Opinionated bare simple logger for NodeJS (with TypeScript typings).
BSLogger has been created after being disapointed not finding a matching logger on the internet. Not that others aren't good, they just did not fit what I was looking for.
Here is what I was looking for (and tried to implemented in BSLogger):
- light memory usage
- easily extendable (see
child
method) - as few dependencies as possible
- ability to define all targets in a string (so that
ENV
vars can be used) - when using file targets, not re-opening them
- reasonable defautls:
- logs warnings and above to
stderr
- logs JSON to files
- logs warnings and above to
- no overhead if it's not going to log anywhere
Install:
npm install --save bs-logger
# or
yarn add bs-logger
Use:
const { logger } = require('bs-logger');
// or
// import logger from 'bs-logger';
// or
// import { logger } from 'bs-logger';
// as default exports the logger
logger('foo');
logger.debug('bar');
logger.warn({foo: 'bar'}, 'dummy', 'other'/*, ...*/);
More complex example:
// env MY_LOG_TARGETS="debug.log:trace,stderr:warn%json"
import { createLogger } from 'bs-logger';
const logger = createLogger({
context: {namespace: 'http'},
targets: process.env.MY_LOG_TARGETS,
translate: (m) => {
if (process.env.NODE_ENV === 'production') {
m.context = { ...m.context, secret: null };
}
return m;
},
});
// [...]
logger.debug({secret: 'xyz'}, 'trying to login')
// will log into debug.log `trying to login` with secret in the context except in prod
const login = logger.wrap(function login() {
// your login code
})
// [...]
login();
// will log `calling login` with the arguments in context
BSLogger exports a global logger lazyly created on first use, but it is advised to create your own using the createLogger()
helper:
-
If you are using it in a library wich is meant to be re-distributed:
import { createLogger, LogContexts } 'bs-logger'; const logger = createLogger({ [LogContexts.package]: 'my-pacakge' });
-
If you are using it in an application of your own:
import { createLogger, LogContexts } 'bs-logger'; const logger = createLogger({ [LogContexts.application]: 'my-app' });
Child loggers extends the context, targets and message translators from their parent. You create a child logger using the child
method:
const childLogger = logger.child({ [LogContexts.namespace]: 'http' })
// childLogger becomes a new logger
Any helper to log within BSLogger is a function which has the same signature as console.log()
, and also accepts an optional first argument being the context. A context is any object
, with some specific (but optional) properties which we'll see later.
logMethod(message: string, ...args: any[]): void
// or
logMethod(context: LogContext, message: string, ...args: any[]): void
You can log using any logger as a function directly (if the logger or its possible parent(s) has not been created with any log level in its context, no level will be attached):
import { createLogger } from 'bs-logger'
const logger = createLogger()
// [...]
logger('my message');
BSLogger is aware of 6 log levels (trace
, debug
, info
, warn
, error
and fatal
) but you can create your owns. A log level is basically a number. The higher it is, the more important will be the message. You can find log levels constants in LogLevels
export:
import { LogLevels } from 'bs-logger';
const traceLevelValue = LogLevels.trace;
const debugLevelValue = LogLevels.debug;
// etc.
For each log level listed above, a logger will have a helper method to directly log using this level:
import { createLogger } from 'bs-logger'
const logger = createLogger()
// [...]
logger.trace('foo')
logger.debug('bar')
// etc.
Those helpers are the equivalent to
logger({ [LogContexts.logLevel]: level }, 'foo')
...except that they'll be replaced with an empty function on the first call if their level will not be handled by any target.
Each logger has a wrap
method which you can use to wrap a function. If there is no matching log target, the wrap
method will simply return your function, else it'll wrap it in another function of same signature. The wrapper will, before calling your function, log a message with received arguments in the context.
// With `F` being the type of your funciton:
logger.wrap(func: F): F
// or
logger.wrap(message: string, func: F): F
// or
logger.wrap(context: LogContext, messages: string, func: F): F
Each root logger (created using createLogger
helper) is attached to 0 or more "target". A target is responsible of writing a log entry somewhere. It is an object with the following properties:
- minLevel
string
: The minimum log level this target's strem writer will be called for - stream
{ write: (str: string) => void }
: An object with a write function (like node'sstream.Writable
) which will be used to write log entries - format
(msg: LogMessage) => string
: A formatter which will be used to transform a log entry (message object) into a string
When using the global logger, or if no targets
specified when creating a logger, calling log methods will output to STDERR anything which has log level higher or equal to warn
. This can be modified as follow by defineing the LOG_TARGETS
environment variable or passing the targets
option to createLogger
. The targets
can be an array of LogTarget
(see above) or a string
defining a list of one or more targets separated by comma (,
). A string
target is composed as follow:
- The file path, absolute or relative to CWD. It can also be the specials
stdout
orstderr
strings (case insensitive). When giving a path to a file, if it ends with the plus sign ( - An optional minimum log level after a colon (
:
). It should be anumber
or the log level name (ietrace
,error
, ...). - An optional formatter name after a percent sign (
%
). There are 2 included formatter:json
(used for files by default) andsimple
(used forstdout
andstderr
by default). See below to define your own.
