MOPAC/internal/events/normalize.go

package events

import (
	"bytes"
	"crypto/sha256"
	"encoding/hex"
	"encoding/json"
	"fmt"
	"net/http"
	"strings"
	"time"
)

// Provider delivery/event id headers used for dedup before falling back to
// the payload digest.
const (
	giteaDeliveryHeader     = "X-Gitea-Delivery"
	discourseEventIDHeader  = "X-Discourse-Event-Id"
	redmineDeliveryFallback = "X-Redmine-Delivery" // some plugins send it
)

// Normalize maps a verified provider payload to the internal Event record.
// It is deliberately tolerant: providers (and their plugin ecosystems)
// disagree on payload shapes, so extraction tries the known variants and
// leaves anything missing empty rather than rejecting. Unparseable bodies
// are the only hard error.
func Normalize(source string, header http.Header, body []byte, now time.Time) (Event, error) {
	var root map[string]any
	dec := json.NewDecoder(bytes.NewReader(body))
	dec.UseNumber()
	if err := dec.Decode(&root); err != nil {
		return Event{}, fmt.Errorf("payload is not a JSON object: %w", err)
	}

	sum := sha256.Sum256(body)
	ev := Event{
		Source:        source,
		PayloadDigest: hex.EncodeToString(sum[:]),
		ReceivedAt:    now.UTC(),
	}

	switch source {
	case SourceRedmine:
		normalizeRedmine(&ev, header, root)
	case SourceDiscourse:
		normalizeDiscourse(&ev, header, root)
	case SourceGitea:
		normalizeGitea(&ev, header, root)
	default:
		return Event{}, fmt.Errorf("unknown event source %q", source)
	}

	// Dedup identity: the provider's event id, else the payload digest
	// (identical replays dedup; different payloads never collide).
	if ev.ProviderID == "" {
		ev.ProviderID = ev.PayloadDigest
	}
	if ev.Actor == "" {
		ev.Actor = "unknown"
	}
	if ev.Kind == "" {
		ev.Kind = "unknown"
	}
	ev.Action = MapAction(ev.Source, ev.Kind)
	return ev, nil
}

// MapAction implements the DESIGN trigger -> action mapping:
// redmine issue update/note -> dispatch_turn; discourse post reply ->
// respond_turn; gitea PR approved/merged -> pipeline_step; else ignore.
func MapAction(source, kind string) string {
	switch source {
	case SourceRedmine:
		if kind == "issue_updated" || kind == "issue_edited" ||
			strings.Contains(kind, "note") || strings.Contains(kind, "journal") {
			return ActionDispatchTurn
		}
	case SourceDiscourse:
		if kind == "post_created" || kind == "post_edited" || kind == "post" {
			return ActionRespondTurn
		}
	case SourceGitea:
		if kind == "pr_approved" || kind == "pr_merged" {
			return ActionPipelineStep
		}
	}
	return ActionIgnore
}

// normalizeRedmine handles the redmine_webhooks shape
// {"event_name": "...", "payload": {"issue": {...}}} and flatter plugin
// payloads that put "issue" (and the actor) at the top level.
func normalizeRedmine(ev *Event, header http.Header, root map[string]any) {
	p := obj(root, "payload")
	issue := obj(p, "issue")
	if issue == nil {
		issue = obj(root, "issue")
	}
	ev.Kind = firstNonEmpty(
		str(root, "event_name"),
		str(p, "event_name"),
		header.Get("X-Redmine-Event"),
	)
	ev.Actor = firstNonEmpty(
		person(p, "user", "author"),
		person(root, "user", "author"),
		person(issue, "user", "author"),
	)
	if issue != nil {
		if ev.Kind == "" {
			ev.Kind = "issue_updated"
		}
		if id := numStr(issue, "id"); id != "" {
			ev.SubjectID = "redmine:issue:" + id
		}
		ev.Subject = str(issue, "subject")
	}
	ev.ProviderID = firstNonEmpty(header.Get(redmineDeliveryFallback))
}

