ingest.py

import os
from supabase.client import create_client
from langchain_community.vectorstores import SupabaseVectorStore
from langchain_huggingface import HuggingFaceEmbeddings

# 1. Connect to Supabase
supabase = create_client(
    os.environ["SUPABASE_URL"],
    os.environ.get("SUPABASE_SERVICE_KEY") or os.environ["SUPABASE_ANON_KEY"],
)

# 2. Create embeddings object
embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-mpnet-base-v2")

# 3. Add texts and metadata to load
texts = [
    "Hello world, this is my first document.",
    "Second doc on another topic.",
    "More example content here.",
]
metadatas = [
    {"source": "doc1"},
    {"source": "doc2"},
    {"source": "doc3"},
]

# 4. Ingest texts into Supabase
vector_store = SupabaseVectorStore.from_texts(
    texts=texts,
    embedding=embeddings,
    metadatas=metadatas,
    client=supabase,
    table_name="documents",
    query_name="match_documents_langchain",
)

# 5. OPTIONAL: Test retrieval
results = vector_store.similarity_search("Hello world", k=1)
print("Retrieved:", results)