descrybe/apps/api/cmd/_final_verify_quick_tmp/main.go

// Temporary slim LIVE verify — 2 products, short timeouts. Do not commit.
package main

import (
	"context"
	"encoding/json"
	"fmt"
	"io"
	"log"
	"net/http"
	"os"
	"path/filepath"
	"strings"
	"time"

	"github.com/descrybe/descrybe-v2/apps/api/internal/aiprompts"
	"github.com/descrybe/descrybe-v2/apps/api/internal/aiprovider"
	"github.com/descrybe/descrybe-v2/apps/api/internal/billing"
	"github.com/descrybe/descrybe-v2/apps/api/internal/config"
	"github.com/descrybe/descrybe-v2/apps/api/internal/db"
	"github.com/descrybe/descrybe-v2/apps/api/internal/eprel"
	"github.com/descrybe/descrybe-v2/apps/api/internal/logredact"
	"github.com/descrybe/descrybe-v2/apps/api/internal/platformsettings"
	"github.com/descrybe/descrybe-v2/apps/api/internal/processing"
	"github.com/google/uuid"
)

func main() {
	out := `F:\laragon\www\_MY\descrybe-v2\.codehelper\_final_verify_20260817`
	_ = os.MkdirAll(out, 0o755)
	logPath := filepath.Join(out, "inproc_run_quick.log")
	logFile, err := os.Create(logPath)
	if err != nil {
		log.Fatalf("log: %v", err)
	}
	defer logFile.Close()
	log.SetOutput(logredact.Writer(io.MultiWriter(os.Stderr, logFile)))

	repoRoot := `F:\laragon\www\_MY\descrybe-v2`
	wpPath := filepath.Join(repoRoot, "scripts", "seed", "wp_product_categories.sql")
	wpBytes, err := os.ReadFile(wpPath)
	if err != nil {
		failReport(out, "FAIL", "repo seed missing: "+err.Error(), nil)
	}
	_ = os.Setenv("SEED_A1_WP_CATEGORIES", wpPath)

	eans := []string{
		"4548736132597", // Slušalke
		"195348253666",  // Gaming monitorji
	}
	companyID := uuid.MustParse("604f23a8-b66e-4b21-8b45-0d72b68f4790")
	userID := uuid.MustParse("6bf00877-a693-4d77-b28e-8c8292adac98")

	cfg, err := config.Load()
	if err != nil {
		failReport(out, "FAIL", "config: "+err.Error(), nil)
	}
	ctx := context.Background()
	pool, err := db.NewPool(ctx, cfg.DatabaseURL, db.PoolOptions{
		MaxConns: int32(cfg.DBMaxConns), MinConns: int32(cfg.DBMinConns),
		MaxConnLifetime: cfg.DBMaxConnLifetime, MaxConnLifetimeJitter: cfg.DBMaxConnLifetimeJitter,
		MaxConnIdleTime: cfg.DBMaxConnIdleTime, HealthCheckPeriod: cfg.DBHealthCheckPeriod,
		StatementTimeout: cfg.DBStatementTimeout,
	})
	if err != nil {
		failReport(out, "FAIL", "db: "+err.Error(), nil)
	}
	defer pool.Close()

	// --- Sync A1 from repo seed only ---
	var legacy string
	_ = pool.QueryRow(ctx, `SELECT COALESCE(legacy_company_id,'') FROM companies WHERE id=$1`, companyID).Scan(&legacy)
	a1Cohort := billing.IsA1CohortCompany(legacy, "A1 Slovenija")
	syncRes, serr := processing.SyncCompanyA1(ctx, pool, companyID, "A1 Slovenija", a1Cohort, processing.SyncCompanyA1Opts{
		SkipDumpBackfill: true,
		FixCompanyCatalogOpts: processing.FixCompanyCatalogOpts{
			BackfillCategories: true,
			WPCategoriesSQL:    wpBytes,
			AIPrompts:          aiprompts.NewService(pool),
		},
	})
	syncInfo := map[string]any{"wp_path": wpPath, "wp_bytes": len(wpBytes), "error": nil, "result": syncRes}
	if serr != nil {
		syncInfo["error"] = serr.Error()
		writeJSON(out, "sync_quick.json", syncInfo)
		failReport(out, "FAIL", "SyncCompanyA1: "+serr.Error(), syncInfo)
	}
	writeJSON(out, "sync_quick.json", syncInfo)
	log.Printf("sync ok prompts_already_ok=%d wp_entries=%d", syncRes.CategoryPromptsAlreadyOK, syncRes.WPCategoriesEntries)

