📑 Lightweight, extensible, configurable logging library written in Golang.
Output in console:
- Simple, directly available without configuration
- Support common log level processing.
- eg:
trace
debug
info
notice
warn
error
fatal
panic
- eg:
- Support any extension of
Handler
Formatter
as needed - Supports adding multiple
Handler
log processing at the same time, outputting logs to different places - Support to custom log message
Formatter
- Built-in
json
text
two log record formattingFormatter
- Built-in
- Support to custom build log messages
Handler
- The built-in
handler.Config
handler.Builder
can easily and quickly build the desired log handler
- The built-in
- Has built-in common log write handler program
console
output logs to the console, supports color outputwriter
output logs to the specifiedio.Writer
file
output log to the specified file, optionally enablebuffer
to buffer writessimple
output log to the specified file, write directly to the file without bufferingrotate_file
outputs logs to the specified file, and supports splitting files by time and size, andbuffer
buffered writing is enabled by default- See ./handler folder for more built-in implementations
- Benchmark performance test please see Benchmarks
- Support enabling
buffer
for log writing - Support splitting log files by
time
andsize
- Support configuration to compress log files via
gzip
- Support clean old log files by
BackupNum
BackupTime
- The
rotatefile
subpackage is a stand-alone tool library with file splitting, cleaning, and compressing backups rotatefile.Writer
can also be directly wrapped and used in other logging libraries. For example:log
,glog
,zap
, etc.rotatefile.FilesClear
is an independent file cleaning backup tool, which can be used in other places (such as other program log cleaning such as PHP)- For more usage, please see rotatefile
中文说明请阅读 README.zh-CN
go get github.com/gookit/slog
slog
is very simple to use and can be used without any configuration
package main
import (
"github.com/gookit/slog"
)
func main() {
slog.Info("info log message")
slog.Warn("warning log message")
slog.Infof("info log %s", "message")
slog.Debugf("debug %s", "message")
}
Output:
[2020/07/16 12:19:33] [application] [INFO] [main.go:7] info log message
[2020/07/16 12:19:33] [application] [WARNING] [main.go:8] warning log message
[2020/07/16 12:19:33] [application] [INFO] [main.go:9] info log message
[2020/07/16 12:19:33] [application] [DEBUG] [main.go:10] debug message
You can enable color on output logs to console. This is default
package main
import (
"github.com/gookit/slog"
)
func main() {
slog.Configure(func(logger *slog.SugaredLogger) {
f := logger.Formatter.(*slog.TextFormatter)
f.EnableColor = true
})
slog.Trace("this is a simple log message")
slog.Debug("this is a simple log message")
slog.Info("this is a simple log message")
slog.Notice("this is a simple log message")
slog.Warn("this is a simple log message")
slog.Error("this is a simple log message")
slog.Fatal("this is a simple log message")
}
Output:
Above is the Formatter
setting that changed the default logger.
You can also create your own logger and append
ConsoleHandler
to support printing logs to the console:
h := handler.NewConsoleHandler(slog.AllLevels)
l := slog.NewWithHandlers(h)
l.Trace("this is a simple log message")
l.Debug("this is a simple log message")
Change the default logger log output style:
h.Formatter().(*slog.TextFormatter).SetTemplate(slog.NamedTemplate)
Output:
Note:
slog.TextFormatter
uses a template string to format the output log, so the new field output needs to adjust the template at the same time.
slog
also has a built-in Formatter
for JSON format. If not specified, the default is to use TextFormatter
to format log records.
package main
import (
"github.com/gookit/slog"
)
func main() {
// use JSON formatter
slog.SetFormatter(slog.NewJSONFormatter())
slog.Info("info log message")
slog.Warn("warning log message")
slog.WithData(slog.M{
"key0": 134,
"key1": "abc",
}).Infof("info log %s", "message")
r := slog.WithFields(slog.M{
"category": "service",
"IP": "127.0.0.1",
})
r.Infof("info %s", "message")
r.Debugf("debug %s", "message")
}
Output:
{"channel":"application","data":{},"datetime":"2020/07/16 13:23:33","extra":{},"level":"INFO","message":"info log message"}
{"channel":"application","data":{},"datetime":"2020/07/16 13:23:33","extra":{},"level":"WARNING","message":"warning log message"}
{"channel":"application","data":{"key0":134,"key1":"abc"},"datetime":"2020/07/16 13:23:33","extra":{},"level":"INFO","message":"info log message"}
{"IP":"127.0.0.1","category":"service","channel":"application","datetime":"2020/07/16 13:23:33","extra":{},"level":"INFO","message":"info message"}
{"IP":"127.0.0.1","category":"service","channel":"application","datetime":"2020/07/16 13:23:33","extra":{},"level":"DEBUG","message":"debug message"}
Logger
- log dispatcher. One logger can register multipleHandler
,Processor
Record
- log records, each log is aRecord
instance.Processor
- enables extended processing of log records. It is called before the logRecord
is processed by theHandler
.- You can use it to perform additional operations on
Record
, such as: adding fields, adding extended information, etc.
