go-migrate/scripts/mssql-copy-in/seed-manzana.go

package main

import (
	"context"
	"database/sql"
	"fmt"
	"math/rand"
	"sync"
	"time"

	"github.com/gaspardle/go-mssqlclrgeo"
	"github.com/google/uuid"
	log "github.com/sirupsen/logrus"
)

var manzanaJob = MigrationJob{
	Schema: "Cartografia",
	Table:  "MANZANA",
}

func getMaxGDBArchiveOid(ctx context.Context, db *sql.DB) (int, error) {
	var maxOid sql.NullInt64

	query := fmt.Sprintf(`
		SELECT ISNULL(MAX(GDB_ARCHIVE_OID), 0)
		FROM [%s].[%s]
	`, manzanaJob.Schema, manzanaJob.Table)

	err := db.QueryRowContext(ctx, query).Scan(&maxOid)
	if err != nil && err != sql.ErrNoRows {
		return 0, err
	}

	if !maxOid.Valid {
		return 0, nil
	}

	return int(maxOid.Int64), nil
}

func generateManzanaRows(
	ctx context.Context,
	startOid int,
	totalRows int,
	chunkSize int,
	out chan<- []UnknownRowValues,
) {
	defer close(out)

	rowsGenerated := 0
	currentChunk := make([]UnknownRowValues, 0, chunkSize)

	for i := range totalRows {
		gdbArchiveOid := startOid + i + 1
		row := generateManzanaRow(gdbArchiveOid)
		currentChunk = append(currentChunk, row)
		rowsGenerated++

		if len(currentChunk) == chunkSize {
			select {
			case out <- currentChunk:
				log.Debugf("Sent chunk with %d rows", len(currentChunk))
			case <-ctx.Done():
				log.Info("Context cancelled, stopping row generation")
				return
			}
			currentChunk = make([]UnknownRowValues, 0, chunkSize)
		}

		if rowsGenerated%100_000 == 0 {
			logManzanaSampleRow(rowsGenerated, row)
		}
	}

	if len(currentChunk) > 0 {
		select {
		case out <- currentChunk:
			log.Debugf("Sent final chunk with %d rows", len(currentChunk))
		case <-ctx.Done():
			log.Info("Context cancelled, stopping row generation")
		}
	}

	log.Infof("Finished generating %d rows", rowsGenerated)
}

func generateManzanaRow(gdbArchiveOid int) UnknownRowValues {
	dateLowerLimit, _ := time.Parse(time.RFC3339, "2020-12-31T23:59:59Z")
	dateUpperLimit, _ := time.Parse(time.RFC3339, "2025-12-31T23:59:59Z")

	rowID := gdbArchiveOid
	distrito := fmt.Sprintf("D%d", rand.Intn(100))
	nombre := generateRandomString(15)
	codigo := generateRandomString(15)
	cantidadTotal := rand.Intn(1000)
	ocupacionResidencial := rand.Intn(1000)
	ocupacionNegocio := rand.Intn(1000)
	ocupacionDepartamento := rand.Intn(1000)
	indicador := rand.Intn(10000)
	fechaAlta := generateRandomTimestamp(dateLowerLimit, dateUpperLimit)
	fechaAct := generateRandomTimestamp(dateLowerLimit, dateUpperLimit)
	shapeWKB := generateRandomPolygonWKB()
	geoData := []byte{}
	globalID, _ := uuid.New().MarshalBinary()
	gdbFromDate := fechaAct
	gdbToDate, _ := time.Parse(time.RFC3339, "9999-12-31T23:59:59Z")
	objectID := gdbArchiveOid

	shapeMssql, err := mssqlclrgeo.WkbToUdtGeo(shapeWKB, false)
	if err != nil {
		log.Errorf("Error convirtiendo WKB a formato MSSQL: %v", err)
		shapeMssql = []byte{}
	}

	return UnknownRowValues{
		gdbArchiveOid,
		rowID,
		distrito,
		nombre,
		codigo,
		cantidadTotal,
		ocupacionResidencial,
		ocupacionNegocio,
		ocupacionDepartamento,
		indicador,
		fechaAlta,
		fechaAct,
		shapeMssql,
		geoData,
		globalID,
		gdbFromDate,
		gdbToDate,
		objectID,
	}
}

func logManzanaSampleRow(id int, rowValues UnknownRowValues) {
	log.Infof(`
Sample row #%d:
GDB_ARCHIVE_OID (%T): %v
ID_MANZANA (%T): %v
ID_DISTRITO (%T): %v
NOMBRE (%T): %v
CODIGO (%T): %v
CANTIDAD_TOTAL (%T): %v
OCUPACION_RESIDENCIAL (%T): %v
OCUPACION_NEGOCIO (%T): %v
OCUPACION_DEPARTAMENTO (%T): %v
INDICADOR (%T): %v
FECHA_ALTA (%T): %v
FECHA_ACT (%T): %v
Shape (%T): %v
GDB_GEOMATTR_DATA (%T): %v
GlobalID (%T): %v
GDB_FROM_DATE (%T): %v
GDB_TO_DATE (%T): %v
OBJECTID (%T): %v
`,
		id,
		rowValues[0], rowValues[0],
		rowValues[1], rowValues[1],
		rowValues[2], rowValues[2],
		rowValues[3], rowValues[3],
		rowValues[4], rowValues[4],
		rowValues[5], rowValues[5],
		rowValues[6], rowValues[6],
		rowValues[7], rowValues[7],
		rowValues[8], rowValues[8],
		rowValues[9], rowValues[9],
		rowValues[10], rowValues[10],
		rowValues[11], rowValues[11],
		rowValues[12], rowValues[12],
		rowValues[13], rowValues[13],
		rowValues[14], rowValues[14],
		rowValues[15], rowValues[15],
		rowValues[16], rowValues[16],
		rowValues[17], rowValues[17],
	)
}

func seedManzanas(ctx context.Context, db *sql.DB) error {
	maxOid, err := getMaxGDBArchiveOid(ctx, db)
	if err != nil {
		log.Fatal("Error getting max GDB_ARCHIVE_OID: ", err)
	}

	log.Infof("Starting data generation from GDB_ARCHIVE_OID: %d", maxOid+1)

	rowsChan := make(chan []UnknownRowValues, queueSize)

	var wgRowGenerator sync.WaitGroup

	wgRowGenerator.Go(func() {
		generateManzanaRows(ctx, maxOid, totalRows, chunkSize, rowsChan)
	})

	columns := []string{
		"GDB_ARCHIVE_OID",
		"ID_MANZANA",
		"ID_DISTRITO",
		"NOMBRE",
		"CODIGO",
		"CANTIDAD_TOTAL",
		"OCUPACION_RESIDENCIAL",
		"OCUPACION_NEGOCIO",
		"OCUPACION_DEPARTAMENTO",
		"INDICADOR",
		"FECHA_ALTA",
		"FECHA_ACT",
		"Shape",
		"GDB_GEOMATTR_DATA",
		"GlobalID",
		"GDB_FROM_DATE",
		"GDB_TO_DATE",
		"OBJECTID",
	}

	if err := loadRowsMssql(ctx, manzanaJob, columns, db, rowsChan); err != nil {
		return fmt.Errorf("Error loading rows (MANZANA): %w", err)
	}

	log.Info("Data generation and loading completed successfully (MANZANA)")
	wgRowGenerator.Wait()

	return nil
}