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

package processing

import (
	"encoding/json"
	"fmt"
	"sort"
	"strconv"
	"strings"

	"github.com/descrybe/descrybe-v2/apps/api/internal/aiprompts"
	"github.com/descrybe/descrybe-v2/apps/api/internal/company"
)

// AppendFormulaConstraints appends language-agnostic title/description/meta formula
// guidance to the enhance user template (before {{var}} render). Empty templates
// are no-ops. Shared skeleton stays in company/built-in prompts; formulas only
// constrain structure for the active language via {{language}} elsewhere.
// Meta instructions come from description_template.metaTitle / metaDescription
// (legacy A1 / cats.json) and are distinct from HTML description sections.
// Attribute allowlist / formula-key guidance is AppendAttributeConstraints.
// When omitSEOMeta is true (A1 cohort), meta formula blocks are skipped.
func AppendFormulaConstraints(userTpl string, titleTemplate, descriptionTemplate any, omitSEOMeta bool) string {
	userTpl = strings.TrimSpace(userTpl)
	blocks := []string{
		FormatTitleFormulaConstraint(titleTemplate),
		FormatDescriptionFormulaConstraint(descriptionTemplate),
	}
	if !omitSEOMeta {
		blocks = append(blocks, FormatMetaFormulaConstraint(descriptionTemplate))
	}
	var joined strings.Builder
	for _, block := range blocks {
		if block == "" {
			continue
		}
		if joined.Len() > 0 {
			joined.WriteString("\n\n")
		}
		joined.WriteString(block)
	}
	if joined.Len() == 0 {
		return userTpl
	}
	if userTpl == "" {
		return joined.String()
	}
	return userTpl + "\n\n" + joined.String()
}

// FormatTitleSectionConstraint turns free-form --- Title --- section instructions
// into a required enhance constraint when no structured title_template elements exist.
func FormatTitleSectionConstraint(categoryPrompt string) string {
	body := strings.TrimSpace(aiprompts.TitleSectionInstructions(categoryPrompt))
	if body == "" || isBuiltInTitleRoleBoilerplate(body) {
		return ""
	}
	var b strings.Builder
	b.WriteString("Title instructions (REQUIRED — overrides any shorter \"short retail title\" rule):\n")
	b.WriteString(body)
	b.WriteString("\nBuild JSON \"name\" exactly per these instructions in {{language}}.")
	b.WriteString(" Never copy the supplier Name unchanged when instructions ask for a new name or formula.")
	return b.String()
}

func isBuiltInTitleRoleBoilerplate(body string) bool {
	lower := strings.ToLower(strings.TrimSpace(body))
	return strings.HasPrefix(lower, "role: title")
}

// AppendTitleSectionConstraint appends FormatTitleSectionConstraint when the
// category Title role has text and structured title_template is empty/unusable.
func AppendTitleSectionConstraint(userTpl, categoryPrompt string, titleTemplate any) string {
	if FormatTitleFormulaConstraint(titleTemplate) != "" {
		return strings.TrimSpace(userTpl)
	}
	block := FormatTitleSectionConstraint(categoryPrompt)
	if block == "" {
		return strings.TrimSpace(userTpl)
	}
	userTpl = strings.TrimSpace(userTpl)
	if strings.Contains(userTpl, "Title instructions (REQUIRED") {
		return userTpl
	}
	if userTpl == "" {
		return block
	}
	return userTpl + "\n\n" + block
}

// MaxAttrAllowlistPromptKeys caps Allowed attribute keys listed in enhance prompts.
const MaxAttrAllowlistPromptKeys = 40

// AppendAttributeConstraints appends category_attributes allowlist + title-formula
// variable keys so the model can emit JSON "attrs" validated via AttrsForEnhance.
// No-op when both allowlist and formula keys are empty.
func AppendAttributeConstraints(userTpl string, allowed map[string]struct{}, titleTemplate any) string {
	userTpl = strings.TrimSpace(userTpl)
	block := FormatAttributeAllowlistConstraint(allowed, titleTemplate)
	if block == "" {
		return userTpl
	}
	if userTpl == "" {
		return block
	}
	if strings.Contains(userTpl, `Allowed attribute keys (JSON "attrs" object`) {
		return userTpl
	}
	return userTpl + "\n\n" + block
}

