37 lines
1.5 KiB
Python
37 lines
1.5 KiB
Python
from qdrant_client import QdrantClient
|
|
from qdrant_client.http import models
|
|
from langchain_ollama import OllamaEmbeddings
|
|
|
|
class QdrantStore:
|
|
def __init__(self, url="http://localhost:6333", collection_name="rag_collection"):
|
|
self.client = QdrantClient(url=url)
|
|
self.collection_name = collection_name
|
|
self._ensure_collection()
|
|
|
|
def _ensure_collection(self):
|
|
if self.collection_name not in self.client.get_collections().collections:
|
|
self.client.recreate_collection(
|
|
collection_name=self.collection_name,
|
|
vectors_config=models.VectorParams(size=384, distance=models.Distance.COSINE),
|
|
)
|
|
|
|
def add_documents(self, documents, titles):
|
|
embeddings = OllamaEmbeddings(model="nomic-embed-text")
|
|
vectors = embeddings.embed_documents(documents)
|
|
payload = [{"title": t} for t in titles]
|
|
self.client.upsert(
|
|
collection_name=self.collection_name,
|
|
points=models.Batch(points=[models.PointStruct(id=i, vector=v, payload=p) for i, (v, p) in enumerate(zip(vectors, payload))])
|
|
)
|
|
|
|
def search(self, query, limit=5):
|
|
embeddings = OllamaEmbeddings(model="nomic-embed-text")
|
|
query_vector = embeddings.embed_query(query)
|
|
results = self.client.search(
|
|
collection_name=self.collection_name,
|
|
query_vector=query_vector,
|
|
limit=limit,
|
|
with_payload=True,
|
|
)
|
|
return [r.payload for r in results]
|