feat: update target database type to sqlserver and clean up unused code in mssql.go

.gitignore

@@ -4,6 +4,7 @@
 *.dll
 *.so
 *.dylib
+bin/
 
 # Test binary, built with `go test -c`
 *.test

Makefile

@@ -0,0 +1,80 @@
+.PHONY: build build-linux build-windows build-all clean help
+
+# Variables
+BINARY_NAME=go-migrate
+CMD_PATH=./cmd/go_migrate
+OUTPUT_DIR=bin
+VERSION?=$(shell git describe --tags --always --dirty 2>/dev/null || echo "dev")
+BUILD_TIME=$(shell date -u '+%Y-%m-%d_%H:%M:%S')
+GIT_COMMIT=$(shell git rev-parse --short HEAD 2>/dev/null || echo "unknown")
+
+# Flags de compilación
+LD_FLAGS=-ldflags="-s -w -X main.Version=$(VERSION) -X main.BuildTime=$(BUILD_TIME) -X main.GitCommit=$(GIT_COMMIT)"
+
+# Default: compilar para el SO actual
+build: build-$(OS)
+
+ifeq ($(OS),Windows_NT)
+build-native: build-windows
+else
+build-native: build-linux
+endif
+
+# Compilar para Linux (sin CGO para máxima compatibilidad)
+build-linux:
+	@echo "Compilando para Linux..."
+	@mkdir -p $(OUTPUT_DIR)
+	CGO_ENABLED=0 GOOS=linux GOARCH=amd64 go build \
+		$(LD_FLAGS) \
+		-o $(OUTPUT_DIR)/$(BINARY_NAME)-linux-amd64 \
+		$(CMD_PATH)
+	@echo "Binario creado: $(OUTPUT_DIR)/$(BINARY_NAME)-linux-amd64"
+
+# Compilar para Windows
+build-windows:
+	@echo "Compilando para Windows..."
+	@mkdir -p $(OUTPUT_DIR)
+	CGO_ENABLED=0 GOOS=windows GOARCH=amd64 go build \
+		$(LD_FLAGS) \
+		-o $(OUTPUT_DIR)/$(BINARY_NAME)-windows-amd64.exe \
+		$(CMD_PATH)
+	@echo "Binario creado: $(OUTPUT_DIR)/$(BINARY_NAME)-windows-amd64.exe"
+
+# Compilar para ambas plataformas
+build-all: build-linux build-windows
+	@echo ""
+	@echo "Binarios compilados:"
+	@ls -lh $(OUTPUT_DIR)/$(BINARY_NAME)*
+
+# Compilar para Linux arm64 (opcional, para Raspberry Pi, etc.)
+build-linux-arm64:
+	@echo "Compilando para Linux ARM64..."
+	@mkdir -p $(OUTPUT_DIR)
+	CGO_ENABLED=0 GOOS=linux GOARCH=arm64 go build \
+		$(LD_FLAGS) \
+		-o $(OUTPUT_DIR)/$(BINARY_NAME)-linux-arm64 \
+		$(CMD_PATH)
+	@echo "Binario creado: $(OUTPUT_DIR)/$(BINARY_NAME)-linux-arm64"
+
+# Limpiar binarios
+clean:
+	@echo "Limpiando binarios..."
+	@rm -rf $(OUTPUT_DIR)
+	@echo "Limpieza completada"
+
+# Ayuda
+help:
+	@echo "Comandos disponibles:"
+	@echo ""
+	@echo "  make build              - Compilar para el SO actual (Linux/Windows)"
+	@echo "  make build-linux        - Compilar para Linux x86_64"
+	@echo "  make build-windows      - Compilar para Windows x86_64"
+	@echo "  make build-linux-arm64  - Compilar para Linux ARM64 (opcional)"
+	@echo "  make build-all          - Compilar para Linux y Windows"
+	@echo "  make clean              - Eliminar binarios compilados"
+	@echo "  make help               - Mostrar esta ayuda"
+	@echo ""
+	@echo "Ejemplos de uso:"
+	@echo "  make build-all          # Crear binarios para ambas plataformas"
+	@echo "  make build-linux OS=    # Crear solo para Linux"
+	@echo ""

cmd/go_migrate/batch-generator.go

@@ -1,117 +0,0 @@
-package main
-
-import (
-	"context"
-	"database/sql"
-	"fmt"
-	"time"
-
-	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/config"
-	"github.com/google/uuid"
-)
-
-type Batch struct {
-	Id                    uuid.UUID
-	ParentId              uuid.UUID
-	LowerLimit            int64
-	UpperLimit            int64
-	IsLowerLimitInclusive bool
-	ShouldUseRange        bool
-	RetryCounter          int
-}
-
-func estimateTotalRowsMssql(ctx context.Context, db *sql.DB, tableInfo config.SourceTableInfo) (int64, error) {
-	query := `
-SELECT
-	SUM(p.rows) AS count
-FROM sys.tables t
-JOIN sys.schemas s ON t.schema_id = s.schema_id
-JOIN sys.partitions p ON t.object_id = p.object_id
-WHERE s.name = @schema AND t.name = @table AND p.index_id IN (0, 1)
-GROUP BY t.name`
-
-	ctxTimeout, cancel := context.WithTimeout(ctx, time.Second*20)
-	defer cancel()
-
-	var rowsCount int64
-	err := db.QueryRowContext(ctxTimeout, query, sql.Named("schema", tableInfo.Schema), sql.Named("table", tableInfo.Table)).Scan(&rowsCount)
-	if err != nil {
-		return 0, err
-	}
-
-	return rowsCount, nil
-}
-
-func calculateBatchesMssql(ctx context.Context, db *sql.DB, tableInfo config.SourceTableInfo, batchCount int64) ([]Batch, error) {
-	query := fmt.Sprintf(`
-SELECT
-	MIN([%s]) AS lower_limit,
-	MAX([%s]) AS upper_limit
-FROM
-	(SELECT [%s], NTILE(@batchCount) OVER (ORDER BY [%s]) AS batch_id FROM [%s].[%s]) AS T
-GROUP BY batch_id
-ORDER BY batch_id`,
-		tableInfo.PrimaryKey,
-		tableInfo.PrimaryKey,
-		tableInfo.PrimaryKey,
-		tableInfo.PrimaryKey,
-		tableInfo.Schema,
-		tableInfo.Table)
-
-	ctxTimeout, cancel := context.WithTimeout(ctx, time.Second*20)
-	defer cancel()
-
-	rows, err := db.QueryContext(ctxTimeout, query, sql.Named("batchCount", batchCount))
-	if err != nil {
-		return nil, err
-	}
-	defer rows.Close()
-
-	batches := make([]Batch, 0, batchCount)
-
-	for rows.Next() {
-		batch := Batch{
-			Id:                    uuid.New(),
-			ShouldUseRange:        true,
-			RetryCounter:          0,
-			IsLowerLimitInclusive: true,
-		}
-
-		if err := rows.Scan(&batch.LowerLimit, &batch.UpperLimit); err != nil {
-			return nil, err
-		}
-
-		batches = append(batches, batch)
-	}
-
-	if err := rows.Err(); err != nil {
-		return nil, err
-	}
-
-	return batches, nil
-}
-
-func batchGeneratorMssql(ctx context.Context, db *sql.DB, tableInfo config.SourceTableInfo, rowsPerBatch int64) ([]Batch, error) {
-	rowsCount, err := estimateTotalRowsMssql(ctx, db, tableInfo)
-	if err != nil {
-		return nil, err
-	}
-
-	var batchCount int64 = 1
-	if rowsCount > rowsPerBatch {
-		batchCount = rowsCount / rowsPerBatch
-	} else {
-		return []Batch{{
-			Id:             uuid.New(),
-			ShouldUseRange: false,
-			RetryCounter:   0,
-		}}, nil
-	}
-
-	batches, err := calculateBatchesMssql(ctx, db, tableInfo, batchCount)
-	if err != nil {
-		return nil, err
-	}
-
-	return batches, nil
-}

cmd/go_migrate/build-extract-query.go

@@ -1,75 +0,0 @@
-package main
-
-import (
-	"fmt"
-	"strings"
-
-	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/config"
-)
-
-func buildExtractQueryMssql(sourceDbInfo config.SourceTableInfo, columns []ColumnType, includeRange bool, isMinInclusive bool) string {
-	var sbQuery strings.Builder
-
-	sbQuery.WriteString("SELECT ")
-
-	if len(columns) == 0 {
-		sbQuery.WriteString("*")
-	} else {
-		for i, col := range columns {
-			fmt.Fprintf(&sbQuery, "[%s]", col.name)
-
-			if col.unifiedType == "GEOMETRY" {
-				fmt.Fprintf(&sbQuery, ".STAsBinary() AS [%s]", col.name)
-			}
-
-			if i < len(columns)-1 {
-				sbQuery.WriteString(", ")
-			}
-		}
-	}
-
-	fmt.Fprintf(&sbQuery, " FROM [%s].[%s]", sourceDbInfo.Schema, sourceDbInfo.Table)
-
-	if includeRange {
-		fmt.Fprintf(&sbQuery, " WHERE [%s]", sourceDbInfo.PrimaryKey)
-		if isMinInclusive {
-			sbQuery.WriteString(" >=")
-		} else {
-			sbQuery.WriteString(" >")
-		}
-
-		fmt.Fprintf(&sbQuery, " @min AND [%s] <= @max", sourceDbInfo.PrimaryKey)
-	}
-
-	fmt.Fprintf(&sbQuery, " ORDER BY [%s] ASC", sourceDbInfo.PrimaryKey)
-
-	return sbQuery.String()
-}
-
-func buildExtractQueryPostgres(sourceDbInfo config.SourceTableInfo, columns []ColumnType) string {
-	var sbColumns strings.Builder
-
-	if len(columns) == 0 {
-		sbColumns.WriteString("*")
-	} else {
-		for i, col := range columns {
-			if col.unifiedType == "GEOMETRY" {
-				sbColumns.WriteString(`ST_AsEWKB("`)
-				sbColumns.WriteString(col.name)
-				sbColumns.WriteString(`") AS "`)
-				sbColumns.WriteString(col.name)
-				sbColumns.WriteString(`"`)
-			} else {
-				sbColumns.WriteString(`"`)
-				sbColumns.WriteString(col.name)
-				sbColumns.WriteString(`"`)
-			}
-
-			if i < len(columns)-1 {
-				sbColumns.WriteString(", ")
-			}
-		}
-	}
-
-	return fmt.Sprintf(`SELECT %s FROM "%s"."%s"  ORDER BY "%s" ASC`, sbColumns.String(), sourceDbInfo.Schema, sourceDbInfo.Table, sourceDbInfo.PrimaryKey)
-}

cmd/go_migrate/colum-type.go

@@ -1,44 +0,0 @@
-package main
-
-type ColumnType struct {
-	name string
-
-	hasMaxLength      bool
-	hasPrecisionScale bool
-
-	userType    string
-	systemType  string
-	unifiedType string
-	nullable    bool
-	maxLength   int64
-	precision   int64
-	scale       int64
-}
-
-func (c *ColumnType) Name() string {
-	return c.name
-}
-
-func (c *ColumnType) UserType() string {
-	return c.userType
-}
-
-func (c *ColumnType) SystemType() string {
-	return c.systemType
-}
-
-func (c *ColumnType) Length() (length int64, ok bool) {
-	return c.maxLength, c.hasMaxLength
-}
-
-func (c *ColumnType) DecimalSize() (precision, scale int64, ok bool) {
-	return c.precision, c.scale, c.hasPrecisionScale
-}
-
-func (c *ColumnType) Nullable() bool {
-	return c.nullable
-}
-
-func (c *ColumnType) Type() string {
-	return c.unifiedType
-}

cmd/go_migrate/connect.go

@@ -1,77 +0,0 @@
-package main
-
-import (
-	"context"
-	"database/sql"
-	"errors"
-	"fmt"
-	"sync"
-	"time"
-
-	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/config"
-	"github.com/jackc/pgx/v5/pgxpool"
-	_ "github.com/microsoft/go-mssqldb"
-	log "github.com/sirupsen/logrus"
-)
-
-func connectToSqlServer() (*sql.DB, error) {
-	db, err := sql.Open("sqlserver", config.App.SourceDbUrl)
-	if err != nil {
-		return nil, fmt.Errorf("Unable to connect to sqlserver: %w", err)
-	}
-
-	ctx, cancel := context.WithTimeout(context.Background(), 20*time.Second)
-	defer cancel()
-
-	if err := db.PingContext(ctx); err != nil {
-		return nil, fmt.Errorf("Unable to ping sqlserver: %w", err)
-	}
-
-	return db, nil
-}
-
-func connectToPostgres() (*pgxpool.Pool, error) {
-	ctx, cancel := context.WithTimeout(context.Background(), 20*time.Second)
-	defer cancel()
-
-	pool, err := pgxpool.New(ctx, config.App.TargetDbUrl)
-	if err != nil {
-		return nil, fmt.Errorf("Unable to connect to postgres: %w", err)
-	}
-
-	if err := pool.Ping(ctx); err != nil {
-		pool.Close()
-		return nil, fmt.Errorf("Unable to ping postgres: %w", err)
-	}
-
-	return pool, nil
-}
-
-func connectToDatabases() (*sql.DB, *pgxpool.Pool, error) {
-	var sourceDbErr, targetDbErr error
-	var sourceDb *sql.DB
-	var targetDb *pgxpool.Pool
-	var wg sync.WaitGroup
-
-	wg.Go(func() {
-		sourceDb, sourceDbErr = connectToSqlServer()
-		if sourceDbErr != nil {
-			log.Error("Unable to connect to source db: ", sourceDbErr)
-		}
-	})
-
-	wg.Go(func() {
-		targetDb, targetDbErr = connectToPostgres()
-		if targetDbErr != nil {
-			log.Error("Unable to connect to target db: ", targetDbErr)
-		}
-	})
-
-	wg.Wait()
-
-	if sourceDbErr != nil || targetDbErr != nil {
-		return nil, nil, errors.New("Unable to connect to databases")
-	}
-
-	return sourceDb, targetDb, nil
-}

