.readme-partials.yml

introduction: |-
  This module provides an easy to use, higher-level layer for working with [Cloud Logging](https://cloud.google.com/logging/docs),
  compatible with [Bunyan](https://www.npmjs.com/package/bunyan). Simply attach this as a transport to your existing Bunyan loggers.
body: |-
  ### Using as an express middleware

  ***NOTE: this feature is experimental. The API may change in a backwards
  incompatible way until this is deemed stable. Please provide us feedback so
  that we can better refine this express integration.***

  We provide a middleware that can be used in an express application. Apart from
  being easy to use, this enables some more powerful features of Cloud
  Logging: request bundling. Any application logs emitted on behalf of a specific
  request will be shown nested inside the request log as you see in this
  screenshot:

  ![Request Bundling Example](https://raw.githubusercontent.com/googleapis/nodejs-logging-bunyan/master/doc/images/request-bundling.png)

  The middleware adds a `bunyan`-style log function to the `request` object. You
  can use this wherever you have access to the `request` object (`req` in the
  sample below). All log entries that are made on behalf of a specific request are
  shown bundled together in the Cloud Logging UI.

  ```javascript
  const lb = require('@google-cloud/logging-bunyan');

  // Import express module and create an http server.
  const express = require('express');

  async function startServer() {
    const {logger, mw} = await lb.express.middleware();
    const app = express();

    // Install the logging middleware. This ensures that a Bunyan-style `log`
    // function is available on the `request` object. Attach this as one of the
    // earliest middleware to make sure that log function is available in all the
    // subsequent middleware and routes.
    app.use(mw);

    // Setup an http route and a route handler.
    app.get('/', (req, res) => {
      // `req.log` can be used as a bunyan style log method. All logs generated
      // using `req.log` use the current request context. That is, all logs
      // corresponding to a specific request will be bundled in the Cloud UI.
      req.log.info('this is an info log message');
      res.send('hello world');
    });

    // `logger` can be used as a global logger, one not correlated to any specific
    // request.
    logger.info({port: 8080}, 'bonjour');

    // Start listening on the http server.
    app.listen(8080, () => {
      console.log('http server listening on port 8080');
    });
  }

  startServer();
  ```

  ### Error Reporting

  Any `Error` objects you log at severity `error` or higher can automatically be picked up by [Cloud Error Reporting](https://cloud.google.com/error-reporting/) if you have specified a `serviceContext.service` when instantiating a `LoggingBunyan` instance:

  ```javascript
  const loggingBunyan = new LoggingBunyan({
    serviceContext: {
      service: 'my-service', // required to report logged errors
                             // to the Google Cloud Error Reporting
                             // console
      version: 'my-version'
    }
  });
  ```

  It is an error to specify a `serviceContext` but not specify `serviceContext.service`.

  Make sure to add logs to your [uncaught exception](https://nodejs.org/api/process.html#process_event_uncaughtexception) and [unhandled rejection](https://nodejs.org/api/process.html#process_event_unhandledrejection) handlers if you want to see those errors too.

  You may also want to see the [@google-cloud/error-reporting][@google-cloud/error-reporting] module which provides direct access to the Error Reporting API.

  ### Special Payload Fields in LogEntry

  There are some fields that are considered special by Google cloud logging and will be extracted into the LogEntry structure. For example, `severity`, `message` and `labels` can be extracted to LogEntry if included in the bunyan log payload. These [special JSON fields](https://cloud.google.com/logging/docs/structured-logging#special-payload-fields) will be used to set the corresponding fields in the `LogEntry`. Please be aware of these special fields to avoid unexpected logging behavior. 
  
  ### LogEntry Labels

  If the bunyan log record contains a label property where all the values are strings, we automatically promote that
  property to be the [`LogEntry.labels`](https://cloud.google.com/logging/docs/reference/v2/rpc/google.logging.v2#logentry) value rather
  than being one of the properties in the `payload` fields. This makes it easier to filter the logs in the UI using the labels.

  ```javascript
  logger.info({labels: {someKey: 'some value'}}, 'test log message');
  ```

  All the label values must be strings for this automatic promotion to work. Otherwise the labels are left in the payload.

