descrybe/apps/api/internal/aiprompts/title_formula.go

package aiprompts

import (
	"encoding/json"
	"fmt"
	"regexp"
	"strings"
)

var (
	reQuotedNameSlot = regexp.MustCompile(`"([^"]+)"`)
	// Slovenian legacy ("… po formuli: … Ne uporabljaj vejic.") and the English
	// platform-default translation ("… using this formula: … Do not use commas.").
	reNameFormulaTail   = regexp.MustCompile(`(?is)formuli\s*:\s*(.*?)(?:\.?\s*Ne uporabljaj|\.?$)`)
	reNameFormulaTailEN = regexp.MustCompile(`(?is)formula\s*:\s*(.*?)(?:\.?\s*Do not use|\.?$)`)
)

// TitleFormulaElement is one ordered slot in categories.title_template.
type TitleFormulaElement struct {
	ID          string `json:"id"`
	Type        string `json:"type"` // "text" | "variable"
	Label       string `json:"label,omitempty"`
	Value       string `json:"value"`
	Description string `json:"description,omitempty"`
	Example     string `json:"example,omitempty"`
}

// TitleFormula is the JSON shape stored in categories.title_template.
type TitleFormula struct {
	Separator string                `json:"separator"`
	Elements  []TitleFormulaElement `json:"elements"`
}

// DefaultRetailTitleFormula matches the dominant old A1/PHP <name> pattern:
// product type + brand + full model (never brand alone).
func DefaultRetailTitleFormula() TitleFormula {
	return TitleFormula{
		Separator: " ",
		Elements: []TitleFormulaElement{
			{ID: "0-variable-product_type", Type: "variable", Label: "Product type", Value: "product_type", Description: "Specific product type (sentence case)", Example: "Gaming monitor"},
			{ID: "1-variable-brand", Type: "variable", Label: "Brand", Value: "brand", Description: "Brand with correct capitalization", Example: "Samsung"},
			{ID: "2-variable-product_model", Type: "variable", Label: "Model", Value: "product_model", Description: "Full product model / ID", Example: "Odyssey G5"},
		},
	}
}

// TitleRoleInstruction is the shared Title-section body (CategoryEnhanceUserTemplate
// and BuildCategoryEnhanceOverlay). Drives enhance JSON "name"; never brand-only.
const TitleRoleInstruction = `Role: title. Build JSON "name": short retail title from Title formula + Attrs — never brand-only. Include product type and full model when evidence exists; follow any Title formula constraints that follow; use Attrs.`

// TitleTemplateIsBrandOnly reports stubs that only encode brand (the migrated A1
// default) so enhance collapses to brand-only names.
func TitleTemplateIsBrandOnly(template any) bool {
	els, ok := titleFormulaElements(template)
	if !ok || len(els) == 0 {
		return false
	}
	vars := 0
	brandOnly := true
	for _, el := range els {
		switch strings.ToLower(strings.TrimSpace(el.Type)) {
		case "variable":
			vars++
			v := strings.ToLower(strings.TrimSpace(el.Value))
			v = strings.ReplaceAll(v, "-", "_")
			if v != "brand" && v != "znamka" {
				brandOnly = false
			}
		case "text":
			// Free-form naming-rule blobs are not brand-only stubs.
			if strings.TrimSpace(el.Value) != "" {
				brandOnly = false
			}
		}
	}
	return vars > 0 && brandOnly
}

// TitleTemplateNeedsRepair is true when template is missing, unparseable, or
// brand-only (insufficient vs old name/title rules).
func TitleTemplateNeedsRepair(template any) bool {
	els, ok := titleFormulaElements(template)
	if !ok || len(els) == 0 {
		return true
	}
	return TitleTemplateIsBrandOnly(template)
}

// DeriveTitleFormulaFromLegacyNameRules turns PHP/A1 <name> formula text
// (e.g. "tip izdelka …", "znamka …", "poln model …") into a structured
// title_template. Falls back to DefaultRetailTitleFormula when parsing yields
// fewer than two variable slots.
func DeriveTitleFormulaFromLegacyNameRules(titleRules string) TitleFormula {
	slots := extractLegacyNameSlots(titleRules)
	out := TitleFormula{Separator: " ", Elements: make([]TitleFormulaElement, 0, len(slots))}
	seen := map[string]struct{}{}
	for _, slot := range slots {
		el, ok := mapLegacyNameSlot(slot)
		if !ok {
			continue
		}
		key := el.Type + ":" + strings.ToLower(el.Value)
		if _, dup := seen[key]; dup {
			continue
		}
		seen[key] = struct{}{}
		el.ID = fmt.Sprintf("%d-%s-%s", len(out.Elements), el.Type, sanitizeFormulaID(el.Value))
		out.Elements = append(out.Elements, el)
	}
	if countFormulaVariables(out) < 2 {
		return ensureRetailTitleCoverage(out)
	}
	return ensureRetailTitleCoverage(out)
}

