| 1 | import json |
| 2 | import os |
| 3 | import shutil |
| 4 | import subprocess |
| 5 | import textwrap |
| 6 | import unittest |
| 7 | from pathlib import Path |
| 8 | from unittest import mock |
| 9 | |
| 10 | from lib.bird_x import parse_bird_response |
| 11 | |
| 12 | REPO_ROOT = Path(__file__).resolve().parents[1] |
| 13 | VENDORED_BIRD = REPO_ROOT / "skills" / "last30days" / "scripts" / "lib" / "vendor" / "bird-search" / "bird-search.mjs" |
| 14 | |
| 15 | |
| 16 | class TestSubprocessEnv(unittest.TestCase): |
| 17 | """_subprocess_env() passes only the vendored client's env surface (issue #1063).""" |
| 18 | |
| 19 | def _ambient(self, **overrides): |
| 20 | base = { |
| 21 | "PATH": "/usr/bin:/bin", |
| 22 | "HOME": "/home/test", |
| 23 | "NODE_ENV": "production", |
| 24 | "AUTH_TOKEN": "ambient-token", |
| 25 | "CT0": "ambient-ct0", |
| 26 | "TWITTER_AUTH_TOKEN": "ambient-tw-token", |
| 27 | "TWITTER_CT0": "ambient-tw-ct0", |
| 28 | "LAST30DAYS_DISABLE_BROWSER_COOKIES": "0", |
| 29 | "BIRD_QUERY_IDS_CACHE": "/tmp/ids.json", |
| 30 | "SCRAPECREATORS_API_KEY": "sc-secret", |
| 31 | "AWS_SECRET_ACCESS_KEY": "aws-secret", |
| 32 | } |
| 33 | base.update(overrides) |
| 34 | return base |
| 35 | |
| 36 | def _call(self, ambient, credentials=None): |
| 37 | from lib import bird_x |
| 38 | old = bird_x._credentials |
| 39 | try: |
| 40 | bird_x._credentials = dict(credentials or {}) |
| 41 | # None means "absent" - os.environ cannot hold None values. |
| 42 | patch_env = {k: v for k, v in ambient.items() if v is not None} |
| 43 | with mock.patch.dict(os.environ, patch_env, clear=True): |
| 44 | return bird_x._subprocess_env() |
| 45 | finally: |
| 46 | bird_x._credentials = old |
| 47 | |
| 48 | def test_unrelated_ambient_secrets_are_excluded(self): |
| 49 | out = self._call(self._ambient()) |
| 50 | self.assertNotIn("SCRAPECREATORS_API_KEY", out) |
| 51 | self.assertNotIn("AWS_SECRET_ACCESS_KEY", out) |
| 52 | |
| 53 | def test_runtime_and_client_vars_pass_through(self): |
| 54 | out = self._call(self._ambient()) |
| 55 | self.assertEqual("/usr/bin:/bin", out.get("PATH")) |
| 56 | self.assertEqual("/home/test", out.get("HOME")) |
| 57 | self.assertEqual("production", out.get("NODE_ENV")) |
| 58 | self.assertEqual("ambient-token", out.get("AUTH_TOKEN")) |
| 59 | self.assertEqual("ambient-ct0", out.get("CT0")) |
| 60 | self.assertEqual("ambient-tw-token", out.get("TWITTER_AUTH_TOKEN")) |
| 61 | self.assertEqual("ambient-tw-ct0", out.get("TWITTER_CT0")) |
| 62 | self.assertEqual("0", out.get("LAST30DAYS_DISABLE_BROWSER_COOKIES")) |
| 63 | self.assertEqual("/tmp/ids.json", out.get("BIRD_QUERY_IDS_CACHE")) |
| 64 | |
| 65 | def test_absent_vars_are_omitted(self): |
| 66 | out = self._call(self._ambient(NODE_ENV=None)) |
| 67 | self.assertNotIn("NODE_ENV", out) |
| 68 | out = self._call(self._ambient(LAST30DAYS_DISABLE_BROWSER_COOKIES=None)) |
| 69 | self.assertNotIn("LAST30DAYS_DISABLE_BROWSER_COOKIES", out) |
| 70 | |
| 71 | def test_injected_credentials_override_ambient(self): |
| 72 | out = self._call(self._ambient(), credentials={"AUTH_TOKEN": "injected", "CT0": "injected-ct0"}) |
| 73 | self.assertEqual("injected", out.get("AUTH_TOKEN")) |
| 74 | self.assertEqual("injected-ct0", out.get("CT0")) |
| 75 | |
| 76 | def test_disable_flag_always_hard_set(self): |
| 77 | out = self._call(self._ambient(), credentials={"AUTH_TOKEN": "t", "CT0": "c"}) |
| 78 | self.assertEqual("1", out.get("BIRD_DISABLE_BROWSER_COOKIES")) |
| 79 | # ambient 0 is overridden to 1 |
| 80 | out = self._call(self._ambient(BIRD_DISABLE_BROWSER_COOKIES="0")) |
| 81 | self.assertEqual("1", out.get("BIRD_DISABLE_BROWSER_COOKIES")) |