- You can use it to perform additional operations on
Handler
- log handler, each log will be processed byHandler.Handle()
.- Here you can send logs to console, file, remote server, etc.
Formatter
- logging data formatting process.- Usually set in
Handler
, it can be used to format log records, convert records into text, JSON, etc.,Handler
then writes the formatted data to the specified place. Formatter
is not required. You can do without it and handle logging directly inHandler.Handle()
.
- Usually set in
Simple structure of log scheduler:
Processors
Logger --{
Handlers --|- Handler0 With Formatter0
|- Handler1 With Formatter1
|- Handler2 (can also without Formatter)
|- ... more
Note: Be sure to remember to add
Handler
,Processor
to the logger instance and log records will be processed byHandler
.
Processor
interface:
// Processor interface definition
type Processor interface {
// Process record
Process(record *Record)
}
// ProcessorFunc definition
type ProcessorFunc func(record *Record)
// Process record
func (fn ProcessorFunc) Process(record *Record) {
fn(record)
}
You can use it to perform additional operations on the Record before the log
Record
reaches theHandler
for processing, such as: adding fields, adding extended information, etc.
Add processor to logger:
slog.AddProcessor(slog.AddHostname())
// or
l := slog.New()
l.AddProcessor(slog.AddHostname())
The built-in processor slog.AddHostname
is used here as an example, which can add a new field hostname
on each log record.
slog.AddProcessor(slog.AddHostname())
slog.Info("message")
Output, including new fields "hostname":"InhereMac"
:
{"channel":"application","level":"INFO","datetime":"2020/07/17 12:01:35","hostname":"InhereMac","data":{},"extra":{},"message":"message"}
Handler
interface:
You can customize any
Handler
you want, just implement theslog.Handler
interface.
// Handler interface definition
type Handler interface {
io.Closer
Flush() error
// IsHandling Checks whether the given record will be handled by this handler.
IsHandling(level Level) bool
// Handle a log record.
// all records may be passed to this method, and the handler should discard
// those that it does not want to handle.
Handle(*Record) error
}
Formatter
interface:
// Formatter interface
type Formatter interface {
Format(record *Record) ([]byte, error)
}
Function wrapper type:
// FormatterFunc wrapper definition
type FormatterFunc func(r *Record) ([]byte, error)
// Format a log record
func (fn FormatterFunc) Format(r *Record) ([]byte, error) {
return fn(r)
}
JSON formatter
type JSONFormatter struct {
// Fields exported log fields.
Fields []string
// Aliases for output fields. you can change export field name.
// item: `"field" : "output name"`
// eg: {"message": "msg"} export field will display "msg"
Aliases StringMap
// PrettyPrint will indent all json logs
PrettyPrint bool
// TimeFormat the time format layout. default is time.RFC3339
TimeFormat string
}
Text formatter
Default templates:
const DefaultTemplate = "[{{datetime}}] [{{channel}}] [{{level}}] [{{caller}}] {{message}} {{data}} {{extra}}\n"
const NamedTemplate = "{{datetime}} channel={{channel}} level={{level}} [file={{caller}}] message={{message}} data={{data}}\n"
Change template:
myTemplate := "[{{datetime}}] [{{level}}] {{message}}"
f := slog.NewTextFormatter()
f.SetTemplate(myTemplate)
Custom Processor
and Formatter
are relatively simple, just implement a corresponding method.
slog.Info, slog.Warn
and other methods use the default logger and output logs to the console by default.
