Get or initialize the appropriate vector database client for a specific endpoint. Uses a cache to avoid creating duplicate client ins

Print section header

Send a message with appropriate metadata and routing.

Get current cache size

Upload documents to the database using the configured write endpoint. Args: documents: List of document objects to upload en

Extract product names from store data.

Get the Azure AI Search client for a specific index Args: index_name: Name of the index (defaults to the configured inde

Add standard metadata fields to a message if not already present. Args: message: The message dictionary to add fields to

Close the MCP client. No cleanup needed for now.

Delete all documents matching the specified site. Args: site: Site identifier **kwargs: Additional parameter

Disconnect WebSocket.

Disconnect WebSocket.

* Escapes HTML special characters in a string * * @param {string} str - The string to escape * @returns {string} - The escaped string

Convert string to dict if needed.

Print narration with pause

Process either a sitemap URL or a domain. If given a domain, checks robots.txt for sitemaps.

Sanitize value for safe logging by removing newlines and control characters.

Send a message through WebSocket.

Consolidate all fast track abort conditions into a single method. Returns True if fast track should be aborted, False otherwise.

Extract simplified site information from a /who response item. Args: item: A single item from the /who response Returns:

Save current status to file.

Cleanup resources

Close the Elasticsearch client connections

Check if a collection exists in Qdrant. Args: collection_name: Name of the collection to check Returns:

Create a collection in Qdrant if it doesn't exist. Args: collection_name: Name of the collection to create v

Delete a specific conversation. Args: conversation_id: ID of the conversation to delete user_id: Optional us

Main entry point following NLWeb module pattern.

Extract all links from HTML

* Formats an ISO 8601 duration string to human-readable format * * @param {string} duration - ISO 8601 duration string (e.g., PT1H30M)

Get message delivery statistics.

Get or create the singleton logging configuration

Get or create the storage client instance. Returns: StorageProvider: The storage provider instance

Load test cases from JSON file. Args: file_path: Path to JSON file Returns: List of test case dicti

Normalize rubric features to zero mean, unit variance per feature.

Extract rubric features grouped by category from examples.

Asynchronously query the cricket statistics API and return formatted results. Args: query: Natural language query for cricket statis

Record current memory usage.

* Registers a specific renderer with a JSON renderer * * @param {Function} RendererClass - The renderer class * @param {JsonRenderer} jsonRe

Run multiple test cases and return summary. Args: test_cases: List of test case dictionaries Returns:

Load JSON file, returning None if file is being written or corrupted.

Fetch content from a URL and save to a temporary file. Args: url: URL to fetch Returns: Tuple of (file_path, file_type)

Search the Azure AI Search index for records filtered by site and ranked by vector similarity Args: query: The search qu

Returns True if exceptions should be raised instead of caught (for testing and development).

Print text with typewriter effect

Check if the current request uses v0.55 protocol.

Resolves a path, considering the base output directory if set. If path is absolute, returns it unchanged. If path is relative

Perform search for a single site. Args: query: Search query string site: Site to search within (or "all" for

Close the Elasticsearch client connection

Crawl multiple URLs one at a time

* Creates default HTML for a JSON item * * @param {Object} item - The item data * @returns {HTMLElement} - The HTML element

Create a conversation for load testing.

Delete all entries for a specific site from the configured database. Args: site: Site identifier to delete database: Specifi

Detect the type of a file based on its extension and content. Also determine if it already contains embeddings. Args: file_path:

Forward a request to the NLWeb MCP endpoint

* Get all messages (for building conversation list)

Get the canonical (preferred) mapping for an ontology type.

Fetch webpage content

Look up a template by ID.

Count how many of model's top-k have ref_score >= threshold.

Returns True if the system is running in development mode.

Load embedding model configuration.

* Creates a span element with the given content * * @param {string} content - The content for the span * @returns {HTMLElement} - The span e

Map extracted NL slot values to ontology filter specs. Returns dict of slot_name → list of filter specs.

Match a natural language question to the best template. Returns dict with: template_id, score, slots (extracted values).

Merge two JSON objects according to specific rules. Args: obj1: First JSON object (can be string, dict, or list of dicts) ob

Compute metrics per group (site, difficulty, etc.).

* Adds an explanation to an item * * @param {Object} item - The item data * @param {HTMLElement} contentDiv - The content div * @param {b

Record a sent message.

Reverse-map an ontology enum value to source system's native value(s). Returns: - A single source value (string) if exactly one s

* Sanitize DOM element to remove potentially harmful content * @param {Element} element - The DOM element to sanitize

Search the Milvus collection for records filtered by site and ranked by vector similarity. Args: query: The search query

Search for documents matching the query and site using Bing Web Search. Supports using rewritten queries from handler for better resu

Search conversations using a query string. Args: query: The search query string user_id: Optional user ID to

Send end/complete message at the end of query processing. Args: error: If True, indicates the query ended with an error

Send a message in the conversation.

Set the application mode (development, production, or testing).