mesh-sdk/go/delivery/composed_compare.go

package delivery

import (
	"bytes"
	"encoding/json"
	"strconv"
	"strings"

	"git.novox.be/novox/mesh-sdk/go/catalogue"
)

// Comparison is what CompareComposed finds, module by module: the same, differing, added by what was composed (not run
// now) and removed by it (run now, not composed). Each list is sorted by module.
type Comparison struct {
	Same    []string
	Differs []ModuleDiffers
	Added   []ModuleAdded
	Removed []string
}

// CompareComposed compares the compositions a node composed (by key, as compose.Compositions answers them) with the ones
// it runs, as its agent would (ADR 0559 §4: a node's answer to a placement candidate). A module is the same when its
// composition's digest is the one it runs (ComposedDigestOf, the signature left out), so the agent would apply nothing
// of it; it differs when the digests differ (both named); it is added when composed and not run, removed when run and
// not composed. The node's own composition is named HaveNodeWord. The bus's user list (catalogue.BusUsersID) is left out
// on both sides: the bus node follows it alone and never applies it from a composition it made (ADR 0547 §2, §8). A
// difference names the first resource, by its id, whose content differs or that one side alone holds; when every
// resource is equal it says what else differs (their order, the files filled, the data, the node's own parts, a file
// carried by digest).
func CompareComposed(composed map[string]Composed, running []Composed) Comparison {
	return CompareComposedWith(composed, running, nil)
}

// CompareComposedWith is CompareComposed showing, for each module that differs, every differing resource with both
// bodies masked (MaskedPair); sameValue, when given, says whether two sealed values open to the same value (the node
// opens them with its own key), so a value sealed again reads as such.
func CompareComposedWith(composed map[string]Composed, running []Composed, sameValue func(a, b string) (same, known bool)) Comparison {
	name := func(c Composed) string {
		if c.Module == "" {
			return HaveNodeWord
		}
		return c.Module
	}
	ours := map[string]Composed{}
	for _, c := range composed {
		ours[name(c)] = c
	}
	theirs := map[string]Composed{}
	for _, c := range running {
		theirs[name(c)] = c
	}
	modules := map[string]bool{}
	for m := range ours {
		modules[m] = true
	}
	for m := range theirs {
		modules[m] = true
	}
	var out Comparison
	for _, m := range sortedKeys(modules) {
		a, composedHere := ours[m]
		b, runs := theirs[m]
		switch {
		case !runs:
			dig, _ := ComposedDigestOf(withoutBusUsers(a))
			out.Added = append(out.Added, ModuleAdded{Module: m, Composed: dig})
		case !composedHere:
			out.Removed = append(out.Removed, m)
		default:
			if d, ok := compareOne(m, a, b, sameValue); ok {
				out.Same = append(out.Same, m)
			} else {
				out.Differs = append(out.Differs, d)
			}
		}
	}
	return out
}

// compareOne is one module's comparison: true when the digests agree, else what differs first.
func compareOne(module string, composed, running Composed, sameValue func(a, b string) (bool, bool)) (ModuleDiffers, bool) {
	composed, running = withoutBusUsers(composed), withoutBusUsers(running)
	dc, errC := ComposedDigestOf(composed)
	dr, errR := ComposedDigestOf(running)
	if errC != nil {
		return ModuleDiffers{Module: module, Running: dr, Why: "the composition has no digest: " + errC.Error()}, false
	}
	if errR == nil && dc == dr {
		return ModuleDiffers{}, true
	}
	differs := func(resource, why string) (ModuleDiffers, bool) {
		return ModuleDiffers{Module: module, Composed: dc, Running: dr, Resource: resource, Why: why}, false
	}
	// Words for the difference: resource by resource, each side with its large files by digest, so a file carried whole
	// and the same file carried by digest read as one.
	a, _, errA := composed.ByDigest()
	b, _, errB := running.ByDigest()
	if errA != nil || errB != nil {
		return differs("", "the compositions differ and cannot be read resource by resource")
	}
	ra, rb := resourcesByID(a.Resources), resourcesByID(b.Resources)
	ids := map[string]bool{}
	for id := range ra {
		ids[id] = true
	}
	for id := range rb {
		ids[id] = true
	}
	var first ModuleDiffers
	found := false
	var bodies []ResourceBodies
	for _, id := range sortedKeys(ids) {
		x, inA := ra[id]
		y, inB := rb[id]
		why := ""
		switch {
		case !inB:
			why = "composed, and not in what runs"
		case !inA:
			why = "in what runs, and not composed"
		case !bytes.Equal(x, y):
			why = "the resource differs"
		default:
			continue
		}
		if !found {
			first, _ = differs(id, why)
			found = true
		}
		if len(bodies) < MaxBodies {
			c, r := MaskedPair(x, y, sameValue)
			bodies = append(bodies, ResourceBodies{ID: id, Why: why, Composed: cut(c), Running: cut(r)})
		}
	}
	if found {
		first.Bodies = bodies
		return first, false
	}
	switch {
	case !jsonEqual(a.Filled, b.Filled):
		return differs("", "the files the provisioner fills differ")
	case a.Data != b.Data:
		return differs("", "whether it declares data differs")
	case !jsonEqual(a.Own, b.Own):
		return differs("", "the node's own parts differ")
	case a.FP != b.FP:
		return differs("", "the fingerprint differs")
	}
	if dA, err := ComposedDigestOf(a); err == nil {
		if dB, err := ComposedDigestOf(b); err == nil && dA == dB {
			return differs("", "the same content, a large file carried whole on one side and by digest on the other")
		}
	}
	return differs("", "the same resources in another order")
}

