| 1 | """Tests for GitHub source module.""" |
| 2 | |
| 3 | import json |
| 4 | import unittest |
| 5 | from unittest.mock import patch, MagicMock |
| 6 | |
| 7 | from lib import github |
| 8 | |
| 9 | |
| 10 | class TestResolveToken(unittest.TestCase): |
| 11 | def test_explicit_token(self): |
| 12 | self.assertEqual(github._resolve_token("my-token"), "my-token") |
| 13 | |
| 14 | @patch.dict("os.environ", {"GITHUB_TOKEN": "env-token"}) |
| 15 | def test_env_token(self): |
| 16 | self.assertEqual(github._resolve_token(), "env-token") |
| 17 | |
| 18 | @patch.dict("os.environ", {}, clear=True) |
| 19 | @patch("subprocess.run") |
| 20 | def test_gh_cli_fallback(self, mock_run): |
| 21 | mock_run.return_value = MagicMock(returncode=0, stdout="gh-token\n") |
| 22 | # Clear GITHUB_TOKEN from env for this test |
| 23 | result = github._resolve_token() |
| 24 | self.assertEqual(result, "gh-token") |
| 25 | |
| 26 | @patch.dict("os.environ", {}, clear=True) |
| 27 | @patch("subprocess.run", side_effect=FileNotFoundError) |
| 28 | def test_no_token_available(self, mock_run): |
| 29 | result = github._resolve_token() |
| 30 | self.assertIsNone(result) |
| 31 | |
| 32 | |
| 33 | class TestParseRepoFromUrl(unittest.TestCase): |
| 34 | def test_issue_url(self): |
| 35 | url = "https://github.com/facebook/react/issues/123" |
| 36 | self.assertEqual(github._parse_repo_from_url(url), "facebook/react") |
| 37 | |
| 38 | def test_pr_url(self): |
| 39 | url = "https://github.com/vercel/next.js/pull/456" |
| 40 | self.assertEqual(github._parse_repo_from_url(url), "vercel/next.js") |
| 41 | |
| 42 | def test_empty(self): |
| 43 | self.assertEqual(github._parse_repo_from_url(""), "") |
| 44 | |
| 45 | |
| 46 | class TestParseDate(unittest.TestCase): |
| 47 | def test_iso_date(self): |
| 48 | self.assertEqual(github._parse_date("2026-03-15T12:00:00Z"), "2026-03-15") |
| 49 | |
| 50 | def test_none(self): |
| 51 | self.assertIsNone(github._parse_date(None)) |
| 52 | |
| 53 | def test_empty(self): |
| 54 | self.assertIsNone(github._parse_date("")) |
| 55 | |
| 56 | def test_rejects_garbage(self): |
| 57 | """The old naive slicing returned 'hello worl' for 'hello world'. Reject it.""" |
| 58 | self.assertIsNone(github._parse_date("hello world")) |
| 59 | self.assertIsNone(github._parse_date("not-a-date")) |
| 60 | self.assertIsNone(github._parse_date("abcdefghij")) |
| 61 | |
| 62 | def test_rejects_invalid_date_values(self): |
| 63 | """An out-of-range date like 2026-99-99 is not a real date.""" |
| 64 | self.assertIsNone(github._parse_date("2026-99-99")) |
| 65 | |
| 66 | def test_iso_with_offset(self): |
| 67 | self.assertEqual(github._parse_date("2026-03-15T12:00:00+00:00"), "2026-03-15") |
| 68 | |
| 69 | def test_iso_with_no_colon_offset(self): |
| 70 | self.assertEqual(github._parse_date("2026-03-15T12:00:00+0000"), "2026-03-15") |
| 71 | |
| 72 | |
| 73 | class TestSearchGithub(unittest.TestCase): |
| 74 | @patch.dict("os.environ", {}, clear=True) |
| 75 | @patch("subprocess.run", side_effect=FileNotFoundError) |
| 76 | @patch("lib.github._fetch_json", return_value=None) |
| 77 | def test_no_token_unauth_rate_limited_sets_error(self, mock_fetch, mock_run): |
| 78 | # No token -> unauthenticated request; on failure (likely anon rate |
| 79 | # limit) the envelope carries a clear error instead of being silent. |
| 80 | result = github.search_github("react", "2026-03-01", "2026-03-31", token=None) |
