descrybe/apps/api/cmd/seed-a1/category_prompts.go

package main

import (
	"context"
	"encoding/json"
	"fmt"
	"log"
	"os"
	"path/filepath"
	"regexp"
	"sort"
	"strings"
	"unicode"

	"github.com/descrybe/descrybe-v2/apps/api/internal/aiprompts"
	"github.com/descrybe/descrybe-v2/apps/api/internal/catalog"
	"github.com/descrybe/descrybe-v2/apps/api/internal/security"
	"github.com/google/uuid"
	"github.com/jackc/pgx/v5/pgxpool"
)

// MaxCategoryPromptRunes bounds stored category.prompt (aligned with campaign prompts).
const MaxCategoryPromptRunes = security.MaxCampaignPromptRunes

// categoryPromptsFile is the committed A1 overlay of legacy Name → Prompt pairs.
type categoryPromptsFile struct {
	Version int                   `json:"version"`
	Source  string                `json:"source"`
	Entries []categoryPromptEntry `json:"entries"`
}

type categoryPromptEntry struct {
	Name     string `json:"name"`
	UniqueID string `json:"unique_id,omitempty"`
	Prompt   string `json:"prompt"`
}

var (
	// Legacy v1 placeholders → v2 {{variables}} used by aiprompts.Render.
	reLegacyDesc = regexp.MustCompile(`(?i)\{\s*""?\s*OPIS\s+IZDELKA\s*""?\s*\}`)
	reLegacyName = regexp.MustCompile(`(?i)\{\s*""?\s*STARO\s+IME\s+IZDELKA\s*""?\s*\}`)
)

func defaultCategoryPromptsPath(archivePath string) string {
	// Prefer live WP dump when present (Sync A1 / seed share the same source of truth).
	if p := catalog.ResolveWPCategoryPromptsPath(""); p != "" {
		return p
	}
	if strings.TrimSpace(archivePath) != "" {
		return filepath.Join(filepath.Dir(archivePath), "a1-category-prompts.json")
	}
	return filepath.Join("..", "..", "scripts", "seed", "a1-category-prompts.json")
}

func loadCategoryPromptsFile(path string) (categoryPromptsFile, error) {
	path = filepath.Clean(strings.TrimSpace(path))
	if path == "" || path == "." {
		return categoryPromptsFile{}, fmt.Errorf("category prompts path is empty")
	}
	base := strings.ToLower(filepath.Base(path))
	if strings.HasPrefix(base, "wp_product_categories") && strings.HasSuffix(base, ".sql") {
		return loadCategoryPromptsFromWPSQL(path)
	}
	// Allowlist the committed seed filename (blocks accidental reads of unrelated dumps).
	if filepath.Base(path) != "a1-category-prompts.json" {
		return categoryPromptsFile{}, fmt.Errorf("refusing unexpected category prompts file name %q", filepath.Base(path))
	}
	raw, err := os.ReadFile(path)
	if err != nil {
		return categoryPromptsFile{}, fmt.Errorf("read category prompts: %w", err)
	}
	if len(raw) > 8<<20 {
		return categoryPromptsFile{}, fmt.Errorf("category prompts file too large (%d bytes)", len(raw))
	}
	var f categoryPromptsFile
	if err := json.Unmarshal(raw, &f); err != nil {
		return categoryPromptsFile{}, fmt.Errorf("parse category prompts JSON: %w", err)
	}
	if len(f.Entries) == 0 {
		return categoryPromptsFile{}, fmt.Errorf("category prompts file has no entries")
	}
	if len(f.Entries) > 5000 {
		return categoryPromptsFile{}, fmt.Errorf("category prompts file has too many entries (%d)", len(f.Entries))
	}
	return f, nil
}