cmd/go_migrate/extractor-error-handler.go

@@ -1,102 +0,0 @@
-package main
-
-import (
-	"context"
-	"fmt"
-	"sync"
-
-	"github.com/google/uuid"
-)
-
-type ExtractorError struct {
-	Batch
-	LastId    int64
-	HasLastId bool
-	Msg       string
-}
-
-func (e *ExtractorError) Error() string {
-	return e.Msg
-}
-
-const maxRetryAttempts = 3
-
-func extractorErrorHandler(
-	ctx context.Context,
-	chErrorsIn <-chan ExtractorError,
-	chBatchesOut chan<- Batch,
-	chJobErrorsOut chan<- JobError,
-	wgActiveBatches *sync.WaitGroup,
-) {
-	for {
-		if ctx.Err() != nil {
-			return
-		}
-
-		select {
-		case <-ctx.Done():
-			return
-
-		case err, ok := <-chErrorsIn:
-			if !ok {
-				return
-			}
-
-			if err.RetryCounter >= maxRetryAttempts {
-				jobError := JobError{
-					ShouldCancelJob: false,
-					Msg:             fmt.Sprintf("batch %v reached max retries (%d)", err.Id, maxRetryAttempts),
-					Prev:            &err,
-				}
-
-				select {
-				case chJobErrorsOut <- jobError:
-				case <-ctx.Done():
-					return
-				}
-
-				wgActiveBatches.Done()
-				continue
-			}
-
-			newBatch := err.Batch
-			newBatch.RetryCounter++
-
-			if err.HasLastId {
-				newBatch.ParentId = err.Id
-				newBatch.Id = uuid.New()
-				newBatch.LowerLimit = err.LastId
-				newBatch.IsLowerLimitInclusive = false
-			}
-
-			select {
-			case chBatchesOut <- newBatch:
-			case <-ctx.Done():
-				return
-			}
-		}
-	}
-}
-
-func ExtractorErrorFromLastRowMssql(lastRow UnknownRowValues, indexPrimaryKey int, batch *Batch, previousError error) ExtractorError {
-	lastIdRawValue := lastRow[indexPrimaryKey]
-
-	lastId, ok := ToInt64(lastIdRawValue)
-	if !ok {
-		currentBatch := *batch
-		currentBatch.RetryCounter = maxRetryAttempts
-		return ExtractorError{
-			Batch:     currentBatch,
-			HasLastId: true,
-			Msg:       fmt.Sprintf("Couldn't cast last id value as int: %s", previousError.Error()),
-		}
-
-	}
-
-	return ExtractorError{
-		Batch:     *batch,
-		HasLastId: true,
-		LastId:    lastId,
-		Msg:       previousError.Error(),
-	}
-}

cmd/go_migrate/extractor.go

@@ -1,251 +0,0 @@
-package main
-
-import (
-	"context"
-	"database/sql"
-	"errors"
-	"slices"
-	"strings"
-	"sync"
-	"sync/atomic"
-	"time"
-
-	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/config"
-	"github.com/google/uuid"
-	"github.com/jackc/pgx/v5/pgxpool"
-	_ "github.com/microsoft/go-mssqldb"
-	log "github.com/sirupsen/logrus"
-)
-
-type UnknownRowValues = []any
-
-type Chunk struct {
-	Id           uuid.UUID
-	BatchId      uuid.UUID
-	Data         []UnknownRowValues
-	RetryCounter int
-}
-
-func extractFromMssql(
-	ctx context.Context,
-	db *sql.DB,
-	tableInfo config.SourceTableInfo,
-	columns []ColumnType,
-	chunkSize int,
-	chBatchesIn <-chan Batch,
-	chChunksOut chan<- Chunk,
-	chErrorsOut chan<- ExtractorError,
-	chJobErrorsOut chan<- JobError,
-	wgActiveBatches *sync.WaitGroup,
-	rowsRead *int64,
-) {
-	indexPrimaryKey := slices.IndexFunc(columns, func(col ColumnType) bool {
-		return strings.EqualFold(col.name, tableInfo.PrimaryKey)
-	})
-
-	if indexPrimaryKey == -1 {
-		jobError := JobError{
-			ShouldCancelJob: true,
-			Msg:             "Primary key not found in provided columns",
-		}
-
-		select {
-		case <-ctx.Done():
-			return
-		case chJobErrorsOut <- jobError:
-		}
-
-		return
-	}
-
-	for {
-		if ctx.Err() != nil {
-			return
-		}
-
-		select {
-		case <-ctx.Done():
-			return
-		case batch, ok := <-chBatchesIn:
-			if !ok {
-				return
-			}
-
-			if abort := processBatch(ctx, db, tableInfo, columns, chunkSize, batch, indexPrimaryKey, chChunksOut, chErrorsOut, wgActiveBatches, rowsRead); abort {
-				return
-			}
-		}
-	}
-}
-
-func processBatch(
-	ctx context.Context,
-	db *sql.DB,
-	tableInfo config.SourceTableInfo,
-	columns []ColumnType,
-	chunkSize int,
-	batch Batch,
-	indexPrimaryKey int,
-	chChunksOut chan<- Chunk,
-	chErrorsOut chan<- ExtractorError,
-	wgActiveBatches *sync.WaitGroup,
-	rowsRead *int64,
-) (abort bool) {
-	query := buildExtractQueryMssql(tableInfo, columns, batch.ShouldUseRange, batch.IsLowerLimitInclusive)
-	log.Debug("Query used to extract data from mssql: ", query)
-
-	var queryArgs []any
-	if batch.ShouldUseRange {
-		queryArgs = append(queryArgs,
-			sql.Named("min", batch.LowerLimit),
-			sql.Named("max", batch.UpperLimit),
-		)
-	}
-
-	queryStartTime := time.Now()
-	rows, err := db.QueryContext(ctx, query, queryArgs...)
-	if err != nil {
-		select {
-		case chErrorsOut <- ExtractorError{Batch: batch, HasLastId: false, Msg: err.Error()}:
-		case <-ctx.Done():
-			return true
-		}
-		return false
-	}
-	defer rows.Close()
-	log.Debugf("Query executed in %v", time.Since(queryStartTime))
-
-	rowsChunk := make([]UnknownRowValues, 0, chunkSize)
-	totalRowsExtracted := 0
-	chunkStartTime := time.Now()
-
-	for rows.Next() {
-		values := make([]any, len(columns))
-		scanArgs := make([]any, len(columns))
-
-		for i := range values {
-			scanArgs[i] = &values[i]
-		}
-
-		if err := rows.Scan(scanArgs...); err != nil {
-			if len(rowsChunk) == 0 {
-				select {
-				case chErrorsOut <- ExtractorError{Batch: batch, HasLastId: false, Msg: err.Error()}:
-				case <-ctx.Done():
-					return true
-				}
-				return false
-			}
-
-			lastRow := rowsChunk[len(rowsChunk)-1]
-			select {
-			case chErrorsOut <- ExtractorErrorFromLastRowMssql(lastRow, indexPrimaryKey, &batch, err):
-			case <-ctx.Done():
-				return true
-			}
-
-			select {
-			case chChunksOut <- Chunk{Id: uuid.New(), BatchId: batch.Id, Data: rowsChunk, RetryCounter: 0}:
-			case <-ctx.Done():
-				return true
-			}
-
-			atomic.AddInt64(rowsRead, int64(len(rowsChunk)))
-
-			return false
-		}
-
-		rowsChunk = append(rowsChunk, values)
-		totalRowsExtracted++
-
-		if len(rowsChunk) >= chunkSize {
-			chunkDuration := time.Since(chunkStartTime)
-			rowsPerSec := float64(chunkSize) / chunkDuration.Seconds()
-			log.Infof("Extracted chunk: %d rows in %v (%.0f rows/sec) - Total: %d rows", len(rowsChunk), chunkDuration, rowsPerSec, totalRowsExtracted)
-
-			select {
-			case chChunksOut <- Chunk{Id: uuid.New(), BatchId: batch.Id, Data: rowsChunk, RetryCounter: 0}:
-			case <-ctx.Done():
-				return true
-			}
-
-			atomic.AddInt64(rowsRead, int64(len(rowsChunk)))
-			rowsChunk = make([]UnknownRowValues, 0, chunkSize)
-			chunkStartTime = time.Now()
-		}
-	}
-
-	if err := rows.Err(); err != nil {
-		if errors.Is(err, ctx.Err()) {
-			return true
-		}
-
-		if len(rowsChunk) == 0 {
-			select {
-			case chErrorsOut <- ExtractorError{Batch: batch, HasLastId: false, Msg: err.Error()}:
-			case <-ctx.Done():
-				return true
-			}
-			return false
-		}
-
-		lastRow := rowsChunk[len(rowsChunk)-1]
-		select {
-		case chErrorsOut <- ExtractorErrorFromLastRowMssql(lastRow, indexPrimaryKey, &batch, err):
-		case <-ctx.Done():
-			return true
-		}
-		return false
-	}
-
-	if len(rowsChunk) > 0 {
-		chunkDuration := time.Since(chunkStartTime)
-		rowsPerSec := float64(len(rowsChunk)) / chunkDuration.Seconds()
-		log.Infof("Extracted final chunk: %d rows in %v (%.0f rows/sec) - Total: %d rows", len(rowsChunk), chunkDuration, rowsPerSec, totalRowsExtracted)
-		select {
-		case chChunksOut <- Chunk{Id: uuid.New(), BatchId: batch.Id, Data: rowsChunk, RetryCounter: 0}:
-		case <-ctx.Done():
-			return true
-		}
-
-		atomic.AddInt64(rowsRead, int64(len(rowsChunk)))
-	}
-
-	wgActiveBatches.Done()
-	return false
-}
-
-func extractFromPostgres(ctx context.Context, tableInfo config.SourceTableInfo, columns []ColumnType, chunkSize int, db *pgxpool.Pool, out chan<- []UnknownRowValues) error {
-	query := buildExtractQueryPostgres(tableInfo, columns)
-	log.Debug("Query used to extract data from postgres: ", query)
-
-	rows, err := db.Query(ctx, query)
-	if err != nil {
-		return err
-	}
-	defer rows.Close()
-
-	rowsChunk := make([]UnknownRowValues, 0, chunkSize)
-
-	for rows.Next() {
-		values, err := rows.Values()
-		if err != nil {
-			return err
-		}
-
-		rowsChunk = append(rowsChunk, values)
-
-		if len(rowsChunk) >= chunkSize {
-			out <- rowsChunk
-			rowsChunk = make([]UnknownRowValues, 0, chunkSize)
-			log.Infof("Chunk send... %+v", tableInfo)
-		}
-	}
-
-	if len(rowsChunk) > 0 {
-		out <- rowsChunk
-		log.Infof("Chunk send... %+v", tableInfo)
-	}
-
-	return nil
-}

cmd/go_migrate/inspect-columns.go

@@ -1,289 +0,0 @@
-package main
-
-import (
-	"context"
-	"database/sql"
-	"errors"
-	"fmt"
-	"strings"
-	"sync"
-	"time"
-
-	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/config"
-	"github.com/jackc/pgx/v5/pgxpool"
-	_ "github.com/microsoft/go-mssqldb"
-	log "github.com/sirupsen/logrus"
-)
-
-func GetUnifiedType(systemType string) string {
-	systemType = strings.ToLower(systemType)
-
-	if systemType == "varchar" || systemType == "char" || systemType == "nvarchar" || systemType == "nchar" || systemType == "text" || systemType == "ntext" {
-		return "STRING"
-	}
-
-	if systemType == "int" || systemType == "int4" || systemType == "integer" || systemType == "smallint" || systemType == "int2" || systemType == "bigint" || systemType == "int8" || systemType == "tinyint" {
-		return "INTEGER"
-	}
-
-	if systemType == "decimal" || systemType == "numeric" {
-		return "DECIMAL"
-	}
-
-	if systemType == "float" || systemType == "real" || systemType == "double precision" {
-		return "FLOAT"
-	}
-
-	if systemType == "bit" || systemType == "boolean" {
-		return "BOOLEAN"
-	}
-
-	if systemType == "date" {
-		return "DATE"
-	}
-	if systemType == "time" || systemType == "time without time zone" {
-		return "TIME"
-	}
-	if systemType == "datetime" || systemType == "datetime2" || systemType == "timestamp" || systemType == "timestamptz" || systemType == "timestamp with time zone" {
-		return "TIMESTAMP"
-	}
-
-	if systemType == "binary" || systemType == "varbinary" || systemType == "image" || systemType == "bytea" {
-		return "BINARY"
-	}
-
-	if systemType == "uniqueidentifier" || systemType == "uuid" {
-		return "UUID"
-	}
-
-	if systemType == "json" {
-		return "JSON"
-	}
-
-	if systemType == "geometry" || systemType == "geography" {
-		return "GEOMETRY"
-	}
-
-	return strings.ToUpper(systemType)
-}
-
-func MapPostgresColumn(column ColumnType, maxLength *int64, precision *int64, scale *int64) ColumnType {
-	stringTypes := map[string]bool{
-		"varchar": true, "char": true, "character": true, "text": true, "character varying": true,
-	}
-
-	decimalTypes := map[string]bool{
-		"decimal": true, "numeric": true,
-	}
-
-	if stringTypes[column.systemType] {
-		if maxLength != nil {
-			column.maxLength = *maxLength
-			column.hasMaxLength = true
-		} else {
-			column.maxLength = -1
-			column.hasMaxLength = false
-		}
-		column.hasPrecisionScale = false
-		column.precision = -1
-		column.scale = -1
-	} else if decimalTypes[column.systemType] {
-		column.hasMaxLength = false
-		column.maxLength = -1
-		if precision != nil && scale != nil {
-			column.precision = *precision
-			column.scale = *scale
-			column.hasPrecisionScale = true
-		} else {
-			column.precision = -1
-			column.scale = -1
-			column.hasPrecisionScale = false
-		}
-	} else {
-		column.hasMaxLength = false
-		column.maxLength = -1
-		column.hasPrecisionScale = false
-		column.precision = -1
-		column.scale = -1
-	}
-
-	column.unifiedType = GetUnifiedType(column.systemType)
-
-	return column
-}
-
-func GetColumnTypesPostgres(db *pgxpool.Pool, tableInfo config.TargetTableInfo) ([]ColumnType, error) {
-	query := `
-SELECT 
-  c.column_name AS name,
-  c.data_type AS user_type,
-  c.udt_name AS system_type,
-  (CASE WHEN c.is_nullable = 'YES' THEN TRUE ELSE FALSE END) AS nullable,
-  c.character_maximum_length AS max_length,
-  c.numeric_precision AS precision,
-  c.numeric_scale AS scale
-FROM information_schema.columns c
-WHERE c.table_schema = $1 AND c.table_name = $2
-ORDER BY c.ordinal_position;
-`
-
-	ctx, cancel := context.WithTimeout(context.Background(), 20*time.Second)
-	defer cancel()
-
-	rows, err := db.Query(ctx, query, tableInfo.Schema, tableInfo.Table)
-	if err != nil {
-		return nil, fmt.Errorf("Error querying column types: %w", err)
-	}
-	defer rows.Close()
-
-	var colTypes []ColumnType
-
-	for rows.Next() {
-		var column ColumnType
-		var scanMaxLength *int64
-		var scanPrecision *int64
-		var scanScale *int64
-
-		if err := rows.Scan(
-			&column.name,
-			&column.userType,
-			&column.systemType,
-			&column.nullable,
-			&scanMaxLength,
-			&scanPrecision,
-			&scanScale,
-		); err != nil {
-			return nil, fmt.Errorf("Error scanning column type results: %w", err)
-		}
-
-		colTypes = append(colTypes, MapPostgresColumn(column, scanMaxLength, scanPrecision, scanScale))
-	}
-
-	return colTypes, nil
-}
-
-func MapMssqlColumn(column ColumnType) ColumnType {
-	stringTypes := map[string]bool{
-		"varchar": true, "char": true, "nvarchar": true, "nchar": true, "text": true, "ntext": true,
-	}
-
-	decimalTypes := map[string]bool{
-		"decimal": true, "numeric": true,
-	}
-
-	if stringTypes[column.systemType] {
-		column.hasMaxLength = true
-		if column.systemType == "nvarchar" || column.systemType == "nchar" {
-			if column.maxLength > 0 {
-				column.maxLength = column.maxLength / 2
-			}
-		}
-		column.hasPrecisionScale = false
-		column.precision = -1
-		column.scale = -1
-	} else if decimalTypes[column.systemType] {
-		column.hasMaxLength = false
-		column.maxLength = -1
-		column.hasPrecisionScale = true
-	} else {
-		column.hasMaxLength = false
-		column.maxLength = -1
-		column.hasPrecisionScale = false
-		column.precision = -1
-		column.scale = -1
-	}
-
-	column.unifiedType = GetUnifiedType(column.systemType)
-
-	return column
-}
-
-func GetColumnTypesMssql(db *sql.DB, tableInfo config.SourceTableInfo) ([]ColumnType, error) {
-	query := `
-SELECT 
-	c.name AS name,
-	t.name AS user_type,
-	CASE WHEN t.is_user_defined = 0 THEN t.name ELSE bt.name END AS system_type,
-	c.is_nullable AS nullable,
-	c.max_length AS max_length,
-	c.precision AS precision,
-	c.scale AS scale
-FROM sys.columns c
-JOIN sys.types t ON c.user_type_id = t.user_type_id
-LEFT JOIN sys.types bt ON t.is_user_defined = 1 AND bt.user_type_id = t.system_type_id
-JOIN sys.tables st ON c.object_id = st.object_id
-JOIN sys.schemas s ON st.schema_id = s.schema_id
-WHERE s.name = @schema AND st.name = @table
-ORDER BY c.column_id;
-`
-
-	ctx, cancel := context.WithTimeout(context.Background(), 20*time.Second)
-	defer cancel()
-
-	rows, err := db.QueryContext(ctx, query, sql.Named("schema", tableInfo.Schema), sql.Named("table", tableInfo.Table))
-	if err != nil {
-		return nil, fmt.Errorf("Error querying column types: %w", err)
-	}
-	defer rows.Close()
-
-	var colTypes []ColumnType
-
-	for rows.Next() {
-		var column ColumnType
-
-		if err := rows.Scan(
-			&column.name,
-			&column.userType,
-			&column.systemType,
-			&column.nullable,
-			&column.maxLength,
-			&column.precision,
-			&column.scale,
-		); err != nil {
-			return nil, fmt.Errorf("Error scanning column type results: %W", err)
-		}
-
-		if strings.HasPrefix(column.name, "graph_id") && column.systemType == "bigint" {
-			continue
-		}
-
-		colTypes = append(colTypes, MapMssqlColumn(column))
-	}
-
-	return colTypes, nil
-}
-
-func GetColumnTypes(
-	sourceDb *sql.DB,
-	targetDb *pgxpool.Pool,
-	sourceTable config.SourceTableInfo,
-	targetTable config.TargetTableInfo,
-) ([]ColumnType, []ColumnType, error) {
-	var sourceDbErr error
-	var targetDbErr error
-	var sourceColTypes []ColumnType
-	var targetColTypes []ColumnType
-	var wg sync.WaitGroup
-
-	wg.Go(func() {
-		sourceColTypes, sourceDbErr = GetColumnTypesMssql(sourceDb, sourceTable)
-		if sourceDbErr != nil {
-			log.Error("Error (sourceDb): ", sourceDbErr)
-		}
-	})
-
-	wg.Go(func() {
-		targetColTypes, targetDbErr = GetColumnTypesPostgres(targetDb, targetTable)
-		if targetDbErr != nil {
-			log.Error("Error (targetDb): ", targetDbErr)
-		}
-	})
-
-	wg.Wait()
-
-	if sourceDbErr != nil || targetDbErr != nil {
-		return nil, nil, errors.New("Error querying column types")
-	}
-
-	return sourceColTypes, targetColTypes, nil
-}

