Update app.py

app.py

@@ -53,7 +53,7 @@ def chunk(text, words, splitter_selection, slider_overlap):
             ),
         )
         text_splits = [split.content for split in splits]
-    elif splitter_selection == "Character":
+    elif splitter_selection == "LangChain's CharacterTextSplitter":
         text_splitter = CharacterTextSplitter(
             separator="",
             chunk_size=words,
@@ -63,7 +63,16 @@ def chunk(text, words, splitter_selection, slider_overlap):
         )
         splits = text_splitter.create_documents([text])
         text_splits = [split.page_content for split in splits]
-    elif splitter_selection == "Recursive Character Text Splitter":
+    elif splitter_selection == "Langchain's RecursiveCharacterTextSplitter - vanilla":
+        text_splitter = RecursiveCharacterTextSplitter(
+            chunk_size=words,
+            chunk_overlap=slider_overlap,
+            length_function=len,
+            add_start_index=True,
+        )
+        splits = text_splitter.create_documents([text])
+        text_splits = [split.page_content for split in splits]
+    elif splitter_selection == "Langchain's RecursiveCharacterTextSplitter - with '.'":
         text_splitter = RecursiveCharacterTextSplitter(
             chunk_size=words,
             chunk_overlap=slider_overlap,
@@ -125,10 +134,11 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
     text = gr.Textbox(label="Your text 🪶", value=ESSAY)
     split_selection = gr.Radio(
         [
-            "Word - respect sentence boundaries",
-            "Word - no sentence boundaries",
-            "Recursive Character Text Splitter",
-            "Character",
+            "LangChain's CharacterTextSplitter",
+            "Langchain's RecursiveCharacterTextSplitter - vanilla",
+            "Langchain's RecursiveCharacterTextSplitter - with '.'"
+            "Haystack's PreProcessor - Word level, no sentence boundaries",
+            "Haystack's PreProcessor - Word level, respect sentence boundaries",
         ],
         value="Character",
         label="Chunking method ",