| 81 | self.assertEqual(result.get("items", []), []) |
| 82 | self.assertIn("error", result) |
| 83 | self.assertIn("unauthenticated", result["error"].lower()) |
| 84 | self.assertIn("context", result) |
| 85 | self.assertEqual(result["context"]["from_date"], "2026-03-01") |
| 86 | # Unauth requests are capped to the low-rate tier. |
| 87 | self.assertLessEqual(result["context"]["count"], github.UNAUTH_COUNT_CAP) |
| 88 | # The request was actually attempted without a token (no early return). |
| 89 | mock_fetch.assert_called_once() |
| 90 | self.assertIsNone(mock_fetch.call_args.kwargs.get("token")) |
| 91 | |
| 92 | @patch.dict("os.environ", {}, clear=True) |
| 93 | @patch("subprocess.run", side_effect=FileNotFoundError) |
| 94 | @patch("lib.github._fetch_json", return_value={"items": [{"id": 1, "title": "x"}]}) |
| 95 | def test_no_token_unauth_success_returns_items(self, mock_fetch, mock_run): |
| 96 | result = github.search_github("react", "2026-03-01", "2026-03-31", token=None) |
| 97 | self.assertEqual(len(result["items"]), 1) |
| 98 | self.assertNotIn("error", result) |
| 99 | |
| 100 | def test_resolve_token_public_alias(self): |
| 101 | """resolve_token is the public entry point pipeline uses; _resolve_token stays |
| 102 | private. Both should return the same value for the same input.""" |
| 103 | self.assertEqual( |
| 104 | github.resolve_token("explicit-token"), |
| 105 | github._resolve_token("explicit-token"), |
| 106 | ) |
| 107 | self.assertEqual(github.resolve_token("explicit-token"), "explicit-token") |
| 108 | |
| 109 | @patch.object(github, "_fetch_json") |
| 110 | @patch.object(github, "_resolve_token", return_value="test-token") |
| 111 | def test_search_returns_raw_envelope(self, mock_token, mock_fetch): |
| 112 | mock_fetch.return_value = { |
| 113 | "total_count": 1, |
| 114 | "items": [ |
| 115 | { |
| 116 | "html_url": "https://github.com/facebook/react/issues/42", |
| 117 | "title": "React Server Components bug", |
| 118 | "body": "There is a bug when using RSC with streaming...", |
| 119 | "created_at": "2026-03-15T10:00:00Z", |
| 120 | "state": "open", |
| 121 | "comments": 12, |
| 122 | "reactions": {"total_count": 8}, |
| 123 | "labels": [{"name": "bug"}, {"name": "rsc"}], |
| 124 | "user": {"login": "testuser"}, |
| 125 | }, |
| 126 | ], |
| 127 | } |
| 128 | # Search returns raw envelope; parse normalizes. |
| 129 | response = github.search_github("react", "2026-03-01", "2026-03-31") |
| 130 | self.assertEqual(len(response["items"]), 1) |
| 131 | self.assertEqual(response["items"][0]["title"], "React Server Components bug") |
| 132 | self.assertEqual(response["context"]["from_date"], "2026-03-01") |
| 133 | |
| 134 | items = github.parse_github_response(response) |
| 135 | self.assertEqual(len(items), 1) |
| 136 | item = items[0] |
| 137 | self.assertEqual(item["source"], "github") |
| 138 | self.assertEqual(item["container"], "facebook/react") |
| 139 | self.assertEqual(item["title"], "React Server Components bug") |
| 140 | self.assertEqual(item["date"], "2026-03-15") |
| 141 | self.assertEqual(item["author"], "testuser") |
| 142 | self.assertIn("bug", item["metadata"]["labels"]) |
| 143 | self.assertEqual(item["metadata"]["state"], "open") |
| 144 | self.assertEqual(item["metadata"]["comment_count"], 12) |
| 145 | self.assertEqual(item["metadata"]["reactions"], 8) |
| 146 | self.assertEqual(item["engagement"]["reactions"], 8) |
| 147 | self.assertEqual(item["engagement"]["comments"], 12) |
