Add bounded sources and strict YAML parsing

This commit is contained in:
2026-09-06 19:30:31 +00:00
parent 14b10cb6a5
commit 7c78da9cd2
17 changed files with 1287 additions and 44 deletions

View File

@@ -0,0 +1,62 @@
// Package document defines ConfigKit's parser-independent configuration tree.
package document
// Kind identifies a configuration node category.
type Kind uint8
const (
Mapping Kind = iota + 1
Sequence
String
Boolean
Number
Null
)
// Location identifies a source coordinate. Root depth and unavailable coordinates
// are represented independently by callers; line and column are one-based when set.
type Location struct {
SourceID string
Line int
Column int
}
// Entry is an ordered mapping entry with separate key and value locations.
type Entry struct {
Key string
KeyLocation Location
ValueLocation Location
Value *Node
}
// Node is ConfigKit's owned configuration representation. Number Text retains its
// exact validated decimal spelling; Boolean Text is "true" or "false".
type Node struct {
Kind Kind
Location Location
Text string
Mapping []Entry
Sequence []*Node
}
// Clone returns a detached copy of a node tree.
func Clone(node *Node) *Node {
if node == nil {
return nil
}
cloned := *node
if node.Mapping != nil {
cloned.Mapping = make([]Entry, len(node.Mapping))
for i, entry := range node.Mapping {
cloned.Mapping[i] = entry
cloned.Mapping[i].Value = Clone(entry.Value)
}
}
if node.Sequence != nil {
cloned.Sequence = make([]*Node, len(node.Sequence))
for i, child := range node.Sequence {
cloned.Sequence[i] = Clone(child)
}
}
return &cloned
}

274
internal/yaml/parse.go Normal file
View File

@@ -0,0 +1,274 @@
// Package yaml adapts the private upstream parser to ConfigKit's document model.
package yaml
import (
"errors"
"io"
"regexp"
"strings"
"go.yaml.in/yaml/v4"
"go.yaml.in/yaml/v4/plugin/limit"
"gitea.maximumdirect.net/eric/ConfigKit/internal/document"
)
const (
codeInvalid = "yaml.invalid"
codeEmpty = "yaml.empty"
codeMultipleDocuments = "yaml.multiple_documents"
codeRootType = "yaml.root_type"
codeDuplicateKey = "yaml.duplicate_key"
codeTag = "yaml.tag"
codeAnchor = "yaml.anchor"
codeAlias = "yaml.alias"
codeMergeKey = "yaml.merge_key"
codeScalar = "yaml.scalar"
codeDepthLimit = "yaml.depth_limit"
codeNodeLimit = "yaml.node_limit"
)
var (
jsonNumberPattern = regexp.MustCompile(`^-?(0|[1-9][0-9]*)(\.[0-9]+)?([eE][+-]?[0-9]+)?$`)
decimalLikePattern = regexp.MustCompile(`^[+-]?[0-9][0-9_]*(\.[0-9_]*)?([eE][+-]?[0-9_]*)?$`)
prefixedNumberPattern = regexp.MustCompile(`^[+-]?0[xXoO][0-9A-Fa-f_]+$`)
leadingDecimalPattern = regexp.MustCompile(`^[+-]?\.[0-9_]+$`)
nonFiniteNumberPattern = regexp.MustCompile(`(?i)^[+-]?\.(inf|nan)$`)
errParserDepth = errors.New("configkit yaml parser depth limit")
)
// Limits controls private parser and translated-tree resource bounds.
type Limits struct {
Depth int
Nodes int
ParserDepth int
}
// Issue is a parser-independent, value-free YAML diagnostic.
type Issue struct {
Code string
Message string
Line int
Column int
Related []document.Location
}
// Parse loads exactly one policy-valid mapping document.
