Fix cross-year GitHub activity ranges

app/sources/github.py

@@ -11,6 +11,15 @@ class GitHubSourceError(RuntimeError):
     pass
 
 
+def _iter_year_ranges(from_date: date, to_date: date) -> list[tuple[date, date]]:
+    ranges: list[tuple[date, date]] = []
+    for year in range(from_date.year, to_date.year + 1):
+        range_start = max(from_date, date(year, 1, 1))
+        range_end = min(to_date, date(year, 12, 31))
+        ranges.append((range_start, range_end))
+    return ranges
+
+
 def _extract_attr(tag: str, attr: str) -> str | None:
     match = re.search(rf'{attr}="([^"]+)"', tag)
     return match.group(1) if match else None
@@ -42,7 +51,7 @@ def _parse_public_contributions_html(html: str, from_date: date, to_date: date)
     return normalized
 
 
-async def _fetch_github_activity_public(
+async def _fetch_github_activity_public_range(
     username: str,
     from_date: date,
     to_date: date,
@@ -66,21 +75,32 @@ async def _fetch_github_activity_public(
     return _parse_public_contributions_html(response.text, from_date, to_date)
 
 
-async def fetch_github_activity(
+async def _fetch_github_activity_public(
     username: str,
-    token: str | None,
     from_date: date,
     to_date: date,
-    timeout_seconds: float = 20.0,
+    timeout_seconds: float,
 ) -> dict[str, int]:
-    if not token:
+    normalized: dict[str, int] = {}
-        return await _fetch_github_activity_public(
+    for range_start, range_end in _iter_year_ranges(from_date, to_date):
-            username=username,
+        normalized.update(
-            from_date=from_date,
+            await _fetch_github_activity_public_range(
-            to_date=to_date,
+                username=username,
-            timeout_seconds=timeout_seconds,
+                from_date=range_start,
+                to_date=range_end,
+                timeout_seconds=timeout_seconds,
+            )
         )
+    return normalized
 
+
+async def _fetch_github_activity_graphql_range(
+    username: str,
+    token: str,
+    from_date: date,
+    to_date: date,
+    timeout_seconds: float,
+) -> dict[str, int]:
     query = """
     query($login: String!, $from: DateTime!, $to: DateTime!) {
       user(login: $login) {
@@ -103,9 +123,10 @@ async def fetch_github_activity(
         "to": datetime.combine(to_date, datetime.max.time(), tzinfo=timezone.utc).isoformat(),
     }
 
-    headers: dict[str, str] = {"Accept": "application/json"}
+    headers: dict[str, str] = {
-    if token:
+        "Accept": "application/json",
-        headers["Authorization"] = f"bearer {token}"
+        "Authorization": f"bearer {token}",
+    }
 
     async with httpx.AsyncClient(timeout=timeout_seconds) as client:
         response = await client.post(
@@ -140,3 +161,32 @@ async def fetch_github_activity(
             normalized[date_key] = int(day.get("contributionCount", 0))
 
     return normalized
+
+
+async def fetch_github_activity(
+    username: str,
+    token: str | None,
+    from_date: date,
+    to_date: date,
+    timeout_seconds: float = 20.0,
+) -> dict[str, int]:
+    if not token:
+        return await _fetch_github_activity_public(
+            username=username,
+            from_date=from_date,
+            to_date=to_date,
+            timeout_seconds=timeout_seconds,
+        )
+
+    normalized: dict[str, int] = {}
+    for range_start, range_end in _iter_year_ranges(from_date, to_date):
+        normalized.update(
+            await _fetch_github_activity_graphql_range(
+                username=username,
+                token=token,
+                from_date=range_start,
+                to_date=range_end,
+                timeout_seconds=timeout_seconds,
+            )
+        )
+    return normalized

tests/test_github_source.py

@@ -0,0 +1,88 @@
+from datetime import date
+
+import pytest
+
+from app.sources import github as github_source
+
+
+def test_iter_year_ranges_splits_cross_year_window() -> None:
+    ranges = github_source._iter_year_ranges(date(2025, 6, 1), date(2026, 6, 1))
+
+    assert ranges == [
+        (date(2025, 6, 1), date(2025, 12, 31)),
+        (date(2026, 1, 1), date(2026, 6, 1)),
+    ]
+
+
+@pytest.mark.asyncio
+async def test_fetch_github_activity_without_token_merges_year_slices(monkeypatch) -> None:
+    requested_ranges: list[tuple[date, date]] = []
+
+    async def fake_fetch_public_range(username, from_date, to_date, timeout_seconds):
+        requested_ranges.append((from_date, to_date))
+        return {
+            from_date.isoformat(): 1,
+            to_date.isoformat(): 2,
+        }
+
+    monkeypatch.setattr(
+        github_source,
+        "_fetch_github_activity_public_range",
+        fake_fetch_public_range,
+    )
+
+    activity = await github_source.fetch_github_activity(
+        username="octocat",
+        token=None,
+        from_date=date(2025, 6, 1),
+        to_date=date(2026, 6, 1),
+        timeout_seconds=1.0,
+    )
+
+    assert requested_ranges == [
+        (date(2025, 6, 1), date(2025, 12, 31)),
+        (date(2026, 1, 1), date(2026, 6, 1)),
+    ]
+    assert activity == {
+        "2025-06-01": 1,
+        "2025-12-31": 2,
+        "2026-01-01": 1,
+        "2026-06-01": 2,
+    }
+
+
+@pytest.mark.asyncio
+async def test_fetch_github_activity_with_token_merges_year_slices(monkeypatch) -> None:
+    requested_ranges: list[tuple[date, date, str]] = []
+
+    async def fake_fetch_graphql_range(username, token, from_date, to_date, timeout_seconds):
+        requested_ranges.append((from_date, to_date, token))
+        return {
+            from_date.isoformat(): 3,
+            to_date.isoformat(): 4,
+        }
+
+    monkeypatch.setattr(
+        github_source,
+        "_fetch_github_activity_graphql_range",
+        fake_fetch_graphql_range,
+    )
+
+    activity = await github_source.fetch_github_activity(
+        username="octocat",
+        token="secret",
+        from_date=date(2025, 6, 1),
+        to_date=date(2026, 6, 1),
+        timeout_seconds=1.0,
+    )
+
+    assert requested_ranges == [
+        (date(2025, 6, 1), date(2025, 12, 31), "secret"),
+        (date(2026, 1, 1), date(2026, 6, 1), "secret"),
+    ]
+    assert activity == {
+        "2025-06-01": 3,
+        "2025-12-31": 4,
+        "2026-01-01": 3,
+        "2026-06-01": 4,
+    }