|
| 1 | +""" |
| 2 | +Headless-browser XSS regression tests (issue #11 / sprint item #3). |
| 3 | +
|
| 4 | +Exercises the production render path: Marked.js + DOMPurify via |
| 5 | +renderMarkdownSafe() in static/js/app.js, then DOM insertion via innerHTML |
| 6 | +(the same pattern as templates/workspace.html). |
| 7 | +
|
| 8 | +Pages are served with the same CSP as production; inline event handlers are |
| 9 | +blocked by the header, so these tests assert dangerous markup is stripped from |
| 10 | +the sink (and use a probe where execution is still observable). |
| 11 | +
|
| 12 | +Static source greps live in tests/test_xss_sanitization.py. |
| 13 | +
|
| 14 | +Run: |
| 15 | + playwright install chromium |
| 16 | + pytest -q tests/test_xss_browser.py |
| 17 | +""" |
| 18 | + |
| 19 | +from __future__ import annotations |
| 20 | + |
| 21 | +import threading |
| 22 | +from typing import TYPE_CHECKING, Any, Generator |
| 23 | + |
| 24 | +import pytest |
| 25 | +from werkzeug.serving import make_server |
| 26 | + |
| 27 | +from app import create_app |
| 28 | + |
| 29 | +if TYPE_CHECKING: |
| 30 | + from playwright.sync_api import Page |
| 31 | + |
| 32 | +# Representative vectors from the sprint issue. |
| 33 | +XSS_VECTORS: list[tuple[str, str]] = [ |
| 34 | + ("img_onerror", '<img src=x onerror="window.__xssProbe=1">'), |
| 35 | + ("script_tag", "<script>window.__xssProbe=1</script>"), |
| 36 | + ("javascript_uri", "[x](javascript:window.__xssProbe=1)"), |
| 37 | + ( |
| 38 | + "data_uri", |
| 39 | + "[x](data:text/html,<script>window.__xssProbe=1</script>)", |
| 40 | + ), |
| 41 | + ( |
| 42 | + "svg_onload", |
| 43 | + '<svg xmlns="http://www.w3.org/2000/svg" onload="window.__xssProbe=1"></svg>', |
| 44 | + ), |
| 45 | +] |
| 46 | + |
| 47 | +_PROBE_SETTLE_MS = 150 |
| 48 | + |
| 49 | +_INSPECT_XSS_SINK = f""" |
| 50 | +async ({{ payload, useSafeRender }}) => {{ |
| 51 | + window.__xssProbe = 0; |
| 52 | + const host = document.getElementById('xss-browser-test-host'); |
| 53 | + if (host) {{ |
| 54 | + host.remove(); |
| 55 | + }} |
| 56 | + const el = document.createElement('div'); |
| 57 | + el.id = 'xss-browser-test-host'; |
| 58 | + document.body.appendChild(el); |
| 59 | + let sanitizeCalls = 0; |
| 60 | + let restoreSanitize = null; |
| 61 | + if (useSafeRender) {{ |
| 62 | + const originalSanitize = DOMPurify.sanitize.bind(DOMPurify); |
| 63 | + DOMPurify.sanitize = (...args) => {{ |
| 64 | + sanitizeCalls += 1; |
| 65 | + return originalSanitize(...args); |
| 66 | + }}; |
| 67 | + restoreSanitize = () => {{ |
| 68 | + DOMPurify.sanitize = originalSanitize; |
| 69 | + }}; |
| 70 | + }} |
| 71 | + try {{ |
| 72 | + if (useSafeRender) {{ |
| 73 | + if (typeof renderMarkdownSafe !== 'function') {{ |
| 74 | + throw new Error('renderMarkdownSafe is not defined — is app.js loaded?'); |
| 75 | + }} |
| 76 | + el.innerHTML = renderMarkdownSafe(payload); |
| 77 | + }} else {{ |
| 78 | + const html = marked.parse(payload, {{ breaks: true, gfm: true }}); |
| 79 | + el.innerHTML = html; |
| 80 | + }} |
| 81 | + await new Promise((resolve) => setTimeout(resolve, {_PROBE_SETTLE_MS})); |
| 82 | + const result = {{ |
| 83 | + probe: window.__xssProbe || 0, |
| 84 | + onerrorAttr: el.querySelector('[onerror]') !== null, |
| 85 | + scriptTag: el.querySelector('script') !== null, |
| 86 | + jsHref: el.querySelector('[href^="javascript:"]') !== null, |
| 87 | + dataHref: el.querySelector('[href^="data:"]') !== null, |
| 88 | + svgOnload: el.querySelector('svg[onload]') !== null, |
| 89 | + sanitizeCalls: useSafeRender ? sanitizeCalls : 0, |
| 90 | + }}; |
| 91 | + if (!useSafeRender) {{ |
| 92 | + result.html = el.innerHTML; |
| 93 | + }} |
| 94 | + return result; |
| 95 | + }} finally {{ |
| 96 | + if (restoreSanitize) {{ |
| 97 | + restoreSanitize(); |
| 98 | + }} |
| 99 | + }} |
| 100 | +}} |
| 101 | +""" |
| 102 | + |
| 103 | + |
| 104 | +def _assert_sink_neutralized(result: dict[str, Any], vector_name: str) -> None: |
