2024-09-10 22:46:31 +00:00
|
|
|
package bytes
|
|
|
|
|
|
|
|
import (
|
2024-09-10 22:52:08 +00:00
|
|
|
"errors"
|
2024-09-10 22:46:31 +00:00
|
|
|
"fmt"
|
2024-09-10 22:52:08 +00:00
|
|
|
"io"
|
2024-09-10 22:46:31 +00:00
|
|
|
"regexp"
|
|
|
|
"strings"
|
2024-09-10 22:52:08 +00:00
|
|
|
"unicode/utf8"
|
2024-09-10 22:46:31 +00:00
|
|
|
|
|
|
|
"git.codemonkeysoftware.net/b/gigaparsec"
|
|
|
|
"git.codemonkeysoftware.net/b/gigaparsec/cursor"
|
|
|
|
)
|
|
|
|
|
2024-09-10 22:52:08 +00:00
|
|
|
// RuneReader is an io.RuneReader backed by a Cursor, for compatibility
|
|
|
|
// with the regexp package.
|
|
|
|
type RuneReader struct {
|
|
|
|
cursor cursor.Cursor[byte]
|
|
|
|
}
|
|
|
|
|
|
|
|
func NewRuneReader(c cursor.Cursor[byte]) *RuneReader {
|
|
|
|
return &RuneReader{cursor: c}
|
|
|
|
}
|
|
|
|
|
|
|
|
func (rr *RuneReader) ReadRune() (r rune, size int, err error) {
|
|
|
|
var b [4]byte
|
|
|
|
s := b[:]
|
|
|
|
n, next, err := rr.cursor.Read(s)
|
|
|
|
if err != nil && !errors.Is(err, io.EOF) {
|
|
|
|
rr.cursor = next
|
|
|
|
return 0, 0, fmt.Errorf("ReadRune: %w", err)
|
|
|
|
}
|
|
|
|
s = s[:n]
|
|
|
|
r, size = utf8.DecodeRune(s)
|
|
|
|
rr.cursor = rr.cursor.At(rr.cursor.Pos() + uint64(size))
|
|
|
|
return r, size, err
|
|
|
|
}
|
|
|
|
|
|
|
|
func (rr *RuneReader) Cursor() cursor.Cursor[byte] {
|
|
|
|
return rr.cursor
|
|
|
|
}
|
|
|
|
|
2024-09-10 22:46:31 +00:00
|
|
|
func Regexp(str string) gigaparsec.Parser[byte, []byte] {
|
2024-09-11 00:39:54 +00:00
|
|
|
if !strings.HasPrefix(str, "^") && !strings.HasPrefix(str, `\A`) {
|
2024-09-10 22:46:31 +00:00
|
|
|
str = "^" + str
|
|
|
|
}
|
|
|
|
re := regexp.MustCompile(str)
|
|
|
|
expected := fmt.Sprintf("match `%s`", str)
|
|
|
|
return func(input gigaparsec.State[byte]) (consumed bool, result gigaparsec.Result[byte, []byte], err error) {
|
2024-09-10 22:52:08 +00:00
|
|
|
r := NewRuneReader(input.Cursor())
|
2024-09-10 22:46:31 +00:00
|
|
|
idx := re.FindReaderIndex(r)
|
|
|
|
// TODO Check error from r; this requires an Error() method on cursor.RuneReader.
|
|
|
|
if idx == nil {
|
|
|
|
return false, gigaparsec.Result[byte, []byte]{}, gigaparsec.ParseError{
|
|
|
|
Pos: input.Pos(),
|
|
|
|
Expected: []string{expected},
|
|
|
|
// TODO Not having a Got is unsatisfactory, but how do I extract useful information?
|
|
|
|
// Maybe just read a fixed number of bytes or to the end, whichever comes first?
|
|
|
|
// I could add extra methods to cursor.RuneReader to figure out how much it had read.
|
|
|
|
}
|
|
|
|
}
|
|
|
|
// Alas, this is a little wasteful because a Regexp can only return indices
|
|
|
|
// when searching a RuneReader.
|
|
|
|
dst := make([]byte, idx[1]-idx[0])
|
|
|
|
n, _, err := input.Cursor().Read(dst)
|
|
|
|
if err != nil {
|
|
|
|
// If we can't access those same bytes again, something is wrong.
|
|
|
|
return false, gigaparsec.Result[byte, []byte]{}, fmt.Errorf("Regex: unexpected error: %w", err)
|
|
|
|
}
|
|
|
|
result = gigaparsec.Result[byte, []byte]{
|
|
|
|
State: input.At(input.Pos() + n),
|
|
|
|
Value: dst,
|
|
|
|
Message: gigaparsec.MessageOK(input.Pos()),
|
|
|
|
}
|
|
|
|
return true, result, nil
|
|
|
|
}
|
|
|
|
}
|