descrybe/apps/api/internal/processing/meta.go

package processing

import (
	"fmt"
	"regexp"
	"strings"
	"unicode"
)

// Meta title/description caps mirror seo.MetaTitleMaxChars / MetaDescriptionMaxChars.
// Duplicated here so processing can fill meta without importing seo (seo imports processing).
const (
	metaTitleMaxChars       = 60
	metaDescriptionMaxChars = 155
)

// Bare "| <digits>" suffix = legacy unique_id leak (e.g. "VOX EBR700 | 50").
// Matches upsertProcessedProductSQL refresh predicate.
var poisonedMetaTitleUniqueIDRe = regexp.MustCompile(`\|\s+[0-9]+$`)

// isPoisonedMetaTitle reports meta_title that must be refreshed: bare unique_id
// suffix ("… | 50") or enhance-prompt instruction leakage ("short retail title",
// "Title formula", …). Matches upsert / BackfillMissingMeta SQL predicates.
func isPoisonedMetaTitle(s string) bool {
	s = strings.TrimSpace(s)
	if s == "" {
		return false
	}
	if poisonedMetaTitleUniqueIDRe.MatchString(s) {
		return true
	}
	return isPromptLeakageTitle(s)
}

// fillMetaFromResult builds free template meta from a StepResult (FillMetaTemplate-equivalent).
// Does not call AI / does not spend credits.
func fillMetaFromResult(result StepResult) (title, description string) {
	name := strings.TrimSpace(result.ProcessedName)
	if name == "" {
		name = strings.TrimSpace(result.Name)
	}
	if name == "" {
		name = "Product"
	}
	cat := categoryDisplayLabel(result)
	brand := metaBrandFromAttrs(result.ProcessedAttributes, result.Attributes)

	parts := make([]string, 0, 3)
	if brand != "" && !strings.Contains(strings.ToLower(name), strings.ToLower(brand)) {
		parts = append(parts, brand)
	}
	parts = append(parts, name)
	if cat != "" && !strings.Contains(strings.ToLower(name), strings.ToLower(cat)) {
		parts = append(parts, cat)
	}
	title = truncateMetaRunes(strings.Join(parts, " | "), metaTitleMaxChars)

	body := strings.TrimSpace(result.ProcessedDescription)
	if body == "" {
		body = strings.TrimSpace(result.Description)
	}
	body = stripMetaTags(body)
	if body == "" {
		var bits []string
		if brand != "" {
			bits = append(bits, brand)
		}
		bits = append(bits, name)
		if cat != "" {
			bits = append(bits, "in "+cat)
		}
		body = strings.Join(bits, " ") + ". Shop quality products with clear specs and fast delivery."
	}
	description = truncateMetaDescription(body, metaDescriptionMaxChars)
	return title, description
}

// truncateMetaDescription collapses whitespace and truncates at a word boundary when possible.
func truncateMetaDescription(s string, max int) string {
	s = collapseMetaSpace(s)
	if max <= 0 {
		return ""
	}
	r := []rune(s)
	if len(r) <= max {
		return s
	}
	cut := r[:max]
	lastSpace := -1
	for i := len(cut) - 1; i >= 0; i-- {
		if unicode.IsSpace(cut[i]) {
			lastSpace = i
			break
		}
	}
	if lastSpace > max/2 {
		return strings.TrimSpace(string(cut[:lastSpace]))
	}
	return string(cut)
}

func truncateMetaRunes(s string, max int) string {
	s = collapseMetaSpace(s)
	r := []rune(s)
	if max <= 0 || len(r) <= max {
		return s
	}
	if max <= 3 {
		return string(r[:max])
	}
	return string(r[:max-1]) + "…"
}

func collapseMetaSpace(s string) string {
	return strings.Join(strings.Fields(s), " ")
}

func stripMetaTags(s string) string {
	var b strings.Builder
	inTag := false
	for _, r := range s {
		switch {
		case r == '<':
			inTag = true
		case r == '>':
			inTag = false
		case !inTag:
			b.WriteRune(r)
		}
	}
	return b.String()
}

func metaBrandFromAttrs(bags ...map[string]any) string {
	keys := []string{"brand", "Brand", "manufacturer"}
	for _, m := range bags {
		if m == nil {
			continue
		}
		for _, k := range keys {
			if v, ok := m[k]; ok {
				if s, ok := v.(string); ok {
					if t := strings.TrimSpace(s); t != "" {
						return t
					}
				}
			}
		}
	}
	return ""
}

// metaFieldsFromEnhanceObj extracts plain SEO meta from an enhance JSON object.
// Accepts snake_case and camelCase keys (legacy A1 <metaDescription> / cats.json).
// Strips HTML and rejects prompt-leakage titles; empty when missing.
func metaFieldsFromEnhanceObj(obj map[string]any) (title, description string) {
	if obj == nil {
		return "", ""
	}
	title = strings.TrimSpace(SanitizeOutput(fmt.Sprint(obj["meta_title"])))
	if title == "" || title == "<nil>" {
		title = strings.TrimSpace(SanitizeOutput(fmt.Sprint(obj["metaTitle"])))
	}
	description = strings.TrimSpace(SanitizeOutput(fmt.Sprint(obj["meta_description"])))
	if description == "" || description == "<nil>" {
		description = strings.TrimSpace(SanitizeOutput(fmt.Sprint(obj["metaDescription"])))
	}
	if title == "<nil>" {
		title = ""
	}
	if description == "<nil>" {
		description = ""
	}
	title = stripMetaTags(title)
	description = stripMetaTags(description)
	if isPromptLabelTitle(title) || isPromptLeakageTitle(title) {
		title = ""
	}
	if isPromptLabelTitle(description) || isPromptLeakageTitle(description) {
		description = ""
	}
	if title != "" {
		title = truncateMetaRunes(title, metaTitleMaxChars)
	}
	if description != "" {
		description = truncateMetaDescription(description, metaDescriptionMaxChars)
	}
	return title, description
}

// enhanceMetaFromRaw reads meta_title / meta_description stashed on enhance raw meta.
func enhanceMetaFromRaw(raw any) (title, description string) {
	m, ok := raw.(map[string]any)
	if !ok || m == nil {
		return "", ""
	}
	title = strings.TrimSpace(fmt.Sprint(m["meta_title"]))
	description = strings.TrimSpace(fmt.Sprint(m["meta_description"]))
	if title == "<nil>" {
		title = ""
	}
	if description == "<nil>" {
		description = ""
	}
	return title, description
}