9.6 KiB
goparsify 
A parser-combinator library for building easy to test, read and maintain parsers using functional composition.
Everything should be unicode safe by default, but you can opt out of unicode whitespace for a decent ~%20 performance boost.
Run(parser, input, ASCIIWhitespace)
benchmarks
I dont have many benchmarks set up yet, but the json parser is very promising. Nearly keeping up with the stdlib for raw speed:
$ go test -bench=. -benchtime=2s -benchmem ./json
BenchmarkUnmarshalParsec-8 20000 65682 ns/op 50460 B/op 1318 allocs/op
BenchmarkUnmarshalParsify-8 30000 51292 ns/op 45104 B/op 334 allocs/op
BenchmarkUnmarshalStdlib-8 30000 46522 ns/op 13953 B/op 262 allocs/op
PASS
ok github.com/vektah/goparsify/json 10.840s
debugging parsers
When a parser isnt working as you intended you can build with debugging and enable logging to get a detailed log of exactly what the parser is doing.
- First build with debug using
-tags debug - enable logging by passing a runtime flag -parselogs or calling
EnableLogging(os.Stdout)in your code.
This works great with tests, eg in the goparsify source tree
$ cd html
$ go test -tags debug -parselogs
html.go:50 | <body>hello <p | | tag
html.go:45 | <body>hello <p | | tstart
html.go:45 | body>hello <p c | < | <
html.go:20 | >hello <p color | body | identifier
html.go:35 | >hello <p color | | attrs
html.go:34 | >hello <p color | | attr
html.go:20 | >hello <p color | fail | identifier
html.go:45 | hello <p color= | > | >
html.go:26 | hello <p color= | | elements
html.go:25 | hello <p color= | | element
html.go:21 | <p color="blue" | hello | text
html.go:25 | <p color="blue" | | element
html.go:21 | <p color="blue" | fail | text
html.go:50 | <p color="blue" | | tag
html.go:45 | <p color="blue" | | tstart
html.go:45 | p color="blue"> | < | <
html.go:20 | color="blue">w | p | identifier
html.go:35 | color="blue">w | | attrs
html.go:34 | color="blue">w | | attr
html.go:20 | ="blue">world</ | color | identifier
html.go:34 | "blue">world</p | = | =
html.go:34 | >world</p></bod | | string literal
html.go:34 | >world</p></bod | | attr
html.go:20 | >world</p></bod | fail | identifier
html.go:45 | world</p></body | > | >
html.go:26 | world</p></body | | elements
html.go:25 | world</p></body | | element
html.go:21 | </p></body> | world | text
html.go:25 | </p></body> | | element
html.go:21 | </p></body> | fail | text
html.go:50 | </p></body> | | tag
html.go:45 | </p></body> | | tstart
html.go:45 | /p></body> | < | <
html.go:20 | /p></body> | fail | identifier
html.go:46 | </p></body> | | tend
html.go:46 | p></body> | </ | </
html.go:20 | ></body> | p | identifier
html.go:46 | </body> | > | >
html.go:25 | </body> | | element
html.go:21 | </body> | fail | text
html.go:50 | </body> | | tag
html.go:45 | </body> | | tstart
html.go:45 | /body> | < | <
html.go:20 | /body> | fail | identifier
html.go:46 | </body> | | tend
html.go:46 | body> | </ | </
html.go:20 | > | body | identifier
html.go:46 | | > | >
PASS
ok github.com/vektah/goparsify/html 0.118s
debugging performance
If you build the parser with -tags debug it will instrument each parser and a call to DumpDebugStats() will show stats:
Any() 415.7136ms 87000 calls json.go:35
Map() 309.6569ms 12000 calls json.go:31
Seq() 298.6519ms 12000 calls json.go:23
Some() 290.6462ms 12000 calls json.go:13
Seq() 272.6392ms 81000 calls json.go:13
Seq() 78.0404ms 13000 calls json.go:15
Map() 78.0404ms 13000 calls json.go:21
Some() 77.0401ms 1000 calls json.go:15
string literal 7.5053ms 81000 calls json.go:13
string literal 4.5031ms 84000 calls json.go:11
, 4.0008ms 81000 calls json.go:13
false 2.0018ms 85000 calls json.go:10
null 2.0005ms 87000 calls json.go:8
true 1.501ms 87000 calls json.go:9
: 500.8µs 81000 calls json.go:13
[ 0s 13000 calls json.go:15
} 0s 12000 calls json.go:23
{ 0s 12000 calls json.go:23
number literal 0s 31000 calls json.go:12
] 0s 1000 calls json.go:15
Nil 0s 0 calls profile/json.go:148
, 0s 5000 calls json.go:15
All times are cumulative, it would be nice to break this down into a parse tree with relative times. This is a nice addition to pprof as it will break down the parsers based on where they are used instead of grouping them all by type.
