feat: add split_by_row feature to CSVDocumentSplitter (#9031)

* Add split by row feature

haystack/components/preprocessors/csv_document_splitter.py

@@ -3,7 +3,7 @@
 # SPDX-License-Identifier: Apache-2.0
 
 from io import StringIO
-from typing import Any, Dict, List, Literal, Optional, Tuple
+from typing import Any, Dict, List, Literal, Optional, Tuple, get_args
 
 from haystack import Document, component, logging
 from haystack.lazy_imports import LazyImport
@@ -13,14 +13,19 @@ with LazyImport("Run 'pip install pandas'") as pandas_import:
 
 logger = logging.getLogger(__name__)
 
+SplitMode = Literal["threshold", "row-wise"]
+
 
 @component
 class CSVDocumentSplitter:
     """
-    A component for splitting CSV documents into sub-tables based on empty rows and columns.
+    A component for splitting CSV documents into sub-tables based on split arguments.
 
-    The splitter identifies consecutive empty rows or columns that exceed a given threshold
+    The splitter supports two modes of operation:
+    - identify consecutive empty rows or columns that exceed a given threshold
     and uses them as delimiters to segment the document into smaller tables.
+    - split each row into a separate sub-table, represented as a Document.
+
     """
 
     def __init__(
@@ -28,6 +33,7 @@ class CSVDocumentSplitter:
         row_split_threshold: Optional[int] = 2,
         column_split_threshold: Optional[int] = 2,
         read_csv_kwargs: Optional[Dict[str, Any]] = None,
+        split_mode: SplitMode = "threshold",
     ) -> None:
         """
         Initializes the CSVDocumentSplitter component.
@@ -40,8 +46,16 @@ class CSVDocumentSplitter:
             - `skip_blank_lines=False` to preserve blank lines
             - `dtype=object` to prevent type inference (e.g., converting numbers to floats).
             See https://pandas.pydata.org/docs/reference/api/pandas.read_csv.html for more information.
+        :param split_mode:
+            If `threshold`, the component will split the document based on the number of
+            consecutive empty rows or columns that exceed the `row_split_threshold` or `column_split_threshold`.
+            If `row-wise`, the component will split each row into a separate sub-table.
         """
         pandas_import.check()
+        if split_mode not in get_args(SplitMode):
+            raise ValueError(
+                f"Split mode '{split_mode}' not recognized. Choose one among: {', '.join(get_args(SplitMode))}."
+            )
         if row_split_threshold is not None and row_split_threshold < 1:
             raise ValueError("row_split_threshold must be greater than 0")
 
@@ -54,6 +68,7 @@ class CSVDocumentSplitter:
         self.row_split_threshold = row_split_threshold
         self.column_split_threshold = column_split_threshold
         self.read_csv_kwargs = read_csv_kwargs or {}
+        self.split_mode = split_mode
 
     @component.output_types(documents=List[Document])
     def run(self, documents: List[Document]) -> Dict[str, List[Document]]:
@@ -89,6 +104,7 @@ class CSVDocumentSplitter:
         resolved_read_csv_kwargs = {"header": None, "skip_blank_lines": False, "dtype": object, **self.read_csv_kwargs}
 
         split_documents = []
+        split_dfs = []
         for document in documents:
             try:
                 df = pd.read_csv(StringIO(document.content), **resolved_read_csv_kwargs)  # type: ignore
@@ -97,19 +113,32 @@ class CSVDocumentSplitter:
                 split_documents.append(document)
                 continue
 
-            if self.row_split_threshold is not None and self.column_split_threshold is None:
-                # split by rows
-                split_dfs = self._split_dataframe(df=df, split_threshold=self.row_split_threshold, axis="row")
-            elif self.column_split_threshold is not None and self.row_split_threshold is None:
-                # split by columns
-                split_dfs = self._split_dataframe(df=df, split_threshold=self.column_split_threshold, axis="column")
-            else:
-                # recursive split
-                split_dfs = self._recursive_split(
-                    df=df,
-                    row_split_threshold=self.row_split_threshold,  # type: ignore
-                    column_split_threshold=self.column_split_threshold,  # type: ignore
+            if self.split_mode == "row-wise":
+                # each row is a separate sub-table
+                split_dfs = self._split_by_row(df=df)
+
+            elif self.split_mode == "threshold":
+                if self.row_split_threshold is not None and self.column_split_threshold is None:
+                    # split by rows
+                    split_dfs = self._split_dataframe(df=df, split_threshold=self.row_split_threshold, axis="row")
+                elif self.column_split_threshold is not None and self.row_split_threshold is None:
+                    # split by columns
+                    split_dfs = self._split_dataframe(df=df, split_threshold=self.column_split_threshold, axis="column")
+                else:
+                    # recursive split
+                    split_dfs = self._recursive_split(
+                        df=df,
+                        row_split_threshold=self.row_split_threshold,  # type: ignore
+                        column_split_threshold=self.column_split_threshold,  # type: ignore
+                    )
+
+            # check if no sub-tables were found
+            if len(split_dfs) == 0:
+                logger.warning(
+                    "No sub-tables found while splitting CSV Document with id {doc_id}. Skipping document.",
+                    doc_id=document.id,
                 )
+                continue
 