| 105 | + assert result["probe"] == 0, ( |
| 106 | + f"XSS probe fired for vector {vector_name!r}; " |
| 107 | + "renderMarkdownSafe must neutralize this payload" |
| 108 | + ) |
| 109 | + assert not result["onerrorAttr"], ( |
| 110 | + f"onerror attribute survived sanitization for {vector_name!r}" |
| 111 | + ) |
| 112 | + assert not result["scriptTag"], ( |
| 113 | + f"<script> survived sanitization for {vector_name!r}" |
| 114 | + ) |
| 115 | + assert not result["jsHref"], ( |
| 116 | + f"javascript: URI survived sanitization for {vector_name!r}" |
| 117 | + ) |
| 118 | + assert not result["dataHref"], ( |
| 119 | + f"data: URI survived sanitization for {vector_name!r}" |
| 120 | + ) |
| 121 | + assert not result["svgOnload"], ( |
| 122 | + f"svg onload survived sanitization for {vector_name!r}" |
| 123 | + ) |
| 124 | + assert result.get("sanitizeCalls", 0) >= 1, ( |
| 125 | + f"DOMPurify.sanitize was not called for {vector_name!r}; " |
| 126 | + "renderMarkdownSafe must not take the escapeHtml-only path for these payloads" |
| 127 | + ) |
| 128 | + |
| 129 | + |
| 130 | +@pytest.fixture(scope="module") |
| 131 | +def playwright_browser(): |
| 132 | + pytest.importorskip("playwright") |
| 133 | + from playwright.sync_api import sync_playwright |
| 134 | + |
| 135 | + with sync_playwright() as playwright: |
| 136 | + browser = playwright.chromium.launch(headless=True) |
| 137 | + yield browser |
| 138 | + browser.close() |
| 139 | + |
| 140 | + |
| 141 | +@pytest.fixture |
| 142 | +def browser_page(playwright_browser) -> Generator["Page", None, None]: |
| 143 | + page = playwright_browser.new_page() |
| 144 | + try: |
| 145 | + yield page |
| 146 | + finally: |
| 147 | + page.close() |
| 148 | + |
| 149 | + |
| 150 | +@pytest.fixture |
| 151 | +def live_server_url(workspace_storage: str) -> Generator[str, None, None]: |
| 152 | + app = create_app() |
| 153 | + app.config["TESTING"] = True |
| 154 | + app.config["EXCLUSION_RULES"] = [] |
| 155 | + server = make_server("127.0.0.1", 0, app) |
| 156 | + thread = threading.Thread(target=server.serve_forever, daemon=True) |
| 157 | + thread.start() |
| 158 | + host, port = server.server_address[:2] |
| 159 | + try: |
| 160 | + yield f"http://{host}:{port}" |
| 161 | + finally: |
| 162 | + server.shutdown() |
| 163 | + |
| 164 | + |
| 165 | +@pytest.fixture |
| 166 | +def app_page(browser_page: "Page", live_server_url: str) -> "Page": |
| 167 | + """Any HTML page that loads base.html scripts (marked, DOMPurify, app.js).""" |
| 168 | + response = browser_page.goto(f"{live_server_url}/", wait_until="networkidle") |
| 169 | + assert response is not None and response.ok |
| 170 | + browser_page.wait_for_function( |
| 171 | + "() => typeof renderMarkdownSafe === 'function' && typeof DOMPurify !== 'undefined'" |
| 172 | + ) |
| 173 | + return browser_page |
| 174 | + |
| 175 | + |
| 176 | +@pytest.mark.browser |
| 177 | +@pytest.mark.parametrize("vector_name,payload", XSS_VECTORS, ids=[v[0] for v in XSS_VECTORS]) |
| 178 | +def test_render_markdown_safe_neutralizes_xss_vector( |
| 179 | + app_page: "Page", vector_name: str, payload: str |
| 180 | +) -> None: |
| 181 | + result = app_page.evaluate( |
| 182 | + _INSPECT_XSS_SINK, {"payload": payload, "useSafeRender": True} |
| 183 | + ) |
| 184 | + _assert_sink_neutralized(result, vector_name) |
| 185 | + |
| 186 | + |
| 187 | +@pytest.mark.browser |
| 188 | +def test_bare_marked_parse_leaves_dangerous_markup_negative_control( |
| 189 | + app_page: "Page", |
| 190 | +) -> None: |
| 191 | + """Without DOMPurify.sanitize, marked output still carries exploitable markup.""" |
| 192 | + payload = XSS_VECTORS[0][1] |
| 193 | + result = app_page.evaluate( |
| 194 | + _INSPECT_XSS_SINK, {"payload": payload, "useSafeRender": False} |
| 195 | + ) |
| 196 | + assert result["onerrorAttr"] or result["scriptTag"], ( |
| 197 | + "negative control: bare marked.parse should leave at least one dangerous " |
| 198 | + "node/attribute in the DOM sink" |
| 199 | + ) |
0 commit comments