feat: added split by page to DocumentSplitter (#6753)

* feat-added-split-by-page-to-DocumentSplitter

* added test case and the suggested changes

* Update document_splitter.py

* Update haystack/components/preprocessors/document_splitter.py

* Update test_document_splitter.py

---------

Co-authored-by: Sebastian Husch Lee <[email protected]>

haystack/components/preprocessors/document_splitter.py

@@ -14,18 +14,21 @@ class DocumentSplitter:
  """
 
  def __init__(
- self, split_by: Literal["word", "sentence", "passage"] = "word", split_length: int = 200, split_overlap: int = 0
+ self,
+ split_by: Literal["word", "sentence", "page", "passage"] = "word",
+ split_length: int = 200,
+ split_overlap: int = 0,
  ):
  """
  :param split_by: The unit by which the document should be split. Choose from "word" for splitting by " ",
- "sentence" for splitting by ".", or "passage" for splitting by "\\n\\n".
+ "sentence" for splitting by ".", "page" for splitting by "\f" or "passage" for splitting by "\\n\\n".
  :param split_length: The maximum number of units in each split.
  :param split_overlap: The number of units that each split should overlap.
  """
 
  self.split_by = split_by
- if split_by not in ["word", "sentence", "passage"]:
- raise ValueError("split_by must be one of 'word', 'sentence' or 'passage'.")
+ if split_by not in ["word", "sentence", "page", "passage"]:
+ raise ValueError("split_by must be one of 'word', 'sentence', 'page' or 'passage'.")
  if split_length <= 0:
  raise ValueError("split_length must be greater than 0.")
  self.split_length = split_length
@@ -60,16 +63,18 @@ def run(self, documents: List[Document]):
  split_docs += [Document(content=txt, meta=metadata) for txt in text_splits]
  return {"documents": split_docs}
 
- def _split_into_units(self, text: str, split_by: Literal["word", "sentence", "passage"]) -> List[str]:
- if split_by == "passage":
+ def _split_into_units(self, text: str, split_by: Literal["word", "sentence", "passage", "page"]) -> List[str]:
+ if split_by == "page":
+ split_at = "\f"
+ elif split_by == "passage":
  split_at = "\n\n"
  elif split_by == "sentence":
  split_at = "."
  elif split_by == "word":
  split_at = " "
  else:
  raise NotImplementedError(
- "DocumentSplitter only supports 'passage', 'sentence' or 'word' split_by options."
+ "DocumentSplitter only supports 'word', 'sentence', 'page' or 'passage' split_by options."
  )
  units = text.split(split_at)
  # Add the delimiter back to all units except the last one

releasenotes/notes/add-split-by-page-to-DocumentSplitter-63232c17d858d787.yaml

@@ -0,0 +1,4 @@
+---
+enhancements:
+ - |
+ Added split by page to DocumentSplitter, which will split the document at \f

test/components/preprocessors/test_document_splitter.py

@@ -23,7 +23,7 @@ def test_empty_list(self):
  assert res == {"documents": []}
 
  def test_unsupported_split_by(self):
- with pytest.raises(ValueError, match="split_by must be one of 'word', 'sentence' or 'passage'."):
+ with pytest.raises(ValueError, match="split_by must be one of 'word', 'sentence', 'page' or 'passage'."):
  DocumentSplitter(split_by="unsupported")
 
  def test_unsupported_split_length(self):
@@ -94,6 +94,20 @@ def test_split_by_passage(self):
  assert result["documents"][1].content == "And there is a third sentence.\n\n"
  assert result["documents"][2].content == " And another passage."
 
+ def test_split_by_page(self):
+ splitter = DocumentSplitter(split_by="page", split_length=1)
+ result = splitter.run(
+ documents=[
+ Document(
+ content="This is a text with some words. There is a second sentence.\f And there is a third sentence.\f And another passage."
+ )
+ ]
+ )
+ assert len(result["documents"]) == 3
+ assert result["documents"][0].content == "This is a text with some words. There is a second sentence.\x0c"
+ assert result["documents"][1].content == " And there is a third sentence.\x0c"
+ assert result["documents"][2].content == " And another passage."
+
  def test_split_by_word_with_overlap(self):
  splitter = DocumentSplitter(split_by="word", split_length=10, split_overlap=2)
  result = splitter.run(