From 96d178efefbe2357725d5b1fcb9649f4ec56487e Mon Sep 17 00:00:00 2001
From: Brandon Dyck <brandon@dyck.us>
Date: Tue, 10 Sep 2024 16:46:31 -0600
Subject: [PATCH] Implement most of Regexp parser

---
 bytes/bytes.go | 46 ++++++++++++++++++++++++++++++++++++++++++++++
 gigaparsec.go  |  4 ++++
 parser_test.go |  4 ++++
 3 files changed, 54 insertions(+)
 create mode 100644 bytes/bytes.go

diff --git a/bytes/bytes.go b/bytes/bytes.go
new file mode 100644
index 0000000..e810345
--- /dev/null
+++ b/bytes/bytes.go
@@ -0,0 +1,46 @@
+package bytes
+
+import (
+	"fmt"
+	"regexp"
+	"strings"
+
+	"git.codemonkeysoftware.net/b/gigaparsec"
+	"git.codemonkeysoftware.net/b/gigaparsec/cursor"
+)
+
+func Regexp(str string) gigaparsec.Parser[byte, []byte] {
+	if !strings.HasPrefix(str, "^") {
+		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) {
+		r := cursor.NewRuneReader(input.Cursor())
+		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
+	}
+}
diff --git a/gigaparsec.go b/gigaparsec.go
index f13e9bd..1d2eaf8 100644
--- a/gigaparsec.go
+++ b/gigaparsec.go
@@ -57,6 +57,10 @@ func (s State[In]) Pos() uint64 {
 	return s.cursor.Pos()
 }
 
+func (s State[In]) At(pos uint64) State[In] {
+	return State[In]{cursor: s.cursor.At(pos)}
+}
+
 type Parser[In, Out any] func(State[In]) (consumed bool, reply Result[In, Out], err error)
 
 func Return[In, Out any](value Out) Parser[In, Out] {
diff --git a/parser_test.go b/parser_test.go
index a56ede5..d1de26a 100644
--- a/parser_test.go
+++ b/parser_test.go
@@ -62,3 +62,7 @@ func TestSlice(t *testing.T) {
 		test.EqOp(t, uint64(len(s)), result.State.Pos())
 	}))
 }
+
+func TestChoose(t *testing.T) {
+	Todo(t)
+}