// withoutBusUsers is c without the bus's user list resource (its id `bus-users`, or `<module>.bus-users`).
func withoutBusUsers(c Composed) Composed {
	out := c
	out.Resources = nil
	for _, raw := range c.Resources {
		var head struct {
			ID string `json:"id"`
		}
		if json.Unmarshal(raw, &head) == nil && (head.ID == catalogue.BusUsersID() || strings.HasSuffix(head.ID, "."+catalogue.BusUsersID())) {
			continue
		}
		out.Resources = append(out.Resources, raw)
	}
	return out
}

// resourcesByID is each resource canonical, by its id (a resource without one by its place).
func resourcesByID(resources []json.RawMessage) map[string][]byte {
	out := make(map[string][]byte, len(resources))
	for i, raw := range resources {
		var head struct {
			ID string `json:"id"`
		}
		_ = json.Unmarshal(raw, &head)
		id := head.ID
		if id == "" {
			id = "#" + strconv.Itoa(i)
		}
		canon, err := canonicalJSON(raw)
		if err != nil {
			canon = raw
		}
		out[id] = canon
	}
	return out
}

// cut is s cut at MaxBody bytes, saying so.
func cut(s string) string {
	if len(s) <= MaxBody {
		return s
	}
	return s[:MaxBody] + "… (cut at " + strconv.Itoa(MaxBody) + " bytes)"
}

// secretWords are the words of a key whose value may hold a secret or a sealed one: masked in a shown body.
var secretWords = []string{"sealed", "secret", "password", "passwd", "token", "private", "key", "credential", "jwt", "salt", "hash"}

func secretKey(k string) bool {
	k = strings.ToLower(k)
	for _, w := range secretWords {
		if strings.Contains(k, w) {
			return true
		}
	}
	return false
}

// MaskedPair is two bodies of one resource (either may be nil) as JSON text with every value under a key a secret could
// be in (secretWords, at any depth) masked: "<masked: the same on both sides>", "<masked: another seal of the same
// value>" or "<masked: another value>" when sameValue says so, "<masked: differs>" otherwise, "<masked>" on one side
// alone. A body that is not JSON is shown as its length only.
func MaskedPair(composed, running json.RawMessage, sameValue func(a, b string) (same, known bool)) (string, string) {
	var a, b any
	okA := composed != nil && json.Unmarshal(composed, &a) == nil
	okB := running != nil && json.Unmarshal(running, &b) == nil
	if okA && okB {
		a, b = maskBoth(a, b, sameValue)
	} else if okA {
		a, _ = maskBoth(a, nil, nil)
	} else if okB {
		_, b = maskBoth(nil, b, nil)
	}
	show := func(v any, raw json.RawMessage, ok bool) string {
		if raw == nil {
			return ""
		}
		if !ok {
			return "<not JSON: " + strconv.Itoa(len(raw)) + " bytes>"
		}
		var out bytes.Buffer
		enc := json.NewEncoder(&out)
		enc.SetEscapeHTML(false)
		_ = enc.Encode(v)
		return strings.TrimSuffix(out.String(), "\n")
	}
	return show(a, composed, okA), show(b, running, okB)
}

func maskBoth(a, b any, sameValue func(a, b string) (bool, bool)) (any, any) {
	ma, isA := a.(map[string]any)
	mb, isB := b.(map[string]any)
	if isA || isB {
		outA, outB := map[string]any{}, map[string]any{}
		keys := map[string]bool{}
		for k := range ma {
			keys[k] = true
		}
		for k := range mb {
			keys[k] = true
		}
		for k := range keys {
			va, inA := ma[k]
			vb, inB := mb[k]
			if secretKey(k) {
				word := maskWord(va, vb, inA && inB, sameValue)
				if inA {
					outA[k] = word
				}
				if inB {
					outB[k] = word
				}
				continue
			}
			x, y := maskBoth(va, vb, sameValue)
			if inA {
				outA[k] = x
			}
			if inB {
				outB[k] = y
			}
		}
		var ra, rb any
		if isA {
			ra = outA
		} else {
			ra = a
		}
		if isB {
			rb = outB
		} else {
			rb = b
		}
		return ra, rb
	}
	la, isLA := a.([]any)
	lb, isLB := b.([]any)
	if isLA || isLB {
		outA := make([]any, len(la))
		outB := make([]any, len(lb))
		for i := range max(len(la), len(lb)) {
			var x, y any
			if i < len(la) {
				x = la[i]
			}
			if i < len(lb) {
				y = lb[i]
			}
			x, y = maskBoth(x, y, sameValue)
			if i < len(la) {
				outA[i] = x
			}
			if i < len(lb) {
				outB[i] = y
			}
		}
		var ra, rb any = a, b
		if isLA {
			ra = outA
		}
		if isLB {
			rb = outB
		}
		return ra, rb
	}
	return a, b
}

func maskWord(a, b any, both bool, sameValue func(a, b string) (bool, bool)) string {
	if !both {
		return "<masked>"
	}
	x, _ := json.Marshal(a)
	y, _ := json.Marshal(b)
	if bytes.Equal(x, y) {
		return "<masked: the same on both sides>"
	}
	if sameValue != nil {
		sa, okA := a.(string)
		sb, okB := b.(string)
		if !okA {
			sa = string(x)
		}
		if !okB {
			sb = string(y)
		}
		if same, known := sameValue(sa, sb); known {
			if same {
				return "<masked: another seal of the same value>"
			}
			return "<masked: another value>"
		}
	}
	return "<masked: differs>"
}