mesh-delivery/modules/mesh-delivery-view/internal/view/reach.go

package view

import (
	"fmt"
	"maps"
	"slices"
	"strings"

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

// How far each repository reached (novox/hq ADR 0556 clauses 2 and 3, issue 1280). The view keeps, per repository the gate's
// source names, how far each of its modules reached on the nodes that want it, and derives the repository's reached head: the
// newest head of the repository in the source's history, not past the merge the indexer took in, at which every wanted module's
// floor is at or after its last change at or before that head. It writes that as its state reach and says each new reached
// head as its event delivered, per repository. Nothing here records a merge as delivered (ADR 0429 §1): a merge at or before
// the reached head is on the machines, derived by whoever reads it.

// ReachHistory names the watch of a bucket with its history, as Initial is told it: the modules' and the source's history
// give each fingerprint's first commit and each head's source revision.
func ReachHistory(bucket string) string { return bucket + "#history" }

// reachBuckets are the watches the derivation waits for before it writes or says anything, so a start does not write a reach
// derived from half its inputs.
var reachBuckets = []string{delivery.ModulesBucket, delivery.AssignerWantsBucket, delivery.HaveBucket, delivery.SourceBucket,
	delivery.TakenBucket, ReachHistory(delivery.ModulesBucket), ReachHistory(delivery.SourceBucket)}

// ReachWrite is a reach value to write under its key in the view's state reach.
type ReachWrite struct {
	Key   string
	Value []byte
}

// SetReachWriter sets where the view's reach values go; nothing is written until it is set. A value it answers an error for is
// written again at the next change the view takes in (sendReach).
func (m *Model) SetReachWriter(put func(ReachWrite) error) {
	m.mu.Lock()
	defer m.mu.Unlock()
	m.reachPut = put
}

// applyHistory takes one revision of a watch with history: of the indexer's modules (each module's values in order) and of
// the gate's source (each revision's heads).
func (m *Model) applyHistory(e Entry) {
	switch e.Bucket {
	case delivery.ModulesBucket:
		if e.Deleted {
			delete(m.modHist, e.Key)
			return
		}
		mod, err := delivery.DecodeModule(e.Value)
		if err != nil || mod.FP2 == "" || mod.Commit == "" {
			return
		}
		h := m.modHist[e.Key]
		if n := len(h); n > 0 && h[n-1].FP2 == mod.FP2 && h[n-1].Commit == mod.Commit {
			if mod.Src != 0 {
				h[n-1].Src = mod.Src
			}
			return
		}
		h = append(h, mod)
		if len(h) > 2*delivery.MostChanges {
			h = h[len(h)-delivery.MostChanges:]
		}
		m.modHist[e.Key] = h
	case delivery.SourceBucket:
		if e.Key != delivery.SourceMainKey || e.Deleted || e.Revision == 0 {
			return
		}
		src, err := delivery.DecodeSource(e.Value)
		if err != nil {
			return
		}
		m.srcHist[e.Revision] = src.Heads
	}
}

// headAt is one head of a repository and the first source revision that named it.
type headAt struct {
	commit string
	rev    uint64
}

// headsOf are the heads of a repository in the source's history, oldest first, each with the first revision naming it.
func (m *Model) headsOf(repository string) []headAt {
	var out []headAt
	for _, rev := range slices.Sorted(maps.Keys(m.srcHist)) {
		for r, head := range m.srcHist[rev] {
			if !delivery.SameRepository(r, repository) || head == "" {
				continue
			}
			if n := len(out); n == 0 || !sameCommit(out[n-1].commit, head) {
				out = append(out, headAt{commit: head, rev: rev})
			}
		}
	}
	return out
}

// revOf is the first source revision naming a commit as the repository's head, or 0 when the history holds none.
func revOf(heads []headAt, commit string) uint64 {
	for _, h := range heads {
		if sameCommit(h.commit, commit) {
			return h.rev
		}
	}
	return 0
}

// sameCommit is whether two spellings of a commit are one (either may be abbreviated to at least seven characters).
func sameCommit(a, b string) bool {
	a, b = strings.ToLower(a), strings.ToLower(b)
	if len(a) > len(b) {
		a, b = b, a
	}
	return len(a) >= 7 && strings.HasPrefix(b, a)
}

// noBound is a module's bound when its floor is its newest change: it holds no head back.
const noBound = ^uint64(0)

// wantKeys are the wants' keys by module, each sorted: read once for every repository a derivation looks at.
func (m *Model) wantKeys() map[string][]string {
	out := map[string][]string{}
	for _, key := range slices.Sorted(maps.Keys(m.wants)) {
		_, module := splitKey(key)
		out[module] = append(out[module], key)
	}
	return out
}