func Parse(input io.Reader, sourceID string, limits Limits) (*document.Node, []Issue) {
parserDepth := limits.ParserDepth
if parserDepth <= 0 {
parserDepth = limits.Depth + 2
}
depthExceeded := false
plugin := limit.New(limit.DepthFunc(func(depth int, _ *limit.DepthContext) error {
if depth > parserDepth {
depthExceeded = true
return errParserDepth
}
return nil
}))
loader, err := yaml.NewLoader(input, yaml.WithPlugin(plugin))
if err != nil {
return nil, []Issue{{Code: codeInvalid, Message: "YAML parser could not be initialized"}}
}
var parsed yaml.Node
if err := loader.Load(&parsed); err != nil {
if errors.Is(err, io.EOF) {
return nil, []Issue{{Code: codeEmpty, Message: "configuration input is empty"}}
}
return nil, []Issue{parserIssue(err, depthExceeded)}
}
var trailing yaml.Node
if err := loader.Load(&trailing); err == nil {
return nil, []Issue{{Code: codeMultipleDocuments, Message: "configuration input must contain exactly one document", Line: trailing.Line, Column: trailing.Column}}
} else if !errors.Is(err, io.EOF) {
return nil, []Issue{parserIssue(err, depthExceeded)}
}
root := documentRoot(&parsed)
if root == nil || (root.Kind == yaml.ScalarNode && root.Value == "" && root.ShortTag() == "!!null") {
return nil, []Issue{{Code: codeEmpty, Message: "configuration document is empty", Line: parsed.Line, Column: parsed.Column}}
}
if root.Kind != yaml.MappingNode {
return nil, []Issue{{Code: codeRootType, Message: "configuration document root must be a mapping", Line: root.Line, Column: root.Column}}
}
state := translationState{sourceID: sourceID, limits: limits}
translated, issue := state.translate(root, 0)
if issue != nil {
return nil, []Issue{*issue}
}
return translated, nil
}
type translationState struct {
sourceID string
limits Limits
nodes int
}
func (s *translationState) translate(node *yaml.Node, depth int) (*document.Node, *Issue) {
if depth > s.limits.Depth {
return nil, nodeIssue(codeDepthLimit, "configuration exceeds the nesting depth limit", node)
}
if node.Style&yaml.TaggedStyle != 0 {
return nil, nodeIssue(codeTag, "explicit YAML tags are not supported", node)
}
if node.Anchor != "" {
return nil, nodeIssue(codeAnchor, "YAML anchors are not supported", node)
}
if node.Kind == yaml.AliasNode || node.Alias != nil {
return nil, nodeIssue(codeAlias, "YAML aliases are not supported", node)
}
if issue := s.chargeNode(node); issue != nil {
return nil, issue
}
location := document.Location{SourceID: s.sourceID, Line: node.Line, Column: node.Column}
switch node.Kind {
case yaml.MappingNode:
if len(node.Content)%2 != 0 {
return nil, nodeIssue(codeInvalid, "mapping structure is invalid", node)
}
result := &document.Node{Kind: document.Mapping, Location: location, Mapping: make([]document.Entry, 0, len(node.Content)/2)}
seen := make(map[string]document.Location, len(node.Content)/2)
for i := 0; i < len(node.Content); i += 2 {
keyNode, valueNode := node.Content[i], node.Content[i+1]
if issue := s.chargeNode(keyNode); issue != nil {
return nil, issue
}
key, issue := mappingKey(keyNode)
if issue != nil {
return nil, issue
}
if key == "<<" {
return nil, nodeIssue(codeMergeKey, "YAML merge keys are not supported", keyNode)
}
keyLocation := document.Location{SourceID: s.sourceID, Line: keyNode.Line, Column: keyNode.Column}
if previous, ok := seen[key]; ok {
issue := nodeIssue(codeDuplicateKey, "mapping key is defined more than once", keyNode)
issue.Related = []document.Location{previous}
return nil, issue
}
seen[key] = keyLocation
value, issue := s.translate(valueNode, depth+1)
if issue != nil {
return nil, issue
}
result.Mapping = append(result.Mapping, document.Entry{
Key: key,
KeyLocation: keyLocation,
ValueLocation: value.Location,
Value: value,
})
}
return result, nil
case yaml.SequenceNode:
result := &document.Node{Kind: document.Sequence, Location: location, Sequence: make([]*document.Node, 0, len(node.Content))}
for _, child := range node.Content {
translated, issue := s.translate(child, depth+1)
if issue != nil {
return nil, issue
}
result.Sequence = append(result.Sequence, translated)
}
return result, nil
case yaml.ScalarNode:
kind, text, issue := scalar(node)
if issue != nil {
return nil, issue
}
return &document.Node{Kind: kind, Location: location, Text: text}, nil
default:
return nil, nodeIssue(codeInvalid, "YAML node kind is not supported", node)
}
}
func (s *translationState) chargeNode(node *yaml.Node) *Issue {
s.nodes++
if s.limits.Nodes >= 0 && s.nodes > s.limits.Nodes {
return nodeIssue(codeNodeLimit, "configuration exceeds the node limit", node)
}
return nil
}
func mappingKey(node *yaml.Node) (string, *Issue) {
if node.Kind == yaml.AliasNode || node.Alias != nil {
return "", nodeIssue(codeAlias, "YAML aliases are not supported", node)
}
if node.Kind != yaml.ScalarNode {
return "", nodeIssue(codeScalar, "mapping keys must be strings", node)
}
if node.Style&yaml.TaggedStyle != 0 {
return "", nodeIssue(codeTag, "explicit YAML tags are not supported", node)
}
if node.Anchor != "" {
return "", nodeIssue(codeAnchor, "YAML anchors are not supported", node)
}
kind, text, issue := scalar(node)
if issue != nil {
return "", issue
}
if kind != document.String {
return "", nodeIssue(codeScalar, "mapping keys must be strings", node)
}
return text, nil
}
func scalar(node *yaml.Node) (document.Kind, string, *Issue) {
if node.Style&(yaml.DoubleQuotedStyle|yaml.SingleQuotedStyle|yaml.LiteralStyle|yaml.FoldedStyle) != 0 {
return document.String, node.Value, nil
}
value := node.Value
switch value {
case "":
if node.ShortTag() == "!!null" {
return document.Null, "", nil
}
case "null", "~":
return document.Null, "", nil
case "true", "false":
return document.Boolean, value, nil
}
lower := strings.ToLower(value)
if lower == "true" || lower == "false" || lower == "null" || lower == "yes" || lower == "no" || lower == "on" || lower == "off" {
return 0, "", nodeIssue(codeScalar, "plain scalar uses an unsupported reserved form", node)
}
if jsonNumberPattern.MatchString(value) {
return document.Number, value, nil
}
if decimalLikePattern.MatchString(value) || prefixedNumberPattern.MatchString(value) ||
leadingDecimalPattern.MatchString(value) || nonFiniteNumberPattern.MatchString(value) {
return 0, "", nodeIssue(codeScalar, "plain scalar uses an unsupported numeric form", node)
}
if tag := node.ShortTag(); tag == "!!int" || tag == "!!float" {
return 0, "", nodeIssue(codeScalar, "plain scalar uses an unsupported numeric form", node)
}
return document.String, value, nil
}
func documentRoot(node *yaml.Node) *yaml.Node {
if node.Kind == yaml.DocumentNode && len(node.Content) == 1 {
return node.Content[0]
}
return node
}
func parserIssue(err error, depthExceeded bool) Issue {
code := codeInvalid
if depthExceeded || errors.Is(err, errParserDepth) {
code = codeDepthLimit
}
issue := Issue{Code: code, Message: "configuration contains invalid YAML"}
var loadError *yaml.LoadError
if errors.As(err, &loadError) {
issue.Line = loadError.Mark.Line
issue.Column = loadError.Mark.Column
}
return issue
}
func nodeIssue(code, message string, node *yaml.Node) *Issue {
return &Issue{Code: code, Message: message, Line: node.Line, Column: node.Column}
}

200
internal/yaml/parse_test.go Normal file
View File

@@ -0,0 +1,200 @@
package yaml
import (
"io"
"strings"
"testing"
"go.yaml.in/yaml/v4"
"go.yaml.in/yaml/v4/plugin/limit"
"gitea.maximumdirect.net/eric/ConfigKit/internal/document"
)
func TestParsePreservesDocumentValuesAndLocations(t *testing.T) {
t.Parallel()
input := "# comment\nwide: 9223372036854775808\nwhen: 2026-09-06\nquoted: \"yes\"\nflow: [true, null, ~, false]\nempty:\n"