             # Sort split_dfs first by row index, then by column index
             split_dfs.sort(key=lambda dataframe: (dataframe.index[0], dataframe.columns[0]))
@@ -242,3 +271,12 @@ class CSVDocumentSplitter:
                 result.append(table)
 
         return result
+
+    def _split_by_row(self, df: "pd.DataFrame") -> List["pd.DataFrame"]:
+        """Split each CSV row into a separate subtable"""
+        split_dfs = []
+        for idx, row in enumerate(df.itertuples(index=False)):
+            split_df = pd.DataFrame(row).T
+            split_df.index = [idx]  # Set the index of the new DataFrame to idx
+            split_dfs.append(split_df)
+        return split_dfs

releasenotes/notes/add-split-by-row-csv-splitter-e810b96b0db287b3.yaml

@@ -0,0 +1,6 @@
+---
+features:
+  - |
+    Added a new parameter `split_mode` to the `CSVDocumentSplitter` component to control the splitting mode.
+    The new parameter can be set to `row-wise` to split the CSV file by rows.
+    The default value is `threshold`, which is the previous behavior.

test/components/preprocessors/test_csv_document_splitter.py

@@ -3,6 +3,7 @@
 # SPDX-License-Identifier: Apache-2.0
 
 import pytest
+import logging
 from pandas import read_csv
 from io import StringIO
 from haystack import Document, Pipeline
@@ -15,6 +16,15 @@ def splitter() -> CSVDocumentSplitter:
     return CSVDocumentSplitter()
 
 
+@pytest.fixture
+def csv_with_four_rows() -> str:
+    return """A,B,C
+1,2,3
+X,Y,Z
+7,8,9
+"""
+
+
 @pytest.fixture
 def two_tables_sep_by_two_empty_rows() -> str:
     return """A,B,C
@@ -255,7 +265,12 @@ E,F,,,G,H
         config_serialized = component_to_dict(splitter, name="CSVDocumentSplitter")
         config = {
             "type": "haystack.components.preprocessors.csv_document_splitter.CSVDocumentSplitter",
-            "init_parameters": {"row_split_threshold": 2, "column_split_threshold": 2, "read_csv_kwargs": {}},
+            "init_parameters": {
+                "row_split_threshold": 2,
+                "column_split_threshold": 2,
+                "read_csv_kwargs": {},
+                "split_mode": "threshold",
+            },
         }
         assert config_serialized == config
 
@@ -268,6 +283,7 @@ E,F,,,G,H
                 "row_split_threshold": 1,
                 "column_split_threshold": None,
                 "read_csv_kwargs": {"sep": ";"},
+                "split_mode": "threshold",
             },
         }
         assert config_serialized == config
@@ -284,6 +300,7 @@ E,F,,,G,H
         assert splitter.row_split_threshold == 2
         assert splitter.column_split_threshold == 2
         assert splitter.read_csv_kwargs == {}
+        assert splitter.split_mode == "threshold"
 
     def test_from_dict_non_defaults(self) -> None:
         splitter = component_from_dict(
@@ -294,6 +311,7 @@ E,F,,,G,H
                     "row_split_threshold": 1,
                     "column_split_threshold": None,
                     "read_csv_kwargs": {"sep": ";"},
+                    "split_mode": "row-wise",
                 },
             },
             name="CSVDocumentSplitter",
@@ -301,3 +319,25 @@ E,F,,,G,H
         assert splitter.row_split_threshold == 1
         assert splitter.column_split_threshold is None
         assert splitter.read_csv_kwargs == {"sep": ";"}
+        assert splitter.split_mode == "row-wise"
+
+    def test_split_by_row(self, csv_with_four_rows: str) -> None:
+        splitter = CSVDocumentSplitter(split_mode="row-wise")
+        doc = Document(content=csv_with_four_rows)
+        result = splitter.run([doc])["documents"]
+        assert len(result) == 4
+        assert result[0].content == "A,B,C\n"
+        assert result[1].content == "1,2,3\n"
+        assert result[2].content == "X,Y,Z\n"
+
+    def test_split_by_row_with_empty_rows(self, caplog) -> None:
+        splitter = CSVDocumentSplitter(split_mode="row-wise")
+        doc = Document(content="")
+        with caplog.at_level(logging.ERROR):
+            result = splitter.run([doc])["documents"]
+            assert len(result) == 1
+            assert result[0].content == ""
+
+    def test_incorrect_split_mode(self) -> None:
+        with pytest.raises(ValueError, match="not recognized"):
+            CSVDocumentSplitter(split_mode="incorrect_mode")