cmd/go_migrate/job-error-handler.go

@@ -1,47 +0,0 @@
-package main
-
-import (
-	"context"
-	"fmt"
-
-	log "github.com/sirupsen/logrus"
-)
-
-type JobError struct {
-	ShouldCancelJob bool
-	Msg             string
-	Prev            error
-}
-
-func (e *JobError) Error() string {
-	if e.Prev != nil {
-		return fmt.Sprintf("%s: %v", e.Msg, e.Prev)
-	}
-
-	return e.Msg
-}
-
-func jobErrorHandler(ctx context.Context, chErrorsIn <-chan JobError) error {
-	for {
-		if ctx.Err() != nil {
-			return nil
-		}
-
-		select {
-		case <-ctx.Done():
-			return nil
-
-		case err, ok := <-chErrorsIn:
-			if !ok {
-				return nil
-			}
-
-			if err.ShouldCancelJob {
-				log.Error(err.Msg, " - ", err.Prev)
-				return &err
-			}
-
-			log.Error(err.Msg, " - ", err.Prev)
-		}
-	}
-}

cmd/go_migrate/loader-error-handler.go

@@ -1,65 +0,0 @@
-package main
-
-import (
-	"context"
-	"fmt"
-	"sync"
-)
-
-type LoaderError struct {
-	Chunk
-	Msg string
-}
-
-func (e *LoaderError) Error() string {
-	return e.Msg
-}
-
-func loaderErrorHandler(
-	ctx context.Context,
-	chErrorsIn <-chan LoaderError,
-	chChunksOut chan<- Chunk,
-	chJobErrorsOut chan<- JobError,
-	wgActiveChunks *sync.WaitGroup,
-) {
-	for {
-		if ctx.Err() != nil {
-			return
-		}
-
-		select {
-		case <-ctx.Done():
-			return
-
-		case err, ok := <-chErrorsIn:
-			if !ok {
-				return
-			}
-
-			if err.RetryCounter >= maxRetryAttempts {
-				jobError := JobError{
-					ShouldCancelJob: false,
-					Msg:             fmt.Sprintf("chunk %v reached max retries (%d)", err.Id, maxRetryAttempts),
-					Prev:            &err,
-				}
-
-				select {
-				case chJobErrorsOut <- jobError:
-				case <-ctx.Done():
-					return
-				}
-
-				wgActiveChunks.Done()
-				continue
-			}
-
-			err.RetryCounter++
-
-			select {
-			case chChunksOut <- err.Chunk:
-			case <-ctx.Done():
-				return
-			}
-		}
-	}
-}

cmd/go_migrate/loader.go

@@ -1,195 +0,0 @@
-package main
-
-import (
-	"context"
-	"database/sql"
-	"errors"
-	"fmt"
-	"sync"
-	"sync/atomic"
-	"time"
-
-	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/config"
-	"github.com/jackc/pgx/v5"
-	"github.com/jackc/pgx/v5/pgconn"
-	"github.com/jackc/pgx/v5/pgxpool"
-	mssql "github.com/microsoft/go-mssqldb"
-	log "github.com/sirupsen/logrus"
-)
-
-func loadRowsPostgres(
-	ctx context.Context,
-	db *pgxpool.Pool,
-	tableInfo config.TargetTableInfo,
-	columns []ColumnType,
-	chChunksIn <-chan Chunk,
-	chErrorsOut chan<- LoaderError,
-	chJobErrorsOut chan<- JobError,
-	wgActiveChunks *sync.WaitGroup,
-	rowsLoaded *int64,
-) {
-	tableId := pgx.Identifier{tableInfo.Schema, tableInfo.Table}
-	colNames := Map(columns, func(col ColumnType) string {
-		return col.name
-	})
-
-	for {
-		if ctx.Err() != nil {
-			return
-		}
-
-		select {
-		case <-ctx.Done():
-			return
-		case chunk, ok := <-chChunksIn:
-			if !ok {
-				return
-			}
-
-			if abort := loadChunkPostgres(ctx, db, tableId, colNames, chunk, chErrorsOut, chJobErrorsOut, wgActiveChunks, rowsLoaded); abort {
-				return
-			}
-		}
-	}
-}
-
-func loadChunkPostgres(
-	ctx context.Context,
-	db *pgxpool.Pool,
-	identifier pgx.Identifier,
-	colNames []string,
-	chunk Chunk,
-	chErrorsOut chan<- LoaderError,
-	chJobErrorsOut chan<- JobError,
-	wgActiveChunks *sync.WaitGroup,
-	rowsLoaded *int64,
-) (abort bool) {
-	chunkStartTime := time.Now()
-	_, err := db.CopyFrom(
-		ctx,
-		identifier,
-		colNames,
-		pgx.CopyFromRows(chunk.Data),
-	)
-
-	if err != nil {
-		var pgErr *pgconn.PgError
-		if errors.As(err, &pgErr) {
-			if pgErr.Code == "23505" {
-				select {
-				case chJobErrorsOut <- JobError{
-					ShouldCancelJob: true,
-					Msg:             fmt.Sprintf("Fatal error in table %s", identifier.Sanitize()),
-					Prev:            err,
-				}:
-				case <-ctx.Done():
-				}
-				wgActiveChunks.Done()
-				return true
-			}
-		}
-
-		select {
-		case chErrorsOut <- LoaderError{Chunk: chunk, Msg: err.Error()}:
-		case <-ctx.Done():
-			return true
-		}
-		return false
-	}
-
-	chunkDuration := time.Since(chunkStartTime)
-	rowsPerSec := float64(len(chunk.Data)) / chunkDuration.Seconds()
-
-	log.Infof("Loaded chunk: %d rows in %v (%.0f rows/sec)", len(chunk.Data), chunkDuration, rowsPerSec)
-
-	atomic.AddInt64(rowsLoaded, int64(len(chunk.Data)))
-	wgActiveChunks.Done()
-	return false
-}
-
-func loadRowsMssql(ctx context.Context, tableInfo config.TargetTableInfo, columns []ColumnType, db *sql.DB, in <-chan []UnknownRowValues) error {
-	chunkCount := 0
-	totalRowsLoaded := 0
-
-	for rows := range in {
-		chunkStartTime := time.Now()
-
-		tx, err := db.BeginTx(ctx, nil)
-		if err != nil {
-			return fmt.Errorf("error starting transaction: %w", err)
-		}
-
-		fullTableName := fmt.Sprintf("[%s].[%s]", tableInfo.Schema, tableInfo.Table)
-		colNames := Map(columns, func(col ColumnType) string {
-			return col.name
-		})
-
-		stmt, err := tx.PrepareContext(ctx, mssql.CopyIn(fullTableName, mssql.BulkOptions{}, colNames...))
-		if err != nil {
-			tx.Rollback()
-			return fmt.Errorf("error preparing bulk copy statement: %w", err)
-		}
-
-		copyStartTime := time.Now()
-
-		for _, row := range rows {
-			_, err = stmt.ExecContext(ctx, row...)
-			if err != nil {
-				stmt.Close()
-				tx.Rollback()
-				return fmt.Errorf("error executing row insert: %w", err)
-			}
-		}
-
-		result, err := stmt.ExecContext(ctx)
-		if err != nil {
-			stmt.Close()
-			tx.Rollback()
-			return fmt.Errorf("error flushing bulk data: %w", err)
-		}
-
-		err = stmt.Close()
-		if err != nil {
-			tx.Rollback()
-			return fmt.Errorf("error closing statement: %w", err)
-		}
-
-		if err := tx.Commit(); err != nil {
-			return fmt.Errorf("error committing transaction: %w", err)
-		}
-
-		rowsAffected, _ := result.RowsAffected()
-		chunkCount++
-		totalRowsLoaded += int(rowsAffected)
-
-		copyDuration := time.Since(copyStartTime)
-		chunkDuration := time.Since(chunkStartTime)
-		rowsPerSec := float64(len(rows)) / chunkDuration.Seconds()
-
-		log.Infof("Loaded chunk #%d (MSSQL): %d rows in %v (copy: %v, %.0f rows/sec) - Total: %d rows", chunkCount, len(rows), chunkDuration, copyDuration, rowsPerSec, totalRowsLoaded)
-	}
-
-	return nil
-}
-
-func Map[T any, V any](input []T, mapper func(T) V) []V {
-	result := make([]V, len(input))
-
-	for i, v := range input {
-		result[i] = mapper(v)
-	}
-
-	return result
-}
-
-func fakeLoader(tableInfo config.TargetTableInfo, columns []ColumnType, in <-chan [][]any) {
-	for rows := range in {
-		log.Debugf("Chunk received, loading data into...")
-
-		for i, rowValues := range rows {
-			if i%100 == 0 {
-				logSampleRow(tableInfo.Schema, tableInfo.Table, columns, rowValues, fmt.Sprintf("row %d", i))
-			}
-		}
-	}
-}

cmd/go_migrate/log.go

@@ -13,5 +13,5 @@ func configureLog() {
 		DisableSorting:  false,
 		PadLevelText:    true,
 	})
-	log.SetLevel(log.InfoLevel)
+	log.SetLevel(log.DebugLevel)
 }

cmd/go_migrate/main.go

@@ -2,13 +2,18 @@ package main
 
 import (
 	"context"
-	"database/sql"
 	"sync"
 	"time"
 
 	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/config"
-	"github.com/jackc/pgx/v5/pgxpool"
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/db-wrapper"
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/etl/extractors"
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/etl/loaders"
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/etl/table_analyzers"
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/etl/transformers"
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/models"
 	log "github.com/sirupsen/logrus"
+	"golang.org/x/sync/errgroup"
 )
 
 func main() {
@@ -28,11 +33,33 @@ func main() {
 
 	log.Info("=== Starting migration ===")
 
-	sourceDb, targetDb, connError := connectToDatabases()
+	var wgConnect errgroup.Group
-	if connError != nil {
+	var sourceDb, targetDb dbwrapper.DbWrapper
-		log.Fatal("Connection error: ", connError)
+
+	wgConnect.Go(func() error {
+		var err error
+		sourceDb, err = connectWithTimeout(ctx, migrationConfig.SourceDbType, config.App.SourceDbUrl, 20*time.Second)
+		if err != nil {
+			return err
 		}
 
+		return nil
+	})
+
+	wgConnect.Go(func() error {
+		var err error
+		targetDb, err = connectWithTimeout(ctx, migrationConfig.TargetDbType, config.App.TargetDbUrl, 20*time.Second)
+		if err != nil {
+			return err
+		}
+
+		return nil
+	})
+
+	if err := wgConnect.Wait(); err != nil {
+		log.Error("Connection error: ", err)
+		return
+	}
 	defer sourceDb.Close()
 	defer targetDb.Close()
 
