This tool is based off trentms excellent node-bunyan utility. It's used for creating structured logs using the JSON format. The output is JSON which can be piped to either the 'bunyan' tool itself, or to another JSON parser or to a file.
To use bash-bunyan in your bash script, source the 'bunyan' script then use one of the level names to log directly. ex:
$ cat example.sh
#!/usr/bin/bash
. includes/bunyan
info "this is an info message"
$
~
$ sh example.sh | json
{
"name": "example.sh",
"hostname": "mac.local",
"pid": 49121,
"level": 30,
"msg": "this is an info message",
"time": "2012-03-24T02:48:21Z",
"v": 0
}
When you include bunyan you will automatically inherit functions which correspond to the log levels. These functions are
- trace (60): logging from external libraries
- debug (50): verbose debug information
- info (40): detail on regular information
- warn (30): something an operation should pay attention to
- error (20): fatal for a request / action
- fatal (10): the application exited because of some error
To change the loglevel set the '__bunyanLevel to the appropriate level you care about. Anything under that level will not be logged. By default, the level is set to 'info'.
bash-bunyan doesn't have nearly the granularity of node-bunyan, but you can set the name of the process reported in bunyan by setting the '__bunyanName' variable. This variable will be set automatically in your script when you source the bunyan include file. ex:
$ cat example2.sh
#!/usr/bin/bash
. includes/bunyan
__bunyanName='super'
info 'hello world'
~
$ sh example2.sh | bunyan
[2012-03-24T02:47:15Z] INFO: super/49105 on mac.local: hello world