Spaces:
Sleeping
Sleeping
File size: 4,308 Bytes
94f31ec | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 | #!/usr/bin/env python3
"""One-time script to create the Pinecone index for the multi-agent RAG system.
This script avoids the need to install the Pinecone CLI by using the Python SDK
directly for this single administrative task.
The index is a plain dense-vector serverless index. Embeddings are generated
by the app using NVIDIA models (nvidia/nv-embed-v1 or similar).
Usage:
uv run python scripts/create_pinecone_index.py
Environment Variables:
PINECONE_API_KEY - Required: Your Pinecone API key
PINECONE_INDEX - Optional: Index name (default: multi-agent-index)
"""
import os
import sys
from pathlib import Path
# Add project root to path so app imports work
sys.path.insert(0, str(Path(__file__).parent.parent))
from dotenv import load_dotenv
load_dotenv()
def get_embedding_dimension() -> int:
"""Probe the configured embeddings model to determine output dimension."""
from app.core.llm_factory import get_embeddings_model
embeddings = get_embeddings_model()
vector = embeddings.embed_query("dimension probe")
return len(vector)
def main() -> int:
api_key = os.getenv("PINECONE_API_KEY")
if not api_key:
print("Error: PINECONE_API_KEY environment variable not set")
return 1
index_name = os.getenv("PINECONE_INDEX", "multi-agent-index")
print("Pinecone index setup")
print(f" Index name : {index_name}")
# Determine embedding dimension from the app's configured model
try:
dimension = get_embedding_dimension()
print(f" Dimension : {dimension} (from configured embedding model)")
except Exception as exc:
print(f"Error: Could not determine embedding dimension: {exc}")
print("You can also set it manually by editing this script.")
return 1
# Create index via SDK (one-time admin task)
try:
from pinecone import Pinecone
pc = Pinecone(api_key=api_key)
if pc.has_index(index_name):
print(f"\nIndex '{index_name}' already exists. Skipping creation.")
info = pc.describe_index(index_name)
if isinstance(info, dict):
print(f" Status : {info.get('status', 'unknown')}")
print(f" Dimension: {info.get('dimension', 'unknown')}")
print(f" Metric : {info.get('metric', 'unknown')}")
else:
print(f" Status : {getattr(info, 'status', 'unknown')}")
print(f" Dimension: {getattr(info, 'dimension', 'unknown')}")
print(f" Metric : {getattr(info, 'metric', 'unknown')}")
return 0
print(f"\nCreating serverless dense-vector index '{index_name}' ...")
print(f" Cloud : aws")
print(f" Region : us-east-1")
print(f" Metric : cosine")
print(f" Dim : {dimension}")
pc.create_index(
name=index_name,
dimension=dimension,
metric="cosine",
spec={
"serverless": {
"cloud": "aws",
"region": "us-east-1",
}
},
)
# Wait briefly and verify
import time
time.sleep(5)
if pc.has_index(index_name):
info = pc.describe_index(index_name)
if isinstance(info, dict):
print(f"\n Status : {info.get('status', 'unknown')}")
print(f" Dimension: {info.get('dimension', 'unknown')}")
print(f" Metric : {info.get('metric', 'unknown')}")
else:
print(f"\n Status : {getattr(info, 'status', 'unknown')}")
print(f" Dimension: {getattr(info, 'dimension', 'unknown')}")
print(f" Metric : {getattr(info, 'metric', 'unknown')}")
print(f"\nIndex '{index_name}' created successfully.")
print("You can now run: uv run python scripts/seed_rag_data.py")
return 0
else:
print("\nWarning: Index creation initiated but not yet visible.")
print("Check the Pinecone console for status.")
return 0
except Exception as exc:
print(f"\nError: Failed to create index: {exc}")
return 1
if __name__ == "__main__":
raise SystemExit(main())
|