descrybe/apps/api/internal/woocommerce/sync.go

package woocommerce

import (
	"context"
	"encoding/json"
	"fmt"
	"strconv"
	"strings"
	"time"

	"github.com/google/uuid"
)

const (
	defaultBatchSize     = 25
	defaultSyncLimit     = 200
	maxBatchSize         = 100
	maxSyncLimit         = 1000
	maxOrdersSyncLimit   = 5000
	maxReviewsSyncLimit  = 5000
	maxProductIDMap      = 5000
	maxScheduleIntervalH = 168 // 7 days
	metaDescrybeID       = "_descrybe_product_id"
)

type SyncOptions struct {
	CategoryMappings      map[string]CategoryMap  `json:"category_mappings"`
	AttributeMappings     map[string]AttributeMap `json:"attribute_mappings"`
	ProductIDs            map[string]int          `json:"product_ids"`
	MatchStrategy         string                  `json:"match_strategy"`
	BatchSize             int                     `json:"batch_size"`
	SyncLimit             int                     `json:"sync_limit"`
	PendingSync           bool                    `json:"pending_sync"`
	PendingOrdersSync     bool                    `json:"pending_orders_sync"`
	PendingReviewsSync    bool                    `json:"pending_reviews_sync"`
	LastSyncStatus        string                  `json:"last_sync_status"`
	LastSyncError         string                  `json:"last_sync_error"`
	LastSyncSummary       *SyncSummary            `json:"last_sync_summary,omitempty"`
	LastOrdersSyncStatus  string                  `json:"last_orders_sync_status"`
	LastOrdersSyncError   string                  `json:"last_orders_sync_error"`
	LastOrdersSyncedAt    *time.Time              `json:"last_orders_synced_at,omitempty"`
	LastReviewsSyncStatus string                  `json:"last_reviews_sync_status"`
	LastReviewsSyncError  string                  `json:"last_reviews_sync_error"`
	LastReviewsSyncedAt   *time.Time              `json:"last_reviews_synced_at,omitempty"`
	OrdersSyncLimit       int                     `json:"orders_sync_limit"`
	ReviewsSyncLimit      int                     `json:"reviews_sync_limit"`
	OrdersModifiedAfter   string                  `json:"orders_modified_after"`
	ScheduleIntervalHours int                     `json:"schedule_interval_hours"`
	SchedulePaused        bool                    `json:"schedule_paused"`
	// One-shot selection for the next product sync (cleared when sync finishes).
	SyncFilterStatus   string   `json:"sync_filter_status,omitempty"`
	SyncFilterCategory string   `json:"sync_filter_category,omitempty"`
	SyncOnlyIDs        []string `json:"sync_only_ids,omitempty"`
}

type CategoryMap struct {
	Name string `json:"name"`
	Slug string `json:"slug"`
	WCID int    `json:"wc_id"`
}

type AttributeMap struct {
	Name string `json:"name"`
	WCID int    `json:"wc_id"`
}

type SyncSummary struct {
	Total   int `json:"total"`
	Created int `json:"created"`
	Updated int `json:"updated"`
	Failed  int `json:"failed"`
	Skipped int `json:"skipped"`
}

type syncProductRow struct {
	ID                   uuid.UUID
	ProductID            *string
	Name                 *string
	Category             *string
	Description          *string
	ProcessedName        *string
	ProcessedDescription *string
	MetaDescription      *string
	Attributes           []byte
	ProcessedAttributes  []byte
	GTIN                 *string
	MappedData           []byte
}