@@ -45,8 +72,10 @@ func main() {
 		status := "OK"
 		if res.Error != nil {
 			status = "FAILED"
-		}
+			log.Infof("[%s] Status: %s | Read: %d | Loaded: %d | Errors: %d | Time: %v | Error: %v", res.JobName, status, res.RowsRead, res.RowsLoaded, res.RowsFailed, res.Duration, res.Error)
+		} else {
 			log.Infof("[%s] Status: %s | Read: %d | Loaded: %d | Errors: %d | Time: %v", res.JobName, status, res.RowsRead, res.RowsLoaded, res.RowsFailed, res.Duration)
+		}
 
 		totalProcessed += res.RowsLoaded
 		if res.Error != nil {
@@ -63,14 +92,14 @@ func main() {
 
 func processMigrationJobs(
 	ctx context.Context,
-	sourceDb *sql.DB,
+	sourceDb dbwrapper.DbWrapper,
-	targetDb *pgxpool.Pool,
+	targetDb dbwrapper.DbWrapper,
 	jobs []config.Job,
 	maxParallelWorkers int,
-) []JobResult {
+) []models.JobResult {
 	if len(jobs) == 0 {
 		log.Info("No migration jobs configured")
-		return []JobResult{}
+		return []models.JobResult{}
 	}
 
 	if maxParallelWorkers <= 0 {
@@ -83,15 +112,31 @@ func processMigrationJobs(
 
 	log.Infof("Starting migration with %d parallel worker(s)", maxParallelWorkers)
 
-	chJobResults := make(chan JobResult, len(jobs))
+	chJobResults := make(chan models.JobResult, len(jobs))
 	chJobs := make(chan config.Job, len(jobs))
 	var wgJobs sync.WaitGroup
 
+	sourceTableAnalyzer := table_analyzers.NewMssqlTableAnalyzer(sourceDb)
+	targetTableAnalyzer := table_analyzers.NewPostgresTableAnalyzer(targetDb)
+	extractor := extractors.NewMssqlExtractor(sourceDb)
+	transformer := transformers.NewMssqlTransformer()
+	loader := loaders.NewGenericLoader(targetDb)
+
 	for i := range maxParallelWorkers {
 		wgJobs.Go(func() {
 			for job := range chJobs {
 				log.Infof("[worker %d] >>> Processing job: %s.%s <<<", i, job.SourceTable.Schema, job.SourceTable.Table)
-				res := processMigrationJob(ctx, sourceDb, targetDb, job)
+				res := processMigrationJob(
+					ctx,
+					targetDb,
+					sourceTableAnalyzer,
+					targetTableAnalyzer,
+					extractor,
+					transformer,
+					loader,
+					job,
+				)
+
 				chJobResults <- res
 			}
 		})
@@ -107,10 +152,26 @@ func processMigrationJobs(
 		close(chJobResults)
 	}()
 
-	var finalResults []JobResult
+	var finalResults []models.JobResult
 	for res := range chJobResults {
 		finalResults = append(finalResults, res)
 	}
 
 	return finalResults
 }
+
+func connectWithTimeout(ctx context.Context, dbType string, dbUrl string, timeout time.Duration) (dbwrapper.DbWrapper, error) {
+	localCtx, cancel := context.WithTimeout(ctx, timeout)
+	defer cancel()
+
+	sourceDb, err := dbwrapper.New(dbType)
+	if err != nil {
+		return nil, err
+	}
+
+	if err = sourceDb.Connect(localCtx, dbUrl); err != nil {
+		return nil, err
+	}
+
+	return sourceDb, nil
+}

cmd/go_migrate/metrics.go

@@ -1,13 +0,0 @@
-package main
-
-import "time"
-
-type JobResult struct {
-	JobName    string
-	StartTime  time.Time
-	Duration   time.Duration
-	RowsRead   int64
-	RowsLoaded int64
-	RowsFailed int64
-	Error      error
-}

cmd/go_migrate/process.go

@@ -2,84 +2,152 @@ package main
 
 import (
 	"context"
-	"database/sql"
+	"fmt"
 	"sync"
 	"sync/atomic"
 	"time"
 
 	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/config"
-	"github.com/jackc/pgx/v5/pgxpool"
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/custom_errors"
-
+	dbwrapper "git.ksdemosapps.com/kylesoda/go-migrate/internal/app/db-wrapper"
-	_ "github.com/microsoft/go-mssqldb"
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/etl"
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/etl/table_analyzers"
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/models"
 	log "github.com/sirupsen/logrus"
+	"golang.org/x/sync/errgroup"
 )
 
 func processMigrationJob(
 	ctx context.Context,
-	sourceDb *sql.DB,
+	targetDbWrapper dbwrapper.DbWrapper,
-	targetDb *pgxpool.Pool,
+	sourceTableAnalyzer etl.TableAnalyzer,
+	targetTableAnalyzer etl.TableAnalyzer,
+	extractor etl.Extractor,
+	transformer etl.Transformer,
+	loader etl.Loader,
 	job config.Job,
-) JobResult {
+) models.JobResult {
-	result := JobResult{
+	localCtx, cancel := context.WithCancel(ctx)
+	defer cancel()
+
+	result := models.JobResult{
 		JobName:   job.Name,
 		StartTime: time.Now(),
 	}
 
 	var rowsRead, rowsLoaded, rowsFailed int64
 
-	sourceColTypes, targetColTypes, err := GetColumnTypes(sourceDb, targetDb, job.SourceTable, job.TargetTable)
+	var wgQueryColumnTypes errgroup.Group
+	var sourceColTypes, targetColTypes []models.ColumnType
+
+	wgQueryColumnTypes.Go(func() error {
+		var err error
+		sourceColTypes, err = sourceTableAnalyzer.QueryColumnTypes(localCtx, job.SourceTable.TableInfo)
+		if err != nil {
+			return err
+		}
+
+		return nil
+	})
+
+	wgQueryColumnTypes.Go(func() error {
+		var err error
+		targetColTypes, err = targetTableAnalyzer.QueryColumnTypes(localCtx, job.TargetTable.TableInfo)
+		if err != nil {
+			return err
+		}
+
+		return nil
+	})
+
+	err := wgQueryColumnTypes.Wait()
 	if err != nil {
 		result.Error = err
 		return result
 	}
 
-	logColumnTypes(sourceColTypes, "Source col types")
+	for _, query := range job.PreSQL {
-	logColumnTypes(targetColTypes, "Target col types")
+		if _, err := targetDbWrapper.Exec(localCtx, query); err != nil {
+			result.Error = err
+			return result
+		}
+	}
 
-	jobCtx, cancel := context.WithCancel(ctx)
+	partitions, err := table_analyzers.PartitionRangeGenerator(
-	defer cancel()
+		localCtx,
-
+		sourceTableAnalyzer,
-	batches, err := batchGeneratorMssql(jobCtx, sourceDb, job.SourceTable, job.RowsPerBatch)
+		job.SourceTable.TableInfo,
+		job.SourceTable.PrimaryKey,
+		job.RowsPerPartition,
+	)
 	if err != nil {
 		log.Error("Unexpected error calculating batch ranges: ", err)
 	}
 
-	chJobErrors := make(chan JobError, job.QueueSize)
+	chJobErrors := make(chan custom_errors.JobError, job.QueueSize)
-	chBatches := make(chan Batch, job.QueueSize)
+	chExtractorErrors := make(chan custom_errors.ExtractorError, job.QueueSize)
-	chExtractorErrors := make(chan ExtractorError, job.QueueSize)
+	chLoadersErrors := make(chan custom_errors.LoaderError, job.QueueSize)
-	chChunksRaw := make(chan Chunk, job.QueueSize)
+	chPartitions := make(chan models.Partition, job.QueueSize)
-	chChunksTransformed := make(chan Chunk, job.QueueSize)
+	chBatchesRaw := make(chan models.Batch, job.QueueSize)
-	chLoadersErrors := make(chan LoaderError, job.QueueSize)
+	chBatchesTransformed := make(chan models.Batch, job.QueueSize)
 
+	var wgActivePartitions sync.WaitGroup
 	var wgActiveBatches sync.WaitGroup
-	var wgActiveChunks sync.WaitGroup
 	var wgExtractors sync.WaitGroup
 	var wgTransformers sync.WaitGroup
 	var wgLoaders sync.WaitGroup
 
 	go func() {
-		if err := jobErrorHandler(jobCtx, chJobErrors); err != nil {
+		if err := custom_errors.JobErrorHandler(localCtx, chJobErrors); err != nil {
+			log.Error("Fatal error received from JobErrorHandler, canceling job... - ", err)
 			cancel()
 			result.Error = err
 		}
 	}()
 
-	go extractorErrorHandler(jobCtx, chExtractorErrors, chBatches, chJobErrors, &wgActiveBatches)
+	go custom_errors.ExtractorErrorHandler(
-	go loaderErrorHandler(jobCtx, chLoadersErrors, chChunksTransformed, chJobErrors, &wgActiveChunks)
+		localCtx,
+		job.Retry,
+		job.MaxPartitionErrrors,
+		chExtractorErrors,
+		chPartitions,
+		chJobErrors,
+		&wgActivePartitions,
+	)
+	go custom_errors.LoaderErrorHandler(
+		localCtx,
+		job.Retry,
+		job.MaxChunkErrors,
+		chLoadersErrors,
+		chBatchesTransformed,
+		chJobErrors,
+		&wgActiveBatches,
+	)
 
-	maxExtractors := min(job.MaxExtractors, len(batches))
+	maxExtractors := min(job.MaxExtractors, len(partitions))
 	log.Infof("Starting %d extractor(s)...", maxExtractors)
 
 	for range maxExtractors {
 		wgExtractors.Go(func() {
-			extractFromMssql(jobCtx, sourceDb, job.SourceTable, sourceColTypes, job.ChunkSize, chBatches, chChunksRaw, chExtractorErrors, chJobErrors, &wgActiveBatches, &rowsRead)
+			extractor.Exec(
+				localCtx,
+				job.SourceTable,
+				sourceColTypes,
+				job.BatchSize,
+				chPartitions,
+				chBatchesRaw,
+				chExtractorErrors,
+				chJobErrors,
+				&wgActivePartitions,
+				&rowsRead,
+			)
 		})
 	}
 
-	wgActiveBatches.Add(len(batches))
+	wgActivePartitions.Add(len(partitions))
 	go func() {
-		for _, batch := range batches {
+		for _, batch := range partitions {
-			chBatches <- batch
+			chPartitions <- batch
 		}
 	}()
 
@@ -87,7 +155,14 @@ func processMigrationJob(
 
 	for range maxExtractors {
 		wgTransformers.Go(func() {
-			transformRowsMssql(jobCtx, sourceColTypes, chChunksRaw, chChunksTransformed, chJobErrors, &wgActiveChunks)
+			transformer.Exec(
+				localCtx,
+				sourceColTypes,
+				chBatchesRaw,
+				chBatchesTransformed,
+				chJobErrors,
+				&wgActiveBatches,
+			)
 		})
 	}
 
@@ -95,30 +170,60 @@ func processMigrationJob(
 
 	for range job.MaxLoaders {
 		wgLoaders.Go(func() {
-			loadRowsPostgres(jobCtx, targetDb, job.TargetTable, targetColTypes, chChunksTransformed, chLoadersErrors, chJobErrors, &wgActiveChunks, &rowsLoaded)
+			loader.Exec(
+				localCtx,
+				job.TargetTable,
+				targetColTypes,
+				chBatchesTransformed,
+				chLoadersErrors,
+				chJobErrors,
+				&wgActiveBatches,
+				&rowsLoaded,
+			)
 		})
 	}
 
 	go func() {
-		wgActiveBatches.Wait()
+		log.Debugf("Waiting for goroutines (%v)", job.Name)
-		close(chBatches)
+
+		wgActivePartitions.Wait()
+		log.Debugf("wgActivePartitions is empty (%v)", job.Name)
+		close(chPartitions)
+		log.Debugf("chPartitions is closed (%v)", job.Name)
 		close(chExtractorErrors)
+		log.Debugf("chExtractorErrors is closed (%v)", job.Name)
 
 		wgExtractors.Wait()
-		close(chChunksRaw)
+		log.Debugf("wgExtractors is empty (%v)", job.Name)
+		close(chBatchesRaw)
+		log.Debugf("chBatchesRaw is closed (%v)", job.Name)
 
 		wgTransformers.Wait()
+		log.Debugf("wgTransformers is empty (%v)", job.Name)
 
-		wgActiveChunks.Wait()
+		wgActiveBatches.Wait()
-		close(chChunksTransformed)
+		log.Debugf("wgActiveBatches is empty (%v)", job.Name)
+		close(chBatchesTransformed)
+		log.Debugf("chBatchesTransformed is empty (%v)", job.Name)
 		close(chLoadersErrors)
+		log.Debugf("chLoadersErrors is empty (%v)", job.Name)
 
 		wgLoaders.Wait()
+		log.Debugf("wgLoaders is empty (%v)", job.Name)
 
 		cancel()
 	}()
 
-	<-jobCtx.Done()
+	for _, query := range job.PostSQL {
+		if _, err := targetDbWrapper.Exec(localCtx, query); err != nil {
+			result.Error = err
+			return result
+		}
+	}
+
+	log.Debugf("waiting for local context to be done (%v)", job.Name)
+	<-localCtx.Done()
+	log.Debugf("local context done (%v)", job.Name)
 
 	if ctx.Err() != nil {
 		result.Error = ctx.Err()
@@ -129,26 +234,9 @@ func processMigrationJob(
 	result.RowsLoaded = atomic.LoadInt64(&rowsLoaded)
 	result.RowsFailed = atomic.LoadInt64(&rowsFailed)
 