This is free when the debug tag isnt used.
example calculator
Lets say we wanted to build a calculator that could take an expression and calculate the result.
Lets start with test:
func TestNumbers(t *testing.T) {
result, err := Calc(`1`)
require.NoError(t, err)
require.EqualValues(t, 1, result)
}
Then define a parser for numbers
var number = Map(NumberLit(), func(n Result) Result {
switch i := n.Result.(type) {
case int64:
return Result{Result: float64(i)}
case float64:
return Result{Result: i}
default:
panic(fmt.Errorf("unknown value %#v", i))
}
})
func Calc(input string) (float64, error) {
result, err := Run(y, input)
if err != nil {
return 0, err
}
return result.(float64), nil
}
This parser will return numbers either as float64 or int depending on the literal, for this calculator we only want floats so we Map the results and type cast.
Run the tests and make sure everything is ok.
Time to add addition
func TestAddition(t *testing.T) {
result, err := Calc(`1+1`)
require.NoError(t, err)
require.EqualValues(t, 2, result)
}
var sumOp = Chars("+-", 1, 1)
sum = Map(Seq(number, Some(And(sumOp, number))), func(n Result) Result {
i := n.Child[0].Result.(float64)
for _, op := range n.Child[1].Child {
switch op.Child[0].Token {
case "+":
i += op.Child[1].Result.(float64)
case "-":
i -= op.Child[1].Result.(float64)
}
}
return Result{Result: i}
})
// and update Calc to point to the new root parser -> `result, err := ParseString(sum, input)`
This parser will match number ([+-] number)+, then map its to be the sum. See how the Child map directly to the positions in the parsers? n is the result of the and, n.Child[0] is its first argument, n.Child[1] is the result of the Some parser, n.Child[1].Child[0] is the result of the first And and so fourth. Given how closely tied the parser and the Map are it is good to keep the two together.
You can continue like this and add multiplication and parenthesis fairly easily. Eventually if you keep adding parsers you will end up with a loop, and go will give you a handy error message like:
typechecking loop involving value = goparsify.Any(number, groupExpr)
we need to break the loop using a pointer, then set its value in init
var (
value Parser
prod = Seq(&value, Some(And(prodOp, &value)))
)
func init() {
value = Any(number, groupExpr)
}
Take a look at calc for a full example.
preventing backtracking with cuts
A cut is a marker that prevents backtracking past the point it was set. This greatly improves error messages when used correctly:
alpha := Chars("a-z")
// without a cut if the close tag is left out the parser will backtrack and ignore the rest of the string
nocut := Many(Any(Seq("<", alpha, ">"), alpha))
_, err := Run(nocut, "asdf <foo")
fmt.Println(err.Error())
// Outputs: left unparsed: <foo
// with a cut, once we see the open tag we know there must be a close tag that matches it, so the parser will error
cut := Many(Any(Seq("<", Cut, alpha, ">"), alpha))
_, err = Run(cut, "asdf <foo")
fmt.Println(err.Error())
// Outputs: offset 9: expected >
prior art
Inspired by https://github.com/prataprc/goparsec