RSpace MCP Server

Instructions

Implementation Reference

• main.py:172-272 (handler)

  The primary handler function for the 'search_documents' MCP tool. Decorated with @mcp.tool for automatic registration. Implements both simple and advanced search capabilities using the RSpace ELN client library, including optional full content retrieval for matching documents.
  @mcp.tool(tags={"rspace", "search"}) def search_documents( query: str, search_type: Literal["simple", "advanced"] = "simple", query_types: List[Literal["global", "fullText", "tag", "name", "created", "lastModified", "form", "attachment"]] = None, operator: Literal["and", "or"] = "and", order_by: str = "lastModified desc", page_number: int = 0, page_size: int = 20, include_content: bool = False ) -> dict: """ Generic search tool for RSpace documents with flexible search options Usage: Search across all your RSpace documents using various criteria Parameters: - query: The search term(s) to look for - search_type: "simple" for basic search, "advanced" for multi-criteria search - query_types: List of search types to use (for advanced search): - "global": Search across all document content and metadata - "fullText": Search within document text content - "tag": Search by document tags - "name": Search by document names/titles - "created": Search by creation date (use ISO format like "2024-01-01") - "lastModified": Search by modification date - "form": Search by form type - "attachment": Search by attachments - operator: "and" (all criteria must match) or "or" (any criteria can match) - order_by: Sort results by field (e.g., "lastModified desc", "name asc") - page_number: Page number for pagination (0-based) - page_size: Number of results per page (max 200) - include_content: Whether to fetch full document content (slower but more complete) Returns: Dictionary with search results and metadata Examples: - Simple text search: search_documents("PCR protocol") - Search by tags: search_documents("experiment", search_type="advanced", query_types=["tag"]) - Multi-criteria search: search_documents("DNA", search_type="advanced", query_types=["fullText", "tag"], operator="or") """ if page_size > 200: raise ValueError("page_size must be 200 or less") if search_type == "simple": # Use simple search - works like RSpace's "All" search results = eln_cli.get_documents( query=query, order_by=order_by, page_number=page_number, page_size=page_size ) else: # Use advanced search with AdvancedQueryBuilder if query_types is None: query_types = ["global"] # Default to global search builder = AdvancedQueryBuilder(operator=operator) # Add search terms for each specified query type for query_type in query_types: if query_type == "global": builder.add_term(query, AdvancedQueryBuilder.QueryType.GLOBAL) elif query_type == "fullText": builder.add_term(query, AdvancedQueryBuilder.QueryType.FULL_TEXT) elif query_type == "tag": builder.add_term(query, AdvancedQueryBuilder.QueryType.TAG) elif query_type == "name": builder.add_term(query, AdvancedQueryBuilder.QueryType.NAME) elif query_type == "created": builder.add_term(query, AdvancedQueryBuilder.QueryType.CREATED) elif query_type == "lastModified": builder.add_term(query, AdvancedQueryBuilder.QueryType.LAST_MODIFIED) elif query_type == "form": builder.add_term(query, AdvancedQueryBuilder.QueryType.FORM) elif query_type == "attachment": builder.add_term(query, AdvancedQueryBuilder.QueryType.ATTACHMENT) advanced_query = builder.get_advanced_query() results = eln_cli.get_documents_advanced_query( advanced_query=advanced_query, order_by=order_by, page_number=page_number, page_size=page_size ) # Optionally fetch full content for each document if include_content and 'documents' in results: for doc in results['documents']: try: full_doc = eln_cli.get_document(doc['globalId']) # Add concatenated content to the document content = '' for field in full_doc.get('fields', []): content += field.get('content', '') doc['fullContent'] = content except Exception as e: doc['fullContent'] = f"Error fetching content: {str(e)}" return results
• main.py:172-172 (registration)

  The @mcp.tool decorator registers the search_documents function as an MCP tool with tags for categorization.
  @mcp.tool(tags={"rspace", "search"})