You can create a brand-new instance of slog.Logger
:
Method 1:
l := slog.New()
// add handlers ...
h1 := handler.NewConsoleHandler(slog.AllLevels)
l.AddHandlers(h1)
Method 2:
l := slog.NewWithName("myLogger")
// add handlers ...
h1 := handler.NewConsoleHandler(slog.AllLevels)
l.AddHandlers(h1)
Method 3:
package main
import (
"github.com/gookit/slog"
"github.com/gookit/slog/handler"
)
func main() {
l := slog.NewWithHandlers(handler.NewConsoleHandler(slog.AllLevels))
l.Info("message")
}
You only need to implement the slog.Handler
interface to create a custom Handler
.
You can quickly assemble your own Handler through the built-in handler.LevelsWithFormatter
handler.LevelWithFormatter
and other fragments of slog.
Examples:
Use
handler.LevelsWithFormatter
, only need to implementClose, Flush, Handle
methods
type MyHandler struct {
handler.LevelsWithFormatter
Output io.Writer
}
func (h *MyHandler) Handle(r *slog.Record) error {
// you can write log message to file or send to remote.
}
func (h *MyHandler) Flush() error {}
func (h *MyHandler) Close() error {}
Add Handler
to the logger to use:
// add to default logger
slog.AddHander(&MyHandler{})
// or, add to custom logger:
l := slog.New()
l.AddHander(&MyHandler{})
./handler package has built-in common log handlers, which can basically meet most scenarios.
// Output logs to console, allow render color.
func NewConsoleHandler(levels []slog.Level) *ConsoleHandler
// Send logs to email
func NewEmailHandler(from EmailOption, toAddresses []string) *EmailHandler
// Send logs to syslog
func NewSysLogHandler(priority syslog.Priority, tag string) (*SysLogHandler, error)
// A simple handler implementation that outputs logs to a given io.Writer
func NewSimpleHandler(out io.Writer, level slog.Level) *SimpleHandler
Output log to file:
// Output log to the specified file, without buffering by default
func NewFileHandler(logfile string, fns ...ConfigFn) (h *SyncCloseHandler, err error)
// Output logs to the specified file in JSON format, without buffering by default
func JSONFileHandler(logfile string, fns ...ConfigFn) (*SyncCloseHandler, error)
// Buffered output log to specified file
func NewBuffFileHandler(logfile string, buffSize int, fns ...ConfigFn) (*SyncCloseHandler, error)
TIP:
NewFileHandler
JSONFileHandler
can also enable write buffering by passing in fnshandler.WithBuffSize(buffSize)
Output log to file and rotate automatically:
// Automatic rotating according to file size
func NewSizeRotateFile(logfile string, maxSize int, fns ...ConfigFn) (*SyncCloseHandler, error)
// Automatic rotating according to time
func NewTimeRotateFile(logfile string, rt rotatefile.RotateTime, fns ...ConfigFn) (*SyncCloseHandler, error)
// It supports configuration to rotate according to size and time.
// The default setting file size is 20M, and the default automatic splitting time is 1 hour (EveryHour).
func NewRotateFileHandler(logfile string, rt rotatefile.RotateTime, fns ...ConfigFn) (*SyncCloseHandler, error)
TIP: By passing in
fns ...ConfigFn
, more options can be set, such as log file retention time, log write buffer size, etc. For detailed settings, see thehandler.Config
structure
Output log to the specified file, buffer
buffered writing is not enabled by default. Buffering can also be enabled by passing in a parameter.
package mypkg
import (
"github.com/gookit/slog"
"github.com/gookit/slog/handler"
)
func main() {
defer slog.MustClose()
// DangerLevels contains: slog.PanicLevel, slog.ErrorLevel, slog.WarnLevel
h1 := handler.MustFileHandler("/tmp/error.log", handler.WithLogLevels(slog.DangerLevels))
// custom log format
// f := h1.Formatter().(*slog.TextFormatter)
f := slog.AsTextFormatter(h1.Formatter())
f.SetTemplate("your template format\n")
// NormalLevels contains: slog.InfoLevel, slog.NoticeLevel, slog.DebugLevel, slog.TraceLevel
h2 := handler.MustFileHandler("/tmp/info.log", handler.WithLogLevels(slog.NormalLevels))
// register handlers
slog.PushHandler(h1)
slog.PushHandler(h2)
// add logs
slog.Info("info message text")
slog.Error("error message text")
}
Note: If write buffering
buffer
is enabled, be sure to calllogger.Close()
at the end of the program to flush the contents of the buffer to the file.
slog/handler
also has a built-in output log to a specified file, and supports splitting files by time and size at the same time.
