Added cursor tests

cursor/cursor.go

@@ -17,8 +17,9 @@ type Cursor[Datum any] interface {
 	// source. It returns the number of data it read and a new Cursor for
 	// the position at which the read ended, or an error if the read failed.
 	// All calls to a given Cursor will return data from the same position.
-	// If n < len(dst), Read will return an error explaining why it read fewer
-	// bytes than requested. If Read tried to read past the end of the source,
+	// If n < len(dst) or if the cursor's position is at the end of the data source,
+	// Read will return an error explaining why it read fewer bytes than requested.
+	// If the error was due to the cursor reaching the end of the data source,
 	// err will be io.EOF.
 	Read(dst []Datum) (n uint64, next Cursor[Datum], err error)
 
@@ -31,9 +32,17 @@ type SliceCursor[Datum any] struct {
 	offset uint64
 }
 
-func NewSlice[Datum any]([]Datum) SliceCursor[Datum] { panic("not implemented") }
+func NewSlice[Datum any](data []Datum) SliceCursor[Datum] {
+	return SliceCursor[Datum]{
+		data:   data,
+		offset: 0,
+	}
+}
 
 func (sc SliceCursor[Datum]) Read(dst []Datum) (n uint64, next Cursor[Datum], err error) {
+	if sc.offset == uint64(len(sc.data)) {
+		return 0, sc, io.EOF
+	}
 	copied := copy(dst, sc.data[sc.offset:])
 	if copied < len(dst) {
 		err = io.EOF
@@ -81,6 +90,9 @@ func NewString(s string) StringCursor {
 }
 
 func (sc StringCursor) Read(dst []byte) (n uint64, next Cursor[byte], err error) {
+	if sc.offset == uint64(len(sc.source)) {
+		return 0, sc, io.EOF
+	}
 	copied := copy(dst, sc.source[sc.offset:])
 	if copied < len(dst) {
 		err = io.EOF

cursor/cursor_test.go

@@ -0,0 +1,92 @@
+package cursor_test
+
+import (
+	"bytes"
+	"io"
+	"testing"
+
+	"git.codemonkeysoftware.net/b/gigaparsec/cursor"
+	"github.com/shoenig/test"
+	"github.com/shoenig/test/must"
+	"pgregory.net/rapid"
+)
+
+func Todo(t *testing.T) {
+	t.Errorf("TODO")
+}
+
+func SliceOfNZero[T any](minLen, maxLen int) *rapid.Generator[[]T] {
+	return rapid.Map(rapid.IntRange(minLen, maxLen), func(n int) []T {
+		return make([]T, n)
+	})
+}
+
+func testCursor[C cursor.Cursor[byte]](t *testing.T, makeCursor func([]byte) C) {
+	t.Helper()
+	t.Run("cursor reads the same position every time", rapid.MakeCheck(func(t *rapid.T) {
+		data := rapid.SliceOfN(rapid.Byte(), 1, 100).Draw(t, "data")
+		dst := SliceOfNZero[byte](0, len(data)-1).Draw(t, "dst")
+		expected := data[:len(dst)]
+		c := makeCursor(data)
+
+		_, next, err := c.Read(dst)
+		must.NoError(t, err)
+		must.SliceEqOp(t, expected, dst)
+
+		next.Read(dst)
+		_, _, err = c.Read(dst)
+		must.NoError(t, err)
+		must.SliceEqOp(t, expected, dst)
+	}))
+	t.Run("Read returns io.EOF iff it overruns source", rapid.MakeCheck(func(t *rapid.T) {
+		data := rapid.SliceOfN(rapid.Byte(), 0, 100).Draw(t, "data")
+		dst := SliceOfNZero[byte](0, 200).Draw(t, "dst")
+		c := makeCursor(data)
+
+		n, _, err := c.Read(dst)
+		t.Logf("n=%d", n)
+		must.EqOp(t, min(len(data), len(dst)), int(n))
+		if len(dst) > len(data) || c.Pos() == uint64(len(data)) {
+			must.ErrorIs(t, err, io.EOF)
+		} else {
+			must.NoError(t, err)
+		}
+	}))
+	t.Run("next cursor reads next input", rapid.MakeCheck(func(t *rapid.T) {
+		const maxLen = 100
+		data := rapid.SliceOfN(rapid.Byte(), 1, maxLen).Draw(t, "data")
+		skip := rapid.IntRange(0, len(data)-1).Draw(t, "data")
+		c := makeCursor(data)
+
+		_, next, err := c.Read(make([]byte, skip))
+		must.NoError(t, err)
+		must.EqOp(t, skip, int(next.Pos()))
+		dst := make([]byte, maxLen)
+		n, _, _ := next.Read(dst)
+		must.SliceEqOp(t, data[skip:skip+int(n)], dst[:n])
+	}))
+	t.Run("Read returns an error if n is less than requested", rapid.MakeCheck(func(t *rapid.T) {
+		data := rapid.SliceOfN(rapid.Byte(), 0, 100).Draw(t, "data")
+		c := makeCursor(data)
+
+		n, _, err := c.Read(make([]byte, len(data)+1))
+		test.ErrorIs(t, err, io.EOF)
+		test.EqOp(t, len(data), int(n))
+	}))
+}
+
+func TestSliceCursor(t *testing.T) {
+	testCursor(t, cursor.NewSlice[byte])
+}
+
+func TestStringCursor(t *testing.T) {
+	testCursor(t, func(b []byte) cursor.StringCursor {
+		return cursor.NewString(string(b))
+	})
+}
+
+func TestReaderAtCursor(t *testing.T) {
+	testCursor(t, func(b []byte) cursor.ReaderAtCursor {
+		return cursor.NewReaderAt(bytes.NewReader(b))
+	})
+}

