internal/encoding/yaml: reject trailing input in Unmarshal

To be consistent with encoding/json.Unmarshal, and so that both
pkg/encoding/json and pkg/encoding/yaml's Unmarshal APIs reject it.

Signed-off-by: Daniel Martí <mvdan@mvdan.cc>
Change-Id: Iee7dabf35d578d461c7aecd5ff6282774a51d017
Reviewed-on: https://review.gerrithub.io/c/cue-lang/cue/+/1193242
Reviewed-by: Roger Peppe <rogpeppe@gmail.com>
TryBot-Result: CUEcueckoo <cueckoo@cuelang.org>

internal/encoding/yaml/decode.go

@@ -158,15 +158,23 @@ func (d *decoder) Decode() (ast.Expr, error) {
 // Unmarshal parses a single YAML value to a CUE expression.
 func Unmarshal(filename string, data []byte) (ast.Expr, error) {
 	d := NewDecoder(filename, data)
-	x, err := d.Decode()
+	n, err := d.Decode()
 	if err != nil {
 		if err == io.EOF {
 			return nil, nil // empty input
 		}
 		return nil, err
 	}
-	// TODO(mvdan): fail if there are more documents or garbage in the input
-	return x, nil
+	// TODO(mvdan): decoding the entire next value is unnecessary;
+	// consider either a "More" or "Done" method to tell if we are at EOF,
+	// or splitting the Decode method into two variants.
+	// This should use proper error values with positions as well.
+	if n2, err := d.Decode(); err == nil {
+		return nil, fmt.Errorf("%s: expected a single YAML document", n2.Pos())
+	} else if err != io.EOF {
+		return nil, fmt.Errorf("expected a single YAML document: %v", err)
+	}
+	return n, nil
 }
 
 func (d *decoder) extract(yn *yaml.Node) (ast.Expr, error) {

internal/encoding/yaml/decode_test.go

@@ -26,13 +26,15 @@ import (
 	"strings"
 	"testing"
 
+	"github.com/go-quicktest/qt"
+	"github.com/google/go-cmp/cmp"
+
 	"cuelang.org/go/cue/ast"
 	"cuelang.org/go/cue/format"
 	"cuelang.org/go/internal"
 	"cuelang.org/go/internal/cueexperiment"
 	"cuelang.org/go/internal/cuetest"
 	"cuelang.org/go/internal/encoding/yaml"
-	"github.com/google/go-cmp/cmp"
 )
 
 // These tests are only for the new YAML decoder.
@@ -764,13 +766,6 @@ a:
 "Override anchor":   "Bar"
 "Reuse anchor":      "Bar"`,
 	},
-	// Single document with garbage following it.
-	// TODO(mvdan): This should work for a single Decoder.Decode call,
-	// but it should be an error with Unmarshal.
-	{
-		"---\nhello\n...\n}not yaml",
-		`"hello"`,
-	},
 }
 
 type M map[interface{}]interface{}
@@ -785,7 +780,7 @@ func cueStr(node ast.Node) string {
 
 func newDecoder(t *testing.T, data string) yaml.Decoder {
 	t.Helper()
-	t.Logf("input yaml:\n%s", data)
+	t.Logf("  yaml:\n%s", data)
 	return yaml.NewDecoder("test.yaml", []byte(data))
 }
 
@@ -978,3 +973,26 @@ func TestFiles(t *testing.T) {
 		})
 	}
 }
+
+func TestTrailingInput(t *testing.T) {
+	for _, input := range []string{
+		"---\nfirst\n...\n}invalid yaml",
+		"---\nfirst\n---\nsecond\n",
+	} {
+		t.Run("", func(t *testing.T) {
+			// Unmarshal should fail as it expects one value.
+			wantErr := ".*expected a single YAML document.*"
+			_, err := callUnmarshal(t, input)
+			qt.Assert(t, qt.ErrorMatches(err, wantErr))
+
+			// A single Decode call should succeed, no matter whether there is any valid or invalid trailing input.
+			wantFirst := `"first"`
+			dec := newDecoder(t, input)
+			expr, err := dec.Decode()
+			qt.Assert(t, qt.IsNil(err))
+			gotFirst := cueStr(expr)
+			qt.Assert(t, qt.Equals(gotFirst, wantFirst))
+		})
+	}
+
+}

pkg/encoding/yaml/testdata/gen.txtar

@@ -45,6 +45,10 @@ t4: error in call to encoding/yaml.ValidatePartial: invalid value 4 (out of boun
     ./in.cue:6:5
     ./in.cue:6:49
     yaml.ValidatePartial:3:4
+unmarshalTrailingInput.valid: error in call to encoding/yaml.Unmarshal: 3:1: expected a single YAML document:
+    ./in.cue:27:11
+unmarshalTrailingInput.invalid: error in call to encoding/yaml.Unmarshal: expected a single YAML document: :2: did not find expected node content:
+    ./in.cue:28:11
 
 Result:
 t1: _|_ // t1: error in call to encoding/yaml.Validate: a: invalid value 4 (out of bound <3)
@@ -81,6 +85,6 @@ unmarshalStream: {
 	null2: [1, null, 2]
 }
 unmarshalTrailingInput: {
-	valid:   1
-	invalid: 1
+	valid:   _|_ // unmarshalTrailingInput.valid: error in call to encoding/yaml.Unmarshal: 3:1: expected a single YAML document
+	invalid: _|_ // unmarshalTrailingInput.invalid: error in call to encoding/yaml.Unmarshal: expected a single YAML document: :2: did not find expected node content
 }