mesh-sdk/go/catalogue/unknown_field.go

package catalogue

import (
	"bytes"
	"encoding/json"
	"errors"
	"fmt"
	"regexp"
)

// UnknownFieldError is a key a manifest has that this controller does not know, wherever it is: at the
// top of the manifest or inside a block (a state, an offer, a data item's backup, an own secret, a verb).
// Typed, because the blocks' decoders wrap the decoder's words in their own.
type UnknownFieldError struct {
	Field string
	err   error
}

func (e *UnknownFieldError) Error() string { return e.err.Error() }
func (e *UnknownFieldError) Unwrap() error { return e.err }

// unknownFieldText is how the JSON decoder words a key its target does not have.
var unknownFieldText = regexp.MustCompile(`^json: unknown field "([^"]*)"$`)

// typedUnknown is err as an UnknownFieldError when it is the decoder refusing an unknown key, else err.
func typedUnknown(err error) error {
	if err == nil {
		return nil
	}
	if m := unknownFieldText.FindStringSubmatch(err.Error()); m != nil {
		return &UnknownFieldError{Field: m[1], err: err}
	}
	return err
}

// asUnknownField is the unknown key err is about, at any depth, or nil.
func asUnknownField(err error) *UnknownFieldError {
	var u *UnknownFieldError
	if errors.As(typedUnknown(err), &u) {
		return u
	}
	return nil
}

// prunedFields is a stored manifest's fields with every key this controller does not know taken out,
// and the keys taken out (novox/hq ADR 0262). A second pass, after the strict one refused: each
// top-level field is decoded alone, and where an entry inside it has an unknown key, the one
// occurrence whose removal moves the decoder past it is removed — never a key of the same name that
// the entry around it knows. So a left-out module still provides what it provides, and its data and
// directories are still read, which its backup lines are made from.
func prunedFields(keys map[string]json.RawMessage) (manifestFields, []string, error) {
	var removed []string
	tree := map[string]any{}
	for k, raw := range keys {
		var v any
		dec := json.NewDecoder(bytes.NewReader(raw))
		dec.UseNumber()
		if err := dec.Decode(&v); err != nil {
			return manifestFields{}, nil, err
		}
		tree[k] = v
	}
	for _, k := range sortedAnyKeys(tree) {
		for tries := 0; ; tries++ {
			err := decodesAlone(k, tree[k])
			if err == nil {
				break
			}
			unknown := asUnknownField(err)
			if unknown == nil || tries > 64 {
				return manifestFields{}, nil, err
			}
			if unknown.Field == k {
				delete(tree, k)
				removed = append(removed, k)
				break
			}
			path, ok := removalThatHelps(k, tree[k], unknown.Field, err.Error())
			if !ok {
				// The same key unknown in two entries alike: no one removal changes the words.
				// Every occurrence goes, and the field is judged again.
				if removeEvery(tree[k], unknown.Field) == 0 {
					return manifestFields{}, nil, err
				}
				path = "…." + unknown.Field
			}
			removed = append(removed, k+path)
		}
	}
	raw, err := json.Marshal(tree)
	if err != nil {
		return manifestFields{}, nil, err
	}
	var fields manifestFields
	dec := json.NewDecoder(bytes.NewReader(raw))
	dec.DisallowUnknownFields()
	if err := dec.Decode(&fields); err != nil {
		return manifestFields{}, nil, err
	}
	return fields, removed, nil
}

// decodesAlone is whether one top-level field decodes strictly on its own.
func decodesAlone(k string, v any) error {
	raw, err := json.Marshal(map[string]any{k: v})
	if err != nil {
		return err
	}
	var fields manifestFields
	dec := json.NewDecoder(bytes.NewReader(raw))
	dec.DisallowUnknownFields()
	return dec.Decode(&fields)
}

// removalThatHelps removes, from v, the one occurrence of key whose removal changes what the strict
// decoder says of field k, and says where it was. Every other occurrence is left as it was.
func removalThatHelps(k string, v any, key, said string) (string, bool) {
	var found bool
	var where string
	var walk func(node any, at string) bool
	walk = func(node any, at string) bool {
		switch n := node.(type) {
		case map[string]any:
			if value, has := n[key]; has {
				delete(n, key)
				// Accepted only when the decoder is past it: nothing left, or an unknown key said
				// elsewhere. A different kind of error means the removal broke the entry; the same
				// words mean this was not the occurrence it refused.
				err := decodesAlone(k, v)
				if err == nil || (asUnknownField(err) != nil && err.Error() != said) {
					found, where = true, at+"."+key
					return true
				}
				n[key] = value
			}
			for _, sub := range sortedAnyKeys(n) {
				if walk(n[sub], at+"."+sub) {
					return true
				}
			}
		case []any:
			for i, item := range n {
				if walk(item, fmt.Sprintf("%s[%d]", at, i)) {
					return true
				}
			}
		}
		return false
	}
	walk(v, "")
	return where, found
}

// removeEvery removes key from every object in v, and says how many it removed.
func removeEvery(v any, key string) int {
	n := 0
	switch node := v.(type) {
	case map[string]any:
		if _, has := node[key]; has {
			delete(node, key)
			n++
		}
		for _, sub := range node {
			n += removeEvery(sub, key)
		}
	case []any:
		for _, item := range node {
			n += removeEvery(item, key)
		}
	}
	return n
}