internal/encoding/yaml: encode YAML anchors as CUE definitions

This commits supports encoding YAML documents such as:
a: &a 3
b: *a

To this CUE document:
 #a: 3
a: #a
b: #a

Fixes #3818

Signed-off-by: Omri Steiner <[email protected]>

Author: OmriSteiner

internal/encoding/yaml/decode.go

@@ -64,6 +64,14 @@ type decoder struct {
 
 	// forceNewline ensures that the next position will be on a new line.
 	forceNewline bool
+
+	// Anchor fields which are gathered while we walk the YAML nodes,
+	// but are only added to the AST when we're done processing the whole document.
+	anchorFields []ast.Field
+	// Map from anchor nodes to their names.
+	anchorNames map[*yaml.Node]string
+	// Keeps track of anchor names that have been taken. Used to ensure unique anchor names.
+	anchorTakenNames map[string]struct{}
 }
 
 // TODO(mvdan): this can be io.Reader really, except that token.Pos is offset-based,
@@ -83,9 +91,11 @@ func NewDecoder(filename string, b []byte) *decoder {
 	tokFile := token.NewFile(filename, 0, len(b)+1)
 	tokFile.SetLinesForContent(b)
 	return &decoder{
-		tokFile:     tokFile,
-		tokLines:    append(tokFile.Lines(), len(b)),
-		yamlDecoder: *yaml.NewDecoder(bytes.NewReader(b)),
+		tokFile:          tokFile,
+		tokLines:         append(tokFile.Lines(), len(b)),
+		yamlDecoder:      *yaml.NewDecoder(bytes.NewReader(b)),
+		anchorNames:      make(map[*yaml.Node]string),
+		anchorTakenNames: make(map[string]struct{}),
 	}
 }
 
@@ -176,24 +186,35 @@ func Unmarshal(filename string, data []byte) (ast.Expr, error) {
 	return n, nil
 }
 
-func (d *decoder) extract(yn *yaml.Node) (ast.Expr, error) {
-	d.addHeadCommentsToPending(yn)
-	var expr ast.Expr
-	var err error
+func (d *decoder) extract_no_anchor(yn *yaml.Node) (ast.Expr, error) {
 	switch yn.Kind {
 	case yaml.DocumentNode:
-		expr, err = d.document(yn)
+		return d.document(yn)
 	case yaml.SequenceNode:
-		expr, err = d.sequence(yn)
+		return d.sequence(yn)
 	case yaml.MappingNode:
-		expr, err = d.mapping(yn)
+		return d.mapping(yn)
 	case yaml.ScalarNode:
-		expr, err = d.scalar(yn)
+		return d.scalar(yn)
 	case yaml.AliasNode:
-		expr, err = d.alias(yn)
+		return d.reference_alias(yn)
 	default:
 		return nil, d.posErrorf(yn, "unknown yaml node kind: %d", yn.Kind)
 	}
+}
+
+func (d *decoder) extract(yn *yaml.Node) (ast.Expr, error) {
+	d.addHeadCommentsToPending(yn)
+
+	var expr ast.Expr
+	var err error
+
+	if yn.Anchor == "" {
+		expr, err = d.extract_no_anchor(yn)
+	} else {
+		expr, err = d.anchor(yn)
+	}
+
 	if err != nil {
 		return nil, err
 	}
@@ -324,7 +345,39 @@ func (d *decoder) document(yn *yaml.Node) (ast.Expr, error) {
 	if n := len(yn.Content); n != 1 {
 		return nil, d.posErrorf(yn, "yaml document nodes are meant to have one content node but have %d", n)
 	}
-	return d.extract(yn.Content[0])
+
+	expr, err := d.extract(yn.Content[0])
+	if err != nil {
+		return nil, err
+	}
+
+	return d.addAnchorNodes(expr)
+}
+
+// Adds anchors nodes at the top of the document.
+func (d *decoder) addAnchorNodes(expr ast.Expr) (ast.Expr, error) {
+	elements := []ast.Decl{}
+
+	for _, field := range d.anchorFields {
+		elements = append(elements, &field)
+	}
+
+	switch x := expr.(type) {
+	case *ast.StructLit:
+		x.Elts = append(elements, x.Elts...)
+		break
+	case *ast.ListLit:
+		expr = &ast.StructLit{
+			Elts: append(elements, x),
+		}
+		break
+	default:
+		// If the whole YAML document is not a map / seq, then it can't have anchors.
+		// maybe assert that `anchorNodes` is empty?
+		break
+	}
+
+	return expr, nil
 }
 
 func (d *decoder) sequence(yn *yaml.Node) (ast.Expr, error) {
@@ -458,7 +511,7 @@ func (d *decoder) label(yn *yaml.Node) (ast.Label, error) {
 		if yn.Alias.Kind != yaml.ScalarNode {
 			return nil, d.posErrorf(yn, "invalid map key: %v", yn.Alias.ShortTag())
 		}
-		expr, err = d.alias(yn)
+		expr, err = d.inline_alias(yn)
 		value = yn.Alias.Value
 	default:
 		return nil, d.posErrorf(yn, "invalid map key: %v", yn.ShortTag())
@@ -639,7 +692,10 @@ func (d *decoder) makeNum(yn *yaml.Node, val string, kind token.Token) (expr ast
 	return expr
 }
 