Examples:
debug.log%simple,stdout:fatal
- Log everything to
debug.log
file in CWD dir (re-creates the file for each run). Uses thesimple
formatter. - Log only messages with level >=
fatal
to the standard out.
- Log everything to
errors.log :error,debug.log:15
- Log only messages with level >=
error
toerrors.log
file (without re-creating the file at each run). - Log only messages with level >= 15 to
debug.log
file (re-creates the file for each run).
- Log only messages with level >=
A custom formatter is a function that takes a LogMessage
object and returns a string
. It can be registered giving it a name using the registerLogFormatter
helper:
import { registerLogFormatter, createLogger } from 'bs-logger';
registerLogFormatter('foo', m => `${m.sequence} ${new Date(m.tim).toLocaleString()} ${m.message}`);
const logger = createLogger({
targets: 'stdout%foo', // specifying out formatter
});
The whole testing
namespace has useful helpers for using BSLogger while unit testing your product.
In your tests you would usually prefer not having any logging to happen, or you would like to check what has been logged but without actually logging it to any target.
The testing
namespace holds all testing utilities:
import { testing } from 'bs-logger'
- If you use the root logger, here is how to disable its output:
testing.setup()
and the logger
(or default
) export will become a LoggerMock
instance (see below).
- If you create logger(s) using
createLogger
, when testing use thetesting.createLoggerMock
instead. It accepts the same first argument, with an extra second argument, optional, being theLogTargetMock
to be used (see below).
Loggers created using the testing
namespace will have one and only one log target being a LogTargetMock
, and that target will be set on the target
extra property of the logger.
Here are the extra properties of LogTargetMock
which you can then use for testing:
- messages
LogMessage[]
: all log message objects which would have normally be logged- last
LogMessage
: the last one being logged - trace
LogMessage[]
: all log message objects withtrace
level- last
LogMessage
: last one withtrace
level
- last
- debug
LogMessage[]
: all log message objects withdebug
level- last
LogMessage
: last one withdebug
level
- last
- ...
- last
- lines
string[]
: all formatted log message lines which would have normally be logged- last
string
: the last one being logged - trace
string[]
: all formatted log message lines withtrace
level- last
string
: last one withtrace
level
- last
- debug
string[]
: all formatted log message lines withdebug
level- last
string
: last one withdebug
level
- last
- ...
- last
- clear
() => void
: method to clear all log message objects and formatted lines - filteredMessages
(level: number | null, untilLevel?: number) => LogMessage[]
: method to filter log message objects - filteredLins
(level: number | null, untilLevel?: number) => string[]
: method to filter formatted log message lines
Let's say you have a logger.js
file in which you create the logger for your app:
// file: logger.js
import { testing, createLogger, LogContexts } from 'bs-logger';
const factory = process.env.TEST ? testing.createLoggerMock : createLogger;
export default factory({ [LogContexts.application]: 'foo' });
In a test you could:
import logger from './logger';
// in `fetch(url)` you'd use the logger like `logger.debug({url}, 'GET')` when the request is actually made
import fetch from './http';
test('it should cache request', () => {
logger.target.clear();
fetch('http://foo.bar/dummy.json');
expect(logger.target.messages.length).toBe(1);
fetch('http://foo.bar/dummy.json');
expect(logger.target.messages.length).toBe(1);
// you can also expect on the message:
expect(logger.target.messages.last.message).toBe('GET')
expect(logger.target.messages.last.context.url).toBe('http://foo.bar/dummy.json')
// or (mock target formater prefix the message with `[level:xxx] ` when there is a level)
expect(logger.target.lines.last).toBe('[level:20] GET')
// or filtering with level:
expect(logger.target.lines.debug.last).toBe('[level:20] GET')
});
Add to your project with npm
:
npm install --save bs-logger
or with yarn
:
yarn add bs-logger
You need to get a copy of the repository to run the tests:
git clone https://github.com/huafu/bs-logger.git
cd bs-logger
npm run test
Pull requests welcome!
We use SemVer for versioning. For the versions available, see the tags on this repository.
- Huafu Gandon - Initial work - huafu
See also the list of contributors who participated in this project.
This project is licensed under the MIT License - see the LICENSE file for details
Hey dude! Help me out for a couple of 🍻!