// DeriveTitleTemplateJSON is the JSON encoding used when repairing categories.title_template.
func DeriveTitleTemplateJSON(titleRules string) string {
	f := DeriveTitleFormulaFromLegacyNameRules(titleRules)
	b, err := json.Marshal(f)
	if err != nil {
		def := DefaultRetailTitleFormula()
		b, _ = json.Marshal(def)
	}
	return string(b)
}

func titleFormulaElements(template any) ([]TitleFormulaElement, bool) {
	if template == nil {
		return nil, false
	}
	switch t := template.(type) {
	case TitleFormula:
		return t.Elements, len(t.Elements) > 0
	case *TitleFormula:
		if t == nil {
			return nil, false
		}
		return t.Elements, len(t.Elements) > 0
	case []byte:
		if len(t) == 0 {
			return nil, false
		}
		var f TitleFormula
		if err := json.Unmarshal(t, &f); err != nil {
			return nil, false
		}
		return f.Elements, len(f.Elements) > 0
	case string:
		s := strings.TrimSpace(t)
		if s == "" || s == "null" || s == "{}" {
			return nil, false
		}
		var f TitleFormula
		if err := json.Unmarshal([]byte(s), &f); err != nil {
			return nil, false
		}
		return f.Elements, len(f.Elements) > 0
	case map[string]any:
		b, err := json.Marshal(t)
		if err != nil {
			return nil, false
		}
		var f TitleFormula
		if err := json.Unmarshal(b, &f); err != nil {
			return nil, false
		}
		return f.Elements, len(f.Elements) > 0
	default:
		b, err := json.Marshal(t)
		if err != nil {
			return nil, false
		}
		var f TitleFormula
		if err := json.Unmarshal(b, &f); err != nil {
			return nil, false
		}
		return f.Elements, len(f.Elements) > 0
	}
}

func extractLegacyNameSlots(rules string) []string {
	rules = strings.TrimSpace(rules)
	if rules == "" {
		return nil
	}
	body := rules
	if m := reNameFormulaTail.FindStringSubmatch(rules); len(m) == 2 {
		body = m[1]
	} else if m := reNameFormulaTailEN.FindStringSubmatch(rules); len(m) == 2 {
		body = m[1]
	}
	raw := reQuotedNameSlot.FindAllStringSubmatch(body, -1)
	out := make([]string, 0, len(raw))
	for _, m := range raw {
		s := strings.TrimSpace(m[1])
		s = strings.Trim(s, `",. `)
		if s == "" || s == "," {
			continue
		}
		out = append(out, s)
	}
	return out
}

