fix: compare enum entries by JSON value

utils/enum.go

@@ -0,0 +1,131 @@
+package utils
+
+import (
+	"math/big"
+	"sort"
+	"strconv"
+	"strings"
+
+	"github.com/pb33f/go-yaml"
+)
+
+type enumValueIndex struct {
+	values map[string]int
+	nodes  map[*yaml.Node]int
+	active map[*yaml.Node]bool
+}
+
+// CheckEnumForDuplicates returns each occurrence after the first of an equal
+// JSON value, preserving input order and node identity. Numbers compare by value,
+// arrays by ordered elements, and objects by members irrespective of key order.
+// YAML aliases are followed; cyclic or non-JSON values are ignored.
+func CheckEnumForDuplicates(seq []*yaml.Node) []*yaml.Node {
+	var duplicates []*yaml.Node
+	seen := make(map[int]bool)
+	index := enumValueIndex{
+		values: make(map[string]int),
+		nodes:  make(map[*yaml.Node]int),
+		active: make(map[*yaml.Node]bool),
+	}
+	for _, node := range seq {
+		id := index.valueID(node)
+		if id == 0 {
+			continue
+		}
+		if seen[id] {
+			duplicates = append(duplicates, node)
+		}
+		seen[id] = true
+	}
+	return duplicates
+}
+
+// valueID interns each value using its children's IDs, keeping repeated aliases
+// compact. Zero denotes a value outside the JSON data model.
+func (index *enumValueIndex) valueID(node *yaml.Node) int {
+	if node == nil || index.active[node] {
+		return 0
+	}
+	if id, ok := index.nodes[node]; ok {
+		return id
+	}
+	index.active[node] = true
+	defer delete(index.active, node)
+	var id int
+	if node.Kind == yaml.AliasNode {
+		id = index.valueID(node.Alias)
+	} else if key, ok := index.valueKey(node); ok {
+		id = index.values[key]
+		if id == 0 {
+			id = len(index.values) + 1
+			index.values[key] = id
+		}
+	}
+	index.nodes[node] = id
+	return id
+}
+
+func (index *enumValueIndex) valueKey(node *yaml.Node) (string, bool) {
+	switch node.Kind {
+	case yaml.ScalarNode:
+		switch node.ShortTag() {
+		case "!!str":
+			return "s" + strconv.Quote(node.Value), true
+		case "!!null":
+			return "null", true
+		case "!!bool":
+			var value bool
+			if err := node.Decode(&value); err != nil {
+				return "", false
+			}
+			return strconv.FormatBool(value), true
+		case "!!int":
+			// YAML accepts base-prefixed integers and legacy octal (012 == 10).
+			value, ok := new(big.Int).SetString(node.Value, 0)
+			if !ok {
+				return "", false
+			}
+			return "n" + value.String(), true
+		case "!!float":
+			value, ok := new(big.Rat).SetString(node.Value)
+			if !ok {
+				return "", false
+			}
+			return "n" + value.RatString(), true
+		}
+	case yaml.SequenceNode:
+		if node.ShortTag() != "!!seq" {
+			return "", false
+		}
+		parts := make([]string, 0, len(node.Content))
+		for _, child := range node.Content {
+			id := index.valueID(child)
+			if id == 0 {
+				return "", false
+			}
+			parts = append(parts, strconv.Itoa(id))
+		}
+		return "[" + strings.Join(parts, ",") + "]", true
+	case yaml.MappingNode:
+		if node.ShortTag() != "!!map" || len(node.Content)%2 != 0 {
+			return "", false
+		}
+		parts := make([]string, 0, len(node.Content)/2)
+		keys := make(map[string]bool)
+		for i := 0; i < len(node.Content); i += 2 {
+			name := node.Content[i]
+			if name == nil || name.Kind != yaml.ScalarNode || name.ShortTag() != "!!str" || keys[name.Value] {
+				return "", false
+			}
+			keys[name.Value] = true
+			id := index.valueID(node.Content[i+1])
+			if id == 0 {
+				return "", false
+			}
+			parts = append(parts, strconv.Quote(name.Value)+":"+strconv.Itoa(id))
+		}
+		sort.Strings(parts)
+		return "{" + strings.Join(parts, ",") + "}", true
+	}
+	return "", false
+}

