Runs export: a truncated one counts the selection into X-Total-Count; the SDK and fluksio export warn how many runs were left out

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01WgBfkm4rCmtd1T4vHWjEPx

backend/fluksio/api/routes/runs.py

@@ -512,15 +512,22 @@ def _selected(
     ids: str,
     since: datetime | None,
     until: datetime | None,
-) -> list[Run]:
-    """The runs an export covers, newest first — the filters the list takes.
+) -> tuple[list[Run], int]:
+    """The runs an export covers, newest first, and how many the filters match.
 
     Capped: the filters bound a *sensible* request and nothing bounded an
     unfiltered one, so asking for everything read every row of the table into
-    memory before a byte was sent. A truncated export says so in a header.
+    memory before a byte was sent. A truncated export says so in a header, and
+    only then is the selection counted, since otherwise the rows are the count.
     """
     statement = _selection(flow, status, group, ids, since, until)
-    return list(session.exec(statement.limit(EXPORT_CAP + 1)))
+    runs = list(session.exec(statement.limit(EXPORT_CAP + 1)))
+    if len(runs) <= EXPORT_CAP:
+        return runs, len(runs)
+    total = session.exec(
+        select(func.count()).select_from(statement.order_by(None).subquery())
+    ).one()
+    return runs[:EXPORT_CAP], total
 
 
 def _cell(value: Any) -> Any:
@@ -613,7 +620,7 @@ def _stream(
     name: str,
     columns: list[str],
     chunks: Iterator[list[dict[str, Any]]],
-    truncated: bool = False,
+    total: int = 0,
 ) -> StreamingResponse:
     """The rows out, a chunk at a time rather than a list.
 
@@ -629,11 +636,12 @@ def _stream(
         body = _csv(columns, chunks)
         media = "text/csv; charset=utf-8"
     headers = {"Content-Disposition": f'attachment; filename="{name}.{fmt}"'}
-    if truncated:
+    if total > EXPORT_CAP:
         # The body is otherwise indistinguishable from a complete one, and an
         # export quietly missing its oldest runs is the wrong thing to hand
-        # somebody who is about to plot it.
+        # somebody who is about to plot it. The total says how many it missed.
         headers["X-Truncated"] = str(EXPORT_CAP)
+        headers["X-Total-Count"] = str(total)
     return StreamingResponse(body, media_type=media, headers=headers)
 
 
@@ -656,9 +664,7 @@ def export_metrics(
     the metrics it lists; ``stride`` thins each curve — per series, so asking
     for every tenth point of two metrics gives every tenth point of both.
     """
-    runs = _selected(session, flow, status, group, ids, since, until)
-    truncated = len(runs) > EXPORT_CAP
-    runs = runs[:EXPORT_CAP]
+    runs, total = _selected(session, flow, status, group, ids, since, until)
     wanted = [part for part in name.split(",") if part]
     step = max(1, stride)
 
@@ -682,7 +688,7 @@ def export_metrics(
                 )
             yield rows
 
-    return _stream(format, "metrics", list(METRIC_COLUMNS), chunks(), truncated)
+    return _stream(format, "metrics", list(METRIC_COLUMNS), chunks(), total)
 
 
 @router.get("/export/runs")
@@ -710,9 +716,7 @@ def export_runs(
     three fields rather than two blobs, and the defaults reach the same
     depth.
     """
-    runs = _selected(session, flow, status, group, ids, since, until)
-    truncated = len(runs) > EXPORT_CAP
-    runs = runs[:EXPORT_CAP]
+    runs, total = _selected(session, flow, status, group, ids, since, until)
     inputs = [part for part in params.split(",") if part] or sorted(
         {path for run in runs for path, _ in _leaves(run.params)}
     )
@@ -730,7 +734,7 @@ def export_runs(
             row.update((f"metric.{k}", _cell(_dig(run.result, k))) for k in scores)
             yield [row]
 
-    return _stream(format, "runs", columns, chunks(), truncated)
+    return _stream(format, "runs", columns, chunks(), total)
 
 
 @router.get("/metrics/names", response_model=list[str])