	platEnv := platformsettings.EnvConfig{
		AppEncryptionKey: cfg.AppEncryptionKey, CredentialsEncryptionKey: cfg.CredentialsEncryptionKey,
		TokenSigningSecret: cfg.TokenSigningSecret, DatabaseURL: cfg.DatabaseURL,
		OpenAIAPIKey: cfg.OpenAIAPIKey, OpenAIBaseURL: cfg.OpenAIBaseURL, OpenAIModel: cfg.OpenAIModel,
	}
	plat := platformsettings.NewService(pool, platEnv)
	aiSvc := aiprovider.NewService(pool, aiprovider.EnvConfig{
		AppEncryptionKey: cfg.AppEncryptionKey, CredentialsEncryptionKey: cfg.CredentialsEncryptionKey,
		TokenSigningSecret: cfg.TokenSigningSecret, DatabaseURL: cfg.DatabaseURL,
		OpenAIAPIKey: cfg.OpenAIAPIKey, OpenAIBaseURL: cfg.OpenAIBaseURL, OpenAIModel: cfg.OpenAIModel,
		ProcessingRPM: cfg.ProcessingRPM, ProcessingMaxRetries: cfg.ProcessingMaxRetries,
	})
	aiSvc.Platform = plat

	roleCfg, rerr := plat.ResolveAIConfig(ctx, platformsettings.AIRoleProcessing)
	if rerr != nil {
		failReport(out, "FAIL", "ResolveAIConfig: "+rerr.Error(), nil)
	}
	completer, modeLabel, byok, cerr := aiSvc.ResolveCompleterForRole(ctx, companyID, aiprovider.RoleProcessing)
	if cerr != nil {
		failReport(out, "FAIL", "ResolveCompleter: "+cerr.Error(), nil)
	}
	client, ok := completer.(*processing.OpenAIClient)
	if !ok || client == nil {
		failReport(out, "FAIL", fmt.Sprintf("completer type %T", completer), nil)
	}
	base := strings.TrimSpace(client.BaseURL)
	model := strings.TrimSpace(client.Model)
	resolve := map[string]any{
		"role_source": roleCfg.Source, "role_provider": roleCfg.Provider,
		"completer_base": base, "completer_model": model,
		"mode": modeLabel, "byok": byok, "env_ignored": cfg.OpenAIBaseURL,
	}
	if processing.IsMockOrLoopbackBaseURL(base) || strings.Contains(base, "18767") {
		resolve["fail_reason"] = "mock/loopback refused"
		writeJSON(out, "resolve_quick.json", resolve)
		failReport(out, "FAIL", "mock/loopback refused — live OverloadedBot only", resolve)
	}

