configkit/internal/yaml/parse.go

// 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"

	"go.maximumdirect.net/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}
}