+	if result.RowsRead != result.RowsLoaded {
+		result.Error = fmt.Errorf("Row count mismatch: extracted %d rows but loaded %d rows (failed: %d)", result.RowsRead, result.RowsLoaded, result.RowsFailed)
+	}
+
 	return result
 }
-
-func logColumnTypes(columnTypes []ColumnType, label string) {
-	log.Debug(label)
-
-	for _, col := range columnTypes {
-		log.Debugf("%+v", col)
-	}
-}
-
-func logSampleRow(
-	schema string,
-	table string,
-	columns []ColumnType,
-	rowValues UnknownRowValues,
-	tag string,
-) {
-	log.Infof("[%s.%s] Sample row: (%s)", schema, table, tag)
-	for i, col := range columns {
-		log.Infof("%s (%T): %v", col.Name(), rowValues[i], rowValues[i])
-	}
-}

cmd/go_migrate/transformer.go

@@ -1,149 +0,0 @@
-package main
-
-import (
-	"context"
-	"errors"
-	"sync"
-	"time"
-
-	log "github.com/sirupsen/logrus"
-)
-
-type transformerFunc func(any) (any, error)
-
-type columnTransformPlan struct {
-	index int
-	fn    transformerFunc
-}
-
-func transformRowsMssql(
-	ctx context.Context,
-	columns []ColumnType,
-	chChunksIn <-chan Chunk,
-	chChunksOut chan<- Chunk,
-	chJobErrorsOut chan<- JobError,
-	wgActiveChunks *sync.WaitGroup,
-) {
-	transformationPlan := computeTransformationPlan(columns)
-
-	for {
-		if ctx.Err() != nil {
-			return
-		}
-
-		select {
-		case <-ctx.Done():
-			return
-
-		case chunk, ok := <-chChunksIn:
-			if !ok {
-				return
-			}
-
-			if len(transformationPlan) == 0 {
-				select {
-				case chChunksOut <- chunk:
-					wgActiveChunks.Add(1)
-					continue
-				case <-ctx.Done():
-					return
-				}
-			}
-
-			chunkStartTime := time.Now()
-
-			err := processChunk(ctx, &chunk, transformationPlan)
-			if err != nil {
-				if errors.Is(err, ctx.Err()) {
-					return
-				}
-
-				select {
-				case chJobErrorsOut <- JobError{ShouldCancelJob: true, Msg: "Transformation failed", Prev: err}:
-				case <-ctx.Done():
-				}
-				return
-			}
-
-			log.Infof("Transformed chunk %s: %d rows in %v", chunk.Id, len(chunk.Data), time.Since(chunkStartTime))
-
-			select {
-			case chChunksOut <- chunk:
-			case <-ctx.Done():
-				return
-			}
-
-			wgActiveChunks.Add(1)
-		}
-	}
-}
-
-func computeTransformationPlan(columns []ColumnType) []columnTransformPlan {
-	var plan []columnTransformPlan
-
-	for i, col := range columns {
-		switch col.SystemType() {
-		case "uniqueidentifier":
-			plan = append(plan, columnTransformPlan{
-				index: i,
-				fn: func(v any) (any, error) {
-					if b, ok := v.([]byte); ok && b != nil {
-						return mssqlUuidToBigEndian(b)
-					}
-					return v, nil
-				},
-			})
-
-		case "geometry", "geography":
-			plan = append(plan, columnTransformPlan{
-				index: i,
-				fn: func(v any) (any, error) {
-					if b, ok := v.([]byte); ok && b != nil {
-						return wkbToEwkbWithSrid(b, 4326)
-					}
-					return v, nil
-				},
-			})
-
-		case "datetime", "datetime2":
-			plan = append(plan, columnTransformPlan{
-				index: i,
-				fn: func(v any) (any, error) {
-					if t, ok := v.(time.Time); ok {
-						return ensureUTC(t), nil
-					}
-					return v, nil
-				},
-			})
-		}
-	}
-
-	return plan
-}
-
-const processChunkCtxCheck = 4096
-
-func processChunk(ctx context.Context, chunk *Chunk, transformationPlan []columnTransformPlan) error {
-	for i, rowValues := range chunk.Data {
-		if i%processChunkCtxCheck == 0 {
-			if err := ctx.Err(); err != nil {
-				return err
-			}
-		}
-
-		for _, task := range transformationPlan {
-			val := rowValues[task.index]
-			if val == nil {
-				continue
-			}
-
-			transformed, err := task.fn(val)
-			if err != nil {
-				return err
-			}
-			rowValues[task.index] = transformed
-		}
-	}
-
-	return nil
-}

config.yaml

@@ -1,15 +1,22 @@
 max_parallel_workers: 4
+source_db_type: sqlserver
+target_db_type: sqlserver
 
 defaults:
   max_extractors: 2
   max_loaders: 4
   queue_size: 8
-  chunk_size: 25000
+  batch_size: 25000
-  chunks_per_batch: 8
+  batches_per_partition: 8
   truncate_target: true
   truncate_method: TRUNCATE # TRUNCATE | DELETE
+  max_partition_errrors: 5
+  max_chunk_errors: 5
   retry:
     attempts: 3
+    base_delay_ms: 500
+    max_delay_ms: 10000
+    max_jitter_ms: 500
 
 jobs:
   - name: cartografia_manzana
@@ -21,6 +28,14 @@ jobs:
     target:
       schema: Cartografia
       table: MANZANA
+    pre_sql: 
+      - 'SELECT 1'
+      # - 'TRUNCATE TABLE "Cartografia"."MANZANA"'
+    range:
+      min: 1000000
+      max: 2000000
+      is_min_inclusive: false
+      is_max_inclusive: true
 
   - name: red_puerto
     enabled: true
@@ -31,3 +46,8 @@ jobs:
     target:
       schema: Red
       table: PUERTO
+    pre_sql: 
+      - 'SELECT 1'
+      # - 'TRUNCATE TABLE "Red"."PUERTO"'
+    post_sql: 
+      - "SELECT 1"

go.mod

@@ -10,6 +10,7 @@ require (
 	github.com/microsoft/go-mssqldb v1.9.8
 	github.com/sirupsen/logrus v1.9.4
 	github.com/twpayne/go-geom v1.6.1
+	golang.org/x/sync v0.19.0
 	gopkg.in/yaml.v3 v3.0.1
 )
 
@@ -23,7 +24,6 @@ require (
 	github.com/rogpeppe/go-internal v1.14.1 // indirect
 	github.com/shopspring/decimal v1.4.0 // indirect
 	golang.org/x/crypto v0.48.0 // indirect
-	golang.org/x/sync v0.19.0 // indirect
 	golang.org/x/sys v0.41.0 // indirect
 	golang.org/x/text v0.34.0 // indirect
 )

internal/app/config/main.go

@@ -9,9 +9,7 @@ import (
 
 type appConfig struct {
 	SourceDbUrl string
-	SourceDbType string
 	TargetDbUrl string
-	TargetDbType string
 }
 
 func loadEnv() {
@@ -29,26 +27,14 @@ func getAppConfig() appConfig {
 		log.Fatal("SOURCE_DB_URL environment variable not set")
 	}
 
-	sourceDbType := os.Getenv("SOURCE_DB_TYPE")
-	if sourceDbType == "" {
-		log.Fatal("SOURCE_DB_TYPE environment variable not set")
-	}
-
 	targetDbUrl := os.Getenv("TARGET_DB_URL")
 	if targetDbUrl == "" {
 		log.Fatal("TARGET_DB_URL environment variable not set")
 	}
 
-	targetDbType := os.Getenv("TARGET_DB_TYPE")
-	if targetDbType == "" {
-		log.Fatal("TARGET_DB_TYPE environment variable not set")
-	}
-
 	return appConfig{
 		SourceDbUrl: sourceDbUrl,
-		SourceDbType: sourceDbType,
 		TargetDbUrl: targetDbUrl,
-		TargetDbType: targetDbType,
 	}
 }
 

internal/app/config/migration.go

@@ -9,28 +9,36 @@ import (
 
 type RetryConfig struct {
 	Attempts    int `yaml:"attempts"`
+	BaseDelayMs int `yaml:"base_delay_ms"`
+	MaxDelayMs  int `yaml:"max_delay_ms"`
+	MaxJitterMs int `yaml:"max_jitter_ms"`
 }
 
 type JobConfig struct {
 	MaxExtractors       int         `yaml:"max_extractors"`
 	MaxLoaders          int         `yaml:"max_loaders"`
 	QueueSize           int         `yaml:"queue_size"`
-	ChunkSize      int `yaml:"chunk_size"`
+	BatchSize           int         `yaml:"batch_size"`
-	ChunksPerBatch int `yaml:"chunks_per_batch"`
+	BatchesPerPartition int         `yaml:"batches_per_partition"`
-	RowsPerBatch   int64
 	TruncateTarget      bool        `yaml:"truncate_target"`
 	TruncateMethod      string      `yaml:"truncate_method"`
+	MaxPartitionErrrors int         `yaml:"max_partition_errrors"`
+	MaxChunkErrors      int         `yaml:"max_chunk_errors"`
 	Retry               RetryConfig `yaml:"retry"`
+	RowsPerPartition    int64
+}
+
+type TableInfo struct {
+	Schema string `yaml:"schema"`
+	Table  string `yaml:"table"`
 }
 
 type TargetTableInfo struct {
-	Schema string `yaml:"schema"`
+	TableInfo `yaml:",inline"`
-	Table  string `yaml:"table"`
 }
 
 type SourceTableInfo struct {
-	Schema     string `yaml:"schema"`
+	TableInfo  `yaml:",inline"`
-	Table      string `yaml:"table"`
 	PrimaryKey string `yaml:"primary_key"`
 }
 
@@ -42,16 +50,26 @@ type Job struct {
 	PreSQL      []string        `yaml:"pre_sql"`
 	PostSQL     []string        `yaml:"post_sql"`
 	JobConfig   `yaml:",inline"`
+	Range       struct {
+		Min            int64 `yaml:"min"`
+		Max            int64 `yaml:"max"`
+		IsMinInclusive bool  `yaml:"is_min_inclusive"`
+		IsMaxInclusive bool  `yaml:"is_max_inclusive"`
+	}
 }
 
 type MigrationConfig struct {
 	MaxParallelWorkers int       `yaml:"max_parallel_workers"`
+	SourceDbType       string    `yaml:"source_db_type"`
+	TargetDbType       string    `yaml:"target_db_type"`
 	Defaults           JobConfig `yaml:"defaults"`
 	Jobs               []Job     `yaml:"jobs"`
 }
 
 type rawConfig struct {
 	MaxParallelWorkers int         `yaml:"max_parallel_workers"`
+	SourceDbType       string      `yaml:"source_db_type"`
+	TargetDbType       string      `yaml:"target_db_type"`
 	Defaults           JobConfig   `yaml:"defaults"`
 	Jobs               []yaml.Node `yaml:"jobs"`
 }
@@ -64,7 +82,9 @@ func (c *MigrationConfig) UnmarshalYAML(value *yaml.Node) error {
 
 	c.MaxParallelWorkers = raw.MaxParallelWorkers
 	c.Defaults = raw.Defaults
-	c.Defaults.RowsPerBatch = int64(raw.Defaults.ChunkSize * raw.Defaults.ChunksPerBatch)
+	c.SourceDbType = raw.SourceDbType
+	c.TargetDbType = raw.TargetDbType
+	c.Defaults.RowsPerPartition = int64(raw.Defaults.BatchSize * raw.Defaults.BatchesPerPartition)
 
 	for _, node := range raw.Jobs {
 		job := Job{
@@ -75,7 +95,7 @@ func (c *MigrationConfig) UnmarshalYAML(value *yaml.Node) error {
 			return err
 		}
 
-		job.RowsPerBatch = int64(job.ChunkSize * job.ChunksPerBatch)
+		job.RowsPerPartition = int64(job.BatchSize * job.BatchesPerPartition)
 
 		c.Jobs = append(c.Jobs, job)
 	}

internal/app/custom_errors/backoff.go

@@ -0,0 +1,61 @@
+package custom_errors
+
+import (
+	"context"
+	"math/rand"
+	"time"
+)
+
+func computeBackoffDelay(retryCounter int, baseDelayMs int, maxDelayMs int, maxJitterMs int) time.Duration {
+	if retryCounter < 0 {
+		retryCounter = 0
+	}
+
+	delay := max(time.Duration(baseDelayMs)*time.Millisecond, 0)
+
+	maxDelay := time.Duration(maxDelayMs) * time.Millisecond
+	for i := 0; i < retryCounter; i++ {
+		if maxDelayMs > 0 && delay >= maxDelay {
+			delay = maxDelay
+			break
+		}
+		if delay == 0 {
+			break
+		}
+		delay *= 2
+	}
+
+	if maxDelayMs > 0 && delay > maxDelay {
+		delay = maxDelay
+	}
+
+	if maxJitterMs > 0 {
+		jitter := time.Duration(rand.Intn(maxJitterMs+1)) * time.Millisecond
+		delay += jitter
+	}
+
+	if delay < 0 {
+		delay = 0
+	}
+
+	return delay
+}
+
+func requeueWithBackoff(ctx context.Context, delay time.Duration, enqueue func()) {
+	if delay <= 0 {
+		enqueue()
+		return
+	}
+
+	go func() {
+		timer := time.NewTimer(delay)
+		defer timer.Stop()
+
+		select {
+		case <-ctx.Done():
+			return
+		case <-timer.C:
+			enqueue()
+		}
+	}()
+}

internal/app/custom_errors/extractor.error.go

@@ -5,12 +5,13 @@ import (
 	"fmt"
 	"sync"
 
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/config"
 	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/models"
 	"github.com/google/uuid"
 )
 
 type ExtractorError struct {
-	Batch     models.Batch
+	Partition models.Partition
 	LastId    int64
 	HasLastId bool
 	Msg       string
@@ -20,15 +21,17 @@ func (e *ExtractorError) Error() string {
 	return e.Msg
 }
 
-const maxRetryAttempts = 3
-
 func ExtractorErrorHandler(
 	ctx context.Context,
+	retryConfig config.RetryConfig,
+	maxPartitionErrors int,
 	chErrorsIn <-chan ExtractorError,
-	chBatchesOut chan<- models.Batch,
+	chPartitionsOut chan<- models.Partition,
 	chJobErrorsOut chan<- JobError,
-	wgActiveBatches *sync.WaitGroup,
+	wgActivePartitions *sync.WaitGroup,
 ) {
+	definitiveErrors := 0
+
 	for {
 		if ctx.Err() != nil {
 			return
@@ -43,10 +46,12 @@ func ExtractorErrorHandler(
 				return
 			}
 
-			if err.Batch.RetryCounter >= maxRetryAttempts {
+			if err.Partition.RetryCounter >= retryConfig.Attempts {
+				wgActivePartitions.Done()
+				definitiveErrors++
 				jobError := JobError{
 					ShouldCancelJob: false,
-					Msg:             fmt.Sprintf("batch %v reached max retries (%d)", err.Batch.Id, maxRetryAttempts),
+					Msg:             fmt.Sprintf("Partition %v reached max retries (%d)", err.Partition.Id, retryConfig.Attempts),
 					Prev:            &err,
 				}
 
@@ -56,25 +61,59 @@ func ExtractorErrorHandler(
 					return
 				}
 
-				wgActiveBatches.Done()
+				if maxPartitionErrors > 0 && definitiveErrors >= maxPartitionErrors {
-				continue
+					fatalError := JobError{
-			}
+						ShouldCancelJob: true,
-
+						Msg:             fmt.Sprintf("Partition error limit reached (%d)", maxPartitionErrors),
-			newBatch := err.Batch
+						Prev:            &err,
-			newBatch.RetryCounter++
-
-			if err.HasLastId {
-				newBatch.ParentId = err.Batch.Id
-				newBatch.Id = uuid.New()
-				newBatch.LowerLimit = err.LastId
-				newBatch.IsLowerLimitInclusive = false
 					}
 
 					select {
-			case chBatchesOut <- newBatch:
+					case chJobErrorsOut <- fatalError:
 					case <-ctx.Done():
 						return
 					}
 				}
+
+				continue
+			} else {
+				jobError := JobError{
+					ShouldCancelJob: false,
+					Msg:             fmt.Sprintf("Temporal error in partition %v (retries: %d)", err.Partition.Id, err.Partition.RetryCounter),
+					Prev:            &err,
+				}
+
+				select {
+				case chJobErrorsOut <- jobError:
+				case <-ctx.Done():
+					return
+				}
+			}
+
+			newPartition := err.Partition
+			newPartition.RetryCounter++
+
+			delay := computeBackoffDelay(
+				newPartition.RetryCounter,
+				retryConfig.BaseDelayMs,
+				retryConfig.MaxDelayMs,
+				retryConfig.MaxJitterMs,
+			)
+
+			if err.HasLastId {
+				newPartition.ParentId = err.Partition.Id
+				newPartition.Id = uuid.New()
+				newPartition.Range.Min = err.LastId
+				newPartition.Range.IsMinInclusive = false
+			}
+
+			requeueWithBackoff(ctx, delay, func() {
+				select {
+				case chPartitionsOut <- newPartition:
+				case <-ctx.Done():
+					return
+				}
+			})
+		}
 	}
 }

internal/app/custom_errors/job.error.go

@@ -37,11 +37,11 @@ func JobErrorHandler(ctx context.Context, chErrorsIn <-chan JobError) error {
 			}
 
 			if err.ShouldCancelJob {
-				log.Error(err.Msg, " - ", err.Prev)
+				log.Errorf("(Fatal job error) - %v - %v", err.Msg, err.Prev)
 				return &err
 			}
 