func parseSyncOptions(raw []byte) SyncOptions {
	opt := SyncOptions{
		CategoryMappings:  map[string]CategoryMap{},
		AttributeMappings: map[string]AttributeMap{},
		ProductIDs:        map[string]int{},
		MatchStrategy:     "sku",
		BatchSize:         defaultBatchSize,
		SyncLimit:         defaultSyncLimit,
	}
	if len(raw) == 0 {
		return opt
	}
	_ = json.Unmarshal(raw, &opt)
	if opt.CategoryMappings == nil {
		opt.CategoryMappings = map[string]CategoryMap{}
	}
	if opt.AttributeMappings == nil {
		opt.AttributeMappings = map[string]AttributeMap{}
	}
	if opt.ProductIDs == nil {
		opt.ProductIDs = map[string]int{}
	}
	pruneStringIntMap(opt.ProductIDs, maxProductIDMap)
	if opt.MatchStrategy == "" {
		opt.MatchStrategy = "sku"
	}
	if opt.BatchSize <= 0 || opt.BatchSize > maxBatchSize {
		opt.BatchSize = defaultBatchSize
	}
	if opt.SyncLimit <= 0 || opt.SyncLimit > maxSyncLimit {
		opt.SyncLimit = defaultSyncLimit
	}
	if opt.OrdersSyncLimit < 0 || opt.OrdersSyncLimit > maxOrdersSyncLimit {
		opt.OrdersSyncLimit = 0
	}
	if opt.ReviewsSyncLimit < 0 || opt.ReviewsSyncLimit > maxReviewsSyncLimit {
		opt.ReviewsSyncLimit = 0
	}
	if opt.ScheduleIntervalHours < 0 || opt.ScheduleIntervalHours > maxScheduleIntervalH {
		opt.ScheduleIntervalHours = 0
	}
	opt.SyncOnlyIDs = pruneSyncOnlyIDs(opt.SyncOnlyIDs, maxSyncOnlyIDs)
	return opt
}

// resolveScheduleInterval maps schedule_interval_hours to a duration (default when hours<=0).
func resolveScheduleInterval(hours int, defaultInterval time.Duration) time.Duration {
	if defaultInterval <= 0 {
		defaultInterval = 6 * time.Hour
	}
	if hours > 0 {
		return time.Duration(hours) * time.Hour
	}
	return defaultInterval
}

// isDueForSchedule reports whether last sync is missing or older than interval.
func isDueForSchedule(last *time.Time, now time.Time, interval time.Duration) bool {
	if last == nil {
		return true
	}
	return now.Sub(last.UTC()) >= interval
}

// shouldEnqueueScheduled reports whether auto product sync should run now.
// Paused schedules never enqueue; interval 0 uses defaultInterval (worker default 6h).
func shouldEnqueueScheduled(paused bool, hours int, last *time.Time, now time.Time, defaultInterval time.Duration) bool {
	if paused {
		return false
	}
	return isDueForSchedule(last, now, resolveScheduleInterval(hours, defaultInterval))
}

func (s *Service) loadProductsForSync(ctx context.Context, companyID uuid.UUID, opt SyncOptions) ([]syncProductRow, error) {
	limit := opt.SyncLimit
	if limit <= 0 || limit > maxSyncLimit {
		limit = defaultSyncLimit
	}
	args := []any{companyID}
	where := []string{"p.company_id = $1"}
	if status := strings.TrimSpace(opt.SyncFilterStatus); status != "" {
		if status == "needs_review" {
			where = append(where, "p.status IN ('needs_review', 'processed')")
		} else {
			args = append(args, status)
			where = append(where, fmt.Sprintf("p.status = $%d", len(args)))
		}
	}
	if cat := strings.TrimSpace(opt.SyncFilterCategory); cat != "" {
		args = append(args, cat)
		n := len(args)
		where = append(where, fmt.Sprintf("(p.category = $%d OR lower(p.category) = lower($%d))", n, n))
	}
	if len(opt.SyncOnlyIDs) > 0 {
		ids := make([]uuid.UUID, 0, len(opt.SyncOnlyIDs))
		for _, raw := range opt.SyncOnlyIDs {
			id, err := uuid.Parse(raw)
			if err != nil {
				continue
			}
			ids = append(ids, id)
		}
		if len(ids) == 0 {
			return nil, nil
		}
		args = append(args, ids)
		where = append(where, fmt.Sprintf("p.id = ANY($%d::uuid[])", len(args)))
	}
	args = append(args, limit)
	limN := len(args)
	q := fmt.Sprintf(`
		SELECT p.id, p.product_id, p.name, p.category, p.description, p.processed_name, p.processed_description,
		       p.meta_description, p.attributes, p.processed_attributes, r.gtin, r.mapped_data
		FROM processed_products p
		LEFT JOIN raw_products r ON r.id = p.raw_product_id AND r.company_id = p.company_id
		WHERE %s
		ORDER BY p.updated_at DESC
		LIMIT $%d`, strings.Join(where, " AND "), limN)
	rows, err := s.Pool.Query(ctx, q, args...)
	if err != nil {
		return nil, err
	}
	defer rows.Close()
	out := make([]syncProductRow, 0)
	for rows.Next() {
		var row syncProductRow
		if err := rows.Scan(
			&row.ID, &row.ProductID, &row.Name, &row.Category, &row.Description,
			&row.ProcessedName, &row.ProcessedDescription, &row.MetaDescription,
			&row.Attributes, &row.ProcessedAttributes, &row.GTIN, &row.MappedData,
		); err != nil {
			return nil, err
		}
		out = append(out, row)
	}
	return out, rows.Err()
}