	// --- /models probe (hard stop on 502; 15s) ---
	modelsURL := strings.TrimRight(base, "/") + "/models"
	modelsCtx, modelsCancel := context.WithTimeout(ctx, 15*time.Second)
	defer modelsCancel()
	req, _ := http.NewRequestWithContext(modelsCtx, http.MethodGet, modelsURL, nil)
	req.Header.Set("Authorization", "Bearer "+client.APIKey)
	req.Header.Set("Accept", "application/json")
	t0 := time.Now()
	resp, merr := http.DefaultClient.Do(req)
	elapsed := time.Since(t0).Round(time.Millisecond)
	if merr != nil {
		resolve["models_error"] = merr.Error()
		writeJSON(out, "resolve_quick.json", resolve)
		failReport(out, "FAIL", "GET /models network: "+merr.Error(), resolve)
	}
	body, _ := io.ReadAll(io.LimitReader(resp.Body, 512))
	_ = resp.Body.Close()
	resolve["models_http_status"] = resp.StatusCode
	resolve["models_elapsed"] = elapsed.String()
	resolve["models_snippet"] = trunc(string(body), 180)
	if resp.StatusCode == http.StatusBadGateway {
		writeJSON(out, "resolve_quick.json", resolve)
		failReport(out, "FAIL", "FAIL 502: GET /v1/models — OverloadedBot unavailable (no mock, no wait)", resolve)
	}
	if resp.StatusCode != http.StatusOK && resp.StatusCode != http.StatusUnauthorized {
		writeJSON(out, "resolve_quick.json", resolve)
		failReport(out, "FAIL", fmt.Sprintf("GET /models HTTP %d", resp.StatusCode), resolve)
	}
	// Prefer advertised green/code-fast if DB model is bare code-fast (gateway list).
	if model == "code-fast" && strings.Contains(string(body), `"green/code-fast"`) {
		client.Model = "green/code-fast"
		model = client.Model
		resolve["completer_model_adjusted"] = model
		log.Printf("adjusted model code-fast -> green/code-fast from /models list")
	}
	// Cap per-call wait (~2.5m) and retries so we never sit on 4m×N.
	if client.HTTPClient != nil {
		client.HTTPClient.Timeout = 150 * time.Second
	}
	client.MaxRetries = 1
	resolve["client_http_timeout"] = "150s"
	resolve["client_max_retries"] = 1
	writeJSON(out, "resolve_quick.json", resolve)
	log.Printf("LIVE OK models=%d base=%s model=%s", resp.StatusCode, base, model)

	pipeline := processing.NewPipeline(pool)
	pipeline.BatchSize = cfg.ProcessingBatchSize
	pipeline.AI = aiSvc
	pipeline.Prompts = aiprompts.NewService(pool)
	pipeline.Engine = &processing.Engine{
		Completer:    nil,
		Vector:       processing.NoopVectorCategorizer{},
		EPREL:        eprel.NewClient(eprel.Options{Enabled: true, Timeout: 15 * time.Second}),
		ProviderMode: processing.AIProviderInternal,
	}

	rawIDs := make([]uuid.UUID, 0, len(eans))
	for _, ean := range eans {
		var id uuid.UUID
		if err := pool.QueryRow(ctx, `SELECT id FROM raw_products WHERE company_id=$1 AND gtin=$2`, companyID, ean).Scan(&id); err != nil {
			failReport(out, "FAIL", fmt.Sprintf("raw %s: %v", ean, err), resolve)
		}
		rawIDs = append(rawIDs, id)
	}
	_, _ = pool.Exec(ctx, `
		UPDATE processed_products pp
		SET field_sources = COALESCE(field_sources, '{}'::jsonb) - 'enhance_input_hash', updated_at=now()
		FROM raw_products rp
		WHERE pp.company_id=$1 AND pp.raw_product_id=rp.id AND rp.gtin=ANY($2::text[])`, companyID, eans)
	_, _ = pool.Exec(ctx, `
		UPDATE processing_jobs SET status='failed', error='yielded to quick final verify', updated_at=now()
		WHERE company_id=$1 AND status IN ('pending','running','processing')`, companyID)

	jobs, err := pipeline.StartJob(ctx, companyID, userID, rawIDs, "full")
	if err != nil || len(jobs) == 0 {
		failReport(out, "FAIL", fmt.Sprintf("StartJob: %v", err), resolve)
	}
	jobID := jobs[0].ID
	var dbStatus string
	var processedCount int
	_ = pool.QueryRow(ctx, `SELECT lower(status), COALESCE(processed_count,0) FROM processing_jobs WHERE id=$1`, jobID).Scan(&dbStatus, &processedCount)
	startOK := (dbStatus == "pending" || dbStatus == "queued") && processedCount == 0
	startSample := map[string]any{
		"process_id": jobID.String(), "status": dbStatus, "processed_items": processedCount,
		"total_items": len(rawIDs), "start_ok": startOK,
	}
	writeJSON(out, "start_sample.json", startSample)