-			log.Error(err.Msg, " - ", err.Prev)
+			log.Errorf("%v - %v", err.Msg, err.Prev)
 		}
 	}
 }

internal/app/custom_errors/loader.error.go

@@ -0,0 +1,107 @@
+package custom_errors
+
+import (
+	"context"
+	"fmt"
+	"sync"
+
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/config"
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/models"
+)
+
+type LoaderError struct {
+	Batch models.Batch
+	Msg   string
+}
+
+func (e *LoaderError) Error() string {
+	return e.Msg
+}
+
+func LoaderErrorHandler(
+	ctx context.Context,
+	retryConfig config.RetryConfig,
+	maxChunkErrors int,
+	chErrorsIn <-chan LoaderError,
+	chBatchesOut chan<- models.Batch,
+	chJobErrorsOut chan<- JobError,
+	wgActiveBatches *sync.WaitGroup,
+) {
+	definitiveErrors := 0
+
+	for {
+		if ctx.Err() != nil {
+			return
+		}
+
+		select {
+		case <-ctx.Done():
+			return
+
+		case err, ok := <-chErrorsIn:
+			if !ok {
+				return
+			}
+
+			if err.Batch.RetryCounter >= retryConfig.Attempts {
+				wgActiveBatches.Done()
+				definitiveErrors++
+				jobError := JobError{
+					ShouldCancelJob: false,
+					Msg:             fmt.Sprintf("Batch %v reached max retries (%d)", err.Batch.Id, retryConfig.Attempts),
+					Prev:            &err,
+				}
+
+				select {
+				case chJobErrorsOut <- jobError:
+				case <-ctx.Done():
+					return
+				}
+
+				if maxChunkErrors > 0 && definitiveErrors >= maxChunkErrors {
+					fatalError := JobError{
+						ShouldCancelJob: true,
+						Msg:             fmt.Sprintf("Chunk error limit reached (%d)", maxChunkErrors),
+						Prev:            &err,
+					}
+
+					select {
+					case chJobErrorsOut <- fatalError:
+					case <-ctx.Done():
+						return
+					}
+				}
+
+				continue
+			} else {
+				jobError := JobError{
+					ShouldCancelJob: false,
+					Msg:             fmt.Sprintf("Temporal error in batch %v (retries: %d)", err.Batch.Id, err.Batch.RetryCounter),
+					Prev:            &err,
+				}
+
+				select {
+				case chJobErrorsOut <- jobError:
+				case <-ctx.Done():
+					return
+				}
+			}
+
+			err.Batch.RetryCounter++
+			delay := computeBackoffDelay(
+				err.Batch.RetryCounter,
+				retryConfig.BaseDelayMs,
+				retryConfig.MaxDelayMs,
+				retryConfig.MaxJitterMs,
+			)
+
+			requeueWithBackoff(ctx, delay, func() {
+				select {
+				case chBatchesOut <- err.Batch:
+				case <-ctx.Done():
+					return
+				}
+			})
+		}
+	}
+}

internal/app/db-wrapper/main.go

@@ -0,0 +1,19 @@
+package dbwrapper
+
+import "fmt"
+
+type Factory func() DbWrapper
+
+var drivers = make(map[string]Factory)
+
+func Register(name string, factory Factory) {
+	drivers[name] = factory
+}
+
+func New(driverType string) (DbWrapper, error) {
+	factory, ok := drivers[driverType]
+	if !ok {
+		return nil, fmt.Errorf("driver not yet supported: %s", driverType)
+	}
+	return factory(), nil
+}

internal/app/db-wrapper/mssql.go

@@ -0,0 +1,176 @@
+package dbwrapper
+
+import (
+	"context"
+	"database/sql"
+	"fmt"
+
+	mssql "github.com/microsoft/go-mssqldb"
+)
+
+func init() {
+	Register("sqlserver", func() DbWrapper {
+		return &mssqlDbWrapper{dialect: "sqlserver"}
+	})
+}
+
+type mssqlRowResult struct {
+	row *sql.Row
+}
+
+func (mr *mssqlRowResult) Scan(dest ...any) error {
+	return mr.row.Scan(dest...)
+}
+
+type mssqlRowsResult struct {
+	columns []string
+	rows    *sql.Rows
+}
+
+func (mr *mssqlRowsResult) Close() error {
+	return mr.rows.Close()
+}
+
+func (mr *mssqlRowsResult) Columns() ([]string, error) {
+	if mr.columns != nil {
+		return mr.columns, nil
+	}
+
+	return mr.rows.Columns()
+}
+
+func (mr *mssqlRowsResult) Err() error {
+	return mr.rows.Err()
+}
+
+func (mr *mssqlRowsResult) Next() bool {
+	return mr.rows.Next()
+}
+
+func (mr *mssqlRowsResult) Scan(dest ...any) error {
+	return mr.rows.Scan(dest...)
+}
+
+func (mr *mssqlRowsResult) Values() ([]any, error) {
+	columns, err := mr.Columns()
+	if err != nil {
+		return nil, err
+	}
+
+	rowValues := make([]any, len(columns))
+	scanArgs := make([]any, len(columns))
+	for i := range rowValues {
+		scanArgs[i] = &rowValues[i]
+	}
+
+	if err := mr.rows.Scan(scanArgs...); err != nil {
+		return nil, err
+	}
+
+	return rowValues, nil
+}
+
+type mssqlDbWrapper struct {
+	db      *sql.DB
+	dialect string
+}
+
+func (mw *mssqlDbWrapper) Connect(ctx context.Context, dbUrl string) error {
+	db, err := sql.Open("sqlserver", dbUrl)
+	if err != nil {
+		return err
+	}
+
+	if err := db.PingContext(ctx); err != nil {
+		if err := db.Close(); err != nil {
+			return err
+		}
+		return err
+	}
+
+	mw.db = db
+	return nil
+}
+
+func (mw *mssqlDbWrapper) Close() error {
+	return mw.db.Close()
+}
+
+func (mw *mssqlDbWrapper) Exec(ctx context.Context, query string, args ...any) (ExecResult, error) {
+	result, execErr := mw.db.ExecContext(ctx, query, args...)
+	if execErr != nil {
+		return ExecResult{}, execErr
+	}
+
+	affectedRows, err := result.RowsAffected()
+	if err != nil {
+		return ExecResult{}, err
+	}
+
+	return ExecResult{AffectedRows: affectedRows}, nil
+}
+
+func (mw *mssqlDbWrapper) GetDialect() string {
+	return mw.dialect
+}
+
+func (mw *mssqlDbWrapper) Query(ctx context.Context, query string, args ...any) (RowsResult, error) {
+	rows, err := mw.db.QueryContext(ctx, query, args...)
+	if err != nil {
+		return nil, err
+	}
+
+	return &mssqlRowsResult{columns: nil, rows: rows}, nil
+}
+
+func (mw *mssqlDbWrapper) QueryRow(ctx context.Context, query string, args ...any) RowResult {
+	row := mw.db.QueryRowContext(ctx, query, args...)
+	return &mssqlRowResult{row: row}
+}
+
+func (mw *mssqlDbWrapper) SaveMassive(ctx context.Context, schema string, table string, columnNames []string, rows [][]any) (int64, error) {
+	tx, err := mw.db.BeginTx(ctx, nil)
+	if err != nil {
+		return 0, err
+	}
+
+	fullTableName := fmt.Sprintf("[%s].[%s]", schema, table)
+
+	stmt, err := tx.PrepareContext(ctx, mssql.CopyIn(fullTableName, mssql.BulkOptions{}, columnNames...))
+	if err != nil {
+		tx.Rollback()
+		return 0, err
+	}
+
+	for _, row := range rows {
+		_, err = stmt.ExecContext(ctx, row...)
+		if err != nil {
+			stmt.Close()
+			tx.Rollback()
+			return 0, err
+		}
+	}
+
+	result, err := stmt.ExecContext(ctx)
+	if err != nil {
+		stmt.Close()
+		tx.Rollback()
+		return 0, err
+	}
+
+	if err := stmt.Close(); err != nil {
+		tx.Rollback()
+		return 0, err
+	}
+
+	if err := tx.Commit(); err != nil {
+		return 0, err
+	}
+
+	rowsAffected, raErr := result.RowsAffected()
+	if raErr != nil {
+		return 0, nil
+	}
+
+	return rowsAffected, nil
+}

internal/app/db-wrapper/postgres.go

@@ -0,0 +1,128 @@
+package dbwrapper
+
+import (
+	"context"
+	"errors"
+
+	"github.com/jackc/pgx/v5"
+	"github.com/jackc/pgx/v5/pgxpool"
+)
+
+func init() {
+	Register("postgres", func() DbWrapper {
+		return &postgresDbWrapper{dialect: "postgres"}
+	})
+}
+
+type postgresRowResult struct {
+	row pgx.Row
+}
+
+func (pr *postgresRowResult) Scan(dest ...any) error {
+	return pr.row.Scan(dest...)
+}
+
+type postgresRowsResult struct {
+	columns []string
+	rows    pgx.Rows
+}
+
+func (pr *postgresRowsResult) Close() error {
+	pr.rows.Close()
+	return nil
+}
+
+func (pr *postgresRowsResult) Columns() ([]string, error) {
+	if pr.columns != nil {
+		return pr.columns, nil
+	}
+
+	rawColumns := pr.rows.FieldDescriptions()
+	if rawColumns == nil {
+		return nil, errors.New("error retrieving columns")
+	}
+
+	columns := make([]string, 0, len(rawColumns))
+	for _, rc := range rawColumns {
+		columns = append(columns, rc.Name)
+	}
+
+	return columns, nil
+}
+
+func (pr *postgresRowsResult) Err() error {
+	return pr.rows.Err()
+}
+
+func (pr *postgresRowsResult) Next() bool {
+	return pr.rows.Next()
+}
+
+func (pr *postgresRowsResult) Scan(dest ...any) error {
+	return pr.rows.Scan(dest...)
+}
+
+func (pr *postgresRowsResult) Values() ([]any, error) {
+	return pr.rows.Values()
+}
+
+type postgresDbWrapper struct {
+	db      *pgxpool.Pool
+	dialect string
+}
+
+func (pw *postgresDbWrapper) Connect(ctx context.Context, dbUrl string) error {
+	pool, err := pgxpool.New(ctx, dbUrl)
+	if err != nil {
+		return err
+	}
+
+	if err := pool.Ping(ctx); err != nil {
+		pool.Close()
+		return err
+	}
+
+	pw.db = pool
+	return nil
+}
+
+func (pw *postgresDbWrapper) Close() error {
+	pw.db.Close()
+	return nil
+}
+
+func (pw *postgresDbWrapper) Exec(ctx context.Context, query string, args ...any) (ExecResult, error) {
+	result, err := pw.db.Exec(ctx, query, args...)
+	if err != nil {
+		return ExecResult{}, err
+	}
+
+	return ExecResult{AffectedRows: result.RowsAffected()}, nil
+}
+
+func (pw *postgresDbWrapper) GetDialect() string {
+	return pw.dialect
+}
+
+func (pw *postgresDbWrapper) Query(ctx context.Context, query string, args ...any) (RowsResult, error) {
+	rows, err := pw.db.Query(ctx, query, args...)
+	if err != nil {
+		return nil, err
+	}
+
+	return &postgresRowsResult{columns: nil, rows: rows}, nil
+}
+
+func (pw *postgresDbWrapper) QueryRow(ctx context.Context, query string, args ...any) RowResult {
+	row := pw.db.QueryRow(ctx, query, args...)
+	return &postgresRowResult{row: row}
+}
+
+func (pw *postgresDbWrapper) SaveMassive(ctx context.Context, schema string, table string, columnNames []string, rows [][]any) (int64, error) {
+	affectedRows, err := pw.db.CopyFrom(ctx, pgx.Identifier{schema, table}, columnNames, pgx.CopyFromRows(rows))
+	if err != nil {
+		return 0, err
+	}
+
+	return affectedRows, nil
+}

internal/app/db-wrapper/types.go

@@ -0,0 +1,35 @@
+package dbwrapper
+
+import (
+	"context"
+	"errors"
+)
+
+var MethodNotSupported error = errors.New("Method not supported by driver... yet :P")
+
+type ExecResult struct {
+	AffectedRows int64
+}
+
+type RowsResult interface {
+	Close() error
+	Columns() ([]string, error)
+	Err() error
+	Next() bool
+	Scan(dest ...any) error
+	Values() ([]any, error)
+}
+
+type RowResult interface {
+	Scan(dest ...any) error
+}
+
+type DbWrapper interface {
+	Close() error
+	Connect(ctx context.Context, dbUrl string) error
+	Exec(ctx context.Context, query string, args ...any) (ExecResult, error)
+	GetDialect() string
+	Query(ctx context.Context, query string, args ...any) (RowsResult, error)
+	QueryRow(ctx context.Context, query string, args ...any) RowResult
+	SaveMassive(ctx context.Context, schema string, table string, columnNames []string, rows [][]any) (int64, error)
+}

internal/app/db/postgres.go

@@ -1,28 +0,0 @@
-package db
-
-import (
-	"context"
-	"fmt"
-
-	"github.com/jackc/pgx/v5/pgxpool"
-)
-
-func Connect(ctx context.Context, dbURL string) (*pgxpool.Pool, error) {
-	pool, err := pgxpool.New(ctx, dbURL)
-	if err != nil {
-		return nil, fmt.Errorf("unable to connect to database: %w", err)
-	}
-
-	if err := pool.Ping(ctx); err != nil {
-		pool.Close()
-		return nil, fmt.Errorf("unable to ping database: %w", err)
-	}
-
-	return pool, nil
-}
-
-func Close(pool *pgxpool.Pool) {
-	if pool != nil {
-		pool.Close()
-	}
-}

internal/app/etl/extractor/main.go

@@ -1,36 +0,0 @@
-package extractor
-
-import (
-	"context"
-	"sync"
-
-	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/config"
-	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/custom_errors"
-	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/models"
-)
-
-type Extractor interface {
-	ProcessBatch(
-		ctx context.Context,
-		tableInfo config.SourceTableInfo,
-		columns []models.ColumnType,
-		chunkSize int,
-		batch models.Batch,
-		indexPrimaryKey int,
-		chChunksOut chan<- models.Chunk,
-		rowsRead *int64,
-	) error
-
-	Exec(
-		ctx context.Context,
-		tableInfo config.SourceTableInfo,
-		columns []models.ColumnType,
-		chunkSize int,
-		chBatchesIn <-chan models.Batch,
-		chChunksOut chan<- models.Chunk,
-		chErrorsOut chan<- custom_errors.ExtractorError,
-		chJobErrorsOut chan<- custom_errors.JobError,
-		wgActiveBatches *sync.WaitGroup,
-		rowsRead *int64,
-	)
-}

internal/app/etl/extractors/mssql.go

@@ -1,4 +1,4 @@
-package extractor
+package extractors
 
 import (
 	"context"
@@ -13,15 +13,17 @@ import (
 	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/config"
 	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/convert"
 	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/custom_errors"
+	dbwrapper "git.ksdemosapps.com/kylesoda/go-migrate/internal/app/db-wrapper"
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/etl"
 	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/models"
 	"github.com/google/uuid"
 )
 
 type MssqlExtractor struct {
-	db *sql.DB
+	db dbwrapper.DbWrapper
 }
 
-func NewMssqlExtractor(db *sql.DB) *MssqlExtractor {
+func NewMssqlExtractor(db dbwrapper.DbWrapper) etl.Extractor {
 	return &MssqlExtractor{db: db}
 }
 
