More complicated test

deepset-ai · Feb 7, 2025 · 1003904 · 1003904
1 parent 4ecaea6
commit 1003904
Showing 1 changed file with 16 additions and 1 deletion.
diff --git a/test/components/preprocessors/test_csv_document_splitter.py b/test/components/preprocessors/test_csv_document_splitter.py
@@ -128,7 +128,7 @@ def test_column_split(self, splitter: CSVDocumentSplitter, two_tables_sep_by_two
         for i, table in enumerate(result):
             assert table.content == expected_tables[i]
 
-    def test_recursive_split(self, splitter: CSVDocumentSplitter) -> None:
+    def test_recursive_split_one_level(self, splitter: CSVDocumentSplitter) -> None:
         csv_content = """A,B,,,X,Y
 1,2,,,7,8
 ,,,,,
@@ -143,6 +143,21 @@ def test_recursive_split(self, splitter: CSVDocumentSplitter) -> None:
         for i, table in enumerate(result):
             assert table.content == expected_tables[i]
 
+    def test_recursive_split_two_levels(self, splitter: CSVDocumentSplitter) -> None:
+        csv_content = """A,B,,,X,Y
+1,2,,,7,8
+,,,,M,N
+,,,,9,10
+P,Q,,,,
+3,4,,,,
+"""
+        doc = Document(content=csv_content)
+        result = splitter.run([doc])["documents"]
+        assert len(result) == 3
+        expected_tables = ["A,B\n1,2\n", "P,Q\n3,4\n", "X,Y\n7,8\nM,N\n9,10\n"]
+        for i, table in enumerate(result):
+            assert table.content == expected_tables[i]
+
     def test_threshold_no_effect(self, two_tables_sep_by_two_empty_rows: str) -> None:
         splitter = CSVDocumentSplitter(row_split_threshold=3)
         doc = Document(content=two_tables_sep_by_two_empty_rows)