Unlink PDF slice tempfiles when slice write fails

haiku_rag_slim/haiku/rag/converters/pdf_split.py

@@ -99,13 +99,14 @@ async def convert_pdf_with_splitting(
             if slice_item is _SENTINEL:
                 break
             start, end, pdf_bytes = slice_item
-            with tempfile.NamedTemporaryFile(
-                mode="wb", suffix=".pdf", delete=False
-            ) as tmp:
-                tmp.write(pdf_bytes)
-                tmp.flush()
-                tmp_path = Path(tmp.name)
+            tmp_path: Path | None = None
             try:
+                with tempfile.NamedTemporaryFile(
+                    mode="wb", suffix=".pdf", delete=False
+                ) as tmp:
+                    tmp_path = Path(tmp.name)
+                    tmp.write(pdf_bytes)
+                    tmp.flush()
                 with logfire.span(
                     "document.convert_slice",
                     uri=source_uri,
@@ -122,7 +123,12 @@ async def convert_pdf_with_splitting(
                         ) from exc
                 converted.append(slice_doc)
             finally:
-                tmp_path.unlink(missing_ok=True)
+                # `delete=False` is required so the converter (which opens
+                # tmp_path itself) sees a fully written, closed file. Unlink
+                # in finally so a mid-write disk-full / mid-convert failure
+                # doesn't leak the slice on disk.
+                if tmp_path is not None:
+                    tmp_path.unlink(missing_ok=True)
     finally:
         # Close the generator under the pdfium lock so src.close() runs
         # even when we abort mid-stream (slice failure, cancellation).

tests/test_pdf_split.py

@@ -69,6 +69,53 @@ def test_iter_pdf_slices_rejects_zero_slice_size(tmp_path):
         list(iter_pdf_slices(src, slice_size=0))
 
 
+@pytest.mark.asyncio
+async def test_convert_unlinks_slice_tempfile_on_write_failure(tmp_path, monkeypatch):
+    """If writing the slice bytes to the tempfile raises (e.g. ENOSPC), the
+    tempfile is created on disk but never reaches the converter. The original
+    error must surface and the orphaned file must be removed."""
+    src = _make_pdf(2, tmp_path)
+
+    monkeypatch.setattr("tempfile.tempdir", str(tmp_path))
+
+    import tempfile as _tempfile
+
+    real_factory = _tempfile.NamedTemporaryFile
+    created: list[Path] = []
+
+    def _make_failing_tempfile(*args, **kwargs):
+        handle = real_factory(*args, **kwargs)
+        created.append(Path(handle.name))
+        original_write = handle.write
+
+        def _raising_write(_data):
+            # Touch the underlying file once so we know the path exists on
+            # disk and the cleanup actually has something to remove.
+            original_write(b"\0")
+            raise OSError("No space left on device")
+
+        handle.write = _raising_write
+        return handle
+
+    monkeypatch.setattr(_tempfile, "NamedTemporaryFile", _make_failing_tempfile)
+
+    class _UnusedConverter:
+        async def convert_file(self, path: Path, *, source_uri):
+            raise AssertionError("converter must not be reached on write failure")
+
+    with pytest.raises(OSError, match="No space left on device"):
+        await convert_pdf_with_splitting(
+            _UnusedConverter(),  # ty: ignore[invalid-argument-type]
+            src,
+            source_uri=None,
+            slice_size=1,
+        )
+
+    assert created, "expected NamedTemporaryFile to be called at least once"
+    leftover = [p for p in created if p.exists()]
+    assert leftover == [], f"tempfiles leaked after write failure: {leftover}"
+
+
 @pytest.mark.asyncio
 async def test_convert_aborts_and_cleans_up_on_mid_stream_slice_failure(
     tmp_path, monkeypatch