Fix cross-year GitHub activity ranges

README.md

@@ -36,6 +36,8 @@ Supported query params:
 - `theme=dark|light`
 - `source=all|github|gitea`
 
+Responses also include an `X-Activity-Stale: true|false` header so image consumers can detect when stale cached data was used after an upstream fetch failure.
+
 Examples:
 
 ```bash

app/main.py

@@ -1,3 +1,4 @@
+import asyncio
 import hashlib
 import json
 import logging
@@ -128,8 +129,14 @@ async def collect_merged_activity(
     github_stale = False
     gitea_stale = False
 
+    fetch_names: list[str] = []
+    fetch_tasks: list[asyncio.Task[tuple[dict[str, int], bool]]] = []
+
     if options.source in ("all", "github"):
-        github_data, github_stale = await _fetch_with_cache(
+        fetch_names.append("github")
+        fetch_tasks.append(
+            asyncio.create_task(
+                _fetch_with_cache(
                     cache,
                     gh_key,
                     lambda: fetch_github_activity(
@@ -138,10 +145,15 @@ async def collect_merged_activity(
                         from_date=from_date,
                         to_date=to_date,
                     ),
+                ),
+            )
         )
 
     if options.source in ("all", "gitea"):
-        gitea_data, gitea_stale = await _fetch_with_cache(
+        fetch_names.append("gitea")
+        fetch_tasks.append(
+            asyncio.create_task(
+                _fetch_with_cache(
                     cache,
                     gt_key,
                     lambda: fetch_gitea_activity(
@@ -151,7 +163,15 @@ async def collect_merged_activity(
                         from_date=from_date,
                         to_date=to_date,
                     ),
+                ),
             )
+        )
+
+    for name, (data, stale) in zip(fetch_names, await asyncio.gather(*fetch_tasks), strict=True):
+        if name == "github":
+            github_data, github_stale = data, stale
+        else:
+            gitea_data, gitea_stale = data, stale
 
     merged = merge_activity(github_data, gitea_data, dates=_date_keys(from_date, to_date))
     merged = filter_activity_source(merged, options.source)
@@ -169,6 +189,10 @@ def _daily_totals(merged: dict[str, dict[str, int]]) -> dict[str, int]:
     return {day: int(payload.get("total", 0)) for day, payload in merged.items()}
 
 
+def _activity_headers(result: ActivityResult) -> dict[str, str]:
+    return {"X-Activity-Stale": "true" if result.stale else "false"}
+
+
 def _image_cache_key(prefix: str, options: QueryOptions, result: ActivityResult) -> str:
     digest = hashlib.sha1(
         json.dumps(result.merged, sort_keys=True, separators=(",", ":")).encode("utf-8")
@@ -203,7 +227,7 @@ async def activity_json(
         "stale": result.stale,
         "activity": result.merged,
     }
-    return JSONResponse(content=payload)
+    return JSONResponse(content=payload, headers=_activity_headers(result))
 
 
 @app.get("/activity.svg")
@@ -222,7 +246,11 @@ async def activity_svg(
     if cache is not None:
         cached = cache.get_json(cache_key)
         if cached is not None:
-            return Response(content=str(cached.value), media_type="image/svg+xml")
+            return Response(
+                content=str(cached.value),
+                media_type="image/svg+xml",
+                headers=_activity_headers(result),
+            )
 
     daily_totals = _daily_totals(result.merged)
     total = sum(daily_totals.values())
@@ -236,7 +264,7 @@ async def activity_svg(
     )
     if cache is not None:
         cache.set_json(cache_key, svg)
-    return Response(content=svg, media_type="image/svg+xml")
+    return Response(content=svg, media_type="image/svg+xml", headers=_activity_headers(result))
 
 
 @app.get("/activity.png")
@@ -256,7 +284,11 @@ async def activity_png(
         cached = cache.get_json(cache_key)
         if cached is not None:
             png_data = bytes.fromhex(str(cached.value))
-            return Response(content=png_data, media_type="image/png")
+            return Response(
+                content=png_data,
+                media_type="image/png",
+                headers=_activity_headers(result),
+            )
 
     daily_totals = _daily_totals(result.merged)
     total = sum(daily_totals.values())
@@ -275,4 +307,4 @@ async def activity_png(
         raise HTTPException(status_code=500, detail="PNG rendering failed") from exc
     if cache is not None:
         cache.set_json(cache_key, png_data.hex())
-    return Response(content=png_data, media_type="image/png")
+    return Response(content=png_data, media_type="image/png", headers=_activity_headers(result))

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:
-        return await _fetch_github_activity_public(
+    normalized: dict[str, int] = {}
+    for range_start, range_end in _iter_year_ranges(from_date, to_date):
+        normalized.update(
+            await _fetch_github_activity_public_range(
                 username=username,
-            from_date=from_date,
-            to_date=to_date,
+                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"}
-    if token:
-        headers["Authorization"] = f"bearer {token}"
+    headers: dict[str, str] = {
+        "Accept": "application/json",
+        "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,
+    }

tests/test_routes.py

@@ -8,6 +8,31 @@ from app.main import ActivityResult, app
 from app.settings import get_settings
 
 
+def test_activity_json_reports_stale_header(monkeypatch) -> None:
+    monkeypatch.setenv("GITHUB_USERNAME", "octocat")
+    monkeypatch.setenv("GITEA_BASE_URL", "https://gitea.example.com")
+    monkeypatch.setenv("GITEA_USERNAME", "octocat")
+    get_settings.cache_clear()
+
+    async def fake_collect_merged_activity(settings, cache, options):
+        return ActivityResult(
+            merged={"2026-01-01": {"github": 1, "gitea": 2, "total": 3}},
+            stale=True,
+            from_date=date(2026, 1, 1),
+            to_date=date(2026, 1, 1),
+            days_count=1,
+        )
+
+    monkeypatch.setattr("app.main.collect_merged_activity", fake_collect_merged_activity)
+
+    client = TestClient(app)
+    response = client.get("/activity.json")
+
+    assert response.status_code == 200
+    assert response.headers["x-activity-stale"] == "true"
+    assert response.json()["stale"] is True
+
+
 def test_activity_svg_returns_svg_content_type(monkeypatch) -> None:
     monkeypatch.setenv("GITHUB_USERNAME", "octocat")
     monkeypatch.setenv("GITEA_BASE_URL", "https://gitea.example.com")
@@ -35,4 +60,18 @@ def test_activity_svg_returns_svg_content_type(monkeypatch) -> None:
 
         assert response.status_code == 200
         assert response.headers["content-type"].startswith("image/svg+xml")
+        assert response.headers["x-activity-stale"] == "false"
         assert "<svg" in response.text
+
+
+def test_activity_json_rejects_year_and_days_together(monkeypatch) -> None:
+    monkeypatch.setenv("GITHUB_USERNAME", "octocat")
+    monkeypatch.setenv("GITEA_BASE_URL", "https://gitea.example.com")
+    monkeypatch.setenv("GITEA_USERNAME", "octocat")
+    get_settings.cache_clear()
+
+    client = TestClient(app)
+    response = client.get("/activity.json?year=2026&days=30")
+
+    assert response.status_code == 400
+    assert response.json() == {"detail": "Provide either year or days, not both"}