| 82 | |
| 83 | def test_ambient_bird_vars_pass_through(self): |
| 84 | out = self._call(self._ambient(BIRD_FEATURES_PATH="/tmp/features.json")) |
| 85 | self.assertEqual("/tmp/features.json", out.get("BIRD_FEATURES_PATH")) |
| 86 | |
| 87 | def test_allowlist_covers_vendored_client_env_reads(self): |
| 88 | """Every process.env name the vendored client reads is reachable. |
| 89 | |
| 90 | Guards the allowlist against vendor drift: a future bird-search bump |
| 91 | that reads a new non-BIRD_ env var must either be added to the |
| 92 | allowlist or fail here, keeping the child env surface explicit |
| 93 | (issue #1063). |
| 94 | """ |
| 95 | import re |
| 96 | |
| 97 | from lib import bird_x |
| 98 | |
| 99 | vendor_dir = REPO_ROOT / "skills" / "last30days" / "scripts" / "lib" / "vendor" / "bird-search" |
| 100 | reads = set() |
| 101 | for path in list(vendor_dir.rglob("*.js")) + list(vendor_dir.rglob("*.mjs")): |
| 102 | text = path.read_text(encoding="utf-8") |
| 103 | for m in re.finditer( |
| 104 | r"process\.env\[\s*['\"]([A-Z0-9_]+)['\"]\s*\]|process\.env\.([A-Z0-9_]+)", |
| 105 | text, |
| 106 | ): |
| 107 | reads.add(m.group(1) or m.group(2)) |
| 108 | # cookies.js reads via helpers with the keys passed as arguments: |
| 109 | # envFlagEnabled('NAME') and readEnvCookie(cookies, ['A', 'B'], ...). |
| 110 | for m in re.finditer(r"envFlagEnabled\(\s*['\"]([A-Z0-9_]+)['\"]\s*\)", text): |
| 111 | reads.add(m.group(1)) |
| 112 | for m in re.finditer(r"readEnvCookie\(\s*\w+\s*,\s*\[([^\]]*)\]", text): |
| 113 | reads.update(re.findall(r"['\"]([A-Z0-9_]+)['\"]", m.group(1))) |
| 114 | self.assertTrue(reads, "vendored client env reads not found") |
| 115 | allowlist = set(bird_x._SUBPROCESS_ENV_ALLOWLIST) |
| 116 | uncovered = { |
| 117 | name for name in reads |
| 118 | if not name.startswith("BIRD_") and name not in allowlist |
| 119 | } |
| 120 | self.assertEqual(set(), uncovered) |
| 121 | |
| 122 | |
| 123 | class TestBirdXEngagementZero(unittest.TestCase): |
| 124 | def test_zero_likes_preserved(self): |
| 125 | tweets = [ |
| 126 | { |
| 127 | "id": "1", |
| 128 | "text": "test", |
| 129 | "permanent_url": "https://x.com/u/status/1", |
| 130 | "likeCount": 0, |
| 131 | "retweetCount": 5, |
| 132 | } |
| 133 | ] |
| 134 | items = parse_bird_response(tweets, "test query") |
| 135 | self.assertEqual(0, items[0]["engagement"]["likes"]) |
| 136 | self.assertEqual(5, items[0]["engagement"]["reposts"]) |
| 137 | |
| 138 | @unittest.skipUnless(shutil.which("node"), "node is required for vendored Bird tests") |
| 139 | class TestVendoredBirdRuntime(unittest.TestCase): |
| 140 | def test_check_uses_env_credentials_without_browser_cookie_dependency(self): |
| 141 | env = os.environ.copy() |
| 142 | env["AUTH_TOKEN"] = "dummy-auth" |
| 143 | env["CT0"] = "dummy-ct0" |
| 144 | |
| 145 | result = subprocess.run( |
| 146 | ["node", str(VENDORED_BIRD), "--check"], |
| 147 | cwd=REPO_ROOT, |
| 148 | env=env, |
| 149 | capture_output=True, |
| 150 | text=True, |
| 151 | check=False, |
| 152 | ) |
| 153 | |
| 154 | self.assertEqual(0, result.returncode, result.stderr) |
| 155 | payload = json.loads(result.stdout) |
| 156 | self.assertTrue(payload["authenticated"]) |
| 157 | self.assertEqual("env AUTH_TOKEN", payload["source"]) |
| 158 | |
| 159 | def test_check_with_browser_lookup_disabled_returns_json_warnings(self): |
| 160 | env = os.environ.copy() |
| 161 | env.pop("AUTH_TOKEN", None) |
| 162 | env.pop("CT0", None) |
| 163 | env["BIRD_DISABLE_BROWSER_COOKIES"] = "1" |
| 164 | |
| 165 | result = subprocess.run( |
| 166 | ["node", str(VENDORED_BIRD), "--check"], |
