fix(tradein/tests): harden parity harness — bool-type mismatch detection + divergence-catch proof (#2304)

Code-reviewer follow-up on the parity harness (aaaf8179): - parity.py: plain `==` in the scalar-equality branch let `True == 1` / `False == 0` silently pass (Python bool is an int subclass). A future migration bug turning a `bool | None` field into a raw 0/1 would slip through undetected. Now any type mismatch where exactly one side is a bool is reported as a diff, regardless of numeric equality. - test_parity.py: unit test for the new bool-vs-int branch. - test_avito_detail_kit_parity.py: the existing smoke test only proved the harness reports "no diff" on two genuinely-identical real dataclass instances — it never proved the harness catches a real divergence on this same 30+-field shape (only the toy fixtures in test_parity.py did). Added a test that mutates `price_rub` via dataclasses.replace() on the real kit DetailEnrichment and asserts assert_parity raises ParityMismatchError naming that field.
2026-07-03 23:45:18 +03:00 · 2026-07-03 23:45:18 +03:00 · d95c9e6e55
commit d95c9e6e55
parent 57963e3a06
3 changed files with 58 additions and 1 deletions
--- a/tradein-mvp/backend/tests/scrapers/test_avito_detail_kit_parity.py
+++ b/tradein-mvp/backend/tests/scrapers/test_avito_detail_kit_parity.py
@ -29,6 +29,8 @@ from __future__ import annotations
 import dataclasses
 import os

+import pytest
+
 # app.services.scrapers.avito_detail импортирует app.core.config.settings=Settings(),
 # которому нужен DATABASE_URL. Офлайн-парсинг БД не трогает — фиктивный DSN достаточен
 # (mirror tests/test_scraper_kit_avito_golden_parity.py — тест должен жить и без
@ -42,7 +44,7 @@ from scraper_kit.providers.avito.detail import (
 from app.services.scrapers.avito_detail import (
    parse_detail_html as legacy_parse_detail_html,
 )
-from tests.support.parity import assert_parity
+from tests.support.parity import ParityMismatchError, assert_parity

 _DETAIL_HTML = """
 <html><body>
@ -100,3 +102,28 @@ def test_parity_harness_confirms_avito_detail_parse_equivalence() -> None:
        kit_fn=kit_parse_detail_html,
        fixtures=[(_DETAIL_HTML, _SOURCE_URL)],
    )
+
+
+def test_parity_harness_catches_real_field_divergence() -> None:
+    """Доказывает, что harness реально ЛОВИТ расхождение на этой же реальной
+    30+-полевой dataclass-форме — а не только сообщает "нет разницы" на двух
+    genuinely-identical инстансах, как предыдущий тест.
+
+    Мутируем один field (``price_rub``) в kit-результате через
+    ``dataclasses.replace`` и убеждаемся, что harness падает с
+    ``ParityMismatchError``, называющим именно этот field.
+    """
+    legacy_result = legacy_parse_detail_html(_DETAIL_HTML, _SOURCE_URL)
+    kit_result = kit_parse_detail_html(_DETAIL_HTML, _SOURCE_URL)
+    assert kit_result.price_rub is not None  # sanity: поле реально распарсилось из фикстуры
+
+    mutated_kit_result = dataclasses.replace(kit_result, price_rub=kit_result.price_rub + 1)
+
+    with pytest.raises(ParityMismatchError) as exc_info:
+        assert_parity(
+            legacy_fn=lambda: legacy_result,
+            kit_fn=lambda: mutated_kit_result,
+            fixtures=[()],
+        )
+
+    assert "price_rub" in str(exc_info.value)
--- a/tradein-mvp/backend/tests/support/parity.py
+++ b/tradein-mvp/backend/tests/support/parity.py
@ -111,6 +111,15 @@ def _diff(
            _diff(f"{path}[{i}]", lv, kv, ignore_fields, tolerance, diffs)
        return

+    if isinstance(legacy_val, bool) != isinstance(kit_val, bool):
+        # bool — подкласс int в Python, поэтому `True == 1` и `False == 0` дают True
+        # при обычном `==` (проверять надо isinstance(x, bool), НЕ isinstance(x, int)).
+        # Миграция, случайно превратившая `bool | None` поле в сырой 0/1 (или наоборот),
+        # должна считаться расхождением, даже если числовое значение совпадает.
+        diffs.append(
+            f"{path}: type mismatch (bool vs non-bool) — legacy={legacy_val!r}, kit={kit_val!r}"
+        )
+        return
    if legacy_val == kit_val:
        return
    if _within_tolerance(legacy_val, kit_val, tolerance):
--- a/tradein-mvp/backend/tests/support/test_parity.py
+++ b/tradein-mvp/backend/tests/support/test_parity.py
@ -98,6 +98,27 @@ def test_dataclass_instances_of_different_classes_compared_structurally() -> Non
    assert compare_outputs(legacy, kit) == []


+def test_bool_vs_int_type_mismatch_is_not_silently_equal() -> None:
+    # Python bool — подкласс int: `True == 1` и `False == 0` дают True при обычном
+    # `==`. Миграция, случайно превратившая bool-поле в сырой 0/1, должна ловиться.
+    def legacy_fn(x: int) -> dict[str, object]:
+        return {"mortgage_available": True}
+
+    def kit_fn(x: int) -> dict[str, object]:
+        return {"mortgage_available": 1}  # намеренно int вместо bool
+
+    with pytest.raises(ParityMismatchError) as exc_info:
+        assert_parity(legacy_fn, kit_fn, fixtures=[1])
+
+    message = str(exc_info.value)
+    assert "mortgage_available" in message
+    assert "bool" in message
+
+    # Обратный случай (False vs 0) и совпадающие типы (bool vs bool) — симметрично.
+    assert compare_outputs({"v": False}, {"v": 0}) != []
+    assert compare_outputs({"v": True}, {"v": True}) == []
+
+
 def test_dataclass_field_mismatch_reported_by_name() -> None:
    @dataclass
    class LegacyResult: