195 lines
3.5 KiB
Go
195 lines
3.5 KiB
Go
package parser
|
|
|
|
import (
|
|
"fmt"
|
|
|
|
"github.com/ChrisTrenkamp/goxpath/internal/lexer"
|
|
)
|
|
|
|
type stateType int
|
|
|
|
const (
|
|
defState stateType = iota
|
|
xpathState
|
|
funcState
|
|
paramState
|
|
predState
|
|
parenState
|
|
)
|
|
|
|
type parseStack struct {
|
|
stack []*Node
|
|
stateTypes []stateType
|
|
cur *Node
|
|
}
|
|
|
|
func (p *parseStack) push(t stateType) {
|
|
p.stack = append(p.stack, p.cur)
|
|
p.stateTypes = append(p.stateTypes, t)
|
|
}
|
|
|
|
func (p *parseStack) pop() {
|
|
stackPos := len(p.stack) - 1
|
|
|
|
p.cur = p.stack[stackPos]
|
|
p.stack = p.stack[:stackPos]
|
|
p.stateTypes = p.stateTypes[:stackPos]
|
|
}
|
|
|
|
func (p *parseStack) curState() stateType {
|
|
if len(p.stateTypes) == 0 {
|
|
return defState
|
|
}
|
|
return p.stateTypes[len(p.stateTypes)-1]
|
|
}
|
|
|
|
type lexFn func(*parseStack, lexer.XItem)
|
|
|
|
var parseMap = map[lexer.XItemType]lexFn{
|
|
lexer.XItemAbsLocPath: xiXPath,
|
|
lexer.XItemAbbrAbsLocPath: xiXPath,
|
|
lexer.XItemAbbrRelLocPath: xiXPath,
|
|
lexer.XItemRelLocPath: xiXPath,
|
|
lexer.XItemEndPath: xiEndPath,
|
|
lexer.XItemAxis: xiXPath,
|
|
lexer.XItemAbbrAxis: xiXPath,
|
|
lexer.XItemNCName: xiXPath,
|
|
lexer.XItemQName: xiXPath,
|
|
lexer.XItemNodeType: xiXPath,
|
|
lexer.XItemProcLit: xiXPath,
|
|
lexer.XItemFunction: xiFunc,
|
|
lexer.XItemArgument: xiFuncArg,
|
|
lexer.XItemEndFunction: xiEndFunc,
|
|
lexer.XItemPredicate: xiPred,
|
|
lexer.XItemEndPredicate: xiEndPred,
|
|
lexer.XItemStrLit: xiValue,
|
|
lexer.XItemNumLit: xiValue,
|
|
lexer.XItemOperator: xiOp,
|
|
lexer.XItemVariable: xiValue,
|
|
}
|
|
|
|
var opPrecedence = map[string]int{
|
|
"|": 1,
|
|
"*": 2,
|
|
"div": 2,
|
|
"mod": 2,
|
|
"+": 3,
|
|
"-": 3,
|
|
"=": 4,
|
|
"!=": 4,
|
|
"<": 4,
|
|
"<=": 4,
|
|
">": 4,
|
|
">=": 4,
|
|
"and": 5,
|
|
"or": 6,
|
|
}
|
|
|
|
//Parse creates an AST tree for XPath expressions.
|
|
func Parse(xp string) (*Node, error) {
|
|
var err error
|
|
c := lexer.Lex(xp)
|
|
n := &Node{}
|
|
p := &parseStack{cur: n}
|
|
|
|
for next := range c {
|
|
if next.Typ != lexer.XItemError {
|
|
parseMap[next.Typ](p, next)
|
|
} else if err == nil {
|
|
err = fmt.Errorf(next.Val)
|
|
}
|
|
}
|
|
|
|
return n, err
|
|
}
|
|
|
|
func xiXPath(p *parseStack, i lexer.XItem) {
|
|
if p.curState() == xpathState {
|
|
p.cur.push(i)
|
|
p.cur = p.cur.next
|
|
return
|
|
}
|
|
|
|
if p.cur.Val.Typ == lexer.XItemFunction {
|
|
p.cur.Right = &Node{Val: i, Parent: p.cur}
|
|
p.cur.next = p.cur.Right
|
|
p.push(xpathState)
|
|
p.cur = p.cur.next
|
|
return
|
|
}
|
|
|
|
p.cur.pushNotEmpty(i)
|
|
p.push(xpathState)
|
|
p.cur = p.cur.next
|
|
}
|
|
|
|
func xiEndPath(p *parseStack, i lexer.XItem) {
|
|
p.pop()
|
|
}
|
|
|
|
func xiFunc(p *parseStack, i lexer.XItem) {
|
|
p.cur.push(i)
|
|
p.cur = p.cur.next
|
|
p.push(funcState)
|
|
}
|
|
|
|
func xiFuncArg(p *parseStack, i lexer.XItem) {
|
|
if p.curState() != funcState {
|
|
p.pop()
|
|
}
|
|
|
|
p.cur.push(i)
|
|
p.cur = p.cur.next
|
|
p.push(paramState)
|
|
p.cur.push(lexer.XItem{Typ: Empty, Val: ""})
|
|
p.cur = p.cur.next
|
|
}
|
|
|
|
func xiEndFunc(p *parseStack, i lexer.XItem) {
|
|
if p.curState() == paramState {
|
|
p.pop()
|
|
}
|
|
p.pop()
|
|
}
|
|
|
|
func xiPred(p *parseStack, i lexer.XItem) {
|
|
p.cur.push(i)
|
|
p.cur = p.cur.next
|
|
p.push(predState)
|
|
p.cur.push(lexer.XItem{Typ: Empty, Val: ""})
|
|
p.cur = p.cur.next
|
|
}
|
|
|
|
func xiEndPred(p *parseStack, i lexer.XItem) {
|
|
p.pop()
|
|
}
|
|
|
|
func xiValue(p *parseStack, i lexer.XItem) {
|
|
p.cur.add(i)
|
|
}
|
|
|
|
func xiOp(p *parseStack, i lexer.XItem) {
|
|
if i.Val == "(" {
|
|
p.cur.push(lexer.XItem{Typ: Empty, Val: ""})
|
|
p.push(parenState)
|
|
p.cur = p.cur.next
|
|
return
|
|
}
|
|
|
|
if i.Val == ")" {
|
|
p.pop()
|
|
return
|
|
}
|
|
|
|
if p.cur.Val.Typ == lexer.XItemOperator {
|
|
if opPrecedence[p.cur.Val.Val] <= opPrecedence[i.Val] {
|
|
p.cur.add(i)
|
|
} else {
|
|
p.cur.push(i)
|
|
}
|
|
} else {
|
|
p.cur.add(i)
|
|
}
|
|
p.cur = p.cur.next
|
|
}
|