From 68bbf945770eb1b2ece44dbf739fad0a830369d9 Mon Sep 17 00:00:00 2001
From: Yiorgis Gozadinos <ggozadinos@gmail.com>
Date: Mon, 22 Jun 2026 11:58:32 +0300
Subject: [PATCH] tighten run-batch manifest replay validation

---
 haiku_rag_slim/haiku/rag/ingester/app.py | 11 ++++-----
 haiku_rag_slim/haiku/rag/ingester/cli.py |  2 +-
 tests/ingester/test_cli.py               | 10 ++++++++
 tests/ingester/test_run_batch.py         | 30 ++++++++++++++++++++++++
 4 files changed, 45 insertions(+), 8 deletions(-)

diff --git a/haiku_rag_slim/haiku/rag/ingester/app.py b/haiku_rag_slim/haiku/rag/ingester/app.py
index 559fd74f..11bde57a 100644
--- a/haiku_rag_slim/haiku/rag/ingester/app.py
+++ b/haiku_rag_slim/haiku/rag/ingester/app.py
@@ -307,16 +307,13 @@ class IngesterApp:
                     "Manifest references unconfigured source(s): " + ", ".join(missing)
                 )
 
-            pending = [
-                source_id
-                for source_id in sorted(manifest_sources)
-                if await self._jobs.has_pending(source_id)
-            ]
+            counts = await self._jobs.counts_by_status()
+            pending = counts.get("queued", 0) + counts.get("claimed", 0)
             if pending:
                 await self._pollers.close_sources()
                 raise ValueError(
-                    "Cannot replay manifest while source(s) have pending work: "
-                    + ", ".join(pending)
+                    "Cannot replay manifest while the queue has pending work: "
+                    f"{pending} queued/claimed job(s)"
                 )
 
             seen: set[tuple[str, str]] = set()
diff --git a/haiku_rag_slim/haiku/rag/ingester/cli.py b/haiku_rag_slim/haiku/rag/ingester/cli.py
index 0fb63bd0..3527edf5 100644
--- a/haiku_rag_slim/haiku/rag/ingester/cli.py
+++ b/haiku_rag_slim/haiku/rag/ingester/cli.py
@@ -229,7 +229,7 @@ def run_batch(
     source's sweep does not complete."""
     if manifest is not None and dry_run:
         raise typer.BadParameter("--manifest cannot be combined with --dry-run")
-    if manifest is not None and output is not None:
+    if output is not None and not dry_run:
         raise typer.BadParameter("--output is only valid with --dry-run")
     asyncio.run(
         _run_batch(
diff --git a/tests/ingester/test_cli.py b/tests/ingester/test_cli.py
index 07ef4e24..d2b3bcf7 100644
--- a/tests/ingester/test_cli.py
+++ b/tests/ingester/test_cli.py
@@ -237,6 +237,16 @@ def test_run_batch_manifest_conflicts_with_output(tmp_path):
     assert "--output is only valid with --dry-run" in result.output
 
 
+def test_run_batch_output_requires_dry_run(tmp_path):
+    result = runner.invoke(
+        cli,
+        ["run-batch", "--output", str(tmp_path / "out.yaml")],
+    )
+
+    assert result.exit_code != 0
+    assert "--output is only valid with --dry-run" in result.output
+
+
 # --- serve ---
 
 
diff --git a/tests/ingester/test_run_batch.py b/tests/ingester/test_run_batch.py
index 636d0d9b..f9f28c98 100644
--- a/tests/ingester/test_run_batch.py
+++ b/tests/ingester/test_run_batch.py
@@ -397,6 +397,36 @@ async def test_run_batch_from_manifest_rejects_pending_work(tmp_path, use_client
         )
 
 
+@pytest.mark.asyncio
+async def test_run_batch_from_manifest_rejects_unrelated_pending_work(
+    tmp_path, use_client
+):
+    (tmp_path / "a.md").write_text("hello")
+    config = _config(tmp_path)
+    client = _mock_client()
+    use_client(client)
+    engine = await open_queue(config.ingester.queue)
+    try:
+        jobs = JobRepo(engine)
+        await jobs.enqueue("other", "file:///outside.md", op=JobOp.UPSERT)
+    finally:
+        await engine.dispose()
+
+    with pytest.raises(ValueError, match="queue has pending work"):
+        await IngesterApp(
+            config=config, db_path=tmp_path / "db.lancedb"
+        ).run_batch_from_manifest(
+            _manifest(
+                BatchChange(
+                    op=JobOp.UPSERT,
+                    source_id="local",
+                    uri=(tmp_path / "a.md").as_uri(),
+                    discovered_at=datetime.now(UTC),
+                )
+            )
+        )
+
+
 @pytest.mark.asyncio
 async def test_run_batch_from_manifest_rejects_duplicate_changes(tmp_path, use_client):
     path = tmp_path / "a.md"