root, issues := Parse(strings.NewReader(input), "source-1", generousLimits())
if len(issues) != 0 {
t.Fatalf("Parse() issues = %#v", issues)
}
if root.Kind != document.Mapping || len(root.Mapping) != 5 {
t.Fatalf("root = %#v", root)
}
if root.Mapping[0].Value.Kind != document.Number || root.Mapping[0].Value.Text != "9223372036854775808" {
t.Fatalf("wide number = %#v", root.Mapping[0].Value)
}
if root.Mapping[0].KeyLocation.Line != 2 || root.Mapping[0].ValueLocation.Column != 7 {
t.Fatalf("locations = %#v", root.Mapping[0])
}
if root.Mapping[1].Value.Kind != document.String || root.Mapping[1].Value.Text != "2026-09-06" {
t.Fatalf("timestamp-like text = %#v", root.Mapping[1].Value)
}
if root.Mapping[3].Value.Sequence[1].Kind != document.Null || root.Mapping[3].Value.Sequence[2].Kind != document.Null {
t.Fatalf("null spellings = %#v", root.Mapping[3].Value.Sequence)
}
if root.Mapping[4].Value.Kind != document.Null {
t.Fatalf("empty value = %#v", root.Mapping[4].Value)
}
}
func TestParsePolicyTable(t *testing.T) {
t.Parallel()
tests := []struct {
name string
input string
code string
}{
{name: "empty input", input: "", code: codeEmpty},
{name: "empty document", input: "---\n", code: codeEmpty},
{name: "sequence root", input: "- value\n", code: codeRootType},
{name: "scalar root", input: "value\n", code: codeRootType},
{name: "second document", input: "a: 1\n---\nb: 2\n", code: codeMultipleDocuments},
{name: "empty second document", input: "a: 1\n---\n", code: codeMultipleDocuments},
{name: "duplicate key", input: "a: 1\na: 2\n", code: codeDuplicateKey},
{name: "nested duplicate", input: "a: {b: 1, b: 2}\n", code: codeDuplicateKey},
{name: "explicit standard tag", input: "a: !!str value\n", code: codeTag},
{name: "explicit custom tag", input: "a: !custom value\n", code: codeTag},
{name: "anchor", input: "a: &name value\n", code: codeAnchor},
{name: "alias source anchor", input: "a: &name value\nb: *name\n", code: codeAnchor},
{name: "merge key", input: "<<: value\n", code: codeMergeKey},
{name: "quoted merge key", input: "\"<<\": value\n", code: codeMergeKey},
{name: "boolean key", input: "true: value\n", code: codeScalar},
{name: "number key", input: "1: value\n", code: codeScalar},
{name: "null key", input: "null: value\n", code: codeScalar},
{name: "legacy yes", input: "a: yes\n", code: codeScalar},
{name: "legacy mixed", input: "a: OfF\n", code: codeScalar},
{name: "reserved case", input: "a: TRUE\n", code: codeScalar},
{name: "hex", input: "a: 0x10\n", code: codeScalar},
{name: "octal", input: "a: 0o10\n", code: codeScalar},
{name: "separator", input: "a: 1_000\n", code: codeScalar},
{name: "leading plus", input: "a: +1\n", code: codeScalar},
{name: "leading zero", input: "a: 01\n", code: codeScalar},
{name: "negative leading zero", input: "a: -01\n", code: codeScalar},
{name: "leading dot", input: "a: .5\n", code: codeScalar},
{name: "trailing dot", input: "a: 1.\n", code: codeScalar},
{name: "incomplete exponent", input: "a: 1e\n", code: codeScalar},
{name: "incomplete signed exponent", input: "a: 1e+\n", code: codeScalar},
{name: "infinity", input: "a: .inf\n", code: codeScalar},
{name: "nan", input: "a: .NaN\n", code: codeScalar},
}
for _, test := range tests {
t.Run(test.name, func(t *testing.T) {
_, issues := Parse(strings.NewReader(test.input), "source-1", generousLimits())
if len(issues) != 1 || issues[0].Code != test.code {
t.Fatalf("issues = %#v, want code %q", issues, test.code)
}
})
}
}
func TestTranslationRejectsAliasNodes(t *testing.T) {
t.Parallel()
node := &yaml.Node{Kind: yaml.AliasNode, Line: 2, Column: 3}
state := translationState{sourceID: "source-1", limits: generousLimits()}
if _, issue := state.translate(node, 0); issue == nil || issue.Code != codeAlias {