	_, _ = pool.Exec(ctx, `UPDATE processing_jobs SET status='running', started_at=COALESCE(started_at,now()), updated_at=now() WHERE id=$1`, jobID)

	// Hard wall: 2 products × ~2.5–3 min ≈ 8 min max
	runCtx, runCancel := context.WithTimeout(ctx, 8*time.Minute)
	defer runCancel()
	procErr := pipeline.ProcessJob(runCtx, jobID)
	logBytes, _ := os.ReadFile(logPath)
	logText := string(logBytes)
	aiEnhance := strings.Contains(logText, "ai_enhance")

	items, loadErr := pipeline.LoadV1ProcessJobItems(ctx, companyID, jobID, "full")
	if loadErr != nil {
		items = nil
	}
	cards := []map[string]any{}
	allPass := startOK && aiEnhance && procErr == nil
	for _, item := range items {
		ean := str(item["ean"])
		if ean == "" {
			ean = str(item["product_id"])
		}
		title := first(str(item["name"]), str(item["title"]))
		desc := str(item["description"])
		catName := str(item["category_name"])
		if catName == "" {
			if m, ok := item["category"].(map[string]any); ok {
				catName = str(m["name"])
			}
		}
		_, hasMetaT := item["meta_title"]
		_, hasMetaD := item["meta_description"]
		_, hasID := item["id"]
		_, hasPP := item["processed_product_id"]
		_, hasRaw := item["raw_product_id"]
		hasHTML := strings.Contains(strings.ToLower(desc), "<h2") || strings.Contains(strings.ToLower(desc), "<p>")
		checks := map[string]bool{
			"name_present":            strings.TrimSpace(title) != "",
			"category_name_present":   strings.TrimSpace(catName) != "",
			"no_meta_title":           !hasMetaT || item["meta_title"] == nil,
			"no_meta_description":     !hasMetaD || item["meta_description"] == nil,
			"no_id":                   !hasID,
			"no_processed_product_id": !hasPP,
			"no_raw_product_id":       !hasRaw,
			"formula_html_or_empty":   hasHTML || strings.TrimSpace(desc) == "" || strings.Contains(strings.ToLower(logText), "empty_or_provider_error") || strings.Contains(strings.ToLower(logText), "refuse"),
		}
		pass := true
		for _, v := range checks {
			if !v {
				pass = false
				break
			}
		}
		if !pass {
			allPass = false
		}
		cards = append(cards, map[string]any{
			"ean": ean, "title": title, "category": catName, "desc_len": len(desc),
			"has_html": hasHTML, "checks": checks, "pass": pass,
		})
	}
	if procErr != nil {
		allPass = false
	}
	overall := "PASS"
	reason := ""
	if !allPass {
		overall = "FAIL"
		if procErr != nil {
			reason = "ProcessJob: " + processing.TruncateError(procErr)
		} else if !aiEnhance {
			reason = "no ai_enhance log lines"
		} else if !startOK {
			reason = "start semantics not pending/0"
		} else {
			reason = "scorecard assertion failed"
		}
	}

	report := map[string]any{
		"generated_at":    time.Now().UTC().Format(time.RFC3339),
		"overall":         overall,
		"blocking_reason": reason,
		"sync_pass":       true,
		"live_llm_pass":   true,
		"process_run":     true,
		"job_id":          jobID.String(),
		"eans":            eans,
		"start_ok":        startOK,
		"ai_enhance_seen": aiEnhance,
		"process_error":   nil,
		"scorecards":      cards,
		"llm":             resolve,
		"sync":            syncInfo,
		"note":            "slim verify: 2 products, 150s/call, 8m wall; repo seed only",
	}
	if procErr != nil {
		report["process_error"] = processing.TruncateError(procErr)
	}
	writeJSON(out, "report.json", report)
	writeMD(out, report, startSample, cards, logText)
	fmt.Printf("OVERALL %s job=%s ai_enhance=%v err=%v\n", overall, jobID, aiEnhance, procErr)
	if overall != "PASS" {
		os.Exit(1)
	}
}