cursor/helper.go

@@ -6,7 +6,9 @@ import "io"
 // with limited backward seeking.
 type BufferedReaderAt struct{}
 
-func NewBufferedReaderAt(r io.Reader, minBuffer uint64) *BufferedReaderAt
+func NewBufferedReaderAt(r io.Reader, minBuffer uint64) *BufferedReaderAt {
+	return nil
+}
 
 // ReadAt reads bytes from the underlying reader. If the offset is after
 // the end of the buffer, ReadAt will first read and ignore bytes from the
@@ -15,7 +17,9 @@ func NewBufferedReaderAt(r io.Reader, minBuffer uint64) *BufferedReaderAt
 //
 // If your parser needs unlimited lookahead, you should probably
 // just read the whole input into a slice and use BytesCursor.
-func (b *BufferedReaderAt) ReadAt(dst []byte, offset int64) (int, error)
+func (b *BufferedReaderAt) ReadAt(dst []byte, offset int64) (int, error) {
+	return 0, nil
+}
 
 // RuneReader is an io.RuneReader backed by a Cursor, for compatibility
 // with the regexp package.

go.mod

@@ -1,3 +1,12 @@
 module git.codemonkeysoftware.net/b/gigaparsec
 
 go 1.23
+
+require (
+	github.com/shoenig/test v1.9.1
+	pgregory.net/rapid v1.1.0
+)
+
+require github.com/google/go-cmp v0.6.0 // indirect
+
+replace github.com/shoenig/test v1.9.1 => ../shoenig-test

go.sum

@@ -0,0 +1,4 @@
+github.com/google/go-cmp v0.6.0 h1:ofyhxvXcZhMsU5ulbFiLKl/XBFqE1GSq7atu8tAmTRI=
+github.com/google/go-cmp v0.6.0/go.mod h1:17dUlkBOakJ0+DkrSSNjCkIjxS6bF9zb3elmeNGIjoY=
+pgregory.net/rapid v1.1.0 h1:CMa0sjHSru3puNx+J0MIAuiiEV4N0qj8/cMWGBBCsjw=
+pgregory.net/rapid v1.1.0/go.mod h1:PY5XlDGj0+V1FCq0o192FdRhpKHGTRIWBgqjDBTrq04=