| 167 | cwd=REPO_ROOT, |
| 168 | env=env, |
| 169 | capture_output=True, |
| 170 | text=True, |
| 171 | check=False, |
| 172 | ) |
| 173 | |
| 174 | self.assertEqual(1, result.returncode, result.stderr) |
| 175 | payload = json.loads(result.stdout) |
| 176 | self.assertFalse(payload["authenticated"]) |
| 177 | self.assertTrue(payload["warnings"]) |
| 178 | self.assertIn("Missing auth_token", " ".join(payload["warnings"])) |
| 179 | |
| 180 | def test_browser_cookie_helpers_lazy_load_sweet_cookie(self): |
| 181 | sweet_cookie_dir = ( |
| 182 | REPO_ROOT |
| 183 | / "skills" |
| 184 | / "last30days" |
| 185 | / "scripts" |
| 186 | / "lib" |
| 187 | / "vendor" |
| 188 | / "bird-search" |
| 189 | / "lib" |
| 190 | / "node_modules" |
| 191 | / "@steipete" |
| 192 | / "sweet-cookie" |
| 193 | ) |
| 194 | if sweet_cookie_dir.exists(): |
| 195 | self.skipTest("vendored sweet-cookie test stub already exists") |
| 196 | |
| 197 | sweet_cookie_dir.mkdir(parents=True) |
| 198 | (sweet_cookie_dir / "package.json").write_text( |
| 199 | json.dumps( |
| 200 | { |
| 201 | "name": "@steipete/sweet-cookie", |
| 202 | "type": "module", |
| 203 | "exports": "./index.js", |
| 204 | } |
| 205 | ), |
| 206 | encoding="utf-8", |
| 207 | ) |
| 208 | (sweet_cookie_dir / "index.js").write_text( |
| 209 | textwrap.dedent( |
| 210 | """ |
| 211 | export async function getCookies(options) { |
| 212 | const browser = options.browsers?.[0] ?? "unknown"; |
| 213 | return { |
| 214 | cookies: [ |
| 215 | { name: "auth_token", value: `${browser}-auth`, domain: "x.com" }, |
| 216 | { name: "ct0", value: `${browser}-ct0`, domain: "x.com" }, |
| 217 | ], |
| 218 | warnings: [], |
| 219 | }; |
| 220 | } |
| 221 | """ |
| 222 | ), |
| 223 | encoding="utf-8", |
| 224 | ) |
| 225 | |
| 226 | try: |
| 227 | result = subprocess.run( |
| 228 | [ |
| 229 | "node", |
| 230 | "--input-type=module", |
| 231 | "-e", |
| 232 | textwrap.dedent( |
| 233 | """ |
| 234 | import { |
| 235 | extractCookiesFromSafari, |
| 236 | extractCookiesFromChrome, |
| 237 | extractCookiesFromFirefox, |
| 238 | } from "./skills/last30days/scripts/lib/vendor/bird-search/lib/cookies.js"; |
| 239 | |
| 240 | const payload = await Promise.all([ |
| 241 | extractCookiesFromSafari(), |
| 242 | extractCookiesFromChrome("Profile 1"), |
| 243 | extractCookiesFromFirefox("default-release"), |
| 244 | ]); |
| 245 | process.stdout.write(JSON.stringify(payload)); |
| 246 | """ |
| 247 | ), |
| 248 | ], |
| 249 | cwd=REPO_ROOT, |
| 250 | capture_output=True, |
| 251 | text=True, |
| 252 | check=False, |
| 253 | ) |
| 254 | |
| 255 | self.assertEqual(0, result.returncode, result.stderr) |
| 256 | payload = json.loads(result.stdout) |
| 257 | self.assertEqual("Safari", payload[0]["cookies"]["source"]) |
| 258 | self.assertEqual('Chrome profile "Profile 1"', payload[1]["cookies"]["source"]) |
| 259 | self.assertEqual( |
| 260 | 'Firefox profile "default-release"', payload[2]["cookies"]["source"] |
| 261 | ) |
| 262 | self.assertEqual("safari-auth", payload[0]["cookies"]["authToken"]) |
| 263 | self.assertEqual("chrome-auth", payload[1]["cookies"]["authToken"]) |
| 264 | self.assertEqual("firefox-auth", payload[2]["cookies"]["authToken"]) |
| 265 | finally: |
| 266 | shutil.rmtree(sweet_cookie_dir, ignore_errors=True) |
| 267 | for path in [sweet_cookie_dir.parent, sweet_cookie_dir.parent.parent]: |
| 268 | try: |
| 269 | path.rmdir() |
| 270 | except OSError: |
| 271 | pass |
| 272 | |
| 273 | def test_none_likes_when_missing(self): |
| 274 | tweets = [ |
| 275 | { |
| 276 | "id": "1", |
| 277 | "text": "test tweet with no engagement fields", |
