func new(g *ast.Grammar, record bool) (*Mem, error) {
simp := interp.NewSimplifier(g)
if record {
simp = simp.OptimizeForRecord()
}
g = simp.Grammar()
refs := ast.NewRefLookup(g)
m := &Mem{
refs: refs,
simplifier: simp,
patterns: newPatternsSet(),
zis: newIntsSet(),
stackElms: newPairSet(),
nullables: newBitsetSet(),
Calls: []*CallNode{},
Returns: []map[int]int{},
Escapables: []bool{},
StateToNullable: []int{},
Accept: []bool{},
}
start := m.patterns.add([]*ast.Pattern{refs["main"]})
e := &exprToFunc{m: make(map[*ast.Expr]funcs.Bool)}
for _, p := range refs {
p.Walk(e)
if e.err != nil {
return nil, e.err
}
}
m.funcs = e.m
m.Start = start
return m, nil
}
func TestExplosionAndSameTree(t *testing.T) {
var input = `(
.A:.B:.DeepLevel:.DeeperLevel:.DeepestLevel:->contains($string,"el") &
(
.A:.B:.Rs:._:->eq("~a",$string) &
(
.A:.B:.NumI32:->contains($int,[]int{28,1,52}) &
(
.A:.B:.NumI64:>= 1 &
(
.A:.B:.NumU32:<= uint(4) &
(
.A:.B:.NumU64:== uint(4) &
(
.A:.B:.YesNo:== true &
(
.A:.B:.BS:== []byte{0x3, 0x2, 0x1, 0x0} &
.A:.B:.Uuid: == []byte{0x3, 0x2, 0x1, 0x0}
)
)
)
)
)
)
)
)`
g, err := parser.ParseGrammar(input)
if err != nil {
t.Fatal(err)
}
// This one causes a state explosion of over 14000 states.
// Since we know field names can't repeat the simplification can be made for record (JSON and proto) like serialization formats, but not for XML.
g = interp.NewSimplifier(g).OptimizeForRecord().Grammar()
t.Logf("%v", g)
a, err := Compile(g)
if err != nil {
t.Fatal(err)
}
t.Logf("number of states %d", len(a.accept))
if len(a.accept) > 1000 {
t.Fatal("number of states exploded")
}
}