// normalizeDiscourse handles Discourse webhook bodies
// {"post": {"id":.., "topic_id":.., "username":.., ...}}; the event type
// rides in the X-Discourse-Event header (e.g. post_created).
func normalizeDiscourse(ev *Event, header http.Header, root map[string]any) {
	ev.Kind = firstNonEmpty(
		header.Get("X-Discourse-Event"),
		header.Get("X-Discourse-Event-Type"),
		str(root, "event_type"),
	)
	post := obj(root, "post")
	if post != nil {
		if ev.Kind == "" {
			ev.Kind = "post_created"
		}
		if id := numStr(post, "topic_id", "topicId"); id != "" {
			ev.SubjectID = "discourse:topic:" + id
		} else if id := numStr(root, "topic_id"); id != "" {
			ev.SubjectID = "discourse:topic:" + id
		}
		ev.Subject = str(post, "topic_title", "topic_slug")
		ev.Actor = firstNonEmpty(str(post, "username"), person(root, "user"), str(root, "username"))
	} else {
		if id := numStr(root, "topic_id"); id != "" {
			ev.SubjectID = "discourse:topic:" + id
		}
		ev.Actor = firstNonEmpty(person(root, "user"), str(root, "username"))
	}
	ev.ProviderID = firstNonEmpty(header.Get(discourseEventIDHeader), str(root, "id"))
}

// normalizeGitea handles Gitea webhook bodies. PR events carry
// {"action": ..., "number": N, "pull_request": {...}, "repository": {...},
// "sender": {...}}; issue events mirror that with "issue". kind vocabulary:
// pr_approved / pr_merged / pr_opened / ... and issue_*.
func normalizeGitea(ev *Event, header http.Header, root map[string]any) {
	ev.ProviderID = header.Get(giteaDeliveryHeader)
	action := str(root, "action")
	if repo := obj(root, "repository"); repo != nil {
		ev.Repo = str(repo, "full_name")
	}
	ev.Actor = person(root, "sender")

	pr := obj(root, "pull_request")
	issue := obj(root, "issue")
	switch {
	case pr != nil:
		kind := action
		if action == "closed" && truthy(pr, "merged") {
			kind = "merged"
		}
		if kind == "" {
			kind = "unknown"
		}
		ev.Kind = "pr_" + kind
		if n := firstNonEmpty(numStr(root, "number"), numStr(pr, "number")); n != "" {
			ev.SubjectID = giteaRef(ev.Repo, "pr", n)
		}
		ev.Subject = str(pr, "title")
	case issue != nil:
		if action == "" {
			action = "updated"
		}
		ev.Kind = "issue_" + action
		if n := firstNonEmpty(numStr(root, "number"), numStr(issue, "number"), numStr(issue, "id")); n != "" {
			ev.SubjectID = giteaRef(ev.Repo, "issue", n)
		}
		ev.Subject = str(issue, "title")
	default:
		ev.Kind = firstNonEmpty(
			header.Get("X-Gitea-Event-Type"),
			header.Get("X-Gitea-Event"),
			action,
			"unknown",
		)
	}
}

func giteaRef(repo, kind, n string) string {
	if repo == "" {
		return "gitea:" + kind + ":" + n
	}
	return "gitea:" + kind + ":" + repo + "#" + n
}

// --- tolerant JSON extraction helpers -------------------------------------

func obj(m map[string]any, key string) map[string]any {
	if m == nil {
		return nil
	}
	if v, ok := m[key].(map[string]any); ok {
		return v
	}
	return nil
}

// str returns the first non-empty string among the keys.
func str(m map[string]any, keys ...string) string {
	if m == nil {
		return ""
	}
	for _, k := range keys {
		if v, ok := m[k].(string); ok && v != "" {
			return v
		}
	}
	return ""
}

// numStr returns the first present numeric (or string) key rendered as a
// decimal string; ids must survive without float mangling.
func numStr(m map[string]any, keys ...string) string {
	if m == nil {
		return ""
	}
	for _, k := range keys {
		switch v := m[k].(type) {
		case json.Number:
			return v.String()
		case string:
			if v != "" {
				return v
			}
		case float64:
			return fmt.Sprintf("%d", int64(v))
		}
	}
	return ""
}

// person extracts an actor from nested user-ish objects: login, name,
// username, full_name in order.
func person(m map[string]any, keys ...string) string {
	for _, k := range keys {
		if o := obj(m, k); o != nil {
			if s := str(o, "login", "username", "name", "full_name"); s != "" {
				return s
			}
		}
	}
	return ""
}

func truthy(m map[string]any, key string) bool {
	v, ok := m[key].(bool)
	return ok && v
}

func firstNonEmpty(vals ...string) string {
	for _, v := range vals {
		if v != "" {
			return v
		}
	}
	return ""
}