Spaces:
Sleeping
Sleeping
| # Ingest script | |
| # File: scripts/ingest.py | |
| import yaml | |
| import sys | |
| from orchestrator.client import MCPClient | |
| def main(): | |
| """ | |
| Ingest papers for a given query into Chroma vector store via MCP. | |
| Usage: python ingest.py "your search query" | |
| """ | |
| if len(sys.argv) < 2: | |
| print("Usage: python ingest.py <query>") | |
| sys.exit(1) | |
| query = sys.argv[1] | |
| cfg = yaml.safe_load(open("config.yaml")) | |
| web = MCPClient(cfg['mcp_servers']['web_search']) | |
| pubmed = MCPClient(cfg['mcp_servers']['pubmed']) | |
| chroma = MCPClient(cfg['mcp_servers']['chroma']) | |
| print(f"Ingesting papers for query: {query}") | |
| results = [] | |
| try: | |
| results += web.call("web_search.search", {"q": query}) | |
| except Exception as e: | |
| print("Web search failed:", e) | |
| try: | |
| results += pubmed.call("metatool.query", {"source": "PubMed", "q": query}) | |
| except Exception as e: | |
| print("PubMed search failed:", e) | |
| for paper in results: | |
| paper_id = paper.get('id') | |
| text = paper.get('abstract', '') | |
| meta = {"title": paper.get('title'), "authors": ",".join(paper.get('authors', []))} | |
| chroma.call("chroma.insert", {"id": paper_id, "text": text, "metadata": meta}) | |
| print("Ingestion complete!") | |
| if __name__ == "__main__": | |
| main() |