-func (d *decoder) alias(yn *yaml.Node) (ast.Expr, error) {
+// Expands an alias node in place, returning the expanded node.
+// Sometimes we have to resort to this, for example when the alias
+// is inside a map key, since CUE does not support structs as map keys.
+func (d *decoder) inline_alias(yn *yaml.Node) (ast.Expr, error) {
 	if d.extractingAliases[yn] {
 		// TODO this could actually be allowed in some circumstances.
 		return nil, d.posErrorf(yn, "anchor %q value contains itself", yn.Value)
@@ -649,11 +705,58 @@ func (d *decoder) alias(yn *yaml.Node) (ast.Expr, error) {
 	}
 	d.extractingAliases[yn] = true
 	var node ast.Expr
-	node, err := d.extract(yn.Alias)
+	node, err := d.extract_no_anchor(yn.Alias)
 	delete(d.extractingAliases, yn)
 	return node, err
 }
 
+// Replace an alias with a reference to the identifier of its anchor.
+func (d *decoder) reference_alias(yn *yaml.Node) (ast.Expr, error) {
+	anchor, ok := d.anchorNames[yn.Alias]
+	if !ok {
+		return nil, d.posErrorf(yn, "anchor %q not found", yn.Alias.Anchor)
+	}
+
+	return &ast.Ident{
+		NamePos: d.pos(yn),
+		Name:    anchor,
+	}, nil
+}
+
+func (d *decoder) anchor(yn *yaml.Node) (ast.Expr, error) {
+	var anchorIdent string
+
+	// Pick a non-conflicting anchor name.
+	for i := 1; ; i++ {
+		if i == 1 {
+			anchorIdent = "#" + yn.Anchor
+		} else {
+			anchorIdent = "#" + yn.Anchor + "_" + strconv.Itoa(i)
+		}
+		if _, ok := d.anchorTakenNames[anchorIdent]; !ok {
+			d.anchorTakenNames[anchorIdent] = struct{}{}
+			break
+		}
+	}
+	d.anchorNames[yn] = anchorIdent
+
+	// Process the node itself, but don't put it into the AST just yet,
+	// store it for later to be used as an anchor identifier.
+	expr, err := d.extract_no_anchor(yn)
+	if err != nil {
+		return nil, err
+	}
+	d.anchorFields = append(d.anchorFields, ast.Field{
+		Label: &ast.Ident{Name: anchorIdent},
+		Value: expr,
+	})
+
+	return &ast.Ident{
+		NamePos: d.pos(yn),
+		Name:    anchorIdent,
+	}, nil
+}
+
 func labelStr(l ast.Label) string {
 	switch l := l.(type) {
 	case *ast.Ident:

internal/encoding/yaml/decode_test.go

@@ -474,25 +474,51 @@ Null: 1
 	// Anchors and aliases.
 	{
 		"a: &x 1\nb: &y 2\nc: *x\nd: *y\n",
-		`a: 1
-b: 2
-c: 1
-d: 2`,
+		`#x: 1
+#y: 2
+a:  #x
+b:  #y
+c:  #x
+d:  #y`,
 	}, {
 		"a: &a {c: 1}\nb: *a",
-		`a: {c: 1}
-b: {
-	c: 1
-}`,
+		`#a: {c: 1}
+a: #a
+b: #a`,
 	}, {
 		"a: &a [1, 2]\nb: *a",
-		"a: [1, 2]\nb: [1, 2]", // TODO: a: [1, 2], b: a
+		"#a: [1, 2]\na: #a\nb: #a",
 	},
 	{
 		`a: &a "b"
 *a : "c"`,
-		`a: "b"
-b: "c"`,
+		`#a: "b"
+a:  #a
+b:  "c"`,
+	},
+	// Test nested anchors
+	{
+		`foo: &a
+  bar: &b
+    baz: 1
+a: *a
+b: *b
+`,
+		`#b: {
+	baz: 1
+}
+#a: {
+	bar: #b
+}
+foo: #a
+a:   #a
+b:   #b`,
+	},
+	{
+		`a:
+  - &b c`,
+		`#b: "c"
+a: [#b]`,
 	},
 
 	{
@@ -778,10 +804,12 @@ a:
 	// yaml-test-suite 3GZX: Spec Example 7.1. Alias Nodes
 	{
 		"First occurrence: &anchor Foo\nSecond occurrence: *anchor\nOverride anchor: &anchor Bar\nReuse anchor: *anchor\n",
-		`"First occurrence":  "Foo"
-"Second occurrence": "Foo"
-"Override anchor":   "Bar"
-"Reuse anchor":      "Bar"`,
+		`#anchor:             "Foo"
+#anchor_2:           "Bar"
+"First occurrence":  #anchor
+"Second occurrence": #anchor
+"Override anchor":   #anchor_2
+"Reuse anchor":      #anchor_2`,
 	},
 }
 

internal/encoding/yaml/testdata/merge.out

@@ -1,15 +1,21 @@
+#CENTER: {
+	x: 1, y: 2
+}
+#LEFT: {
+	x: 0, y: 2
+}
+#BIG: {
+	r: 10
+}
+#SMALL: {
+	r: 1
+}
+
 // From http://yaml.org/type/merge.html
 // Test
 anchors: {
-	list: [{
-		x: 1, y: 2
-	}, {
-		x: 0, y: 2
-	}, {
-		r: 10
-	}, {
-		r: 1
-	}]
+	list: [#CENTER, #LEFT, #BIG, #SMALL,
+	]
 }
 
 // All the following maps are equal: