backend/tests/test_grounding.py

from __future__ import annotations

from backend.core.grounding_checker import _normalize_violation, check_section
from backend.prompts.grounding_prompt import (
    AUDITOR_SYSTEM_PROMPT,
    GROUNDING_SYSTEM,
    GROUNDING_SYSTEM_PROMPT,
    build_grounding_messages,
)


def test_grounding_messages_match_input_contract():
    msgs = build_grounding_messages(
        "Section body with slate tiles noted.",
        ["note bullet"],
        ["The chimney was inspected from ground level."],
    )
    assert msgs[0]["content"] == AUDITOR_SYSTEM_PROMPT.strip()
    assert msgs[0]["content"] == GROUNDING_SYSTEM_PROMPT.strip()
    assert msgs[0]["content"] == GROUNDING_SYSTEM.strip()
    assert "[1] SURVEYOR_MESSY_NOTES" in msgs[1]["content"]
    assert "[2] GENERATED_REPORT_PARAGRAPH" in msgs[1]["content"]
    assert "[3] BASELINE_PARAGRAPH" in msgs[1]["content"]
    assert "* note bullet" in msgs[1]["content"]
    assert "Section body" in msgs[1]["content"]
    assert "read-only semantic audit engine" in msgs[0]["content"].lower()


def test_normalize_structured_violation():
    entry = {
        "violation_type": "invented_defect",
        "offending_text": "asbestos lagging to pipes",
        "reason": "Not stated in surveyor notes.",
    }
    text = _normalize_violation(entry)
    assert "invented_defect" in text
    assert "asbestos lagging" in text


def test_regex_pass_scrubs_email_without_llm():
    result = check_section(
        "Contact john@example.com about the chimney.",
        ["chimney cracked"],
        ["The chimney stack was inspected."],
    )
    assert "john@example.com" not in result.cleaned_text
    assert result.had_pii


def test_clean_text_unchanged_when_grounded():
    notes = ["chimney stack cracked at apex"]
    template = ["The chimney stack was inspected from ground level."]
    body = "The chimney stack is cracked at the apex."
    result = check_section(body, notes, template)
    assert "cracked" in result.cleaned_text.lower()


def test_generic_template_language_allowed_without_notes_match():
    notes = ["minor cracking noted"]
    body = "Inspection was limited to ground level using binoculars. Minor cracking noted."
    result = check_section(body, notes, [])
    assert "binoculars" in result.cleaned_text.lower()