Installation

npm install --save botbuilder-calling-logging

Peer dependencies

npm install --save documentdb azure-storage botbuilder-calling

Usage

TypeScript

import { BotCallLogger } from 'botbuilder-calling-logging';
import { BlobService } from 'azure-storage';
import { DocumentClient } from 'documentdb';
import { UniversalBot } from 'botbuilder-calling';

// your chat bot
const callbot = new UniversalBot(/* params */);

// create logger
const logger = new BotCallLogger(new DocumentClient(/* params */), {
  documents: {
    databaseName: 'logs1', 
    collectionName: 'bot1',
  },
  blobs: { // wav files are stored here
    blobService: new BlobService(/* params */),
    options: { containerName: 'botcalls', }
  },
});

// logs are queued internally to prevent blocking the middleware pipeline
//   so logging errors are not visible to the bot
//   instead, catch logging IO errors in this event handler
logger.events.on('error', console.error);

// attach logger to the chatbot
callbot.use(logger);

Error handling

Any errors encountered by the DocumentClient or BlobService are not returned through the bot middleware service. In order to capture these errors, listen to the logger's error event.

logger.events.on('error', console.error);

Advanced

DocumentDB Partitioning

For large-scale DocumentDB collections (RU > 10K), the recommended partitionKey is /address/conversation/id

Log format

Logs are stored in DocumentDB as JSON documents. Any binary attachments (images, speech) are stored as attachments in DocumentDb, as well as on Azure Blob Storage, if it is configured.

Extending the data store

To persist logs or blobs in arbitrary stores, implement your own DocumentWriter and BlobWriter classes (see the botbuilder-logging package). Then write your own bot middleware that calls BotLogWriter.enqueue. See class BotCallLogger for sample implementation.