elvish/parse/parser.go

149 lines
2.8 KiB
Go
Raw Normal View History

2016-01-21 06:57:18 +08:00
package parse
import (
"bytes"
"errors"
"strings"
"unicode/utf8"
2016-02-06 07:08:39 +08:00
"github.com/elves/elvish/util"
2016-01-21 06:57:18 +08:00
)
2016-02-03 02:39:44 +08:00
// parser maintains some mutable states of parsing.
2016-01-21 06:57:18 +08:00
//
// NOTE: The str member is assumed to be valid UF-8.
2016-02-03 02:39:44 +08:00
type parser struct {
src string
pos int
overEOF int
cutsets []map[rune]int
controls int
errors *util.Errors
2016-01-21 06:57:18 +08:00
}
2016-02-08 06:23:16 +08:00
const eof rune = -1
2016-01-21 06:57:18 +08:00
2016-02-07 06:17:57 +08:00
func (ps *parser) eof() bool {
2016-02-08 06:23:16 +08:00
return ps.peek() == eof
2016-02-07 06:17:57 +08:00
}
2016-02-03 02:39:44 +08:00
func (ps *parser) peek() rune {
if ps.pos == len(ps.src) {
2016-02-08 06:23:16 +08:00
return eof
2016-01-21 06:57:18 +08:00
}
2016-02-03 02:39:44 +08:00
r, _ := utf8.DecodeRuneInString(ps.src[ps.pos:])
2016-02-07 06:17:57 +08:00
if ps.currentCutset()[r] > 0 {
2016-02-08 06:23:16 +08:00
return eof
2016-02-07 06:17:57 +08:00
}
2016-01-21 06:57:18 +08:00
return r
}
2016-02-03 02:39:44 +08:00
func (ps *parser) hasPrefix(prefix string) bool {
return strings.HasPrefix(ps.src[ps.pos:], prefix)
2016-01-21 06:57:18 +08:00
}
// findWord looks ahead for [a-z]* that is also a valid compound. If the
// lookahead fails, it returns an empty string. It is useful for looking for
// command leaders.
func (ps *parser) findPossibleLeader() string {
rest := ps.src[ps.pos:]
i := strings.IndexFunc(rest, func(r rune) bool {
return r < 'a' || r > 'z'
})
if i == -1 {
// The whole rest is just one possible leader.
return rest
}
r, _ := utf8.DecodeRuneInString(rest[i:])
if startsPrimary(r) {
return ""
}
return rest[:i]
}
2016-02-03 02:39:44 +08:00
func (ps *parser) next() rune {
if ps.pos == len(ps.src) {
2016-02-08 06:23:16 +08:00
ps.overEOF++
return eof
2016-01-21 06:57:18 +08:00
}
2016-02-03 02:39:44 +08:00
r, s := utf8.DecodeRuneInString(ps.src[ps.pos:])
2016-02-07 06:17:57 +08:00
if ps.currentCutset()[r] > 0 {
2016-02-08 06:23:16 +08:00
return eof
2016-02-07 06:17:57 +08:00
}
2016-02-03 02:39:44 +08:00
ps.pos += s
2016-01-21 06:57:18 +08:00
return r
}
2016-02-03 02:39:44 +08:00
func (ps *parser) backup() {
if ps.overEOF > 0 {
2016-02-08 06:23:16 +08:00
ps.overEOF--
return
}
2016-02-03 02:39:44 +08:00
_, s := utf8.DecodeLastRuneInString(ps.src[:ps.pos])
ps.pos -= s
2016-01-21 06:57:18 +08:00
}
2016-02-06 07:08:39 +08:00
func (ps *parser) advance(c int) {
ps.pos += c
if ps.pos > len(ps.src) {
ps.overEOF = ps.pos - len(ps.src)
ps.pos = len(ps.src)
}
}
2016-02-06 07:08:39 +08:00
func (ps *parser) error(e error) {
if ps.errors == nil {
ps.errors = &util.Errors{}
2016-02-06 07:08:39 +08:00
}
ps.errors.Append(&util.PosError{ps.pos, ps.pos, e})
2016-02-06 07:08:39 +08:00
}
2016-02-07 00:13:53 +08:00
2016-02-07 06:17:57 +08:00
func (ps *parser) pushCutset(rs ...rune) {
ps.cutsets = append(ps.cutsets, map[rune]int{})
ps.cut(rs...)
}
func (ps *parser) popCutset() {
n := len(ps.cutsets)
ps.cutsets[n-1] = nil
ps.cutsets = ps.cutsets[:n-1]
}
func (ps *parser) currentCutset() map[rune]int {
return ps.cutsets[len(ps.cutsets)-1]
}
func (ps *parser) cut(rs ...rune) {
cutset := ps.currentCutset()
for _, r := range rs {
cutset[r]++
}
}
func (ps *parser) uncut(rs ...rune) {
cutset := ps.currentCutset()
for _, r := range rs {
cutset[r]--
}
}
2016-02-07 00:13:53 +08:00
func newError(text string, shouldbe ...string) error {
if len(shouldbe) == 0 {
return errors.New(text)
}
var buf bytes.Buffer
if len(text) > 0 {
buf.WriteString(text + ", ")
}
buf.WriteString("should be " + shouldbe[0])
for i, opt := range shouldbe[1:] {
if i == len(shouldbe)-2 {
buf.WriteString(" or ")
} else {
buf.WriteString(", ")
}
buf.WriteString(opt)
}
return errors.New(buf.String())
}