
        s = RedactionSummary(ips=2, emails=1, tokens=3, unc_paths=1)
        assert s.total == 7

    def test_to_dict(self):
        s = RedactionSummary(ips=1, emails=2, tokens=0, unc_paths=0)
        d = s.to_dict()
        assert d == {"ips": 1, "emails": 2, "tokens": 0, "unc_paths": 0, "total": 3}


class TestRedactionFooter:
    def test_no_matches(self):
        assert format_redaction_footer(RedactionSummary()) == ""

    def test_single_category(self):
        footer = format_redaction_footer(RedactionSummary(ips=3))
        assert footer == "\n--- Redacted: 3 IPs ---"

    def test_multiple_categories(self):
        footer = format_redaction_footer(RedactionSummary(ips=1, emails=2, tokens=1))
        assert "1 IP" in footer
        assert "2 emails" in footer
        assert "1 token" in footer

    def test_singular_forms(self):
        footer = format_redaction_footer(RedactionSummary(ips=1, emails=1))
        assert "1 IP," in footer or "1 IP ---" in footer
        assert "1 email" in footer