  ### Formatting Request Logs

  To format your request logs you can provide a `httpRequest` property on the bunyan metadata you provide along with the log message. We will treat this as the [`HttpRequest`](https://cloud.google.com/logging/docs/reference/v2/rpc/google.logging.type#google.logging.type.HttpRequest) message and Cloud logging will show this as a request log. Example:

  ![Request Log Example](https://raw.githubusercontent.com/googleapis/nodejs-logging-bunyan/master/doc/images/request-log.png)

  ```js
  logger.info({
    httpRequest: {
      status: res.statusCode,
      requestUrl: req.url,
      requestMethod: req.method,
      remoteIp: req.connection.remoteAddress,
      // etc.
    }
  }, req.path);
  ```

  The `httpRequest` property must be a properly formatted [`HttpRequest`](https://cloud.google.com/logging/docs/reference/v2/rpc/google.logging.type#google.logging.type.HttpRequest) message. (Note: the linked protobuf documentation shows `snake_case` property names, but in JavaScript one needs to provide property names in `camelCase`.)

  ### Correlating Logs with Traces

  If you use [@google-cloud/trace-agent][trace-agent] module, then this module will set the Cloud Logging [LogEntry][LogEntry] `trace` property based on the current trace context when available. That correlation allows you to [view log entries][trace-viewing-log-entries] inline with trace spans in the Cloud Trace Viewer. Example:

  ![Logs in Trace Example](https://raw.githubusercontent.com/googleapis/nodejs-logging-bunyan/master/doc/images/bunyan-logs-in-trace.png)

  If you wish to set the Cloud LogEntry `trace` property with a custom value, then write a Bunyan log entry property for `'logging.googleapis.com/trace'`, which is exported by this module as `LOGGING_TRACE_KEY`. For example:

  ```js
  const bunyan = require('bunyan');
  // Node 6+
  const {LoggingBunyan, LOGGING_TRACE_KEY} = require('@google-cloud/logging-bunyan');
  const loggingBunyan = LoggingBunyan();

  ...

  logger.info({
    [LOGGING_TRACE_KEY]: 'custom-trace-value'
  }, 'Bunyan log entry with custom trace field');
  ```

  ### Error handling with a default callback
  The `LoggingBunyan` class creates an instance of `Logging` which creates the `Log` class from `@google-cloud/logging` package to write log entries. 
  The `Log` class writes logs asynchronously and there are cases when log entries cannot be written when it fails or an error is returned from Logging backend.
  If the error is not handled, it could crash the application. One possible way to handle the error is to provide a default callback
  to the `LoggingBunyan` constructor which will be used to initialize the `Log` object with that callback like in the example below:

  ```js
  // Imports the Google Cloud client library for Bunyan
  const {LoggingBunyan} = require('@google-cloud/logging-bunyan');
  // Creates a client
  const loggingBunyan = new LoggingBunyan({
    projectId: 'your-project-id',
    keyFilename: '/path/to/key.json',
    defaultCallback: err => {
        if (err) {
        console.log('Error occured: ' + err);
        }
    },
  });
  ```

  ### Alternative way to ingest logs in Google Cloud managed environments
  If you use this library with the Cloud Logging Agent, you can configure the handler to output logs to `process.stdout` using
  the [structured logging Json format](https://cloud.google.com/logging/docs/structured-logging#special-payload-fields).
  To do this, add `redirectToStdout: true` parameter to the `LoggingBunyan` constructor as in sample below.
  You can use this parameter when running applications in Google Cloud managed environments such as AppEngine, Cloud Run,
  Cloud Function or GKE. The logger agent installed on these environments can capture `process.stdout` and ingest it into Cloud Logging.
  The agent can parse structured logs printed to `process.stdout` and capture additional log metadata beside the log payload.
  It is recommended to set `redirectToStdout: true` in serverless environments like Cloud Functions since it could 
  decrease logging record loss upon execution termination - since all logs are written to `process.stdout` those
  would be picked up by the Cloud Logging Agent running in Google Cloud managed environment. 
  Note that there is also a `useMessageField` option which controls if "message" field is used to store 
  structured, non-text data inside `jsonPayload` field when `redirectToStdout` is set. By default `useMessageField` is always `true`.
  Set the `skipParentEntryForCloudRun` option to skip creating an entry for the request itself as Cloud Run already automatically creates
  such log entries. This might become the default behaviour in a next major version.

  ```js
  // Imports the Google Cloud client library for Bunyan
  const {LoggingBunyan} = require('@google-cloud/logging-bunyan');

  // Creates a client
  const loggingBunyan = new LoggingBunyan({
    projectId: 'your-project-id',
    keyFilename: '/path/to/key.json',
    redirectToStdout: true,
  });
  ```