Expression evaluation with Lua for Go
glel is an expression evaluation engine for Go using GopherLua
- Safe : glel evaluates expressions in sandboxed environment.
- Flexible : You can write expressions in Lua, most widely used embedded language.
- Fast : glel uses GopherLua as a Lua interperter. GopherLua is a resonably fast for scripting and been used for many years in many projects
- Easy to use : glel integrates Go and Lua using gopher-luar. Values are automatically converted for the languages.
expr := glel.New(
glel.WithAllowedFunctions(glel.DefaultAllowedFunctions+`
string.rep
`),
glel.WithPoolSize(10),
glel.WithEnv(glel.Env{
"hoge": "foo",
}))
defer expr.Close()
evaler, err := expr.Compile(`hoge == "foo" and add(x, y) == 15 and string.rep("ab", 5) == "ababababab" and d.name == "alice" `)
if err != nil {
panic(err)
}
d := &struct {
Name string
}{
Name: "alice",
}
ok, err := evaler.EvalBool(glel.Env{
"y": 10,
"x": lua.LNumber(5),
"d": d,
"add": func(a, b int) int {
return a + b
},
})glel is a sandboxed by default. You can allow glel to use additonal builtin functions using WithAllowedFunctions.
glel pools Lua interpreters for performance. You can set pool size using WithPoolSize. glel does not pool Lua interpreters if WithPoolSize is set to negative.
EvalContextXXX behaves same as EvalXXX, but evaluates in coordination with context.
ctx, cancel := context.WithTimeout(context.TODO(), 1*time.Second)
defer cancel()
ok, err = evaler.EvalContextBool(ctx, nil)expr is a goroutine safe if pooled( == WithPoolSize > 0 ). You can use one expr from multiple goroutines. evaler can be cached. Note that you must not closes expr before evaler evaluates.
Sandboxing has performance disadvantages.
Benchmark_noSandbox 218 ns/op 0 B/op 0 allocs/op
Benchmark_sandbox 3886 ns/op 4000 B/op 17 allocs/op
You can disable a sandboxing by WithDisableSandbox().
Modified version https://github.com/antonmedv/golang-expression-evaluation-comparison .
glel: WithDisableSandbox, WithPoolSize(-1)
Benchmark_bexpr-12 461487 2462 ns/op 784 B/op 43 allocs/op
Benchmark_celgo-12 5900113 201.4 ns/op 24 B/op 2 allocs/op
Benchmark_celgo_startswith-12 3268881 367.5 ns/op 88 B/op 6 allocs/op
Benchmark_evalfilter-12 615469 1888 ns/op 736 B/op 21 allocs/op
Benchmark_expr-12 8480037 139.7 ns/op 32 B/op 1 allocs/op
Benchmark_expr_startswith-12 3973664 301.0 ns/op 128 B/op 4 allocs/op
Benchmark_glel-12 5481398 215.9 ns/op 0 B/op 0 allocs/op
Benchmark_goja-12 3320126 362.4 ns/op 96 B/op 2 allocs/op
Benchmark_govaluate-12 4233823 281.9 ns/op 24 B/op 2 allocs/op
Benchmark_gval-12 1492868 803.0 ns/op 240 B/op 8 allocs/op
Benchmark_otto-12 1415108 828.4 ns/op 336 B/op 7 allocs/op
Benchmark_starlark-12 197761 6169 ns/op 3568 B/op 68 allocs/op
Althouh glel uses turing-complete language, is being resonably fast.
MIT
Yusuke Inuzuka