t.Fatalf("alias issue = %#v", issue)
}
}
func TestParseAcceptsSupportedSyntax(t *testing.T) {
t.Parallel()
inputs := []string{
"plain: text\nquoted_bool: 'ON'\nquoted_number: \"01\"\nblock: |\n text\n",
"{zero: 0, negative: -1, fraction: 1.25, exponent: 1e+3, boolean: false}\n",
"\"true\": value\n\"1\": number key\n\"null\": null key\n",
}
for _, input := range inputs {
if _, issues := Parse(strings.NewReader(input), "source-1", generousLimits()); len(issues) != 0 {
t.Errorf("Parse(%q) issues = %#v", input, issues)
}
}
}
func TestDuplicateKeyReportsBothLocations(t *testing.T) {
t.Parallel()
_, issues := Parse(strings.NewReader("same: first\nsame: second\n"), "source-1", generousLimits())
if len(issues) != 1 || issues[0].Code != codeDuplicateKey || issues[0].Line != 2 || len(issues[0].Related) != 1 || issues[0].Related[0].Line != 1 {
t.Fatalf("duplicate issues = %#v", issues)
}
}
func TestParseEnforcesTreeAndParserLimits(t *testing.T) {
t.Parallel()
for _, input := range []string{"a:\n b: value\n", "{a: {b: value}}\n"} {
if _, issues := Parse(strings.NewReader(input), "source-1", Limits{Depth: 2, Nodes: 20, ParserDepth: 20}); len(issues) != 0 {
t.Fatalf("depth boundary rejected for %q: %#v", input, issues)
}
if _, issues := Parse(strings.NewReader(input), "source-1", Limits{Depth: 1, Nodes: 20, ParserDepth: 20}); len(issues) != 1 || issues[0].Code != codeDepthLimit {
t.Fatalf("depth overflow for %q = %#v", input, issues)
}
}
if _, issues := Parse(strings.NewReader("a: 1\n"), "source-1", Limits{Depth: 2, Nodes: 3, ParserDepth: 20}); len(issues) != 0 {
t.Fatalf("node boundary rejected: %#v", issues)
}
if _, issues := Parse(strings.NewReader("a: 1\n"), "source-1", Limits{Depth: 2, Nodes: 2, ParserDepth: 20}); len(issues) != 1 || issues[0].Code != codeNodeLimit {
t.Fatalf("node overflow = %#v", issues)
}
if _, issues := Parse(strings.NewReader("a:\n b:\n c: value\n"), "source-1", Limits{Depth: 20, Nodes: 20, ParserDepth: 1}); len(issues) != 1 || issues[0].Code != codeDepthLimit {
t.Fatalf("parser depth overflow = %#v", issues)
}
}
func TestParserDepthCalibrationForBlockAndFlow(t *testing.T) {
t.Parallel()
inputs := []string{"a:\n b: value\n", "{a: {b: value}}\n"}
for _, input := range inputs {
maximum := observedParserDepth(t, input)
if maximum > 4 {
t.Fatalf("parser depth %d exceeds root-zero tree depth plus safety offset", maximum)
}
if _, issues := Parse(strings.NewReader(input), "source-1", Limits{Depth: 2, Nodes: 20, ParserDepth: maximum}); len(issues) != 0 {
t.Fatalf("parser boundary %d rejected %q: %#v", maximum, input, issues)
}
if _, issues := Parse(strings.NewReader(input), "source-1", Limits{Depth: 20, Nodes: 20, ParserDepth: maximum - 1}); len(issues) != 1 || issues[0].Code != codeDepthLimit {
t.Fatalf("parser overflow at %d for %q: %#v", maximum-1, input, issues)
}
}
}
func observedParserDepth(t *testing.T, input string) int {
t.Helper()
maximum := 0
depthPlugin := limit.New(limit.DepthFunc(func(depth int, _ *limit.DepthContext) error {
maximum = max(maximum, depth)
return nil
}))
loader, err := yaml.NewLoader(strings.NewReader(input), yaml.WithPlugin(depthPlugin))
if err != nil {
t.Fatal(err)
}
var node yaml.Node
if err := loader.Load(&node); err != nil && err != io.EOF {
t.Fatal(err)
}
return maximum
}
func TestMalformedParserDiagnosticIsSafe(t *testing.T) {
t.Parallel()
const canary = "CANARY_SECRET_VALUE"
_, issues := Parse(strings.NewReader("value: ["+canary+"\n"), "source-1", generousLimits())
if len(issues) != 1 || issues[0].Code != codeInvalid || issues[0].Line == 0 {
t.Fatalf("issues = %#v", issues)
}
if strings.Contains(issues[0].Message, canary) {
t.Fatal("parser diagnostic exposed input")
}
}
func generousLimits() Limits {
return Limits{Depth: 32, Nodes: 1_000, ParserDepth: 64}
}