func failReport(out, overall, reason string, extra map[string]any) {
	report := map[string]any{
		"generated_at": time.Now().UTC().Format(time.RFC3339),
		"overall":      overall, "blocking_reason": reason,
		"process_run": false, "extra": extra,
	}
	writeJSON(out, "report.json", report)
	var b strings.Builder
	b.WriteString("# Final verify 2026-08-17 (quick LIVE)\n\n")
	b.WriteString(fmt.Sprintf("**Overall: %s**\n\n", overall))
	b.WriteString(fmt.Sprintf("Blocking: %s\n\n", reason))
	b.WriteString("Live OverloadedBot only. No mock.\n")
	_ = os.WriteFile(filepath.Join(out, "REPORT.md"), []byte(b.String()), 0o644)
	log.Print(reason)
	os.Exit(1)
}

func writeMD(out string, report map[string]any, start map[string]any, cards []map[string]any, logText string) {
	var b strings.Builder
	b.WriteString("# Final verify 2026-08-17 (quick LIVE)\n\n")
	b.WriteString(fmt.Sprintf("**Overall: %v**\n\n", report["overall"]))
	if r := str(report["blocking_reason"]); r != "" {
		b.WriteString(fmt.Sprintf("Blocking: %s\n\n", r))
	}
	b.WriteString("## Progress note\n\n")
	b.WriteString("Prior hung run killed (4m chat waits on 6 EANs). This pass: live OverloadedBot only, repo seed sync, 2 products, 150s/call, 8m wall.\n\n")
	b.WriteString("## 1. Sync A1 (repo seed only)\n\n")
	b.WriteString("- Source: `scripts/seed/wp_product_categories.sql` (no upload, SkipDumpBackfill)\n")
	b.WriteString("- See `sync_quick.json`\n\n")
	b.WriteString("## 2. Live LLM\n\n")
	if llm, ok := report["llm"].(map[string]any); ok {
		b.WriteString(fmt.Sprintf("- base: `%v`\n- model: `%v`\n- models HTTP: `%v`\n- source: `%v`\n\n",
			llm["completer_base"], llm["completer_model"], llm["models_http_status"], llm["role_source"]))
	}
	b.WriteString("## 3. Start semantics\n\n")
	b.WriteString(fmt.Sprintf("- process_id: `%v`\n- status: `%v`\n- processed_items: `%v`\n- start_ok: `%v`\n\n",
		start["process_id"], start["status"], start["processed_items"], start["start_ok"]))
	b.WriteString(fmt.Sprintf("## 4. ai_enhance\n\n- seen: `%v`\n\n", report["ai_enhance_seen"]))
	b.WriteString("## 5. Scorecards\n\n")
	for _, c := range cards {
		b.WriteString(fmt.Sprintf("- **%v** pass=%v cat=%v html=%v title=%q\n",
			c["ean"], c["pass"], c["category"], c["has_html"], trunc(str(c["title"]), 60)))
	}
	b.WriteString("\n## Log excerpt (ai_enhance)\n\n```\n")
	n := 0
	for _, line := range strings.Split(logText, "\n") {
		if strings.Contains(line, "ai_enhance") || strings.Contains(line, "LIVE") || strings.Contains(line, "OVERALL") {
			b.WriteString(trunc(line, 240) + "\n")
			n++
			if n >= 12 {
				break
			}
		}
	}
	b.WriteString("```\n")
	_ = os.WriteFile(filepath.Join(out, "REPORT.md"), []byte(b.String()), 0o644)
}

func writeJSON(out, name string, v any) {
	raw, _ := json.MarshalIndent(v, "", "  ")
	_ = os.WriteFile(filepath.Join(out, name), raw, 0o644)
}
func str(v any) string {
	if v == nil {
		return ""
	}
	if s, ok := v.(string); ok {
		return s
	}
	return strings.TrimSpace(fmt.Sprint(v))
}
func first(vals ...string) string {
	for _, v := range vals {
		if strings.TrimSpace(v) != "" {
			return v
		}
	}
	return ""
}
func trunc(s string, n int) string {
	s = strings.TrimSpace(s)
	if len(s) <= n {
		return s
	}
	return s[:n] + "…"
}