func loadCategoryPromptsFromWPSQL(path string) (categoryPromptsFile, error) {
	raw, err := os.ReadFile(path)
	if err != nil {
		return categoryPromptsFile{}, fmt.Errorf("read wp category prompts: %w", err)
	}
	if len(raw) > 16<<20 {
		return categoryPromptsFile{}, fmt.Errorf("wp category prompts file too large (%d bytes)", len(raw))
	}
	parsed, err := catalog.ParseWPProductCategoriesSQL(string(raw))
	if err != nil {
		return categoryPromptsFile{}, err
	}
	entries := make([]categoryPromptEntry, 0, len(parsed))
	for _, e := range parsed {
		entries = append(entries, categoryPromptEntry{
			Name:     e.Name,
			UniqueID: e.UniqueID,
			Prompt:   e.Prompt,
		})
	}
	if len(entries) == 0 {
		return categoryPromptsFile{}, fmt.Errorf("wp category prompts file has no entries")
	}
	return categoryPromptsFile{
		Version: 1,
		Source:  filepath.Base(path),
		Entries: entries,
	}, nil
}

// modernizeLegacyPromptPlaceholders rewrites v1 {""OPIS IZDELKA""} tokens to {{description}} / {{name}}.
func modernizeLegacyPromptPlaceholders(prompt string) string {
	prompt = reLegacyDesc.ReplaceAllString(prompt, "{{description}}")
	prompt = reLegacyName.ReplaceAllString(prompt, "{{name}}")
	return prompt
}

func prepareCategoryPrompt(prompt string) string {
	prompt = modernizeLegacyPromptPlaceholders(prompt)
	return security.SanitizePrompt(prompt, MaxCategoryPromptRunes)
}

// normalizeCategoryName keys categories for matching (case/space/diacritic tolerant).
func normalizeCategoryName(s string) string {
	s = strings.TrimSpace(s)
	if s == "" {
		return ""
	}
	s = strings.ToLower(s)
	var b strings.Builder
	b.Grow(len(s))
	prevSpace := false
	for _, r := range s {
		r = foldSloveneRune(r)
		if unicode.IsSpace(r) {
			if prevSpace || b.Len() == 0 {
				continue
			}
			b.WriteByte(' ')
			prevSpace = true
			continue
		}
		prevSpace = false
		if unicode.IsLetter(r) || unicode.IsDigit(r) || r == '-' || r == '/' || r == '&' || r == '+' {
			b.WriteRune(r)
			continue
		}
		// Drop punctuation noise from names.
	}
	return strings.TrimSpace(b.String())
}

func foldSloveneRune(r rune) rune {
	switch r {
	case 'č', 'ć':
		return 'c'
	case 'š':
		return 's'
	case 'ž':
		return 'z'
	case 'đ':
		return 'd'
	default:
		return r
	}
}

type categoryPromptApplyResult struct {
	Updated   int
	Unmatched []string
	Skipped   int // empty prompt after sanitize
}

func applyCategoryPrompts(ctx context.Context, pg *pgxpool.Pool, companyID uuid.UUID, promptsPath string) (categoryPromptApplyResult, error) {
	var out categoryPromptApplyResult
	file, err := loadCategoryPromptsFile(promptsPath)
	if err != nil {
		return out, err
	}

	byNorm := make(map[string]string, len(file.Entries))
	byUnique := make(map[string]string, len(file.Entries))
	// Overlay: split each legacy combined Name+Description prompt into role-sectioned
	// enhance USER text (Title / Description / Meta / Attributes). Unique title /
	// description / meta formulas stay in title_template / description_template.
	// Stored under "sl" + "*" so prompt->>'sl' and LangPromptAny both resolve;
	// language stays via {{language}}.
	for _, e := range file.Entries {
		uid := strings.TrimSpace(e.UniqueID)
		name := strings.TrimSpace(e.Name)
		if uid == "" && name == "" {
			out.Skipped++
			continue
		}
		raw := strings.TrimSpace(e.Prompt)
		var canonical string
		if raw == "" {
			out.Skipped++
			continue
		}
		if aiprompts.IsLegacyCombinedEnhancePrompt(raw) {
			canonical = prepareCategoryPrompt(aiprompts.SplitLegacyCombinedEnhancePrompt(raw))
		} else if aiprompts.CategoryEnhanceHasRoleSections(raw) {
			canonical = prepareCategoryPrompt(raw)
		} else {
			canonical = prepareCategoryPrompt(aiprompts.CategoryEnhanceUserTemplate)
		}
		if canonical == "" {
			out.Skipped++
			continue
		}
		if uid != "" {
			byUnique[strings.ToLower(uid)] = canonical
		}
		if name != "" {
			key := normalizeCategoryName(name)
			if key == "" {
				if uid == "" {
					out.Skipped++
				}
				continue
			}
			byNorm[key] = canonical
		}
	}
	if len(byNorm) == 0 && len(byUnique) == 0 {
		return out, fmt.Errorf("no usable category prompts in %s", promptsPath)
	}