By default, buffer
buffered writing is enabled
func Example_rotateFileHandler() {
h1 := handler.MustRotateFile("/tmp/error.log", handler.EveryHour, handler.WithLogLevels(slog.DangerLevels))
h2 := handler.MustRotateFile("/tmp/info.log", handler.EveryHour, handler.WithLogLevels(slog.NormalLevels))
slog.PushHandler(h1)
slog.PushHandler(h2)
// add logs
slog.Info("info message")
slog.Error("error message")
}
Example of file name sliced by time:
time-rotate-file.log
time-rotate-file.log.20201229_155753
time-rotate-file.log.20201229_155754
Example of a filename cut by size, in the format filename.log.HIS_000N
. For example:
size-rotate-file.log
size-rotate-file.log.122915_0001
size-rotate-file.log.122915_0002
rotatefile.Writer
can also be used with other logging packages, such as: log
, glog
, etc.
For example, using rotatefile
on golang log
:
package main
import (
"log"
"github.com/gookit/slog/rotatefile"
)
func main() {
logFile := "testdata/go_logger.log"
writer, err := rotatefile.NewConfig(logFile).Create()
if err != nil {
panic(err)
}
log.SetOutput(writer)
log.Println("log message")
}
This is config struct for create a Handler:
// Config struct
type Config struct {
// Logfile for write logs
Logfile string `json:"logfile" yaml:"logfile"`
// LevelMode for filter log record. default LevelModeList
LevelMode uint8 `json:"level_mode" yaml:"level_mode"`
// Level value. use on LevelMode = LevelModeValue
Level slog.Level `json:"level" yaml:"level"`
// Levels for log record
Levels []slog.Level `json:"levels" yaml:"levels"`
// UseJSON for format logs
UseJSON bool `json:"use_json" yaml:"use_json"`
// BuffMode type name. allow: line, bite
BuffMode string `json:"buff_mode" yaml:"buff_mode"`
// BuffSize for enable buffer, unit is bytes. set 0 to disable buffer
BuffSize int `json:"buff_size" yaml:"buff_size"`
// RotateTime for rotate file, unit is seconds.
RotateTime rotatefile.RotateTime `json:"rotate_time" yaml:"rotate_time"`
// MaxSize on rotate file by size, unit is bytes.
MaxSize uint64 `json:"max_size" yaml:"max_size"`
// Compress determines if the rotated log files should be compressed using gzip.
// The default is not to perform compression.
Compress bool `json:"compress" yaml:"compress"`
// BackupNum max number for keep old files.
// 0 is not limit, default is 20.
BackupNum uint `json:"backup_num" yaml:"backup_num"`
// BackupTime max time for keep old files. unit is hours
// 0 is not limit, default is a week.
BackupTime uint `json:"backup_time" yaml:"backup_time"`
// RenameFunc build filename for rotate file
RenameFunc func(filepath string, rotateNum uint) string
}
Examples:
testFile := "testdata/error.log"
h := handler.NewEmptyConfig(
handler.WithLogfile(testFile),
handler.WithBuffSize(1024*8),
handler.WithLogLevels(slog.DangerLevels),
handler.WithBuffMode(handler.BuffModeBite),
).
CreateHandler()
l := slog.NewWithHandlers(h)
About BuffMode
Config.BuffMode
The name of the BuffMode type to use. Allow: line, bite
BuffModeBite
: Buffer by bytes, when the number of bytes in the buffer reaches the specified size, write the contents of the buffer to the fileBuffModeLine
: Buffer by line, when the buffer size is reached, always ensure that a complete line of log content is written to the file (to avoid log content being truncated)
Use handler.Builder
to easily and quickly create Handler instances.
testFile := "testdata/info.log"
h := handler.NewBuilder().
WithLogfile(testFile).
WithLogLevels(slog.NormalLevels).
WithBuffSize(1024*8).
WithBuffMode(handler.BuffModeBite).
WithRotateTime(rotatefile.Every30Min).
WithCompress(true).