// deriveReach is the reach of one repository from the model as it stands; wants are the wants' keys by module (wantKeys).
func (m *Model) deriveReach(repository string, wants map[string][]string) delivery.Reach {
	r := delivery.Reach{Repository: delivery.RepositoryName(repository), Modules: map[string]delivery.ModuleReach{}}
	heads := m.headsOf(repository)
	key, _ := delivery.RepositoryKey(repository)
	if t, ok := m.taken[key]; ok {
		r.Indexed = t.Merge
	}
	bound := noBound
	var holds []string
	for _, name := range slices.Sorted(maps.Keys(m.modules)) {
		me := m.modules[name]
		if me.err != "" || me.value.FP2 == "" || !delivery.SameRepository(me.value.Repository, repository) {
			continue
		}
		mr, b, words := m.moduleReach(name, me.value, heads, wants[name])
		r.Modules[name] = mr
		if b < bound {
			bound = b
		}
		if b != noBound {
			holds = append(holds, words)
		}
	}
	if len(r.Modules) == 0 {
		r.Modules = nil
	}
	indexed := uint64(0)
	if r.Indexed != "" {
		indexed = revOf(heads, r.Indexed)
	}
	switch {
	case r.Indexed == "":
		holds = append(holds, "the indexer has taken in no merge of "+r.Repository)
	case indexed == 0:
		holds = append(holds, fmt.Sprintf("the merge the indexer took in, %.12s, is not a head of %s in the source's history", r.Indexed, r.Repository))
	}
	for _, h := range heads {
		if h.rev <= indexed && h.rev < bound {
			r.Reached, r.ReachedSrc = h.commit, h.rev
		}
	}
	if n := len(heads); n > 0 && !sameCommit(heads[n-1].commit, r.Reached) {
		if indexed != 0 && indexed < heads[n-1].rev {
			holds = append(holds, fmt.Sprintf("the indexer has not yet taken in %.12s", heads[n-1].commit))
		}
		slices.Sort(holds)
		r.Holds = slices.Compact(holds)
	}
	return r
}

// moduleReach is how far one module reached, the source revision it holds the reached head before (noBound when none), and
// what holds it, in words. A node counts for the fingerprint its running have runs, healthy or not (ADR 0556 clause 2).
func (m *Model) moduleReach(name string, mod delivery.Module, heads []headAt, wantKeys []string) (delivery.ModuleReach, uint64, string) {
	mr := delivery.ModuleReach{FP: mod.FP2}
	hist := m.modHist[name]
	if n := len(hist); n == 0 || hist[n-1].FP2 != mod.FP2 || hist[n-1].Commit != mod.Commit {
		hist = append(slices.Clone(hist), mod) // the value watched now is the newest, whatever the history watch has given
	}
	for i := len(hist) - 1; i >= 0 && len(mr.Changes) < delivery.MostChanges; i-- {
		c := delivery.Change{Commit: hist[i].Commit, FP: hist[i].FP2, Src: hist[i].Src}
		if c.Src == 0 {
			c.Src = revOf(heads, c.Commit)
		}
		if n := len(mr.Changes); n > 0 && mr.Changes[n-1].FP == c.FP {
			mr.Changes[n-1] = c // the same fingerprint again, back to back: its first appearance is the older commit
			continue
		}
		mr.Changes = append(mr.Changes, c)
	}
	var waiting []string
	floor, floorNode, unset := -1, "", ""
	for _, key := range wantKeys {
		node, _ := splitKey(key)
		w := m.wants[key]
		if w.err != "" || w.value.Removed || w.value.FP == "" {
			continue
		}
		mr.Nodes++
		h, ok := m.haves[key]
		switch {
		case !ok || h.err != "" || h.have.FP == "":
			waiting = append(waiting, fmt.Sprintf("%s on %s runs nothing yet", name, node))
			unset = waiting[len(waiting)-1]
			continue
		case !runs(h.have.Phase):
			waiting = append(waiting, fmt.Sprintf("%s on %s is %s at %s, not running", name, node, h.have.Phase, short(h.have.FP)))
			unset = waiting[len(waiting)-1]
			continue
		case h.have.FP != w.value.FP:
			waiting = append(waiting, fmt.Sprintf("%s on %s runs %s, not yet the %s it wants", name, node, short(h.have.FP), short(w.value.FP)))
		case h.have.Health == delivery.HealthFailed:
			waiting = append(waiting, fmt.Sprintf("%s on %s runs %s and its health check failed", name, node, short(h.have.FP)))
		case h.have.Health != delivery.HealthOK:
			waiting = append(waiting, fmt.Sprintf("%s on %s runs %s, not yet healthy", name, node, short(h.have.FP)))
		}
		i := slices.IndexFunc(mr.Changes, func(c delivery.Change) bool { return c.FP == h.have.FP })
		if i < 0 {
			unset = fmt.Sprintf("%s on %s runs %s, which is not among its %d known changes", name, node, short(h.have.FP), len(mr.Changes))
			waiting = append(waiting, unset)
			continue
		}
		if i > floor {
			floor, floorNode = i, node
		}
	}
	mr.Waiting = strings.Join(waiting, "; ")
	switch {
	case mr.Nodes == 0:
		return mr, noBound, ""
	case unset != "":
		// A node that wants the module runs nothing countable of it (no have, not running, or a build older than the kept
		// history): the module holds no head of the repository back, which would make every apply anywhere, or one module
		// stuck failed, freeze the whole repository (review of issue 1280, ADR 0556 clause 2). Its words name it, and a merge
		// among its changes is not yet on the machines (notYet).
		return mr, noBound, ""
	}
	mr.Floor = mr.Changes[floor].Commit
	if floor == 0 {
		return mr, noBound, ""
	}
	next := mr.Changes[floor-1]
	words := fmt.Sprintf("%s on %s runs the build of %.12s, not yet that of %.12s", name, floorNode, mr.Floor, next.Commit)
	if next.Src == 0 {
		// The next change's order is not known: the head is held at the floor's own revision, or before every head when that is
		// not known either.
		if own := mr.Changes[floor].Src; own != 0 {
			return mr, own + 1, words
		}
		return mr, 0, words
	}
	return mr, next.Src, words
}

