Update utils.py

utils.py

@@ -2,6 +2,7 @@
 # -- Libraries
 from   typing                      import Any, Dict, List, Mapping, Optional
 from   pydantic                    import Extra, Field, root_validator
+from   langchain_community.vectorstores import FAISS
 from   langchain_core.runnables    import RunnablePassthrough
 from   langchain.llms.base         import LLM
 from   langchain.chat_models       import ChatOpenAI
@@ -199,29 +200,32 @@ def get_gpt_response(transcription_path, query, logger):
         | StrOutputParser()
     )
     llm_output = rag_chain.invoke(query)
-    # dataset    = client.create_dataset(dataset_name="Sample LLM dataset", description="A dataset with LLM inputs and outputs", data_type="llm")
-
-    # client.create_example(
-    #   inputs={"input": query},
-    #   outputs={"output": llm_output},
-    #   dataset_id=dataset.id,
-    # )
-
-    # -- Run custom evaluator
-    # evaluation_config = RunEvalConfig(
-    #     custom_evaluators = [RelevanceEvaluator()],
-    # )
-    # eval_output = run_on_dataset(
-    #    dataset_name="Sample LLM dataset",
-    #    llm_or_chain_factory=rag_chain,
-    #    evaluation=evaluation_config,
-    #    client=client,
-    # )
-    # logger.info("Eval output!!!!")
-    # logger.info(eval_output)
-    
     return llm_output
 
+def get_character_info_gpt(text, character):
+    vectorstore = FAISS.from_texts(
+        [text], embedding=OpenAIEmbeddings()
+    )
+    retriever = vectorstore.as_retriever()
+    
+    template = """Responde a la siguiente pregunta basandote unicamente en el siguiente contexto:
+    {context}
+    
+    Pregunta: {question}
+    """
+    prompt = ChatPromptTemplate.from_template(template)
+    
+    model = ChatOpenAI()
+
+    chain = (
+        {"context": retriever, "question": RunnablePassthrough()}
+        | prompt
+        | model
+        | StrOutputParser()
+    )
+    return chain.invoke("¿Quien es {}?".format(character))
+
+    
 # -- Text summarisation with OpenAI (map-reduce technique)
 def summarise_doc(transcription_path, model_name, model=None):
     if model_name == 'gpt':