| 278 | "permanent_url": "https://x.com/u/status/1", |
| 279 | # no likeCount, like_count, or favorite_count |
| 280 | } |
| 281 | ] |
| 282 | items = parse_bird_response(tweets, "test query") |
| 283 | self.assertIsNone(items[0]["engagement"]) |
| 284 | |
| 285 | def test_fallback_to_second_key(self): |
| 286 | tweets = [ |
| 287 | { |
| 288 | "id": "1", |
| 289 | "text": "test", |
| 290 | "permanent_url": "https://x.com/u/status/1", |
| 291 | "like_count": 7, |
| 292 | } |
| 293 | ] |
| 294 | items = parse_bird_response(tweets, "test query") |
| 295 | self.assertEqual(7, items[0]["engagement"]["likes"]) |
| 296 | |
| 297 | def test_zero_does_not_fall_through(self): |
| 298 | """likeCount=0 should not fall through to like_count=10.""" |
| 299 | tweets = [ |
| 300 | { |
| 301 | "id": "1", |
| 302 | "text": "test", |
| 303 | "permanent_url": "https://x.com/u/status/1", |
| 304 | "likeCount": 0, |
| 305 | "like_count": 10, |
| 306 | } |
| 307 | ] |
| 308 | items = parse_bird_response(tweets, "test query") |
| 309 | self.assertEqual(0, items[0]["engagement"]["likes"]) |
| 310 | |
| 311 | def test_engagement_none_when_all_fields_missing(self): |
| 312 | """All-None engagement dict should become None, not propagate.""" |
| 313 | tweets = [ |
| 314 | { |
| 315 | "id": "1", |
| 316 | "text": "test", |
| 317 | "permanent_url": "https://x.com/u/status/1", |
| 318 | } |
| 319 | ] |
| 320 | items = parse_bird_response(tweets, "test query") |
| 321 | self.assertIsNone(items[0]["engagement"]) |
| 322 | |
| 323 | def test_engagement_preserved_when_any_field_present(self): |
| 324 | """Engagement dict kept when at least one metric exists.""" |
| 325 | tweets = [ |
| 326 | { |
| 327 | "id": "1", |
| 328 | "text": "test", |
| 329 | "permanent_url": "https://x.com/u/status/1", |
| 330 | "likeCount": 5, |
| 331 | } |
| 332 | ] |
| 333 | items = parse_bird_response(tweets, "test query") |
| 334 | self.assertIsNotNone(items[0]["engagement"]) |
| 335 | self.assertEqual(5, items[0]["engagement"]["likes"]) |
| 336 | |
| 337 | |
| 338 | class TestRunBirdSearchJsonDecodeRetry(unittest.TestCase): |
| 339 | """When bird-search returns non-JSON stdout, retry the subprocess. |
| 340 | |
| 341 | Twitter's edge sometimes serves an HTML anti-bot interstitial in place of |
| 342 | JSON. Before this fix, that response made json.loads raise JSONDecodeError |
| 343 | and the function returned {"items": []} with no diagnostic — silent-empty |
| 344 | against an orchestrator that can't distinguish "Twitter blocked us" from |
| 345 | "no tweets matched the query." |
| 346 | """ |
| 347 | |
| 348 | def _make_result(self, stdout: str, stderr: str = "", returncode: int = 0): |
| 349 | from lib.subproc import SubprocResult |
| 350 | return SubprocResult(returncode=returncode, stdout=stdout, stderr=stderr) |
| 351 | |
| 352 | def test_retries_subprocess_on_html_interstitial_then_succeeds(self): |
| 353 | """First subprocess attempt returns HTML; second returns JSON → success.""" |
| 354 | from unittest import mock |
| 355 | from lib import bird_x |
| 356 | |
| 357 | html_interstitial = "<!DOCTYPE html><html><body>Rate limited</body></html>" |
| 358 | json_success = '[{"id": "1", "text": "tweet"}]' |
| 359 | |
| 360 | results = [ |
| 361 | (self._make_result(stdout=html_interstitial), None), |
| 362 | (self._make_result(stdout=json_success), None), |
| 363 | ] |
| 364 | |
| 365 | with mock.patch.object(bird_x, "_invoke_bird_subprocess", side_effect=results), \ |
| 366 | mock.patch.object(bird_x.time, "sleep") as mock_sleep: |
| 367 | response = bird_x._run_bird_search("test", count=10, timeout=30) |
| 368 | |
| 369 | self.assertNotIn("error", response) |
| 370 | self.assertEqual(response["items"], [{"id": "1", "text": "tweet"}]) |
| 371 | # Should have slept between the failed first attempt and the retry. |
| 372 | mock_sleep.assert_called_once_with(bird_x.JSON_DECODE_RETRY_DELAY) |
| 373 | |
| 374 | def test_returns_error_after_all_retries_exhausted(self): |
| 375 | """All attempts return HTML → error dict with diagnostic + items=[].""" |
| 376 | from unittest import mock |
| 377 | from lib import bird_x |
| 378 | |
| 379 | html_interstitial = "<!DOCTYPE html><html>blocked</html>" |
| 380 | results = [ |
| 381 | (self._make_result(stdout=html_interstitial), None), |
| 382 | (self._make_result(stdout=html_interstitial), None), |
| 383 | ] |
| 384 | |
| 385 | with mock.patch.object(bird_x, "_invoke_bird_subprocess", side_effect=results), \ |
| 386 | mock.patch.object(bird_x.time, "sleep"): |
| 387 | response = bird_x._run_bird_search("test", count=10, timeout=30) |
| 388 | |
| 389 | self.assertIn("error", response) |
| 390 | self.assertIn("Invalid JSON response", response["error"]) |
| 391 | # Diagnostic message names the anti-bot interstitial so it's |
| 392 | # distinguishable from a genuine no-results case in logs. |
| 393 | self.assertIn("anti-bot interstitial", response["error"].lower()) |
| 394 | self.assertEqual(response["items"], []) |
| 395 | |
| 396 | def test_terminal_subprocess_error_is_not_retried(self): |
| 397 | """Subprocess timeout / spawn failure → terminal error, no retry.""" |
| 398 | from unittest import mock |
| 399 | from lib import bird_x |
| 400 | |
| 401 | timeout_error = {"error": "Search timed out after 30s", "items": []} |
| 402 | results = [(None, timeout_error)] |
| 403 | |
| 404 | with mock.patch.object(bird_x, "_invoke_bird_subprocess", side_effect=results), \ |
| 405 | mock.patch.object(bird_x.time, "sleep") as mock_sleep: |
| 406 | response = bird_x._run_bird_search("test", count=10, timeout=30) |
| 407 | |
| 408 | self.assertEqual(response, timeout_error) |
| 409 | mock_sleep.assert_not_called() |
| 410 | |
| 411 | if __name__ == "__main__": |
| 412 | unittest.main() |
| 413 | |
| 414 | |
| 415 | class TestXFromAndAboutLanes(unittest.TestCase): |
| 416 | """U7/U8: FROM lane drops the topic-AND; ABOUT lane queries @handle and |
| 417 | excludes the handle's own tweets.""" |
| 418 | |
| 419 | def _result(self, body_items): |
| 420 | import json as _j |
| 421 | class _R: |
| 422 | returncode = 0 |
| 423 | stderr = "" |
| 424 | r = _R() |
| 425 | r.stdout = _j.dumps({"items": body_items}) |
| 426 | return r |
| 427 | |
| 428 | def test_from_lane_drops_topic_and(self): |
| 429 | from unittest import mock |
| 430 | from lib import bird_x |
| 431 | captured = [] |
| 432 | |
| 433 | def fake_run(cmd, timeout=None, env=None): |
| 434 | captured.append(cmd[2]) # the query string arg |
| 435 | return self._result([]) |
| 436 | |
| 437 | with mock.patch.object(bird_x.subproc, "run_with_timeout", side_effect=fake_run): |
| 438 | bird_x.search_handles(["xuezhao"], "lan xuezhao", "2026-05-19", count_per=1) |
| 439 | self.assertEqual(captured[0], "from:xuezhao since:2026-05-19") |
| 440 | self.assertNotIn("lan xuezhao", captured[0]) |
| 441 | |
| 442 | def test_mention_lane_queries_at_handle(self): |
| 443 | from unittest import mock |
| 444 | from lib import bird_x |
| 445 | captured = [] |
| 446 | |
| 447 | def fake_run(cmd, timeout=None, env=None): |
| 448 | captured.append(cmd[2]) |
| 449 | return self._result([]) |
| 450 | |
| 451 | with mock.patch.object(bird_x.subproc, "run_with_timeout", side_effect=fake_run): |
| 452 | bird_x.search_mentions(["xuezhao"], "2026-05-19", count_per=1) |
| 453 | self.assertEqual(captured[0], "@xuezhao since:2026-05-19") |
| 454 | |
| 455 | def test_mention_lane_excludes_own_tweets(self): |
| 456 | from unittest import mock |
| 457 | from lib import bird_x |
| 458 | parsed = [ |
| 459 | {"url": "https://x.com/xuezhao/status/1", "title": "own tweet"}, |
| 460 | {"url": "https://twitter.com/xuezhao/status/3", "title": "own legacy-domain tweet"}, |
| 461 | {"url": "https://x.com/fan99/status/2", "title": "mention of them"}, |
| 462 | ] |
| 463 | with mock.patch.object(bird_x.subproc, "run_with_timeout", |
| 464 | return_value=self._result([{"id": "x"}])), \ |
| 465 | mock.patch.object(bird_x, "parse_bird_response", return_value=parsed): |
| 466 | out = bird_x.search_mentions(["xuezhao"], "2026-05-19", count_per=5) |
| 467 | urls = [it["url"] for it in out] |
| 468 | self.assertNotIn("https://x.com/xuezhao/status/1", urls) # own (x.com) excluded |
| 469 | self.assertNotIn("https://twitter.com/xuezhao/status/3", urls) # own (twitter.com) excluded |
| 470 | self.assertIn("https://x.com/fan99/status/2", urls) # mention kept |
| 471 | |
| 472 | def test_mention_lane_empty_when_no_mentions(self): |
| 473 | from unittest import mock |
| 474 | from lib import bird_x |
| 475 | with mock.patch.object(bird_x.subproc, "run_with_timeout", |
| 476 | return_value=self._result([])), \ |
| 477 | mock.patch.object(bird_x, "parse_bird_response", return_value=[]): |
| 478 | out = bird_x.search_mentions(["xuezhao"], "2026-05-19") |
| 479 | self.assertEqual(out, []) |
| 480 | class TestProbeAndDiagnoseHonesty(unittest.TestCase): |
| 481 | """U5: --diagnose probe + true auth lane; X is not reported green when dead.""" |
| 482 | |
| 483 | def setUp(self): |
| 484 | from lib import bird_x |
| 485 | bird_x._probe_cache = "unset" |
| 486 | bird_x._credentials = {"AUTH_TOKEN": "t", "CT0": "c"} # injected creds present |
| 487 | |
| 488 | def tearDown(self): |
| 489 | from lib import bird_x |
| 490 | bird_x._probe_cache = "unset" |
| 491 | bird_x._credentials = {} |
| 492 | |
| 493 | def test_probe_true_when_response_ok(self): |
| 494 | from unittest import mock |
| 495 | from lib import bird_x |
| 496 | with mock.patch.object(bird_x, "_run_bird_search", return_value={"items": [{"id": "1"}]}): |
| 497 | self.assertTrue(bird_x.probe_works()) |
| 498 | |
| 499 | def test_probe_false_on_auth_error(self): |
| 500 | from unittest import mock |
| 501 | from lib import bird_x |
| 502 | with mock.patch.object(bird_x, "_run_bird_search", |
| 503 | return_value={"error": "Missing auth_token", "items": []}): |
| 504 | self.assertIs(bird_x.probe_works(), False) |
| 505 | |
| 506 | def test_probe_none_on_timeout_inconclusive(self): |
| 507 | from unittest import mock |
| 508 | from lib import bird_x |
| 509 | with mock.patch.object(bird_x, "_run_bird_search", |
| 510 | return_value={"error": "Search timed out after 8s", "items": []}): |
| 511 | self.assertIsNone(bird_x.probe_works()) |
| 512 | |
| 513 | def test_probe_false_when_no_credentials(self): |
| 514 | from unittest import mock |
| 515 | from lib import bird_x |
| 516 | bird_x._credentials = {} |
| 517 | with mock.patch.dict(os.environ, {}, clear=True): |
| 518 | self.assertIs(bird_x.probe_works(), False) |
| 519 | |
| 520 | def test_probe_cached_per_process(self): |
| 521 | from unittest import mock |
| 522 | from lib import bird_x |
| 523 | with mock.patch.object(bird_x, "_run_bird_search", |
| 524 | return_value={"items": [{"id": "1"}]}) as m: |
| 525 | bird_x.probe_works() |
| 526 | bird_x.probe_works() |
| 527 | self.assertEqual(m.call_count, 1) # cached, not re-run |
| 528 | |