| 148 | self.assertFalse(item["metadata"]["is_pr"]) |
| 149 | |
| 150 | @patch.object(github, "_fetch_json", return_value=None) |
| 151 | @patch.object(github, "_resolve_token", return_value="test-token") |
| 152 | def test_rate_limit_returns_empty_envelope(self, mock_token, mock_fetch): |
| 153 | """403 rate limit returns envelope with empty items list.""" |
| 154 | response = github.search_github("react", "2026-03-01", "2026-03-31") |
| 155 | self.assertEqual(response["items"], []) |
| 156 | self.assertEqual(github.parse_github_response(response), []) |
| 157 | |
| 158 | @patch.object(github, "_fetch_json") |
| 159 | @patch.object(github, "_resolve_token", return_value="test-token") |
| 160 | def test_pr_detected(self, mock_token, mock_fetch): |
| 161 | mock_fetch.return_value = { |
| 162 | "total_count": 1, |
| 163 | "items": [ |
| 164 | { |
| 165 | "html_url": "https://github.com/vercel/next.js/pull/99", |
| 166 | "title": "Add streaming support", |
| 167 | "body": "This PR adds...", |
| 168 | "created_at": "2026-03-20T10:00:00Z", |
| 169 | "state": "open", |
| 170 | "comments": 5, |
| 171 | "reactions": {"total_count": 3}, |
| 172 | "labels": [], |
| 173 | "user": {"login": "dev"}, |
| 174 | "pull_request": {"url": "..."}, |
| 175 | }, |
| 176 | ], |
| 177 | } |
| 178 | response = github.search_github("next.js", "2026-03-01", "2026-03-31") |
| 179 | items = github.parse_github_response(response) |
| 180 | self.assertEqual(len(items), 1) |
| 181 | self.assertTrue(items[0]["metadata"]["is_pr"]) |
| 182 | |
| 183 | |
| 184 | class TestParseGithubResponse(unittest.TestCase): |
| 185 | """Fixture-driven parse tests: feed a synthetic search_github envelope to |
| 186 | parse_github_response and assert normalized output. |
| 187 | |
| 188 | This contract (search returns dict envelope, parse turns it into a list) |
| 189 | matches every other source adapter. Before this refactor, search_github |
| 190 | returned a bare list and there was no parse step, blocking fixture tests. |
| 191 | """ |
| 192 | |
| 193 | _RAW_ENVELOPE = { |
| 194 | "items": [ |
| 195 | { |
| 196 | "html_url": "https://github.com/facebook/react/issues/42", |
| 197 | "title": "React Server Components bug", |
| 198 | "body": "There is a bug when using RSC with streaming...", |
| 199 | "created_at": "2026-03-15T10:00:00Z", |
| 200 | "state": "open", |
| 201 | "comments": 12, |
| 202 | "reactions": {"total_count": 8}, |
| 203 | "labels": [{"name": "bug"}, {"name": "rsc"}], |
| 204 | "user": {"login": "testuser"}, |
| 205 | }, |
| 206 | { |
| 207 | "html_url": "https://github.com/vercel/next.js/pull/99", |
| 208 | "title": "Add streaming support", |
| 209 | "body": "This PR adds...", |
| 210 | "created_at": "2026-03-20T10:00:00Z", |
| 211 | "state": "open", |
| 212 | "comments": 5, |
| 213 | "reactions": {"total_count": 3}, |
| 214 | "labels": [], |
| 215 | "user": {"login": "dev"}, |
| 216 | "pull_request": {"url": "..."}, |
| 217 | }, |
| 218 | ], |
| 219 | "context": { |
| 220 | "core": "react", |
| 221 | "from_date": "2026-03-01", |
| 222 | "to_date": "2026-03-31", |
| 223 | "count": 25, |
| 224 | }, |
| 225 | } |
| 226 | |
| 227 | def test_normalizes_items(self): |
| 228 | items = github.parse_github_response(self._RAW_ENVELOPE) |
| 229 | self.assertEqual(len(items), 2) |
| 230 | by_url = {i["url"]: i for i in items} |
| 231 | issue = by_url["https://github.com/facebook/react/issues/42"] |