// FormatAttributeAllowlistConstraint lists category allowlist keys and title-formula
// attr slots for the Attributes enhance role. When allowed is nil, only formula
// keys are listed (unit-test / sanitize-only paths).
func FormatAttributeAllowlistConstraint(allowed map[string]struct{}, titleTemplate any) string {
	keys := preferredAttrKeyList(allowed, MaxAttrAllowlistPromptKeys)
	formulaKeys := titleFormulaAttrKeys(titleTemplate)
	if len(keys) == 0 && len(formulaKeys) == 0 {
		return ""
	}
	var b strings.Builder
	b.WriteString("Allowed attribute keys (JSON \"attrs\" object — remap feed labels onto these; omit unknowns):\n")
	if len(keys) > 0 {
		b.WriteString("- category: ")
		b.WriteString(strings.Join(keys, ", "))
		b.WriteByte('\n')
	} else {
		b.WriteString("- category: (core characteristics only — brand, product_model, dims, …)\n")
	}
	if len(formulaKeys) > 0 {
		b.WriteString("- title formula slots: ")
		b.WriteString(strings.Join(formulaKeys, ", "))
		b.WriteByte('\n')
	}
	b.WriteString("Fill missing keys only from Name/Description/Category/Attrs evidence; never invent specs or zero dimensions.")
	return strings.TrimSpace(b.String())
}

func preferredAttrKeyList(allowed map[string]struct{}, maxKeys int) []string {
	if len(allowed) == 0 || maxKeys <= 0 {
		return nil
	}
	prefer := preferredAllowedKeys(allowed)
	seen := map[string]struct{}{}
	out := make([]string, 0, len(prefer))
	for _, k := range prefer {
		k = strings.TrimSpace(k)
		if k == "" {
			continue
		}
		if _, ok := seen[k]; ok {
			continue
		}
		seen[k] = struct{}{}
		out = append(out, k)
	}
	sort.Strings(out)
	if len(out) > maxKeys {
		out = out[:maxKeys]
	}
	return out
}

func titleFormulaAttrKeys(template any) []string {
	elements, _, ok := parseTitleFormula(template)
	if !ok || len(elements) == 0 {
		return nil
	}
	seen := map[string]struct{}{}
	var out []string
	for _, el := range elements {
		if el.Type != "variable" {
			continue
		}
		k := strings.TrimSpace(el.Value)
		if k == "" {
			continue
		}
		canon := canonicalizeAttrKey(k)
		if canon == "" {
			canon = strings.ToLower(k)
		}
		if _, ok := seen[canon]; ok {
			continue
		}
		seen[canon] = struct{}{}
		out = append(out, canon)
	}
	sort.Strings(out)
	return out
}

// FormatTitleFormulaConstraint turns categories.title_template into enhance-prompt
// constraints (element order: literal text + attribute variable keys).
func FormatTitleFormulaConstraint(template any) string {
	elements, separator, ok := parseTitleFormula(template)
	if !ok || len(elements) == 0 {
		return ""
	}
	sep := separator
	if sep == "" {
		sep = " "
	}
	var b strings.Builder
	b.WriteString("Title formula (order matters; join with ")
	b.WriteString(strconv.Quote(sep))
	b.WriteString("). Build name from Attrs using this structure:\n")
	for i, el := range elements {
		switch el.Type {
		case "text":
			fmt.Fprintf(&b, "%d. text %s\n", i+1, strconv.Quote(el.Value))
		case "variable":
			key := strings.TrimSpace(el.Value)
			if key == "" {
				continue
			}
			fmt.Fprintf(&b, "%d. attr [%s]\n", i+1, key)
		default:
			continue
		}
	}
	b.WriteString("Prefer Attrs values for [attr] slots; keep literal text as written; write name in {{language}}; never emit brand-only when product_type or product_model slots exist.")
	return strings.TrimSpace(b.String())
}