Build()
l := slog.NewWithHandlers(h)
Package bufwrite
:
bufwrite.BufIOWriter
additionally implementsSync(), Close()
methods by wrapping go'sbufio.Writer
, which is convenient to usebufwrite.LineWriter
refer to the implementation ofbufio.Writer
in go, which can support flushing the buffer by line, which is more useful for writing log files
Package rotatefile
:
rotatefile.Writer
implements automatic cutting of log files according to size and specified time, and also supports automatic cleaning of log fileshandler/rotate_file
is to use it to cut the log file
Of course, the rotatefile.Writer can be use on other log package, such as: log
, glog
and more.
Examples, use rotatefile on golang log
:
package main
import (
"log"
"github.com/gookit/slog/rotatefile"
)
func main() {
logFile := "testdata/another_logger.log"
writer, err := rotatefile.NewConfig(logFile).Create()
if err != nil {
panic(err)
}
log.SetOutput(writer)
log.Println("log message")
}
run unit tests:
go test ./...
Benchmark code at _example/bench_loglibs_test.go
make test-bench
Benchmarks for slog
and other log packages:
Note: test and record ad 2023.04.13
goos: darwin
goarch: amd64
cpu: Intel(R) Core(TM) i7-3740QM CPU @ 2.70GHz
BenchmarkZapNegative
BenchmarkZapNegative-4 8381674 1429 ns/op 216 B/op 3 allocs/op
BenchmarkZapSugarNegative
BenchmarkZapSugarNegative-4 8655980 1383 ns/op 104 B/op 4 allocs/op
BenchmarkZeroLogNegative
BenchmarkZeroLogNegative-4 14173719 849.8 ns/op 0 B/op 0 allocs/op
BenchmarkPhusLogNegative
BenchmarkPhusLogNegative-4 27456256 451.2 ns/op 0 B/op 0 allocs/op
BenchmarkLogrusNegative
BenchmarkLogrusNegative-4 2550771 4784 ns/op 608 B/op 17 allocs/op
BenchmarkGookitSlogNegative
>>>> BenchmarkGookitSlogNegative-4 8798220 1375 ns/op 120 B/op 3 allocs/op
BenchmarkZapPositive
BenchmarkZapPositive-4 10302483 1167 ns/op 192 B/op 1 allocs/op
BenchmarkZapSugarPositive
BenchmarkZapSugarPositive-4 3833311 3154 ns/op 344 B/op 7 allocs/op
BenchmarkZeroLogPositive
BenchmarkZeroLogPositive-4 14120524 846.7 ns/op 0 B/op 0 allocs/op
BenchmarkPhusLogPositive
BenchmarkPhusLogPositive-4 27152686 434.9 ns/op 0 B/op 0 allocs/op
BenchmarkLogrusPositive
BenchmarkLogrusPositive-4 2601892 4691 ns/op 608 B/op 17 allocs/op
BenchmarkGookitSlogPositive
>>>> BenchmarkGookitSlogPositive-4 8997104 1340 ns/op 120 B/op 3 allocs/op
PASS
ok command-line-arguments 167.095s
- gookit/ini Go config management, use INI files
- gookit/rux Simple and fast request router for golang HTTP
- gookit/gcli Build CLI application, tool library, running CLI commands
- gookit/slog Lightweight, extensible, configurable logging library written in Go
- gookit/color A command-line color library with true color support, universal API methods and Windows support
- gookit/event Lightweight event manager and dispatcher implements by Go
- gookit/cache Generic cache use and cache manager for golang. support File, Memory, Redis, Memcached.
- gookit/config Go config management. support JSON, YAML, TOML, INI, HCL, ENV and Flags
- gookit/filter Provide filtering, sanitizing, and conversion of golang data
- gookit/validate Use for data validation and filtering. support Map, Struct, Form data
- gookit/goutil Some utils for the Go: string, array/slice, map, format, cli, env, filesystem, test and more
- More, please see https://github.com/gookit
The projects is heavily inspired by follow packages:
- https://github.com/phuslu/log
- https://github.com/golang/glog
- https://github.com/sirupsen/logrus
- https://github.com/Seldaek/monolog
- https://github.com/syyongx/llog
- https://github.com/uber-go/zap
- https://github.com/rs/zerolog
- https://github.com/natefinch/lumberjack