fix(sources): correct OpenAlex ID prefixing, citations endpoint, S2 rate-limit

Review-Gate findings:
- openalex: bare DOIs/arXiv IDs need doi:/arxiv: prefix (bare IDs 404);
  add _resolve_id(); fix fetch_citations to use referenced_works field
  instead of non-existent /works/{id}/references endpoint.
- semanticscholar: wait_fixed(1) was inter-retry only; add per-request
  monotonic rate-limiter (_rate_limit()) before each httpx call.
  Add _is_retryable() filter so 404s don't burn 5 retry slots.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

tests/sources/test_openalex.py

@@ -7,6 +7,7 @@ import pytest
 
 from codex.models import Citation, Paper
 from codex.sources import openalex
+from codex.sources.openalex import _resolve_id
 
 # ---------------------------------------------------------------------------
 # Helpers
@@ -27,14 +28,33 @@ _SAMPLE_WORK = {
         "is": [2],
         "great": [3],
     },
+    "referenced_works": [
+        "https://openalex.org/W1",
+        "https://openalex.org/W2",
+    ],
 }
 
-_SAMPLE_REFS = {
-    "results": [
-        {"doi": "10.1000/ref1", "id": "https://openalex.org/W1"},
-        {"doi": "10.1000/ref2", "id": "https://openalex.org/W2"},
-    ]
-}
+
+# ---------------------------------------------------------------------------
+# _resolve_id
+# ---------------------------------------------------------------------------
+
+
+def test_resolve_doi() -> None:
+    assert _resolve_id("10.1145/3592430") == "doi:10.1145/3592430"
+
+
+def test_resolve_arxiv() -> None:
+    assert _resolve_id("2301.07041") == "arxiv:2301.07041"
+
+
+def test_resolve_w_id_unchanged() -> None:
+    assert _resolve_id("W2741809807") == "W2741809807"
+
+
+def test_resolve_prefixed_unchanged() -> None:
+    assert _resolve_id("doi:10.1145/x") == "doi:10.1145/x"
+    assert _resolve_id("arxiv:2301.07041") == "arxiv:2301.07041"
 
 
 # ---------------------------------------------------------------------------
@@ -46,6 +66,7 @@ def test_fetch_paper_success(monkeypatch: pytest.MonkeyPatch) -> None:
     """Successful fetch maps all fields to Paper correctly."""
 
     def mock_get(url: str, params: dict[str, str] | None = None) -> httpx.Response:
+        assert "doi:10.1145/3592430" in url
         return httpx.Response(200, json=_SAMPLE_WORK)
 
     monkeypatch.setattr(openalex, "_get", mock_get)
@@ -61,6 +82,20 @@ def test_fetch_paper_success(monkeypatch: pytest.MonkeyPatch) -> None:
     assert paper.openalex_id == "https://openalex.org/W2741809807"
 
 
+def test_fetch_paper_arxiv_id_prefixed(monkeypatch: pytest.MonkeyPatch) -> None:
+    """Bare arXiv IDs are prefixed with 'arxiv:' in the URL."""
+    called_url: list[str] = []
+
+    def mock_get(url: str, params: dict[str, str] | None = None) -> httpx.Response:
+        called_url.append(url)
+        return httpx.Response(200, json=_SAMPLE_WORK)
+
+    monkeypatch.setattr(openalex, "_get", mock_get)
+    openalex.fetch_paper("2301.07041")
+
+    assert called_url and "arxiv:2301.07041" in called_url[0]
+
+
 # ---------------------------------------------------------------------------
 # fetch_paper — 404 → None
 # ---------------------------------------------------------------------------
@@ -88,7 +123,7 @@ def test_fetch_paper_404_returns_none(monkeypatch: pytest.MonkeyPatch) -> None:
 
 
 def test_fetch_paper_retry_on_429(monkeypatch: pytest.MonkeyPatch) -> None:
-    """On 429 x3 then 200, should return 1 Paper after 3 failed attempts."""
+    """On 429×3 then 200, returns 1 Paper after 3 failed attempts."""
     attempt = [0]
 
     def controlled_get(url: str, params: dict[str, str] | None = None) -> httpx.Response:
@@ -104,7 +139,6 @@ def test_fetch_paper_retry_on_429(monkeypatch: pytest.MonkeyPatch) -> None:
 
     monkeypatch.setattr(openalex, "_get", controlled_get)
 
-    # Call fetch_paper in a loop to simulate retry behaviour
     result: Paper | None = None
     errors = 0
     for _ in range(5):
@@ -116,19 +150,19 @@ def test_fetch_paper_retry_on_429(monkeypatch: pytest.MonkeyPatch) -> None:
 
     assert result is not None
     assert isinstance(result, Paper)
-    assert errors == 3  # 3 failures before success
+    assert errors == 3
 
 
 # ---------------------------------------------------------------------------
-# fetch_citations
+# fetch_citations — uses referenced_works field
 # ---------------------------------------------------------------------------
 
 
 def test_fetch_citations(monkeypatch: pytest.MonkeyPatch) -> None:
-    """fetch_citations returns one Citation per reference entry."""
+    """fetch_citations reads referenced_works from the work object."""
 
     def mock_get(url: str, params: dict[str, str] | None = None) -> httpx.Response:
-        return httpx.Response(200, json=_SAMPLE_REFS)
+        return httpx.Response(200, json=_SAMPLE_WORK)
 
     monkeypatch.setattr(openalex, "_get", mock_get)
 
@@ -138,5 +172,19 @@ def test_fetch_citations(monkeypatch: pytest.MonkeyPatch) -> None:
     assert len(citations) == 2
     assert all(isinstance(c, Citation) for c in citations)
     assert citations[0].citing_id == oa_id
-    assert citations[0].cited_id == "10.1000/ref1"
-    assert citations[1].cited_id == "10.1000/ref2"
+    assert citations[0].cited_id == "https://openalex.org/W1"
+    assert citations[1].cited_id == "https://openalex.org/W2"
+
+
+def test_fetch_citations_404_returns_empty(monkeypatch: pytest.MonkeyPatch) -> None:
+    """404 on the work fetch returns an empty list."""
+
+    def mock_get(url: str, params: dict[str, str] | None = None) -> httpx.Response:
+        raise httpx.HTTPStatusError(
+            "Not Found",
+            request=httpx.Request("GET", url),
+            response=httpx.Response(404, request=httpx.Request("GET", url)),
+        )
+
+    monkeypatch.setattr(openalex, "_get", mock_get)
+    assert openalex.fetch_citations("W999") == []