func mapLegacyNameSlot(slot string) (TitleFormulaElement, bool) {
	s := strings.TrimSpace(slot)
	if s == "" {
		return TitleFormulaElement{}, false
	}
	low := strings.ToLower(s)
	// English platform-default slots (defaults/default-category-prompts.json) are
	// matched before the Slovenian legacy vocabulary. Slovenian rules below are
	// deliberately unchanged: A1's derived templates are already stored, and adding
	// keywords there would churn them on every repair pass.
	switch {
	case strings.Contains(low, "product model"), strings.Contains(low, "device model"),
		strings.Contains(low, "model of the device"):
		return TitleFormulaElement{Type: "variable", Label: "Model", Value: "product_model", Description: s, Example: "EHT6020"}, true
	case strings.Contains(low, "brand"):
		return TitleFormulaElement{Type: "variable", Label: "Brand", Value: "brand", Description: s, Example: "Samsung"}, true
	case strings.Contains(low, "product type"):
		return TitleFormulaElement{Type: "variable", Label: "Product type", Value: "product_type", Description: s, Example: "Bluetooth speaker"}, true
	case strings.Contains(low, "graphics card"):
		return TitleFormulaElement{Type: "variable", Label: "GPU", Value: "graficna_kartica", Description: s}, true
	case strings.Contains(low, "processor"):
		return TitleFormulaElement{Type: "variable", Label: "CPU", Value: "procesor", Description: s}, true
	case strings.Contains(low, "memory"):
		return TitleFormulaElement{Type: "variable", Label: "Memory", Value: "kapaciteta_ram_pomnilnika", Description: s}, true
	case strings.Contains(low, "screen diagonal"), strings.Contains(low, "screen size"):
		return TitleFormulaElement{Type: "variable", Label: "Diagonal", Value: "diagonala_zaslona", Description: s}, true
	case strings.Contains(low, "refresh rate"):
		return TitleFormulaElement{Type: "variable", Label: "Refresh rate", Value: "refresh_rate", Description: s}, true
	case strings.Contains(low, "weight group"):
		return TitleFormulaElement{Type: "variable", Label: "Weight group", Value: "weight_group", Description: s}, true
	case strings.Contains(low, "wash capacity"), strings.Contains(low, "capacity"):
		return TitleFormulaElement{Type: "variable", Label: "Capacity", Value: "kapaciteta", Description: s}, true
	case strings.Contains(low, "connector type"), strings.Contains(low, "socket type"),
		strings.Contains(low, "cable type"):
		return TitleFormulaElement{Type: "variable", Label: "Connector", Value: "connector", Description: s}, true
	case strings.Contains(low, "colour"), strings.Contains(low, "color"):
		return TitleFormulaElement{Type: "variable", Label: "Color", Value: "barva", Description: s}, true
	case strings.Contains(low, "dimensions"):
		return TitleFormulaElement{Type: "variable", Label: "Dimensions", Value: "dimensions", Description: s}, true
	}
	switch {
	case strings.Contains(low, "znamka"):
		return TitleFormulaElement{Type: "variable", Label: "Brand", Value: "brand", Description: s, Example: "Samsung"}, true
	case strings.Contains(low, "poln model"), strings.Contains(low, "model izdelka"),
		strings.HasPrefix(low, "model "), low == "model", strings.Contains(low, "model uppercase"):
		return TitleFormulaElement{Type: "variable", Label: "Model", Value: "product_model", Description: s, Example: "EHT6020"}, true
	case strings.Contains(low, "tip izdelka"):
		return TitleFormulaElement{Type: "variable", Label: "Product type", Value: "product_type", Description: s, Example: "Bluetooth speaker"}, true
	case strings.HasPrefix(low, "barva"):
		return TitleFormulaElement{Type: "variable", Label: "Color", Value: "barva", Description: s}, true
	case strings.Contains(low, "dimenzij"):
		return TitleFormulaElement{Type: "variable", Label: "Dimensions", Value: "dimensions", Description: s}, true
	case strings.Contains(low, "diagonala"):
		return TitleFormulaElement{Type: "variable", Label: "Diagonal", Value: "diagonala_zaslona", Description: s}, true
	case strings.Contains(low, "pomnilnik"):
		return TitleFormulaElement{Type: "variable", Label: "Memory", Value: "kapaciteta_ram_pomnilnika", Description: s}, true
	case strings.Contains(low, "procesor"):
		return TitleFormulaElement{Type: "variable", Label: "CPU", Value: "procesor", Description: s}, true
	case strings.Contains(low, "grafi"):
		return TitleFormulaElement{Type: "variable", Label: "GPU", Value: "graficna_kartica", Description: s}, true
	case strings.Contains(low, "kapaciteta"):
		return TitleFormulaElement{Type: "variable", Label: "Capacity", Value: "kapaciteta", Description: s}, true
	case strings.Contains(low, "skupina po te") || strings.Contains(low, "teži") || strings.Contains(low, "tezi"):
		return TitleFormulaElement{Type: "variable", Label: "Weight group", Value: "weight_group", Description: s}, true
	case strings.Contains(low, "priklju"):
		return TitleFormulaElement{Type: "variable", Label: "Connector", Value: "connector", Description: s}, true
	case strings.Contains(low, "osvež") || strings.Contains(low, "osvez") || strings.Contains(low, "hz"):
		return TitleFormulaElement{Type: "variable", Label: "Refresh rate", Value: "refresh_rate", Description: s}, true
	case strings.Contains(low, "lowercase") || strings.Contains(low, "sentence case") || strings.Contains(low, "uppercase"):
		// Category-specific type word ("cvrtnik lowercase", "monitor lowercase").
		return TitleFormulaElement{Type: "variable", Label: "Product type", Value: "product_type", Description: s}, true
	default:
		// Keep unrecognized instructional slots as text so order/intent survive.
		if len([]rune(s)) > 80 {
			s = string([]rune(s)[:80])
		}
		return TitleFormulaElement{Type: "text", Label: "Naming detail", Value: s, Description: "Legacy name-formula slot"}, true
	}
}

func countFormulaVariables(f TitleFormula) int {
	n := 0
	for _, el := range f.Elements {
		if el.Type == "variable" {
			n++
		}
	}
	return n
}

func ensureRetailTitleCoverage(f TitleFormula) TitleFormula {
	has := map[string]bool{}
	for _, el := range f.Elements {
		if el.Type != "variable" {
			continue
		}
		k := strings.ToLower(strings.ReplaceAll(strings.TrimSpace(el.Value), "-", "_"))
		has[k] = true
	}
	def := DefaultRetailTitleFormula()
	for _, el := range def.Elements {
		k := strings.ToLower(el.Value)
		if has[k] {
			continue
		}
		// Only inject core retail slots when missing.
		if k == "brand" || k == "product_model" || (k == "product_type" && countFormulaVariables(f) < 2) {
			el.ID = fmt.Sprintf("%d-%s-%s", len(f.Elements), el.Type, el.Value)
			f.Elements = append(f.Elements, el)
			has[k] = true
		}
	}
	if countFormulaVariables(f) < 2 {
		return def
	}
	if f.Separator == "" {
		f.Separator = " "
	}
	return f
}

func sanitizeFormulaID(v string) string {
	v = strings.ToLower(strings.TrimSpace(v))
	v = strings.ReplaceAll(v, " ", "_")
	v = strings.ReplaceAll(v, "-", "_")
	if v == "" {
		return "slot"
	}
	if len(v) > 40 {
		v = v[:40]
	}
	return v
}