utils/enum_test.go

@@ -0,0 +1,87 @@
+package utils
+
+import (
+	"testing"
+
+	"github.com/pb33f/go-yaml"
+	"github.com/pb33f/testify/require"
+)
+
+func TestCheckEnumForDuplicates_Values(t *testing.T) {
+	for _, tc := range []struct {
+		name       string
+		yaml       string
+		duplicates []int
+	}{
+		{"distinct arrays", `[[a], [b], [a, b], [b, a], []]`, nil},
+		{"duplicate arrays", `[[a], [b], [a], [a]]`, []int{2, 3}},
+		{"distinct objects", `[{a: 1}, {a: 2}, {b: 1}, {}]`, nil},
+		{"object order", `[{a: 1, b: [2, 3]}, {b: [2.0, 3], a: 1.0}]`, []int{1}},
+		{"nested types", `[{a: [1]}, {a: ["1"]}]`, nil},
+		{"scalar types", `[1, "1", true, "true", null, "null", "", [], {}]`, nil},
+		{"octal integers", `[012, 10, 12, -012, -10, 0o12, 10.0]`, []int{1, 4, 5, 6}},
+		{"numbers", `[1, 1.0, 1e0, -0, 0.0, 0, 0.1, 1e-1]`, []int{1, 2, 4, 5, 7}},
+		{"precision", `[9007199254740992, 9007199254740993, 9007199254740993.0]`, []int{2}},
+		{"yaml scalars", `[true, TRUE, null, ~, 0x10, 16]`, []int{1, 3, 5}},
+		{"aliases", `[&a {x: [1]}, *a, {x: [1.0]}]`, []int{1, 2}},
+		{"repeated aliases", `[&a [1], [*a, *a], [[1], [1]]]`, []int{2}},
+		{"delimiters", `[["a,b", c], [a, "b,c"], {"a:b": c}, {a: "b:c"}]`, nil},
+	} {
+		t.Run(tc.name, func(t *testing.T) {
+			var root yaml.Node
+			require.NoError(t, yaml.Unmarshal([]byte(tc.yaml), &root))
+			values := root.Content[0].Content
+			duplicates := CheckEnumForDuplicates(values)
+			require.Len(t, duplicates, len(tc.duplicates))
+			for i, index := range tc.duplicates {
+				require.Same(t, values[index], duplicates[i])
+			}
+		})
+	}
+}
+
+func TestCheckEnumForDuplicates_NonJSONValues(t *testing.T) {
+	for _, source := range []string{
+		`[&a [*a], &b [*b]]`,
+		`[{1: a}, {1: a}]`,
+		`[{a: 1, a: 2}, {a: 1, a: 2}]`,
+		`[.inf, .inf, .nan, .nan]`,
+		`[!custom a, !custom a]`,
+		`[!custom [], []]`,
+		`[!custom {}, {}]`,
+		`[{x: !custom a}, {x: !custom a}]`,
+	} {
+		t.Run(source, func(t *testing.T) {
+			var root yaml.Node
+			require.NoError(t, yaml.Unmarshal([]byte(source), &root))
+			require.Empty(t, CheckEnumForDuplicates(root.Content[0].Content))
+		})
+	}
+	for _, node := range []*yaml.Node{
+		nil,
+		{Kind: yaml.AliasNode},
+		{Kind: yaml.MappingNode, Content: []*yaml.Node{nil}},
+		{Kind: yaml.MappingNode, Content: []*yaml.Node{nil, nil}},
+		{Kind: yaml.SequenceNode, Content: []*yaml.Node{nil}},
+		{Kind: yaml.ScalarNode, Tag: "!!bool", Value: "invalid"},
+		{Kind: yaml.ScalarNode, Tag: "!!int", Value: "invalid"},
+		{Kind: yaml.DocumentNode},
+	} {
+		require.Empty(t, CheckEnumForDuplicates([]*yaml.Node{node, node}))
+	}
+}
+
+func TestCheckEnumForDuplicates_SharedAliases(t *testing.T) {
+	value := &yaml.Node{Kind: yaml.ScalarNode, Tag: "!!int", Value: "1"}
+	// The graph is linear in size; expanding its aliases would create 2^128 leaves.
+	for range 128 {
+		value = &yaml.Node{Kind: yaml.SequenceNode, Content: []*yaml.Node{
+			{Kind: yaml.AliasNode, Alias: value},
+			{Kind: yaml.AliasNode, Alias: value},
+		}}
+	}
+	alias := &yaml.Node{Kind: yaml.AliasNode, Alias: value}
+	duplicates := CheckEnumForDuplicates([]*yaml.Node{value, alias})
+	require.Len(t, duplicates, 1)
+	require.Same(t, alias, duplicates[0])
+}

utils/utils.go

@@ -1218,21 +1218,6 @@ func DetectCase(input string) Case {
 	return RegularCase
 }
 
-// CheckEnumForDuplicates will check an array of nodes to check if there are any duplicate values.
-func CheckEnumForDuplicates(seq []*yaml.Node) []*yaml.Node {
-	var res []*yaml.Node
-	seen := make(map[string]*yaml.Node)
-
-	for _, enum := range seq {
-		if seen[enum.Value] != nil {
-			res = append(res, enum)
-			continue
-		}
-		seen[enum.Value] = enum
-	}
-	return res
-}
-
 // DetermineWhitespaceLengthBytes determines the minimum leading-space indentation
 // in the input, working directly on []byte without allocating strings or regex matches.
 // Matches the semantics of the regex `\n( +)`: only considers lines after a newline.