| 529 | def test_get_x_source_status_reports_true_lane(self): |
| 530 | from unittest import mock |
| 531 | from lib import env, bird_x |
| 532 | cfg = {"AUTH_TOKEN": "t", "CT0": "c", "_AUTH_TOKEN_SOURCE": "browser"} |
| 533 | with mock.patch.object(bird_x, "get_bird_status", |
| 534 | return_value={"installed": True, "authenticated": True, |
| 535 | "username": "env AUTH_TOKEN", "can_install": True}): |
| 536 | status = env.get_x_source_status(cfg, probe=False) |
| 537 | self.assertEqual(status["bird_username"], "browser AUTH_TOKEN") |
| 538 | |
| 539 | def test_diagnose_probe_downgrades_when_dead(self): |
| 540 | from unittest import mock |
| 541 | from lib import env, bird_x |
| 542 | cfg = {"AUTH_TOKEN": "t", "CT0": "c", "_AUTH_TOKEN_SOURCE": "browser"} |
| 543 | with mock.patch.object(bird_x, "get_bird_status", |
| 544 | return_value={"installed": True, "authenticated": True, |
| 545 | "username": "env AUTH_TOKEN", "can_install": True}), \ |
| 546 | mock.patch.object(bird_x, "probe_works", return_value=False): |
| 547 | status = env.get_x_source_status(cfg, probe=True) |
| 548 | self.assertFalse(status["bird_authenticated"]) |
| 549 | self.assertIn("no working X auth", status["bird_username"]) |
| 550 | |
| 551 | |
| 552 | class TestHandleSearchLogsOnSuccess(unittest.TestCase): |
| 553 | """U6: handle searches log query + count on success, not only on failure.""" |
| 554 | |
| 555 | def test_search_handles_logs_on_success(self): |
| 556 | from unittest import mock |
| 557 | from lib import bird_x |
| 558 | |
| 559 | class _R: |
| 560 | returncode = 0 |
| 561 | stdout = '{"items": [{"id": "1"}]}' |
| 562 | stderr = "" |
| 563 | |
| 564 | logged = [] |
| 565 | with mock.patch.object(bird_x.subproc, "run_with_timeout", return_value=_R()), \ |
| 566 | mock.patch.object(bird_x, "_log", side_effect=lambda m: logged.append(m)): |
| 567 | bird_x.search_handles(["mvanhorn"], "matt van horn", "2026-05-19", count_per=1) |
| 568 | self.assertTrue(any("Searching:" in m for m in logged), |
| 569 | f"expected a Searching: log on success, got {logged}") |
| 570 | |
| 571 | |
| 572 | class TestStrongestTokenRetryAnchored(unittest.TestCase): |
| 573 | """The last-chance retry must keep an entity anchor, not collapse to a bare |
| 574 | generic token (e.g. 'compound') that floods the X pool with off-topic noise. |
| 575 | """ |
| 576 | |
| 577 | def test_last_chance_retry_keeps_entity_anchor(self): |
| 578 | from unittest import mock |
| 579 | from lib import bird_x |
| 580 | |
| 581 | queries = [] |
| 582 | |
| 583 | def fake_run(query, count, timeout): |
| 584 | queries.append(query) |
| 585 | return {"items": []} # always 0 → forces every retry tier |
| 586 | |
| 587 | # extract_compound_terms may run; let it. Force all bird calls empty. |
| 588 | with mock.patch.object(bird_x, "_run_bird_search", side_effect=fake_run): |
| 589 | bird_x.search_x("trevin chow ai agents compound", "2026-05-19", "2026-06-18") |
| 590 | |
| 591 | self.assertTrue(queries, "expected at least one bird query") |
| 592 | last = queries[-1] |
| 593 | # The final (last-chance) query keeps the entity anchor ... |
| 594 | self.assertIn("trevin", last) |
| 595 | # ... and is NOT a bare generic token query. |
| 596 | self.assertFalse(last.startswith("compound "), f"bare generic retry: {last!r}") |
| 597 | self.assertNotEqual(last, "compound since:2026-05-19") |
| 598 | |
| 599 | def test_retry_with_single_distinctive_token_no_crash(self): |
| 600 | from unittest import mock |
| 601 | from lib import bird_x |
| 602 | |
| 603 | queries = [] |
| 604 | |
| 605 | def fake_run(query, count, timeout): |
| 606 | queries.append(query) |