// FormatMetaFormulaConstraint turns description_template.metaTitle / metaDescription
// (cats.json / A1 SEO instructions) into enhance-prompt constraints distinct from
// HTML description sections.
func FormatMetaFormulaConstraint(template any) string {
	metaTitle, metaDesc := parseMetaFormulaInstructions(template)
	if metaTitle == "" && metaDesc == "" {
		return ""
	}
	var b strings.Builder
	b.WriteString("SEO meta formula (REQUIRED — distinct from description HTML; plain text only):\n")
	b.WriteString("Emit meta_title and meta_description as separate JSON fields in {{language}}.\n")
	if metaTitle != "" {
		fmt.Fprintf(&b, "- meta_title (50-60 chars): %s\n", metaTitle)
	} else {
		b.WriteString("- meta_title: 50-60 chars, product name + main benefit or use case\n")
	}
	if metaDesc != "" {
		fmt.Fprintf(&b, "- meta_description (120-155 chars): %s\n", metaDesc)
	} else {
		b.WriteString("- meta_description: 120-155 chars, factual SEO snippet; never HTML\n")
	}
	return strings.TrimSpace(b.String())
}

// FormatDescriptionFormulaConstraint turns categories.description_template sections
// into bullet instructions for the enhance user prompt.
func FormatDescriptionFormulaConstraint(template any) string {
	sections, ok := parseDescriptionFormulaSections(template)
	if !ok || len(sections) == 0 {
		return ""
	}
	var b strings.Builder
	b.WriteString("Description formula (REQUIRED — overrides any shorter \"1-2 sentences\" rule):\n")
	b.WriteString("Emit description as ONE HTML string covering each section in order in {{language}}.\n")
	b.WriteString("Use tags matching section type: h1/h2/h3/h4 → <hN>…</hN>, p → <p>…</p>, ul → <ul><li>…</li></ul>.\n")
	for _, s := range sections {
		typ := strings.TrimSpace(s.Type)
		instr := strings.TrimSpace(s.Instructions)
		if typ == "" && instr == "" {
			continue
		}
		if typ == "" {
			typ = "section"
		}
		if instr == "" {
			fmt.Fprintf(&b, "- %s\n", typ)
			continue
		}
		fmt.Fprintf(&b, "- %s: %s\n", typ, instr)
	}
	return strings.TrimSpace(b.String())
}

// descriptionFormulaSystemOverride is appended to the enhance system template when a
// category description_template is present so company/built-in "1-2 sentences" rules
// cannot override the per-category formula (A1 category definition pages).
const descriptionFormulaSystemOverride = `When the user message includes a "Description formula", obey those sections over any shorter "1-2 sentences" / "1-3 paragraphs" guidance: emit one HTML string using tags matching each section type (h1/h2/h3/h4, p, ul), in order, in {{language}}. Never ignore the Description formula.`

// AppendDescriptionFormulaSystemOverride strengthens the system prompt when a
// description formula is active. No-op when template is empty/unparseable.
func AppendDescriptionFormulaSystemOverride(systemTpl string, descriptionTemplate any) string {
	if FormatDescriptionFormulaConstraint(descriptionTemplate) == "" {
		return strings.TrimSpace(systemTpl)
	}
	systemTpl = strings.TrimSpace(systemTpl)
	if systemTpl == "" {
		return descriptionFormulaSystemOverride
	}
	if strings.Contains(systemTpl, "Description formula") {
		return systemTpl
	}
	return systemTpl + "\n" + descriptionFormulaSystemOverride
}

// titleFormulaSystemOverride is appended to the enhance system template when a
// category title_template is present so company/built-in "short retail title"
// rules cannot ignore the Title formula (mirrors description formula override).
const titleFormulaSystemOverride = `When the user message includes a "Title formula", obey that structure for "name" over any shorter "short retail title" guidance: build name from Attrs in the given element order, keep literal text as written, write name in {{language}}. Never ignore the Title formula. Never emit brand-only when the formula includes product_type or product_model.`

// AppendTitleFormulaSystemOverride strengthens the system prompt when a title
// formula is active. No-op when template is empty/unparseable.
func AppendTitleFormulaSystemOverride(systemTpl string, titleTemplate any) string {
	if FormatTitleFormulaConstraint(titleTemplate) == "" {
		return strings.TrimSpace(systemTpl)
	}
	systemTpl = strings.TrimSpace(systemTpl)
	if systemTpl == "" {
		return titleFormulaSystemOverride
	}
	if strings.Contains(systemTpl, `When the user message includes a "Title formula"`) {
		return systemTpl
	}
	return systemTpl + "\n" + titleFormulaSystemOverride
}

