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

package main

import (
	"context"
	"fmt"
	"log"
	"os"
	"path/filepath"
	"strings"

	"github.com/google/uuid"
	"github.com/jackc/pgx/v5/pgxpool"
)

// resolveMySQLDumpPath picks an explicit path, else the first readable candidate
// under common local locations documented in scripts/seed/README.txt.
func resolveMySQLDumpPath(explicit string) string {
	if p := strings.TrimSpace(explicit); p != "" {
		if st, err := os.Stat(p); err == nil && !st.IsDir() {
			return p
		}
		log.Printf("warning: mysql dump not found at %q — trying auto-detect", p)
	}
	for _, c := range mysqlDumpCandidates() {
		if strings.TrimSpace(explicit) != "" && filepath.Clean(c) == filepath.Clean(strings.TrimSpace(explicit)) {
			continue
		}
		if st, err := os.Stat(c); err == nil && !st.IsDir() {
			return c
		}
	}
	return ""
}

func mysqlDumpCandidates() []string {
	var out []string
	if v := strings.TrimSpace(os.Getenv("SEED_A1_MYSQL_DUMP")); v != "" {
		out = append(out, v)
	}
	home, _ := os.UserHomeDir()
	names := []string{
		"descrybe_new (1).sql",
		"descrybe_new.sql",
		"descrybe_new(1).sql",
	}
	if home != "" {
		for _, n := range names {
			out = append(out, filepath.Join(home, "Downloads", n))
			out = append(out, filepath.Join(home, "downloads", n))
		}
	}
	// Repo-relative guesses (cwd may be apps/api or repo root).
	for _, n := range names {
		out = append(out,
			n,
			filepath.Join("..", "..", n),
			filepath.Join("scripts", "seed", n),
			filepath.Join("..", "..", "scripts", "seed", n),
		)
	}
	return out
}

type mappedCoverage struct {
	Total      int
	WithDesc   int
	WithCat    int
	WithAttrs  int
	Processed  int
	Jobs       int
}

func (c mappedCoverage) pct(n int) float64 {
	if c.Total == 0 {
		return 0
	}
	return 100 * float64(n) / float64(c.Total)
}

// feedAttrsSQL matches catalog.processedHasFeedAttributesSQL (mapped_data aliases).
const feedAttrsSQL = `(
	CASE jsonb_typeof(mapped_data->'specifications')
		WHEN 'string' THEN length(trim(mapped_data->>'specifications')) > 0
		WHEN 'object' THEN mapped_data->'specifications' <> '{}'::jsonb
		WHEN 'array' THEN jsonb_array_length(mapped_data->'specifications') > 0
		ELSE false
	END
	OR CASE jsonb_typeof(mapped_data->'specs')
		WHEN 'string' THEN length(trim(mapped_data->>'specs')) > 0
		WHEN 'object' THEN mapped_data->'specs' <> '{}'::jsonb
		WHEN 'array' THEN jsonb_array_length(mapped_data->'specs') > 0
		ELSE false
	END
	OR COALESCE(NULLIF(trim(mapped_data->>'eprel_id'), ''), '') <> ''
	OR COALESCE(NULLIF(trim(mapped_data->>'eprel'), ''), '') <> ''
	OR COALESCE(NULLIF(trim(mapped_data->>'netwidth'), ''), '') <> ''
	OR COALESCE(NULLIF(trim(mapped_data->>'net_width'), ''), '') <> ''
	OR COALESCE(NULLIF(trim(mapped_data->>'netheight'), ''), '') <> ''
	OR COALESCE(NULLIF(trim(mapped_data->>'net_height'), ''), '') <> ''
	OR COALESCE(NULLIF(trim(mapped_data->>'netdepth'), ''), '') <> ''
	OR COALESCE(NULLIF(trim(mapped_data->>'net_depth'), ''), '') <> ''
	OR COALESCE(NULLIF(trim(mapped_data->>'netmass'), ''), '') <> ''
	OR COALESCE(NULLIF(trim(mapped_data->>'net_mass'), ''), '') <> ''
	OR COALESCE(NULLIF(trim(mapped_data->>'warranty'), ''), '') <> ''
	OR COALESCE(NULLIF(trim(mapped_data->>'productmodel'), ''), '') <> ''
	OR COALESCE(NULLIF(trim(mapped_data->>'product_model'), ''), '') <> ''
)`

func measureMappedCoverage(ctx context.Context, pg *pgxpool.Pool, companyID uuid.UUID) (mappedCoverage, error) {
	var c mappedCoverage
	err := pg.QueryRow(ctx, fmt.Sprintf(`
		SELECT
			count(*)::int,
			count(*) FILTER (WHERE COALESCE(NULLIF(trim(mapped_data->>'description'), ''), '') <> '')::int,
			count(*) FILTER (
				WHERE COALESCE(NULLIF(trim(mapped_data->>'category'), ''), '') <> ''
				  AND lower(trim(mapped_data->>'category')) <> 'none'
			)::int,
			count(*) FILTER (WHERE %s)::int
		FROM raw_products
		WHERE company_id = $1`, feedAttrsSQL), companyID).Scan(
		&c.Total, &c.WithDesc, &c.WithCat, &c.WithAttrs,
	)
	if err != nil {
		return c, fmt.Errorf("measure mapped coverage: %w", err)
	}
	_ = pg.QueryRow(ctx, `SELECT count(*)::int FROM processed_products WHERE company_id = $1`, companyID).Scan(&c.Processed)
	_ = pg.QueryRow(ctx, `SELECT count(*)::int FROM processing_jobs WHERE company_id = $1`, companyID).Scan(&c.Jobs)
	return c, nil
}

func logMappedCoverage(c mappedCoverage, label string) {
	log.Printf("A1 mapped coverage (%s): total=%d desc=%.1f%% (%d) category=%.1f%% (%d) feed_attrs=%.1f%% (%d) processed=%d jobs=%d",
		label, c.Total, c.pct(c.WithDesc), c.WithDesc, c.pct(c.WithCat), c.WithCat, c.pct(c.WithAttrs), c.WithAttrs, c.Processed, c.Jobs)
}