gendesign/tradein-mvp/backend/tests/test_scraper_proxy.py

"""Tests for scraper proxy wiring (#806).

Covers:
  - scraper_proxy_url property: SCRAPER_PROXY_URL takes precedence over AVITO_PROXY_URL
  - AVITO_PROXY_URL fallback works when SCRAPER_PROXY_URL is absent
  - None/empty → direct connection (no proxies dict)
  - _avito_proxies() helper returns correct dict shape
"""

from __future__ import annotations

import os

os.environ.setdefault("DATABASE_URL", "postgresql+psycopg://test:test@localhost:5432/test")

from types import SimpleNamespace
from unittest.mock import patch


def _mock_settings(scraper_proxy_url: str | None) -> SimpleNamespace:
    """Minimal settings stand-in with only the fields the proxy helpers read."""
    return SimpleNamespace(scraper_proxy_url=scraper_proxy_url)


# ── Settings.scraper_proxy_url property ──────────────────────────────────────
# Test the property logic directly via inline replica — avoids instantiating
# pydantic-settings (requires DATABASE_URL) for pure unit tests of the rule.


def _resolve_scraper_proxy_url(
    scraper_proxy_url_env: str | None, avito_proxy_url: str | None
) -> str | None:
    """Inline replica of Settings.scraper_proxy_url property logic."""
    return scraper_proxy_url_env or avito_proxy_url


def test_scraper_proxy_url_uses_scraper_env_when_set():
    result = _resolve_scraper_proxy_url(
        scraper_proxy_url_env="http://scraper-proxy:1234",
        avito_proxy_url="http://avito-proxy:5678",
    )
    assert result == "http://scraper-proxy:1234"


def test_scraper_proxy_url_falls_back_to_avito_proxy_url():
    """AVITO_PROXY_URL used when SCRAPER_PROXY_URL absent — zero env change on prod."""
    result = _resolve_scraper_proxy_url(
        scraper_proxy_url_env=None,
        avito_proxy_url="http://avito-proxy:5678",
    )
    assert result == "http://avito-proxy:5678"


def test_scraper_proxy_url_none_when_both_absent():
    result = _resolve_scraper_proxy_url(scraper_proxy_url_env=None, avito_proxy_url=None)
    assert result is None


def test_scraper_proxy_url_empty_string_falls_through_to_avito():
    """Empty string is falsy — falls through to avito_proxy_url."""
    result = _resolve_scraper_proxy_url(
        scraper_proxy_url_env="",
        avito_proxy_url="http://avito-proxy:5678",
    )
    assert result == "http://avito-proxy:5678"


# ── _avito_proxies() dict shape ───────────────────────────────────────────────


def test_avito_proxies_returns_correct_dict_shape():
    from app.services.scrape_pipeline import _avito_proxies

    proxy_url = "http://user:pass@proxy.example.com:8080"
    with patch("app.services.scrape_pipeline.settings", _mock_settings(proxy_url)):
        result = _avito_proxies()

    assert result == {"http": proxy_url, "https": proxy_url}


def test_avito_proxies_returns_none_when_no_proxy():
    from app.services.scrape_pipeline import _avito_proxies

    with patch("app.services.scrape_pipeline.settings", _mock_settings(None)):
        result = _avito_proxies()

    assert result is None


def test_avito_proxies_fallback_via_avito_proxy_url():
    """_avito_proxies() uses AVITO_PROXY_URL via scraper_proxy_url fallback.

    When settings.scraper_proxy_url resolves to the avito URL (fallback path),
    the dict shape must be {"http": url, "https": url}.
    """
    from app.services.scrape_pipeline import _avito_proxies

    avito_url = "http://mobile-proxy.space:9090"
    with patch("app.services.scrape_pipeline.settings", _mock_settings(avito_url)):
        result = _avito_proxies()

    assert result == {"http": avito_url, "https": avito_url}


# ── Real pydantic env-binding (regression for #806 fixup) ─────────────────────
# The inline-replica tests above prove the OR-logic but NOT that the field
# actually binds to env SCRAPER_PROXY_URL. `validation_alias` is what makes that
# work — without it pydantic-settings reads SCRAPER_PROXY_URL_ENV (the field
# name), so setting SCRAPER_PROXY_URL would be a silent no-op. These tests
# construct a real Settings and assert end-to-end env binding.


def _fresh_settings():
    from app.core.config import Settings

    return Settings(_env_file=None)


def test_settings_binds_scraper_proxy_url_env(monkeypatch):
    monkeypatch.setenv("SCRAPER_PROXY_URL", "http://scraper-proxy:1234")
    monkeypatch.delenv("AVITO_PROXY_URL", raising=False)
    assert _fresh_settings().scraper_proxy_url == "http://scraper-proxy:1234"


def test_settings_scraper_env_overrides_avito(monkeypatch):
    monkeypatch.setenv("SCRAPER_PROXY_URL", "http://scraper-proxy:1234")
    monkeypatch.setenv("AVITO_PROXY_URL", "http://avito-proxy:5678")
    assert _fresh_settings().scraper_proxy_url == "http://scraper-proxy:1234"


def test_settings_falls_back_to_avito_env(monkeypatch):
    monkeypatch.delenv("SCRAPER_PROXY_URL", raising=False)
    monkeypatch.setenv("AVITO_PROXY_URL", "http://avito-proxy:5678")
    assert _fresh_settings().scraper_proxy_url == "http://avito-proxy:5678"


def test_settings_ignores_misnamed_scraper_proxy_url_env(monkeypatch):
    """Regression: the old wrong env name SCRAPER_PROXY_URL_ENV must NOT bind."""
    monkeypatch.delenv("SCRAPER_PROXY_URL", raising=False)
    monkeypatch.delenv("AVITO_PROXY_URL", raising=False)
    monkeypatch.setenv("SCRAPER_PROXY_URL_ENV", "http://wrong-name:1111")
    assert _fresh_settings().scraper_proxy_url is None