Nomad changes

e2e/python/tests/test_code.py

@@ -0,0 +1,41 @@
+# This file is auto-generated by alef — DO NOT EDIT.
+# alef:hash:4e15143f4af1ae8bafbdb1506ef057da924484c66a19483966333558ad437e75
+# To regenerate: alef generate
+# To verify freshness: alef verify --exit-code
+# Issues & docs: https://github.com/kreuzberg-dev/alef
+
+"""E2e tests for category: code."""
+
+import pytest  # noqa: F401
+from kreuzberg import extract_file_sync
+
+
+def _alef_e2e_text(value: object) -> str:
+    return "" if value is None else str(value)
+
+
+def _alef_e2e_item_texts(item: object) -> tuple[str, ...]:
+    raw_items = getattr(item, "items", None)
+    items_text = " ".join(str(value) for value in raw_items) if isinstance(raw_items, list) else ""
+    return (
+        _alef_e2e_text(item),
+        _alef_e2e_text(getattr(item, "kind", None)),
+        _alef_e2e_text(getattr(item, "name", None)),
+        _alef_e2e_text(getattr(item, "source", None)),
+        _alef_e2e_text(getattr(item, "alias", None)),
+        _alef_e2e_text(getattr(item, "text", None)),
+        _alef_e2e_text(getattr(item, "signature", None)),
+        items_text,
+    )
+
+
+def test_code_shebang_detection() -> None:
+    """Test language detection from shebang line via bytes input."""
+    path = "code/script.sh"
+    mime_type = "text/x-source-code"
+
+    result = extract_file_sync(path, mime_type, None)
+    assert result.mime_type.strip() == "text/x-source-code"  # noqa: S101
+    assert len(result.content) >= 10  # noqa: S101
+    assert "build" in result.content  # noqa: S101
+    assert "clean" in result.content  # noqa: S101