	rows, err := pg.Query(ctx, `
		SELECT id, COALESCE(unique_id, ''), name
		FROM categories
		WHERE company_id = $1`, companyID)
	if err != nil {
		return out, fmt.Errorf("list categories: %w", err)
	}
	defer rows.Close()

	ids := make([]uuid.UUID, 0, len(byNorm)+len(byUnique))
	prompts := make([]string, 0, len(byNorm)+len(byUnique))
	matchedNorm := make(map[string]struct{}, len(byNorm))
	matchedUID := make(map[string]struct{}, len(byUnique))

	for rows.Next() {
		var id uuid.UUID
		var uniqueID, name string
		if err := rows.Scan(&id, &uniqueID, &name); err != nil {
			return out, err
		}
		prompt := ""
		if uidKey := strings.ToLower(strings.TrimSpace(uniqueID)); uidKey != "" {
			if p, ok := byUnique[uidKey]; ok {
				prompt = p
				matchedUID[uidKey] = struct{}{}
			}
		}
		if prompt == "" {
			key := normalizeCategoryName(name)
			if p, ok := byNorm[key]; ok {
				prompt = p
				matchedNorm[key] = struct{}{}
			}
		}
		if prompt == "" {
			continue
		}
		ids = append(ids, id)
		prompts = append(prompts, prompt)
	}
	if err := rows.Err(); err != nil {
		return out, err
	}

	for key := range byUnique {
		if _, ok := matchedUID[key]; !ok {
			out.Unmatched = append(out.Unmatched, "uid:"+key)
		}
	}
	for key := range byNorm {
		if _, ok := matchedNorm[key]; !ok {
			// Name unmatched is noise when unique_id matched the same row; still report for seed hygiene.
			out.Unmatched = append(out.Unmatched, "name:"+key)
		}
	}
	sort.Strings(out.Unmatched)

	if len(ids) == 0 {
		return out, fmt.Errorf("no A1 categories matched any of %d seed prompts (check unique_id/names)", len(byNorm)+len(byUnique))
	}

	// Single parameterized batch update — company_id gate prevents cross-tenant writes.
	// sl + * = prompt->>'sl' and company.LangPromptAny ({{language}} at render).
	tag, err := pg.Exec(ctx, `
		UPDATE categories AS c
		SET prompt = jsonb_build_object('sl', v.prompt, '*', v.prompt), updated_at = now()
		FROM (
			SELECT * FROM unnest($1::uuid[], $2::text[]) AS t(id, prompt)
		) AS v
		WHERE c.id = v.id AND c.company_id = $3`, ids, prompts, companyID)
	if err != nil {
		return out, fmt.Errorf("update category prompts: %w", err)
	}
	out.Updated = int(tag.RowsAffected())
	return out, nil
}

func logCategoryPromptResult(res categoryPromptApplyResult, path string) {
	log.Printf("category prompts from %s: updated=%d skipped=%d unmatched=%d",
		path, res.Updated, res.Skipped, len(res.Unmatched))
	if len(res.Unmatched) == 0 {
		return
	}
	const maxShow = 20
	show := res.Unmatched
	if len(show) > maxShow {
		show = show[:maxShow]
	}
	log.Printf("  unmatched seed names (normalized, first %d): %s", len(show), strings.Join(show, ", "))
}