func deref(s *string) string {
	if s == nil {
		return ""
	}
	return strings.TrimSpace(*s)
}

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

func mappedString(mapped []byte, keys ...string) string {
	if len(mapped) == 0 {
		return ""
	}
	var m map[string]any
	if json.Unmarshal(mapped, &m) != nil {
		return ""
	}
	for _, k := range keys {
		if v, ok := m[k]; ok {
			switch t := v.(type) {
			case string:
				if strings.TrimSpace(t) != "" {
					return strings.TrimSpace(t)
				}
			case float64:
				return strconv.FormatFloat(t, 'f', -1, 64)
			case json.Number:
				return t.String()
			default:
				s := strings.TrimSpace(fmt.Sprint(t))
				if s != "" && s != "<nil>" {
					return s
				}
			}
		}
	}
	return ""
}

func mappedImages(mapped []byte) []map[string]string {
	if len(mapped) == 0 {
		return nil
	}
	var m map[string]any
	if json.Unmarshal(mapped, &m) != nil {
		return nil
	}
	raw, ok := m["images"]
	if !ok {
		return nil
	}
	out := make([]map[string]string, 0)
	switch t := raw.(type) {
	case []any:
		for _, item := range t {
			switch u := item.(type) {
			case string:
				if u != "" {
					out = append(out, map[string]string{"src": u})
				}
			case map[string]any:
				if src, ok := u["src"].(string); ok && src != "" {
					out = append(out, map[string]string{"src": src})
				}
			}
		}
	case string:
		if t != "" {
			out = append(out, map[string]string{"src": t})
		}
	}
	return out
}

func (row syncProductRow) toPayload(opt SyncOptions) ProductPayload {
	name := firstNonEmpty(deref(row.ProcessedName), deref(row.Name), "Product")
	desc := firstNonEmpty(deref(row.ProcessedDescription), deref(row.Description))
	shortDesc := deref(row.MetaDescription)
	sku := firstNonEmpty(mappedString(row.MappedData, "sku", "SKU"), deref(row.ProductID), row.ID.String())
	price := mappedString(row.MappedData, "price", "regular_price")
	ean := firstNonEmpty(mappedString(row.MappedData, "ean", "gtin", "EAN"), deref(row.GTIN))

	manage := false
	payload := ProductPayload{
		Name:             name,
		Type:             "simple",
		Status:           "publish",
		Description:      desc,
		ShortDescription: shortDesc,
		SKU:              sku,
		RegularPrice:     price,
		Images:           mappedImages(row.MappedData),
		ManageStock:      &manage,
		StockStatus:      "instock",
		MetaData: []MetaDatum{
			{Key: metaDescrybeID, Value: row.ID.String()},
			{Key: "_descrybe_category", Value: deref(row.Category)},
			{Key: "ean", Value: ean},
		},
	}

	cat := deref(row.Category)
	if cat != "" {
		if m, ok := opt.CategoryMappings[cat]; ok {
			entry := map[string]any{"name": firstNonEmpty(m.Name, cat)}
			if m.WCID > 0 {
				entry["id"] = m.WCID
			}
			if m.Slug != "" {
				entry["slug"] = m.Slug
			}
			payload.Categories = []map[string]any{entry}
		} else {
			payload.Categories = []map[string]any{{"name": cat}}
		}
	}

