integrity_scanner_fuer_statische_Webseiten/tests/test_cloak.py

"""Tests for cloaking detection."""
import pytest
from scanner.differ import compare_cloaking, score_cloak_diff


def _page(url, text="Hallo Welt", links=None, vary_ua=False):
    hdrs = {"Vary": "User-Agent"} if vary_ua else {}
    return {
        "url": url,
        "text": text,
        "links": {"a": links or [], "script": [], "iframe": []},
        "response_headers": hdrs,
        "status": 200,
    }


CFG = {"normalization": {"collapse_whitespace": True, "ignore_patterns": []}, "scoring": {}, "thresholds": {}}


class TestCompareCloaking:
    def test_identical_pages_no_findings(self):
        page = _page("https://x.de/")
        result = compare_cloaking({"https://x.de/": page}, {"https://x.de/": page}, CFG)
        assert result["findings"] == []
        assert result["pages_checked"] == 1

    def test_extra_link_in_bot_version_flagged(self):
        normal = _page("https://x.de/", links=[{"url": "https://x.de/a", "class": "internal"}])
        bot = _page("https://x.de/", links=[
            {"url": "https://x.de/a", "class": "internal"},
            {"url": "https://spam.example.com/evil", "class": "external"},
        ])
        result = compare_cloaking({"https://x.de/": normal}, {"https://x.de/": bot}, CFG)
        assert len(result["findings"]) == 1
        assert "https://spam.example.com/evil" in result["findings"][0]["extra_links"]

    def test_extra_text_in_bot_version_flagged(self):
        normal = _page("https://x.de/", text="Normaler Text")
        bot = _page("https://x.de/", text="Normaler Text " + "x" * 200)
        result = compare_cloaking({"https://x.de/": normal}, {"https://x.de/": bot}, CFG)
        assert len(result["findings"]) == 1
        assert result["findings"][0]["extra_text_chars"] > 100

    def test_vary_ua_header_flagged(self):
        normal = _page("https://x.de/")
        bot = _page("https://x.de/", vary_ua=True)
        result = compare_cloaking({"https://x.de/": normal}, {"https://x.de/": bot}, CFG)
        assert len(result["findings"]) == 1
        assert result["findings"][0]["vary_ua"] is True

    def test_url_only_in_one_crawl_ignored(self):
        """URLs not seen in both crawls are not compared."""
        normal = _page("https://x.de/only-normal")
        bot = _page("https://x.de/only-bot")
        result = compare_cloaking(
            {"https://x.de/only-normal": normal},
            {"https://x.de/only-bot": bot},
            CFG,
        )
        assert result["findings"] == []
        assert result["pages_checked"] == 0


class TestScoreCloakDiff:
    def test_no_findings_green(self):
        result = score_cloak_diff({"findings": [], "pages_checked": 5}, CFG)
        assert result["level"] == "green"
        assert result["score"] == 0
        assert result["exit_code"] == 0

    def test_extra_link_scores_60_red(self):
        diff = {
            "findings": [{"url": "https://x.de/", "vary_ua": False, "extra_links": ["https://spam.com/x"], "extra_text_chars": 0}],
            "pages_checked": 1,
        }
        result = score_cloak_diff(diff, CFG)
        assert result["score"] == 60
        assert result["level"] == "red"
        assert result["exit_code"] == 2

    def test_extra_text_scores_40_yellow(self):
        diff = {
            "findings": [{"url": "https://x.de/", "vary_ua": False, "extra_links": [], "extra_text_chars": 150}],
            "pages_checked": 1,
        }
        result = score_cloak_diff(diff, CFG)
        assert result["score"] == 40
        assert result["level"] == "yellow"

    def test_vary_ua_only_scores_25_yellow(self):
        diff = {
            "findings": [{"url": "https://x.de/", "vary_ua": True, "extra_links": [], "extra_text_chars": 0}],
            "pages_checked": 1,
        }
        result = score_cloak_diff(diff, CFG)
        assert result["score"] == 25
        assert result["level"] == "yellow"

    def test_vary_ua_with_extra_link_not_double_scored(self):
        """vary_ua surcharge is suppressed when extra_links are present (link is already 60)."""
        diff = {
            "findings": [{"url": "https://x.de/", "vary_ua": True, "extra_links": ["https://spam.com/x"], "extra_text_chars": 0}],
            "pages_checked": 1,
        }
        result = score_cloak_diff(diff, CFG)
        assert result["score"] == 60  # only link score, not 60+25