retriever.retrieve("query") or query_engine.query("query") ?from llama_index.core import QueryBundle query = QueryBundle(query_str="unused", embedding=[0.1, ...]) nodes = retriever.retrieve(query) response = query_engine.query(query)
as_query_engine method of the VectorStoreIndex class. This method returns a QueryEngine object that you can use to perform queries. Here's an example of how you can do this:from llama_index.core import VectorStoreIndex
# Assuming you have already created your VectorStoreIndex
index = VectorStoreIndex.from_vector_store(
vector_store,
storage_context=storage_context,
embed_model=self.embeddings
)
# Create a QueryEngine
query_engine = index.as_query_engine(similarity_top_k=3)
# Perform a query
response = query_engine.query("Your query here")"Your query here" with your actual query. The similarity_top_k=3 argument specifies that you want the top 3 results.query method expects a string query. If you have an embedding and want to perform a similarity search with it, you might need to use a different method or convert the embedding back to a string query.VectorStoreQuery class to perform a similarity search in your vector store. Here's an example:from llama_index.core import VectorStoreQuery # Assuming you have your query_embedding query_embedding = ... # Your query embedding here # Create a VectorStoreQuery vector_store_query = VectorStoreQuery(query_embedding=query_embedding, similarity_top_k=3) # Perform the query response = index.as_query_engine().query(vector_store_query)
... with your actual query embedding. The similarity_top_k=3 argument specifies that you want the top 3 results.query method of the QueryEngine class can accept either a string or a VectorStoreQuery object. If you provide a VectorStoreQuery object, it will perform a similarity search using the provided embedding.query method of the QueryEngine class with a VectorStoreQuery object. The error message suggests that the query method is expecting a string or a QueryBundle object, not a VectorStoreQuery object.VectorStoreQuery object with the query method of the QueryEngine class. The examples provided in the knowledge sources use a string as the query.VectorStoreQuery class is not compatible with the query method of the QueryEngine class, or that there's a specific way to use it that's not covered in the provided knowledge sources.VectorStoreQuery and QueryEngine classes for more information on how to use them. If the documentation doesn't help, you might want to reach out to the developers or the community for assistance.QueryEngine from your VectorStoreIndex and then using the query method of the QueryEngine.from llama_index.core import VectorStoreIndex
# Assuming you have already created your VectorStoreIndex
index = VectorStoreIndex.from_vector_store(
vector_store,
storage_context=storage_context,
embed_model=self.embeddings
)
# Create a QueryEngine
query_engine = index.as_query_engine()
# Perform a query
response = query_engine.query("Your query here")"Your query here" with your actual query.query method expects a string query. If you have an embedding and want to perform a similarity search with it, you might need to use a different method or convert the embedding back to a string query.TrafilaturaWebReader is not able to extract text from some of the URLs you're trying to parse. This is causing the Document creation to fail because it's receiving None as the text value, which is not a valid string.import uuid
from llama_index.readers.web import TrafilaturaWebReader
from llama_index.core import Document
def new_uuids(file_name: str):
"""Assign metadata based on the file name."""
return {"id_": uuid.uuid4().hex}
reader = TrafilaturaWebReader()
docs = []
for url in to_parse[:50]:
try:
loaded_docs = reader.load_data([url], show_progress=True)
if loaded_docs and loaded_docs[0].text: # Check if text was successfully extracted
loaded_docs[0].metadata.update(new_uuids(url))
docs.append(loaded_docs[0])
else:
print(f"Failed to extract text from {url}")
except Exception as e:
print(f"Error processing {url}: {str(e)}")
print(f"Successfully loaded {len(docs)} documents")docs list.TrafilaturaWebReader, you can refer to the LlamaIndex documentation on TrafilaturaWebReader ().