// runs is whether a have in this phase runs the fingerprint it names: running, being checked after its apply, or held after a
// failed backup, which keeps running what it has (ADR 0539 clause 16). A have backing up or applying, failed or removed runs
// nothing the derivation can count.
func runs(p delivery.Phase) bool {
	return p == delivery.PhaseRunning || p == delivery.PhaseChecking || p == delivery.PhaseHeld
}

// short is a fingerprint's first twelve characters, after its form.
func short(fp string) string {
	fp = strings.TrimPrefix(fp, delivery.SourcePrefix2)
	if len(fp) > 12 {
		fp = fp[:12]
	}
	return fp
}

// reachReady is whether every watch the derivation reads handed over what stood at its start.
func (m *Model) reachReady() bool {
	for _, b := range reachBuckets {
		if !m.initial[b] {
			return false
		}
	}
	return m.source != nil && m.source.err == ""
}

// reachNow derives the reach of every repository the source names, and answers the values that changed and the reached
// heads not yet said. It is called with the lock held.
func (m *Model) reachNow() ([]ReachWrite, []DeliveredEvent) {
	if (m.reachPut == nil && m.deliveredEvent == nil) || !m.reachReady() {
		return nil, nil
	}
	var writes []ReachWrite
	var events []DeliveredEvent
	wants := m.wantKeys()
	for _, repository := range slices.Sorted(maps.Keys(m.source.heads)) {
		key, err := delivery.RepositoryKey(repository)
		if err != nil {
			continue
		}
		r := m.deriveReach(repository, wants)
		if prev, ok := m.reach[key]; !ok || !delivery.SameReach(prev, r) {
			if data, err := delivery.EncodeReach(r); err == nil {
				m.reach[key] = r
				writes = append(writes, ReachWrite{Key: key, Value: data})
			}
		}
		notYet := notYetOf(r)
		said := strings.Join(notYet, ",")
		if r.Reached != "" && (r.ReachedSrc > m.reachSaid[key] || said != m.reachNotYet[key]) {
			m.reachSaid[key], m.reachNotYet[key] = r.ReachedSrc, said
			events = append(events, DeliveredEvent{Revision: r.ReachedSrc, Heads: map[string]string{r.Repository: r.Reached},
				Repository: r.Repository, NotYet: notYet})
		}
	}
	return writes, events
}

// notYetOf are the merges at or before a repository's reached head that are not yet on the machines: the changes of each module
// a node wants with no floor (a node runs nothing countable of it), sorted. A consumer takes every other merge at or before
// the reached head as on the machines.
func notYetOf(r delivery.Reach) []string {
	var out []string
	for _, mr := range r.Modules {
		if mr.Nodes == 0 || mr.Floor != "" {
			continue
		}
		for _, c := range mr.Changes {
			out = append(out, c.Commit)
		}
	}
	slices.Sort(out)
	return slices.Compact(out)
}

// Reach is the reach of every repository as the view derived it last, by key.
func (m *Model) Reach() map[string]delivery.Reach {
	m.mu.Lock()
	defer m.mu.Unlock()
	return maps.Clone(m.reach)
}