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

package aiprompts

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

var (
	reLegacyHTMLBlock = regexp.MustCompile(`(?is)<\s*(h[1-4]|p|ul|ol|li|b|strong)\s*>\s*\{?\s*(.*?)\s*\}?\s*<\s*/\s*(?:h[1-4]|p|ul|ol|li|b|strong)\s*>`)
	reLegacyBoldUL    = regexp.MustCompile(`(?is)<\s*b\s*>\s*\{?\s*(.*?)\s*\}?\s*<\s*/\s*b\s*>\s*<\s*ul\s*>\s*<\s*li\s*>\s*\{?\s*(.*?)\s*\}?\s*<\s*/\s*li\s*>\s*<\s*/\s*ul\s*>`)
)

// DescriptionFormulaSection is one ordered block in categories.description_template.
type DescriptionFormulaSection struct {
	Type         string `json:"type"`
	Instructions string `json:"instructions"`
}

// DescriptionFormula is the JSON shape stored in categories.description_template
// (sections + optional SEO meta instruction strings).
type DescriptionFormula struct {
	Sections        []DescriptionFormulaSection `json:"sections,omitempty"`
	MetaTitle       string                      `json:"metaTitle,omitempty"`
	MetaDescription string                      `json:"metaDescription,omitempty"`
}

// DescriptionTemplateNeedsRepair is true when template is missing or has no sections
// (and no usable meta instructions).
func DescriptionTemplateNeedsRepair(template any) bool {
	f, ok := parseDescriptionFormula(template)
	if !ok {
		return true
	}
	if len(f.Sections) > 0 {
		return false
	}
	return strings.TrimSpace(f.MetaTitle) == "" && strings.TrimSpace(f.MetaDescription) == ""
}

// DefaultLegacyMetaTitleRule fills description_template.metaTitle for legacy A1
// prompts, which only carried a <metaDescription> instruction. Mirrors the legacy
// name-first Slovenian instruction style with the SEO title length bound.
const DefaultLegacyMetaTitleRule = "Najprej napiši Novo ime izdelka in dodaj glavno prednost. Dolžina 50-60 znakov vključno s presledki."

// DeriveDescriptionFormulaFromLegacyParts builds description_template from split
// legacy DescriptionRules HTML + MetaRules. Legacy dumps never carried a meta
// title rule, so meta intent (MetaRules present) also seeds a default metaTitle
// instruction — the category UI then shows all four prompt areas filled.
func DeriveDescriptionFormulaFromLegacyParts(parts LegacyEnhanceParts) DescriptionFormula {
	out := DescriptionFormula{
		MetaDescription: strings.TrimSpace(parts.MetaRules),
	}
	if out.MetaDescription != "" {
		out.MetaTitle = DefaultLegacyMetaTitleRule
	}
	body := strings.TrimSpace(parts.DescriptionRules)
	if body == "" {
		return out
	}

	// Walk left-to-right so section order matches the legacy HTML template.
	remaining := body
	for remaining != "" {
		boldUL := reLegacyBoldUL.FindStringSubmatchIndex(remaining)
		block := reLegacyHTMLBlock.FindStringSubmatchIndex(remaining)
		switch {
		case boldUL != nil && (block == nil || boldUL[0] <= block[0]):
			m := reLegacyBoldUL.FindStringSubmatch(remaining[boldUL[0]:boldUL[1]])
			heading, instr := "", ""
			if len(m) >= 3 {
				heading = cleanLegacyInstruction(m[1])
				instr = cleanLegacyInstruction(m[2])
			}
			if heading != "" && instr != "" && !strings.Contains(strings.ToLower(instr), strings.ToLower(heading)) {
				instr = heading + ": " + instr
			} else if instr == "" {
				instr = heading
			}
			if strings.TrimSpace(instr) != "" {
				out.Sections = append(out.Sections, DescriptionFormulaSection{Type: "ul", Instructions: instr})
			}
			remaining = remaining[boldUL[1]:]
		case block != nil:
			m := reLegacyHTMLBlock.FindStringSubmatch(remaining[block[0]:block[1]])
			if len(m) >= 3 {
				typ := strings.ToLower(strings.TrimSpace(m[1]))
				instr := cleanLegacyInstruction(m[2])
				if instr != "" {
					switch typ {
					case "b", "strong":
						typ = "h2"
					case "li", "ol":
						typ = "ul"
					}
					out.Sections = append(out.Sections, DescriptionFormulaSection{Type: typ, Instructions: instr})
				}
			}
			remaining = remaining[block[1]:]
		default:
			return out
		}
	}
	return out
}

// DeriveDescriptionTemplateJSON encodes DeriveDescriptionFormulaFromLegacyParts for DB write.
func DeriveDescriptionTemplateJSON(parts LegacyEnhanceParts) string {
	f := DeriveDescriptionFormulaFromLegacyParts(parts)
	if len(f.Sections) == 0 && strings.TrimSpace(f.MetaTitle) == "" && strings.TrimSpace(f.MetaDescription) == "" {
		return ""
	}
	b, err := json.Marshal(f)
	if err != nil {
		return ""
	}
	return string(b)
}

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