| 232 | self.assertEqual(issue["source"], "github") |
| 233 | self.assertEqual(issue["container"], "facebook/react") |
| 234 | self.assertEqual(issue["title"], "React Server Components bug") |
| 235 | self.assertEqual(issue["date"], "2026-03-15") |
| 236 | self.assertEqual(issue["author"], "testuser") |
| 237 | self.assertEqual(issue["engagement"]["reactions"], 8) |
| 238 | self.assertEqual(issue["engagement"]["comments"], 12) |
| 239 | self.assertFalse(issue["metadata"]["is_pr"]) |
| 240 | |
| 241 | def test_detects_pr(self): |
| 242 | items = github.parse_github_response(self._RAW_ENVELOPE) |
| 243 | pr = next(i for i in items if "/pull/" in i["url"]) |
| 244 | self.assertTrue(pr["metadata"]["is_pr"]) |
| 245 | |
| 246 | def test_date_filter_drops_outside_window(self): |
| 247 | envelope = { |
| 248 | "items": [ |
| 249 | { |
| 250 | "html_url": "https://github.com/foo/bar/issues/1", |
| 251 | "title": "Too old", |
| 252 | "created_at": "2026-01-15T10:00:00Z", |
| 253 | "comments": 0, "reactions": {"total_count": 0}, |
| 254 | "labels": [], "user": {"login": "x"}, |
| 255 | }, |
| 256 | { |
| 257 | "html_url": "https://github.com/foo/bar/issues/2", |
| 258 | "title": "In window", |
| 259 | "created_at": "2026-03-15T10:00:00Z", |
| 260 | "comments": 0, "reactions": {"total_count": 0}, |
| 261 | "labels": [], "user": {"login": "x"}, |
| 262 | }, |
| 263 | ], |
| 264 | "context": {"core": "foo", "from_date": "2026-03-01", |
| 265 | "to_date": "2026-03-31", "count": 25}, |
| 266 | } |
| 267 | items = github.parse_github_response(envelope) |
| 268 | self.assertEqual(len(items), 1) |
| 269 | self.assertEqual(items[0]["title"], "In window") |
| 270 | |
| 271 | def test_sorts_by_relevance(self): |
| 272 | items = github.parse_github_response(self._RAW_ENVELOPE) |
| 273 | scores = [i.get("relevance", 0) for i in items] |
| 274 | self.assertEqual(scores, sorted(scores, reverse=True)) |
| 275 | |
| 276 | def test_empty_envelope(self): |
| 277 | self.assertEqual(github.parse_github_response({"items": []}), []) |
| 278 | self.assertEqual(github.parse_github_response({}), []) |
| 279 | |
| 280 | |
| 281 | class TestComputeRelevance(unittest.TestCase): |
| 282 | def test_basic_relevance(self): |
| 283 | score = github._compute_relevance("react hooks", "React Hooks Tutorial", 0, 10, 5) |
| 284 | self.assertGreater(score, 0.5) |
| 285 | self.assertLessEqual(score, 1.0) |
| 286 | |
| 287 | def test_lower_rank_lower_score(self): |
| 288 | high = github._compute_relevance("react", "React", 0, 0, 0) |
| 289 | low = github._compute_relevance("react", "React", 20, 0, 0) |
| 290 | self.assertGreater(high, low) |
| 291 | |
| 292 | class TestPersonPushEventsLane(unittest.TestCase): |
| 293 | """Person mode must not go dark when PR search returns nothing.""" |
| 294 | |
| 295 | @staticmethod |
| 296 | def _event( |
| 297 | event_id, |
| 298 | *, |
| 299 | actor="kurt", |
| 300 | repo="kurt/power-bi-agentic-development", |
| 301 | created_at="2026-07-22T20:28:18Z", |
| 302 | event_type="PushEvent", |
| 303 | ): |
| 304 | return { |
| 305 | "id": str(event_id), |
| 306 | "type": event_type, |
| 307 | "actor": {"login": actor}, |
| 308 | "repo": {"name": repo}, |
| 309 | "created_at": created_at, |
| 310 | } |
| 311 | |
| 312 | def _run(self): |
| 313 | with patch.object(github, "_resolve_token", return_value="t"), \ |
| 314 | patch.object(github, "_enrich_own_repo", return_value={}), \ |
| 315 | patch.object(github, "_fetch_repo_info", return_value={ |