	attrsRaw := row.ProcessedAttributes
	if len(attrsRaw) == 0 {
		attrsRaw = row.Attributes
	}
	var attrs map[string]any
	if len(attrsRaw) > 0 && json.Unmarshal(attrsRaw, &attrs) == nil {
		i := 0
		for key, val := range attrs {
			options := []string{}
			switch t := val.(type) {
			case []any:
				for _, x := range t {
					options = append(options, fmt.Sprint(x))
				}
			default:
				options = []string{fmt.Sprint(val)}
			}
			name := key
			entry := map[string]any{
				"name":      name,
				"visible":   true,
				"variation": false,
				"options":   options,
				"position":  i,
			}
			if m, ok := opt.AttributeMappings[key]; ok {
				if m.Name != "" {
					entry["name"] = m.Name
				}
				if m.WCID > 0 {
					entry["id"] = m.WCID
				}
			}
			payload.Attributes = append(payload.Attributes, entry)
			i++
		}
	}
	return payload
}

func (s *Service) pushProducts(ctx context.Context, client *Client, companyID uuid.UUID, opt *SyncOptions, rows []syncProductRow) (SyncSummary, error) {
	summary := SyncSummary{Total: len(rows)}
	creates := make([]ProductPayload, 0)
	updates := make([]ProductPayload, 0)
	createKeys := make([]string, 0)
	updateKeys := make([]string, 0)

	type pendingSKU struct {
		key     string
		payload ProductPayload
	}
	needSKU := make([]pendingSKU, 0)
	skus := make([]string, 0)

	for _, row := range rows {
		key := row.ID.String()
		payload := row.toPayload(*opt)
		if wcID, ok := opt.ProductIDs[key]; ok && wcID > 0 {
			payload.ID = wcID
			updates = append(updates, payload)
			updateKeys = append(updateKeys, key)
			continue
		}
		if opt.MatchStrategy == "sku" && payload.SKU != "" {
			needSKU = append(needSKU, pendingSKU{key: key, payload: payload})
			skus = append(skus, payload.SKU)
			continue
		}
		creates = append(creates, payload)
		createKeys = append(createKeys, key)
	}

	if len(needSKU) > 0 {
		found, err := client.ListProductsBySKUs(ctx, skus)
		if err != nil {
			summary.Failed += len(needSKU)
		} else {
			for _, item := range needSKU {
				if p, ok := found[item.payload.SKU]; ok && p.ID > 0 {
					item.payload.ID = p.ID
					opt.ProductIDs[item.key] = p.ID
					updates = append(updates, item.payload)
					updateKeys = append(updateKeys, item.key)
					continue
				}
				creates = append(creates, item.payload)
				createKeys = append(createKeys, item.key)
			}
		}
	}

	batchSize := opt.BatchSize
	if batchSize <= 0 {
		batchSize = defaultBatchSize
	}
	for i := 0; i < len(creates); i += batchSize {
		end := i + batchSize
		if end > len(creates) {
			end = len(creates)
		}
		chunk := creates[i:end]
		keys := createKeys[i:end]
		res, err := client.BatchProducts(ctx, BatchRequest{Create: chunk})
		if err != nil {
			summary.Failed += len(chunk)
			continue
		}
		for idx, p := range res.Create {
			if idx < len(keys) && p.ID > 0 {
				opt.ProductIDs[keys[idx]] = p.ID
				summary.Created++
			} else {
				summary.Failed++
			}
		}
		if len(res.Create) < len(chunk) {
			summary.Failed += len(chunk) - len(res.Create)
		}
	}

	for i := 0; i < len(updates); i += batchSize {
		end := i + batchSize
		if end > len(updates) {
			end = len(updates)
		}
		chunk := updates[i:end]
		keys := updateKeys[i:end]
		res, err := client.BatchProducts(ctx, BatchRequest{Update: chunk})
		if err != nil {
			summary.Failed += len(chunk)
			continue
		}
		for idx, p := range res.Update {
			if idx < len(keys) && p.ID > 0 {
				opt.ProductIDs[keys[idx]] = p.ID
				summary.Updated++
			} else {
				summary.Failed++
			}
		}
		if len(res.Update) < len(chunk) {
			summary.Failed += len(chunk) - len(res.Update)
		}
	}

	_ = companyID
	return summary, nil
}

func pruneStringIntMap(m map[string]int, max int) {
	if max <= 0 || len(m) <= max {
		return
	}
	n := len(m) - max
	for k := range m {
		delete(m, k)
		n--
		if n <= 0 {
			return
		}
	}
}