@@ -41,9 +43,9 @@ func buildExtractQueryMssql(
 		for i, col := range columns {
 			fmt.Fprintf(&sbQuery, "[%s]", col.Name())
 
-			if col.Type() == "GEOMETRY" {
+			// if col.Type() == "GEOMETRY" {
-				fmt.Fprintf(&sbQuery, ".STAsBinary() AS [%s]", col.Name())
+			// 	fmt.Fprintf(&sbQuery, ".STAsBinary() AS [%s]", col.Name())
-			}
+			// }
 
 			if i < len(columns)-1 {
 				sbQuery.WriteString(", ")
@@ -69,20 +71,20 @@ func buildExtractQueryMssql(
 	return sbQuery.String()
 }
 
-func extractorErrorFromLastRowMssql(
+func errorFromLastRow(
 	lastRow models.UnknownRowValues,
 	indexPrimaryKey int,
-	batch *models.Batch,
+	partition models.Partition,
 	previousError error,
 ) *custom_errors.ExtractorError {
 	lastIdRawValue := lastRow[indexPrimaryKey]
 
 	lastId, ok := convert.ToInt64(lastIdRawValue)
 	if !ok {
-		currentBatch := *batch
+		currentPartition := partition
-		currentBatch.RetryCounter = 3
+		currentPartition.RetryCounter = 3
 		return &custom_errors.ExtractorError{
-			Batch:     currentBatch,
+			Partition: currentPartition,
 			HasLastId: true,
 			Msg:       fmt.Sprintf("Couldn't cast last id value as int: %s", previousError.Error()),
 		}
@@ -90,117 +92,113 @@ func extractorErrorFromLastRowMssql(
 	}
 
 	return &custom_errors.ExtractorError{
-		Batch:     *batch,
+		Partition: partition,
 		HasLastId: true,
 		LastId:    lastId,
 		Msg:       previousError.Error(),
 	}
 }
 
-func (mssqlEx *MssqlExtractor) ProcessBatch(
+func (mssqlEx *MssqlExtractor) ProcessPartition(
 	ctx context.Context,
 	tableInfo config.SourceTableInfo,
 	columns []models.ColumnType,
-	chunkSize int,
+	batchSize int,
-	batch models.Batch,
+	partition models.Partition,
 	indexPrimaryKey int,
-	chChunksOut chan<- models.Chunk,
+	chBatchesOut chan<- models.Batch,
-	rowsRead *int64,
+) (int, error) {
-) error {
+	query := buildExtractQueryMssql(tableInfo, columns, partition.HasRange, partition.Range.IsMinInclusive)
-	query := buildExtractQueryMssql(tableInfo, columns, batch.ShouldUseRange, batch.IsLowerLimitInclusive)
 
 	var queryArgs []any
-	if batch.ShouldUseRange {
+	if partition.HasRange {
 		queryArgs = append(queryArgs,
-			sql.Named("min", batch.LowerLimit),
+			sql.Named("min", partition.Range.Min),
-			sql.Named("max", batch.UpperLimit),
+			sql.Named("max", partition.Range.Max),
 		)
 	}
 
-	rows, err := mssqlEx.db.QueryContext(ctx, query, queryArgs...)
+	rowsRead := 0
+	rows, err := mssqlEx.db.Query(ctx, query, queryArgs...)
 	if err != nil {
-		return &custom_errors.ExtractorError{Batch: batch, HasLastId: false, Msg: err.Error()}
+		return rowsRead, &custom_errors.ExtractorError{Partition: partition, HasLastId: false, Msg: err.Error()}
 	}
 	defer rows.Close()
 
-	rowsChunk := make([]models.UnknownRowValues, 0, chunkSize)
+	batchRows := make([]models.UnknownRowValues, 0, batchSize)
 
 	for rows.Next() {
-		values := make([]any, len(columns))
+		rowValues := make([]any, len(columns))
 		scanArgs := make([]any, len(columns))
 
-		for i := range values {
+		for i := range rowValues {
-			scanArgs[i] = &values[i]
+			scanArgs[i] = &rowValues[i]
 		}
 
 		if err := rows.Scan(scanArgs...); err != nil {
-			if len(rowsChunk) == 0 {
+			if len(batchRows) == 0 {
-				return &custom_errors.ExtractorError{Batch: batch, HasLastId: false, Msg: err.Error()}
+				return rowsRead, &custom_errors.ExtractorError{Partition: partition, HasLastId: false, Msg: err.Error()}
 			}
 
-			lastRow := rowsChunk[len(rowsChunk)-1]
+			lastRow := batchRows[len(batchRows)-1]
 
 			select {
-			case chChunksOut <- models.Chunk{Id: uuid.New(), BatchId: batch.Id, Data: rowsChunk, RetryCounter: 0}:
+			case chBatchesOut <- models.Batch{Id: uuid.New(), PartitionId: partition.Id, Rows: batchRows, RetryCounter: 0}:
 			case <-ctx.Done():
-				return nil
+				return rowsRead, ctx.Err()
 			}
 
-			atomic.AddInt64(rowsRead, int64(len(rowsChunk)))
+			return rowsRead, errorFromLastRow(lastRow, indexPrimaryKey, partition, err)
-
-			return extractorErrorFromLastRowMssql(lastRow, indexPrimaryKey, &batch, err)
 		}
+		rowsRead++
 
-		rowsChunk = append(rowsChunk, values)
+		batchRows = append(batchRows, rowValues)
-
+		if len(batchRows) >= batchSize {
-		if len(rowsChunk) >= chunkSize {
 			select {
-			case chChunksOut <- models.Chunk{Id: uuid.New(), BatchId: batch.Id, Data: rowsChunk, RetryCounter: 0}:
+			case chBatchesOut <- models.Batch{Id: uuid.New(), PartitionId: partition.Id, Rows: batchRows, RetryCounter: 0}:
 			case <-ctx.Done():
-				return nil
+				return rowsRead, ctx.Err()
 			}
 
-			atomic.AddInt64(rowsRead, int64(len(rowsChunk)))
+			batchRows = make([]models.UnknownRowValues, 0, batchSize)
-			rowsChunk = make([]models.UnknownRowValues, 0, chunkSize)
 		}
+
 	}
 
 	if err := rows.Err(); err != nil {
 		if errors.Is(err, ctx.Err()) {
-			return ctx.Err()
+			return rowsRead, ctx.Err()
 		}
 
-		if len(rowsChunk) == 0 {
+		if len(batchRows) > 0 {
-			return &custom_errors.ExtractorError{Batch: batch, HasLastId: false, Msg: err.Error()}
+			lastRow := batchRows[len(batchRows)-1]
+			return rowsRead, errorFromLastRow(lastRow, indexPrimaryKey, partition, err)
 		}
 
-		lastRow := rowsChunk[len(rowsChunk)-1]
+		return rowsRead, &custom_errors.ExtractorError{Partition: partition, HasLastId: false, Msg: err.Error()}
-		return extractorErrorFromLastRowMssql(lastRow, indexPrimaryKey, &batch, err)
 	}
 
-	if len(rowsChunk) > 0 {
+	if len(batchRows) > 0 {
 		select {
-		case chChunksOut <- models.Chunk{Id: uuid.New(), BatchId: batch.Id, Data: rowsChunk, RetryCounter: 0}:
+		case chBatchesOut <- models.Batch{Id: uuid.New(), PartitionId: partition.Id, Rows: batchRows, RetryCounter: 0}:
 		case <-ctx.Done():
-			return nil
+			return rowsRead, ctx.Err()
+		}
 	}
 
-		atomic.AddInt64(rowsRead, int64(len(rowsChunk)))
+	return rowsRead, nil
-	}
-
-	return nil
 }
 
 func (mssqlEx *MssqlExtractor) Exec(
 	ctx context.Context,
 	tableInfo config.SourceTableInfo,
 	columns []models.ColumnType,
-	chunkSize int,
+	batchSize int,
-	chBatchesIn <-chan models.Batch,
+	chPartitionsIn <-chan models.Partition,
-	chChunksOut chan<- models.Chunk,
+	chBatchesOut chan<- models.Batch,
 	chErrorsOut chan<- custom_errors.ExtractorError,
 	chJobErrorsOut chan<- custom_errors.JobError,
-	wgActiveBatches *sync.WaitGroup,
+	wgActivePartitions *sync.WaitGroup,
 	rowsRead *int64,
 ) {
 	indexPrimaryKey := slices.IndexFunc(columns, func(col models.ColumnType) bool {
@@ -228,42 +226,52 @@ func (mssqlEx *MssqlExtractor) Exec(
 		select {
 		case <-ctx.Done():
 			return
-		case batch, ok := <-chBatchesIn:
+		case partition, ok := <-chPartitionsIn:
 			if !ok {
 				return
 			}
 
-			err := mssqlEx.ProcessBatch(
+			rowsReadResult, err := mssqlEx.ProcessPartition(
 				ctx,
 				tableInfo,
 				columns,
-				chunkSize,
+				batchSize,
-				batch,
+				partition,
 				indexPrimaryKey,
-				chChunksOut,
+				chBatchesOut,
-				rowsRead,
 			)
 
+			if rowsReadResult > 0 {
+				atomic.AddInt64(rowsRead, int64(rowsReadResult))
+			}
+
 			if err != nil {
 				var exError *custom_errors.ExtractorError
+				var jobError *custom_errors.JobError
 				if errors.As(err, &exError) {
 					select {
 					case <-ctx.Done():
 						return
 					case chErrorsOut <- *exError:
 					}
-				}
+				} else if errors.As(err, &jobError) {
-
 					select {
 					case <-ctx.Done():
 						return
-				case chJobErrorsOut <- custom_errors.JobError{ShouldCancelJob: false, Prev: err}:
+					case chJobErrorsOut <- *jobError:
 					}
-
+				} else {
+					select {
+					case <-ctx.Done():
 						return
+					case chErrorsOut <- custom_errors.ExtractorError{Partition: partition, Msg: err.Error()}:
+					}
 				}
 
-			wgActiveBatches.Done()
+				continue
+			}
+
+			wgActivePartitions.Done()
 		}
 	}
 }

internal/app/etl/extractors/postgres.go

@@ -1,4 +1,4 @@
-package extractor
+package extractors
 
 import (
 	"context"
@@ -6,21 +6,21 @@ import (
 	"fmt"
 	"strings"
 	"sync"
-	"sync/atomic"
 
 	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/config"
 	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/custom_errors"
+	dbwrapper "git.ksdemosapps.com/kylesoda/go-migrate/internal/app/db-wrapper"
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/etl"
 	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/models"
 	"github.com/google/uuid"
-	"github.com/jackc/pgx/v5/pgxpool"
 )
 
 type PostgresExtractor struct {
-	db *pgxpool.Pool
+	db dbwrapper.DbWrapper
 }
 
-func NewPostgresExtractor(pool *pgxpool.Pool) *PostgresExtractor {
+func NewPostgresExtractor(db dbwrapper.DbWrapper) etl.Extractor {
-	return &PostgresExtractor{db: pool}
+	return &PostgresExtractor{db: db}
 }
 
 func buildExtractQueryPostgres(sourceDbInfo config.SourceTableInfo, columns []models.ColumnType) string {
@@ -51,77 +51,75 @@ func buildExtractQueryPostgres(sourceDbInfo config.SourceTableInfo, columns []mo
 	return fmt.Sprintf(`SELECT %s FROM "%s"."%s"  ORDER BY "%s" ASC`, sbColumns.String(), sourceDbInfo.Schema, sourceDbInfo.Table, sourceDbInfo.PrimaryKey)
 }
 
-func (postgresEx *PostgresExtractor) ProcessBatch(
+func (postgresEx *PostgresExtractor) ProcessPartition(
 	ctx context.Context,
 	tableInfo config.SourceTableInfo,
 	columns []models.ColumnType,
-	chunkSize int,
+	batchSize int,
-	batch models.Batch,
+	partition models.Partition,
 	indexPrimaryKey int,
-	chChunksOut chan<- models.Chunk,
+	chBatchesOut chan<- models.Batch,
-	rowsRead *int64,
+) (int, error) {
-) error {
 	query := buildExtractQueryPostgres(tableInfo, columns)
 
-	if batch.ShouldUseRange {
+	if partition.HasRange {
-		return errors.New("Batch config not yet supported")
+		return 0, errors.New("Batch config not yet supported")
 	}
 
+	rowsRead := 0
 	rows, err := postgresEx.db.Query(ctx, query)
 	if err != nil {
-		return &custom_errors.ExtractorError{Batch: batch, HasLastId: false, Msg: err.Error()}
+		return rowsRead, &custom_errors.ExtractorError{Partition: partition, HasLastId: false, Msg: err.Error()}
 	}
 	defer rows.Close()
 
-	rowsChunk := make([]models.UnknownRowValues, 0, chunkSize)
+	batchRows := make([]models.UnknownRowValues, 0, batchSize)
 
 	for rows.Next() {
 		values, err := rows.Values()
 		if err != nil {
-			return errors.New("Unexpected error reading rows from source")
+			return rowsRead, errors.New("Unexpected error reading rows from source")
 		}
+		rowsRead++
 
-		rowsChunk = append(rowsChunk, values)
+		batchRows = append(batchRows, values)
 
-		if len(rowsChunk) >= chunkSize {
+		if len(batchRows) >= batchSize {
 			select {
-			case chChunksOut <- models.Chunk{Id: uuid.New(), BatchId: batch.Id, Data: rowsChunk, RetryCounter: 0}:
+			case chBatchesOut <- models.Batch{Id: uuid.New(), PartitionId: partition.Id, Rows: batchRows, RetryCounter: 0}:
 			case <-ctx.Done():
-				return nil
+				return rowsRead, ctx.Err()
 			}
 