| 316 | "stars": 811, |
| 317 | "forks": 119, |
| 318 | "description": "Claude Code plugin marketplace for Power BI", |
| 319 | "language": "Python", |
| 320 | "open_issues": 4, |
| 321 | }): |
| 322 | return github.search_github_person( |
| 323 | "kurt", "2026-06-25", "2026-07-25", token="t", |
| 324 | ) |
| 325 | |
| 326 | def test_unsearchable_account_falls_back_to_actor_push_events(self): |
| 327 | def fetch(url, **kwargs): |
| 328 | if "search/issues" in url: |
| 329 | return None |
| 330 | return [self._event(1)] |
| 331 | |
| 332 | with patch.object(github, "_fetch_json", side_effect=fetch): |
| 333 | items = self._run() |
| 334 | |
| 335 | self.assertEqual(len(items), 1) |
| 336 | self.assertEqual(items[0]["container"], "kurt/power-bi-agentic-development") |
| 337 | self.assertEqual(items[0]["date"], "2026-07-22") |
| 338 | self.assertIn("@kurt pushed", items[0]["title"]) |
| 339 | self.assertIn("recent-push", items[0]["metadata"]["labels"]) |
| 340 | self.assertEqual(items[0]["metadata"]["event_type"], "PushEvent") |
| 341 | |
| 342 | def test_empty_pr_search_falls_back_to_actor_push_events(self): |
| 343 | def fetch(url, **kwargs): |
| 344 | if "search/issues" in url: |
| 345 | return {"total_count": 0, "items": []} |
| 346 | return [self._event(1, actor="KURT")] |
| 347 | |
| 348 | with patch.object(github, "_fetch_json", side_effect=fetch): |
| 349 | items = self._run() |
| 350 | |
| 351 | self.assertEqual(len(items), 1) |
| 352 | self.assertEqual(items[0]["author"], "KURT") |
| 353 | |
| 354 | def test_other_actor_push_is_rejected(self): |
| 355 | def fetch(url, **kwargs): |
| 356 | if "search/issues" in url: |
| 357 | return {"total_count": 0, "items": []} |
| 358 | return [self._event(1, actor="collaborator")] |
| 359 | |
| 360 | with patch.object(github, "_fetch_json", side_effect=fetch): |
| 361 | items = self._run() |
| 362 | |
| 363 | self.assertEqual(items, []) |
| 364 | |
| 365 | def test_discovers_push_on_second_events_page(self): |
| 366 | first_page = [ |
| 367 | self._event( |
| 368 | i, |
| 369 | event_type="WatchEvent", |
| 370 | created_at=f"2026-07-{24 - (i // 25):02d}T12:00:00Z", |
| 371 | ) |
| 372 | for i in range(github.PERSON_EVENTS_PER_PAGE) |
| 373 | ] |
| 374 | requested_urls = [] |
| 375 | |
| 376 | def fetch(url, **kwargs): |
| 377 | requested_urls.append(url) |
| 378 | if "search/issues" in url: |
| 379 | return {"total_count": 0, "items": []} |
| 380 | if "&page=1" in url: |
| 381 | return first_page |
| 382 | if "&page=2" in url: |
| 383 | return [self._event(101, repo="kurt/page-two")] |
| 384 | self.fail(f"Unexpected URL: {url}") |
| 385 | |
| 386 | with patch.object(github, "_fetch_json", side_effect=fetch): |
| 387 | items = self._run() |
| 388 | |
| 389 | self.assertEqual([item["container"] for item in items], ["kurt/page-two"]) |
| 390 | self.assertTrue(any("&page=2" in url for url in requested_urls)) |
| 391 | |
| 392 | def test_requests_page_after_three_full_event_pages(self): |
| 393 | full_page = [ |
| 394 | self._event( |
| 395 | i, |
| 396 | event_type="WatchEvent", |
| 397 | created_at="2026-07-24T12:00:00Z", |
| 398 | ) |
| 399 | for i in range(github.PERSON_EVENTS_PER_PAGE) |
| 400 | ] |
| 401 | requested_pages = [] |
| 402 | |
| 403 | def fetch(url, **kwargs): |
| 404 | if "search/issues" in url: |
| 405 | return {"total_count": 0, "items": []} |
| 406 | page = int(url.rsplit("&page=", 1)[1]) |
