Fix bug in pinecone utils and add test for docubot build knowledge ba…

…se method.

pinecone_utils/pinecone_utils.py

@@ -56,11 +56,13 @@ def create_vector_store(index_name: str, chunks: List[T]) -> Pinecone:
     import pinecone
     from langchain.vectorstores import Pinecone
     from langchain.embeddings.openai import OpenAIEmbeddings
+    from text_utils.text_utils import embedding_cost
 
     # Prompt user whether they want to continue, quit if they don't
     while True:
         user_input = input(
-            f"Would you like to continue? (y/n) (Warning: This will cost approximately ${embedding_cost(chunks):.4f}) "
+            f"Cost Estimate: ${embedding_cost(chunks):.4f}\n"
+            f"Would you like to continue? (y/n)\n"
         )
         if user_input.lower() == "y":
             break

tests/test_docubot.py

@@ -0,0 +1,26 @@
+import unittest
+from docubot import build_kb
+from text_utils.text_utils import tiktoken_len
+from langchain.docstore.document import Document
+
+
+class TestDocuBot(unittest.TestCase):
+    """
+    A class for testing the functionality of the DocuBot application.
+    """
+
+    def test_build_kb(self):
+        chunks = build_kb("test_files")
+        # Test if the function returns a list
+        self.assertIsInstance(chunks, list)
+
+        # Test if a chunk is a document
+        self.assertIsInstance(chunks[0], Document)
+
+        # Test if the function returns a list of chunks with at most 512 tokens per chunk
+        for c in chunks:
+            self.assertLessEqual(tiktoken_len(c.page_content), 512)
+
+
+if __name__ == "__main__":
+    unittest.main()

tests/test_text_utils.py

@@ -6,7 +6,7 @@
 EMBEDDING_COST_PER_TOKEN = 0.0000001
 
 
-class TestDocumentLoaders(unittest.TestCase):
+class TestTextUtils(unittest.TestCase):
     def test_tiktoken_len(self):
         assert tiktoken_len("Hello, world!") == 4
         assert tiktoken_len("This is a sentence.") == 5