| 607 | return {"items": []} |
| 608 | |
| 609 | with mock.patch.object(bird_x, "_run_bird_search", side_effect=fake_run): |
| 610 | # 'trending tools' is all low-signal except nothing distinctive -> |
| 611 | # whatever survives, the retry must not crash and stays anchored. |
| 612 | bird_x.search_x("agentcookie", "2026-05-19", "2026-06-18") |
| 613 | |
| 614 | self.assertTrue(queries) |
| 615 | self.assertIn("agentcookie", queries[-1]) |
| 616 | |
| 617 | |
| 618 | class TestBirdRetryQueryCorrectness(unittest.TestCase): |
| 619 | def test_quoted_topic_only_generates_balanced_retry_queries(self): |
| 620 | from lib import bird_x |
| 621 | |
| 622 | queries = [] |
| 623 | |
| 624 | def fake_run(query, count, timeout): |
| 625 | queries.append(query) |
| 626 | if len(queries) == 1: |
| 627 | return {"items": []} |
| 628 | return {"error": "Bird search failed", "items": []} |
| 629 | |
| 630 | with mock.patch.object( |
| 631 | bird_x, |
| 632 | "_extract_core_subject", |
| 633 | return_value='immobilienmakler(berlin "mixed-use', |
| 634 | ), mock.patch( |
| 635 | "lib.query.extract_compound_terms", |
| 636 | return_value=['"immobilienmakler berlin"'], |
| 637 | ), mock.patch.object(bird_x, "_run_bird_search", side_effect=fake_run): |
| 638 | response = bird_x.search_x( |
| 639 | '"Immobilienmakler Berlin" competitors', |
| 640 | "2026-07-12", |
| 641 | "2026-07-19", |
| 642 | ) |
| 643 | |
| 644 | self.assertGreaterEqual(len(queries), 2) |
| 645 | self.assertEqual( |
| 646 | "immobilienmakler berlin mixed-use since:2026-07-12", |
| 647 | queries[0], |
| 648 | ) |
| 649 | for query in queries: |
| 650 | self.assertEqual(0, query.count('"') % 2, query) |
| 651 | self.assertEqual(query.count("("), query.count(")"), query) |
| 652 | self.assertNotIn("error", response) |
| 653 | self.assertEqual([], response["items"]) |
| 654 | |
| 655 | def test_every_failed_attempt_still_reports_backend_failure(self): |
| 656 | from lib import bird_x |
| 657 | |
| 658 | with mock.patch.object( |
| 659 | bird_x, "_extract_core_subject", return_value="immobilienmakler berlin market" |
| 660 | ), mock.patch.object( |
| 661 | bird_x, |
| 662 | "_run_bird_search", |
| 663 | return_value={"error": "Bird search failed", "items": []}, |
| 664 | ): |
| 665 | response = bird_x.search_x( |
| 666 | "Immobilienmakler Berlin market", "2026-07-12", "2026-07-19" |
| 667 | ) |
| 668 | |
| 669 | self.assertEqual("Bird search failed", response["error"]) |
| 670 | |
| 671 | |
| 672 | class LeadingMentionsTests(unittest.TestCase): |
| 673 | """U5: leading @mentions parsed from post text identify reply targets.""" |
| 674 | |
| 675 | def test_single_leading_mention(self): |
| 676 | from lib import bird_x |
| 677 | self.assertEqual(["alpha"], bird_x._leading_mentions("@alpha thanks so much!")) |
| 678 | |
| 679 | def test_multiple_leading_mentions(self): |
| 680 | from lib import bird_x |
| 681 | self.assertEqual(["alpha", "beta"], bird_x._leading_mentions("@alpha @beta hi")) |
| 682 | |
| 683 | def test_in_body_mention_not_collected(self): |
| 684 | from lib import bird_x |
| 685 | self.assertEqual([], bird_x._leading_mentions("hello @gamma nice work")) |
| 686 | |
| 687 | def test_punctuation_stripped(self): |
| 688 | from lib import bird_x |
| 689 | self.assertEqual(["alpha"], bird_x._leading_mentions("@alpha, nice")) |
| 690 | |
| 691 | def test_empty_text(self): |
| 692 | from lib import bird_x |
| 693 | self.assertEqual([], bird_x._leading_mentions("")) |
| 694 | self.assertEqual([], bird_x._leading_mentions(None)) |
| 695 | |
| 696 | |
| 697 | if __name__ == "__main__": |
| 698 | unittest.main() |
| 699 |