| 407 | requested_pages.append(page) |
| 408 | return full_page if page <= 3 else [] |
| 409 | |
| 410 | with patch.object(github, "_fetch_json", side_effect=fetch): |
| 411 | items = self._run() |
| 412 | |
| 413 | self.assertEqual(items, []) |
| 414 | self.assertEqual(requested_pages, [1, 2, 3, 4]) |
| 415 | |
| 416 | def test_stops_paging_at_event_older_than_window(self): |
| 417 | requested_urls = [] |
| 418 | |
| 419 | def fetch(url, **kwargs): |
| 420 | requested_urls.append(url) |
| 421 | if "search/issues" in url: |
| 422 | return {"total_count": 0, "items": []} |
| 423 | if "&page=1" in url: |
| 424 | return [ |
| 425 | self._event(1, event_type="WatchEvent"), |
| 426 | self._event(2, created_at="2026-06-24T23:59:59Z"), |
| 427 | ] |
| 428 | self.fail("Events paging continued after reaching an old event") |
| 429 | |
| 430 | with patch.object(github, "_fetch_json", side_effect=fetch): |
| 431 | items = self._run() |
| 432 | |
| 433 | self.assertEqual(items, []) |
| 434 | event_urls = [url for url in requested_urls if "/events/public" in url] |
| 435 | self.assertEqual(len(event_urls), 1) |
| 436 | |
| 437 | def test_ranks_all_event_repos_before_applying_depth_cap(self): |
| 438 | events = [ |
| 439 | self._event(1, repo="kurt/newest", created_at="2026-07-24T12:00:00Z"), |
| 440 | self._event(2, repo="kurt/recent", created_at="2026-07-23T12:00:00Z"), |
| 441 | self._event(3, repo="kurt/third", created_at="2026-07-22T12:00:00Z"), |
| 442 | self._event(4, repo="kurt/high-star", created_at="2026-07-21T12:00:00Z"), |
| 443 | ] |
| 444 | stars = { |
| 445 | "kurt/newest": 3, |
| 446 | "kurt/recent": 2, |
| 447 | "kurt/third": 1, |
| 448 | "kurt/high-star": 10_000, |
| 449 | } |
| 450 | |
| 451 | def repo_info(repo, token): |
| 452 | return { |
| 453 | "stars": stars[repo], |
| 454 | "forks": 0, |
| 455 | "description": "", |
| 456 | "language": "Python", |
| 457 | "open_issues": 0, |
| 458 | } |
| 459 | |
| 460 | with patch.object(github, "_fetch_json", return_value=events), \ |
| 461 | patch.object(github, "_fetch_repo_info", side_effect=repo_info), \ |
| 462 | patch.object(github, "_enrich_own_repo", return_value={}) as enrich: |
| 463 | items = github._person_recent_pushes( |
| 464 | "kurt", |
| 465 | "2026-06-25", |
| 466 | "2026-07-25", |
| 467 | {"own_repos": 3}, |
| 468 | "t", |
| 469 | ) |
| 470 | |
| 471 | containers = [item["container"] for item in items] |
| 472 | self.assertIn("kurt/high-star", containers) |
| 473 | self.assertNotIn("kurt/third", containers) |
| 474 | self.assertEqual(enrich.call_count, 3) |
| 475 | |
| 476 | def test_aggregates_each_repo_at_its_latest_matching_push(self): |
| 477 | events = [ |
| 478 | self._event(2, created_at="2026-07-24T12:00:00Z"), |
| 479 | self._event(1, created_at="2026-07-20T12:00:00Z"), |
| 480 | ] |
| 481 | |
| 482 | with patch.object(github, "_fetch_json", return_value=events), \ |
| 483 | patch.object(github, "_fetch_repo_info", return_value={}), \ |
| 484 | patch.object(github, "_enrich_own_repo", return_value={}): |
| 485 | items = github._person_recent_pushes( |
| 486 | "kurt", |
| 487 | "2026-06-25", |
| 488 | "2026-07-25", |
| 489 | {"own_repos": 5}, |
| 490 | "t", |
| 491 | ) |
| 492 | |
| 493 | self.assertEqual(len(items), 1) |
| 494 | self.assertEqual(items[0]["date"], "2026-07-24") |
| 495 | self.assertEqual(items[0]["metadata"]["event_id"], "2") |
| 496 | |
| 497 | |
| 498 | if __name__ == "__main__": |
| 499 | unittest.main() |
| 500 |