// titlePromptRewriteOverride covers free-form Title section instructions
// (categories.prompt --- Title ---) that are not structured title_template.
const titlePromptRewriteOverride = `When the category Title section asks for a new product name or naming formula, "name" MUST be rewritten — never copy the supplier Name unchanged when type/brand/model/color (or other Title cues) are available. Obey the Title section over any "keep short retail title" guidance.`

// AppendTitlePromptRewriteOverride strengthens the system prompt when the
// category Title role requires a rewrite. No-op when Title does not ask for one.
func AppendTitlePromptRewriteOverride(systemTpl, categoryPrompt string) string {
	if !aiprompts.CategoryTitlePromptRequiresRewrite(categoryPrompt) {
		return strings.TrimSpace(systemTpl)
	}
	systemTpl = strings.TrimSpace(systemTpl)
	if systemTpl == "" {
		return titlePromptRewriteOverride
	}
	if strings.Contains(systemTpl, `"name" MUST be rewritten`) {
		return systemTpl
	}
	return systemTpl + "\n" + titlePromptRewriteOverride
}

// categoryEnhanceSystemOverlay is appended when categories.prompt (enhance overlay)
// is active so free-form category copy drives title, description, AND attributes —
// not description alone — while Title/Description formulas keep structural precedence.
const categoryEnhanceSystemOverlay = `When the user message includes category guidance, apply it to "name", "description", and "attrs" (tone/focus for the title and body, plus attribute extraction onto Allowed attribute keys). Structural Title/Description formula blocks and Allowed attribute keys in the user message still take precedence when present.`

// AppendCategoryEnhanceSystemOverlay strengthens the system prompt when a
// per-category enhance overlay (categories.prompt) is set. No-op when empty.
func AppendCategoryEnhanceSystemOverlay(systemTpl, categoryPrompt string) string {
	if strings.TrimSpace(categoryPrompt) == "" {
		return strings.TrimSpace(systemTpl)
	}
	systemTpl = strings.TrimSpace(systemTpl)
	if systemTpl == "" {
		return categoryEnhanceSystemOverlay
	}
	marker := `apply it to "name", "description", and "attrs"`
	if strings.Contains(systemTpl, marker) {
		return systemTpl
	}
	// Upgrade older name+description-only overlay without stacking duplicates.
	legacy := `apply it to BOTH "name" and "description"`
	if strings.Contains(systemTpl, legacy) {
		return strings.Replace(systemTpl, legacy, marker, 1)
	}
	return systemTpl + "\n" + categoryEnhanceSystemOverlay
}

// metaFormulaSystemOverride is appended when description_template carries metaTitle
// / metaDescription instructions so enhance emits SEO fields separately from HTML.
const metaFormulaSystemOverride = `When the user message includes an "SEO meta formula", emit meta_title and meta_description as plain SEO text (not HTML) distinct from description. Obey the character guidance in the formula.`

// AppendMetaFormulaSystemOverride strengthens the system prompt when meta formula
// instructions are present on description_template. No-op when absent.
func AppendMetaFormulaSystemOverride(systemTpl string, descriptionTemplate any) string {
	if FormatMetaFormulaConstraint(descriptionTemplate) == "" {
		return strings.TrimSpace(systemTpl)
	}
	systemTpl = strings.TrimSpace(systemTpl)
	if systemTpl == "" {
		return metaFormulaSystemOverride
	}
	if strings.Contains(systemTpl, "SEO meta formula") {
		return systemTpl
	}
	return systemTpl + "\n" + metaFormulaSystemOverride
}

// descriptionSatisfiesFormula reports whether desc includes the HTML tags required
// by categories.description_template sections. No formula → always true.
func descriptionSatisfiesFormula(desc string, template any) bool {
	sections, ok := parseDescriptionFormulaSections(template)
	if !ok || len(sections) == 0 {
		return true
	}
	desc = strings.TrimSpace(desc)
	if desc == "" || desc == "<nil>" {
		return false
	}
	types := make([]string, 0, len(sections))
	for _, s := range sections {
		types = append(types, s.Type)
	}
	return !company.DescriptionMissingFormulaHTMLTags(desc, types)
}

