tests/test_safety_classifier.py

"""
tests/test_safety_classifier.py
──────────────────────────────
Unit tests for the regex-based SafetyClassifier.
Run with:  python -m pytest tests/ -v
"""

import sys, os
sys.path.insert(0, os.path.join(os.path.dirname(__file__), ".."))

import pytest
from safety_classifier import SafetyClassifier, ClassificationResult

clf = SafetyClassifier()


# ── Helpers ───────────────────────────────────────────────────────────────────

def is_safe(text: str) -> bool:
    return clf.classify(text).label == "SAFE"

def is_dangerous(text: str) -> bool:
    return clf.classify(text).label == "DANGEROUS"

def categories(text: str) -> list[str]:
    return clf.classify(text).hazards


# ── SAFE cases ────────────────────────────────────────────────────────────────

class TestSafeCaptions:
    def test_normal_street(self):
        assert is_safe("A quiet residential street with parked cars and green trees.")

    def test_indoor_office(self):
        assert is_safe("An office with desks, computers, and people working quietly.")

    def test_park_scene(self):
        assert is_safe("Children playing in a sunny park near a fountain.")

    def test_kitchen(self):
        assert is_safe("A clean kitchen with a counter, sink, and hanging pots.")

    def test_empty_caption(self):
        assert is_safe("")

    def test_whitespace_only(self):
        assert is_safe("   ")


# ── DANGEROUS cases ───────────────────────────────────────────────────────────

class TestDangerousCaptions:

    # Fire
    def test_fire(self):
        assert is_dangerous("There is a large fire burning near the building.")
        assert "fire" in categories("Flames are visible through the window.")

    def test_smoke(self):
        assert is_dangerous("The room is filled with thick smoke.")

    # Flood
    def test_flooding(self):
        assert is_dangerous("The street is flooding rapidly after heavy rain.")
        assert "flood" in categories("Water is flooding into the basement.")

    # Weapons
    def test_gun(self):
        assert is_dangerous("A man is holding a pistol near the entrance.")
        assert "weapon" in categories("A rifle is leaning against the wall.")

    def test_knife(self):
        assert is_dangerous("Someone is brandishing a knife in the crowd.")

    # Vehicle crash
    def test_crash(self):
        assert is_dangerous("Two cars have been involved in a collision on the road.")
        assert "crash" in categories("There is an overturned truck blocking the highway.")

    # Oncoming traffic
    def test_oncoming_vehicle(self):
        assert is_dangerous("An oncoming car is speeding directly toward the pedestrian.")

    # Violence
    def test_fight(self):
        assert is_dangerous("A brawl has broken out in front of the store.")

    # Falls & heights
    def test_cliff(self):
        assert is_dangerous("The person is standing at the edge of a cliff.")
        assert "fall" in categories("A steep drop is just ahead on the trail.")

    # Injury / blood
    def test_blood(self):
        assert is_dangerous("There is blood visible on the floor near the door.")

    # Electrical
    def test_exposed_wire(self):
        assert is_dangerous("An exposed wire is hanging from the ceiling.")

    # Slip hazard
    def test_wet_floor(self):
        assert is_dangerous("The wet floor near the entrance presents a slip hazard.")

    # Storm
    def test_storm(self):
        assert is_dangerous("A tornado is approaching from the west.")

    # Chemical
    def test_chemical_spill(self):
        assert is_dangerous("There is a chemical spill near the laboratory bench.")

    # Generic danger keywords
    def test_generic_danger(self):
        assert is_dangerous("This is a dangerous situation requiring immediate evacuation.")


# ── Edge cases ────────────────────────────────────────────────────────────────

class TestEdgeCases:

    def test_case_insensitive(self):
        assert is_dangerous("FIRE is spreading across the building.")
        assert is_dangerous("There is a KNIFE on the ground.")

    def test_multiple_hazards(self):
        r = clf.classify("Flooding and fire are both present, and there is broken glass from the crash.")
        assert r.is_dangerous
        assert len(r.hazards) >= 2

    def test_explain_method(self):
        breakdown = clf.explain("A fire is burning and an exposed wire is visible.")
        assert "fire"       in breakdown
        assert "electrical" in breakdown

    def test_partial_word_no_match(self):
        # "firing" a weapon is different from "fire" — ensure no false positive
        # (pattern uses \b word boundaries)
        r = clf.classify("The CEO is firing employees in the boardroom.")
        # "firing" does NOT match \bfire\b — should be SAFE
        assert r.label == "SAFE"