type titleFormulaElement struct {
	Type  string `json:"type"`
	Value string `json:"value"`
}

type descriptionFormulaSection struct {
	Type         string `json:"type"`
	Instructions string `json:"instructions"`
}

func parseTitleFormula(template any) (elements []titleFormulaElement, separator string, ok bool) {
	if template == nil {
		return nil, "", false
	}
	obj, err := asObjectMap(template)
	if err != nil || obj == nil {
		return nil, "", false
	}
	sep, _ := obj["separator"].(string)
	rawEls, exists := obj["elements"]
	if !exists || rawEls == nil {
		return nil, "", false
	}
	b, err := json.Marshal(rawEls)
	if err != nil {
		return nil, "", false
	}
	var parsed []titleFormulaElement
	if err := json.Unmarshal(b, &parsed); err != nil {
		return nil, "", false
	}
	out := make([]titleFormulaElement, 0, len(parsed))
	for _, el := range parsed {
		t := strings.ToLower(strings.TrimSpace(el.Type))
		v := strings.TrimSpace(el.Value)
		if t != "text" && t != "variable" {
			continue
		}
		if v == "" {
			continue
		}
		out = append(out, titleFormulaElement{Type: t, Value: v})
	}
	if len(out) == 0 {
		return nil, "", false
	}
	return out, sep, true
}

func parseDescriptionFormulaSections(template any) ([]descriptionFormulaSection, bool) {
	if template == nil {
		return nil, false
	}
	obj, err := asObjectMap(template)
	if err != nil || obj == nil {
		return nil, false
	}
	raw, exists := obj["sections"]
	if !exists || raw == nil {
		return nil, false
	}
	b, err := json.Marshal(raw)
	if err != nil {
		return nil, false
	}
	var parsed []descriptionFormulaSection
	if err := json.Unmarshal(b, &parsed); err != nil {
		return nil, false
	}
	out := make([]descriptionFormulaSection, 0, len(parsed))
	for _, s := range parsed {
		t := strings.ToLower(strings.TrimSpace(s.Type))
		instr := strings.TrimSpace(s.Instructions)
		if t == "" && instr == "" {
			continue
		}
		out = append(out, descriptionFormulaSection{Type: t, Instructions: instr})
	}
	return out, len(out) > 0
}

// parseMetaFormulaInstructions reads metaTitle / metaDescription instruction
// strings from description_template (camelCase as stored by the category UI).
func parseMetaFormulaInstructions(template any) (metaTitle, metaDescription string) {
	if template == nil {
		return "", ""
	}
	obj, err := asObjectMap(template)
	if err != nil || obj == nil {
		return "", ""
	}
	metaTitle = strings.TrimSpace(stringFromAny(obj["metaTitle"]))
	if metaTitle == "" {
		metaTitle = strings.TrimSpace(stringFromAny(obj["meta_title"]))
	}
	metaDescription = strings.TrimSpace(stringFromAny(obj["metaDescription"]))
	if metaDescription == "" {
		metaDescription = strings.TrimSpace(stringFromAny(obj["meta_description"]))
	}
	return metaTitle, metaDescription
}

// categoryFormulasFor resolves title/description templates for a category key
// (unique_id or name). Explicit ProductInput fields win over the job cache map.
func categoryFormulasFor(in ProductInput, category string) (title, description any) {
	if in.TitleTemplate != nil || in.DescriptionTemplate != nil {
		return in.TitleTemplate, in.DescriptionTemplate
	}
	key := strings.ToLower(strings.TrimSpace(category))
	if key == "" || len(in.CategoryFormulasByKey) == 0 {
		return nil, nil
	}
	f, ok := in.CategoryFormulasByKey[key]
	if !ok {
		return nil, nil
	}
	return f.TitleTemplate, f.DescriptionTemplate
}

func asObjectMap(v any) (map[string]any, error) {
	switch t := v.(type) {
	case map[string]any:
		return t, nil
	case nil:
		return nil, nil
	default:
		b, err := json.Marshal(t)
		if err != nil {
			return nil, err
		}
		if len(b) == 0 || string(b) == "null" {
			return nil, nil
		}
		var obj map[string]any
		if err := json.Unmarshal(b, &obj); err != nil {
			return nil, err
		}
		return obj, nil
	}
}