-			atomic.AddInt64(rowsRead, int64(len(rowsChunk)))
+			batchRows = make([]models.UnknownRowValues, 0, batchSize)
-			rowsChunk = make([]models.UnknownRowValues, 0, chunkSize)
 		}
 	}
 
 	if err := rows.Err(); err != nil {
-		return errors.New("Unexpected error reading rows from source")
+		return rowsRead, errors.New("Unexpected error reading rows from source")
 	}
 
-	if len(rowsChunk) > 0 {
+	if len(batchRows) > 0 {
 		select {
-		case chChunksOut <- models.Chunk{Id: uuid.New(), BatchId: batch.Id, Data: rowsChunk, RetryCounter: 0}:
+		case chBatchesOut <- models.Batch{Id: uuid.New(), PartitionId: partition.Id, Rows: batchRows, RetryCounter: 0}:
 		case <-ctx.Done():
-			return nil
+			return rowsRead, nil
+		}
 	}
 
-		atomic.AddInt64(rowsRead, int64(len(rowsChunk)))
+	return rowsRead, nil
-	}
-
-	return nil
 }
 
 func (postgresEx *PostgresExtractor) Exec(
 	ctx context.Context,
 	tableInfo config.SourceTableInfo,
 	columns []models.ColumnType,
-	chunkSize int,
+	batchSize int,
-	chBatchesIn <-chan models.Batch,
+	chPartitionsIn <-chan models.Partition,
-	chChunksOut chan<- models.Chunk,
+	chBatchesOut chan<- models.Batch,
 	chErrorsOut chan<- custom_errors.ExtractorError,
 	chJobErrorsOut chan<- custom_errors.JobError,
-	wgActiveBatches *sync.WaitGroup,
+	wgActivePartitions *sync.WaitGroup,
 	rowsRead *int64,
 ) {
 }

internal/app/etl/loaders/main.go

@@ -0,0 +1,117 @@
+package loaders
+
+import (
+	"context"
+	"errors"
+	"fmt"
+	"sync"
+	"sync/atomic"
+
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/config"
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/custom_errors"
+	dbwrapper "git.ksdemosapps.com/kylesoda/go-migrate/internal/app/db-wrapper"
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/etl"
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/models"
+	"github.com/jackc/pgx/v5/pgconn"
+)
+
+type GenericLoader struct {
+	db dbwrapper.DbWrapper
+}
+
+func NewGenericLoader(db dbwrapper.DbWrapper) etl.Loader {
+	return &GenericLoader{db: db}
+}
+
+func (gl *GenericLoader) ProcessBatch(
+	ctx context.Context,
+	tableInfo config.TargetTableInfo,
+	colNames []string,
+	batch models.Batch,
+) (int, error) {
+	_, err := gl.db.SaveMassive(
+		ctx,
+		tableInfo.Schema,
+		tableInfo.Table,
+		colNames,
+		batch.Rows,
+	)
+
+	if err != nil {
+		var pgErr *pgconn.PgError
+		if errors.As(err, &pgErr) {
+			if pgErr.Code == "23505" {
+				return 0, &custom_errors.JobError{
+					ShouldCancelJob: true,
+					Msg:             fmt.Sprintf("Fatal error in table %s.%s", tableInfo.Schema, tableInfo.Table),
+					Prev:            err,
+				}
+			}
+		}
+
+		return 0, &custom_errors.LoaderError{Batch: batch, Msg: err.Error()}
+	}
+
+	return len(batch.Rows), nil
+}
+
+func (gl *GenericLoader) Exec(
+	ctx context.Context,
+	tableInfo config.TargetTableInfo,
+	columns []models.ColumnType,
+	chBatchesIn <-chan models.Batch,
+	chErrorsOut chan<- custom_errors.LoaderError,
+	chJobErrorsOut chan<- custom_errors.JobError,
+	wgActiveBatches *sync.WaitGroup,
+	rowsLoaded *int64,
+) {
+	colNames := mapSlice(columns, func(col models.ColumnType) string {
+		return col.Name()
+	})
+
+	for {
+		if ctx.Err() != nil {
+			return
+		}
+
+		select {
+		case <-ctx.Done():
+			return
+		case batch, ok := <-chBatchesIn:
+			if !ok {
+				return
+			}
+
+			processedRows, err := gl.ProcessBatch(ctx, tableInfo, colNames, batch)
+
+			if err != nil {
+				var ldError *custom_errors.LoaderError
+				var jobError *custom_errors.JobError
+				if errors.As(err, &ldError) {
+					select {
+					case <-ctx.Done():
+						return
+					case chErrorsOut <- *ldError:
+					}
+				} else if errors.As(err, &jobError) {
+					select {
+					case <-ctx.Done():
+						return
+					case chJobErrorsOut <- *jobError:
+					}
+				} else {
+					select {
+					case <-ctx.Done():
+						return
+					case chErrorsOut <- custom_errors.LoaderError{Batch: batch, Msg: err.Error()}:
+					}
+				}
+
+				continue
+			}
+
+			wgActiveBatches.Done()
+			atomic.AddInt64(rowsLoaded, int64(processedRows))
+		}
+	}
+}

internal/app/etl/loaders/utils.go

@@ -0,0 +1,11 @@
+package loaders
+
+func mapSlice[T any, V any](input []T, mapper func(T) V) []V {
+	result := make([]V, len(input))
+
+	for i, v := range input {
+		result[i] = mapper(v)
+	}
+
+	return result
+}

internal/app/etl/table_analyzers/main.go

@@ -0,0 +1,40 @@
+package table_analyzers
+
+import (
+	"context"
+
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/config"
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/etl"
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/models"
+	"github.com/google/uuid"
+)
+
+func PartitionRangeGenerator(
+	ctx context.Context,
+	tableAnalyzer etl.TableAnalyzer,
+	tableInfo config.TableInfo,
+	partitionColumn string,
+	rowsPerPartition int64,
+) ([]models.Partition, error) {
+	rowsCount, err := tableAnalyzer.EstimateTotalRows(ctx, tableInfo)
+	if err != nil {
+		return nil, err
+	}
+
+	if rowsCount <= rowsPerPartition {
+		return []models.Partition{{
+			Id:           uuid.New(),
+			HasRange:     false,
+			RetryCounter: 0,
+		}}, nil
+
+	}
+
+	partitionsCount := rowsCount / rowsPerPartition
+	partitions, err := tableAnalyzer.CalculatePartitionRanges(ctx, tableInfo, partitionColumn, partitionsCount)
+	if err != nil {
+		return nil, err
+	}
+
+	return partitions, nil
+}

internal/app/etl/table_analyzers/mssql.go

@@ -0,0 +1,254 @@
+package table_analyzers
+
+import (
+	"context"
+	"database/sql"
+	"fmt"
+	"strings"
+	"time"
+
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/config"
+	dbwrapper "git.ksdemosapps.com/kylesoda/go-migrate/internal/app/db-wrapper"
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/etl"
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/models"
+	"github.com/google/uuid"
+)
+
+type MssqlTableAnalyzer struct {
+	db dbwrapper.DbWrapper
+}
+
+func NewMssqlTableAnalyzer(db dbwrapper.DbWrapper) etl.TableAnalyzer {
+	return &MssqlTableAnalyzer{db: db}
+}
+
+const mssqlColumnMetadataQuery string = `
+SELECT 
+	c.name AS name,
+	t.name AS user_type,
+	CASE WHEN t.is_user_defined = 0 THEN t.name ELSE bt.name END AS system_type,
+	c.is_nullable AS nullable,
+	c.max_length AS max_length,
+	c.precision AS precision,
+	c.scale AS scale
+FROM sys.columns c
+JOIN sys.types t ON c.user_type_id = t.user_type_id
+LEFT JOIN sys.types bt ON t.is_user_defined = 1 AND bt.user_type_id = t.system_type_id
+JOIN sys.tables st ON c.object_id = st.object_id
+JOIN sys.schemas s ON st.schema_id = s.schema_id
+WHERE s.name = @schema AND st.name = @table AND (c.is_hidden = 0 OR (c.graph_type IS NOT NULL AND c.name LIKE '$%'))
+ORDER BY c.column_id;`
+
+// AND c.name NOT LIKE '$%'
+
+type rawColumnMssql struct {
+	name       string
+	userType   string
+	systemType string
+	nullable   bool
+	maxLength  int64
+	precision  int64
+	scale      int64
+}
+
+func (ta *MssqlTableAnalyzer) systemTypeToUnifiedType(systemType string) string {
+	systemType = strings.ToLower(systemType)
+
+	if systemType == "varchar" || systemType == "char" || systemType == "nvarchar" || systemType == "nchar" || systemType == "text" || systemType == "ntext" {
+		return "STRING"
+	}
+
+	if systemType == "int" || systemType == "int4" || systemType == "integer" || systemType == "smallint" || systemType == "int2" || systemType == "bigint" || systemType == "int8" || systemType == "tinyint" {
+		return "INTEGER"
+	}
+
+	if systemType == "decimal" || systemType == "numeric" {
+		return "DECIMAL"
+	}
+
+	if systemType == "float" || systemType == "real" || systemType == "double precision" {
+		return "FLOAT"
+	}
+
+	if systemType == "bit" || systemType == "boolean" {
+		return "BOOLEAN"
+	}
+
+	if systemType == "date" {
+		return "DATE"
+	}
+	if systemType == "time" || systemType == "time without time zone" {
+		return "TIME"
+	}
+	if systemType == "datetime" || systemType == "datetime2" || systemType == "timestamp" || systemType == "timestamptz" || systemType == "timestamp with time zone" {
+		return "TIMESTAMP"
+	}
+
+	if systemType == "binary" || systemType == "varbinary" || systemType == "image" || systemType == "bytea" {
+		return "BINARY"
+	}
+
+	if systemType == "uniqueidentifier" || systemType == "uuid" {
+		return "UUID"
+	}
+
+	if systemType == "json" {
+		return "JSON"
+	}
+
+	if systemType == "geometry" || systemType == "geography" {
+		return "GEOMETRY"
+	}
+
+	return strings.ToUpper(systemType)
+}
+
+func (ta *MssqlTableAnalyzer) rawColumnToColumnType(rawColumn rawColumnMssql) models.ColumnType {
+	const nullValue int64 = -1
+	stringTypes := map[string]bool{"varchar": true, "char": true, "nvarchar": true, "nchar": true, "text": true, "ntext": true}
+	decimalTypes := map[string]bool{"decimal": true, "numeric": true}
+
+	if stringTypes[rawColumn.systemType] {
+		if rawColumn.systemType == "nvarchar" || rawColumn.systemType == "nchar" {
+			if rawColumn.maxLength > 0 {
+				rawColumn.maxLength = rawColumn.maxLength / 2
+			}
+		}
+
+		rawColumn.precision, rawColumn.scale = nullValue, nullValue
+	} else if decimalTypes[rawColumn.systemType] {
+		rawColumn.maxLength = nullValue
+	} else {
+		rawColumn.maxLength, rawColumn.precision, rawColumn.scale = nullValue, nullValue, nullValue
+	}
+
+	columnType := models.NewColumnType(
+		rawColumn.name,
+		rawColumn.maxLength != nullValue,
+		rawColumn.precision != nullValue || rawColumn.scale != nullValue,
+		rawColumn.userType,
+		rawColumn.systemType,
+		ta.systemTypeToUnifiedType(rawColumn.systemType),
+		rawColumn.nullable,
+		rawColumn.maxLength,
+		rawColumn.precision,
+		rawColumn.scale,
+	)
+
+	return columnType
+}
+
+func (ta *MssqlTableAnalyzer) QueryColumnTypes(
+	ctx context.Context,
+	tableInfo config.TableInfo,
+) ([]models.ColumnType, error) {
+	localCtx, cancel := context.WithTimeout(ctx, 20*time.Second)
+	defer cancel()
+
+	rows, err := ta.db.Query(localCtx, mssqlColumnMetadataQuery, sql.Named("schema", tableInfo.Schema), sql.Named("table", tableInfo.Table))
+	if err != nil {
+		return nil, err
+	}
+	defer rows.Close()
+
+	var columnTypes []models.ColumnType
+
+	for rows.Next() {
+		var rawColumn rawColumnMssql
+
+		if err := rows.Scan(
+			&rawColumn.name,
+			&rawColumn.userType,
+			&rawColumn.systemType,
+			&rawColumn.nullable,
+			&rawColumn.maxLength,
+			&rawColumn.precision,
+			&rawColumn.scale,
+		); err != nil {
+			return nil, err
+		}
+
+		columnTypes = append(columnTypes, ta.rawColumnToColumnType(rawColumn))
+	}
+
+	return columnTypes, nil
+}
+
+func (ta *MssqlTableAnalyzer) EstimateTotalRows(
+	ctx context.Context,
+	tableInfo config.TableInfo,
+) (int64, error) {
+	query := `
+SELECT SUM(p.rows) AS count
+FROM sys.tables t
+JOIN sys.schemas s ON t.schema_id = s.schema_id
+JOIN sys.partitions p ON t.object_id = p.object_id
+WHERE s.name = @schema AND t.name = @table AND p.index_id IN (0, 1)
+GROUP BY t.name`
+
+	ctxTimeout, cancel := context.WithTimeout(ctx, 1*time.Minute)
+	defer cancel()
+
+	var rowsCount int64
+	err := ta.db.QueryRow(ctxTimeout, query, sql.Named("schema", tableInfo.Schema), sql.Named("table", tableInfo.Table)).Scan(&rowsCount)
+	if err != nil {
+		return 0, err
+	}
+
+	return rowsCount, nil
+}
+
+func (ta *MssqlTableAnalyzer) CalculatePartitionRanges(
+	ctx context.Context,
+	tableInfo config.TableInfo,
+	partitionColumn string,
+	maxPartitions int64,
+) ([]models.Partition, error) {
+	query := fmt.Sprintf(`
+SELECT
+	MIN([%s]) AS lower_limit,
+	MAX([%s]) AS upper_limit
+FROM (SELECT [%s], NTILE(@maxPartitions) OVER (ORDER BY [%s]) AS batch_id FROM [%s].[%s]) AS T
+GROUP BY batch_id
+ORDER BY batch_id`,
+		partitionColumn,
+		partitionColumn,
+		partitionColumn,
+		partitionColumn,
+		tableInfo.Schema,
+		tableInfo.Table)
+
+	ctxTimeout, cancel := context.WithTimeout(ctx, 1*time.Minute)
+	defer cancel()
+
+	rows, err := ta.db.Query(ctxTimeout, query, sql.Named("maxPartitions", maxPartitions))
+	if err != nil {
+		return nil, err
+	}
+	defer rows.Close()
+
+	partitions := make([]models.Partition, 0, maxPartitions)
+
+	for rows.Next() {
+		partition := models.Partition{
+			Id:           uuid.New(),
+			HasRange:     true,
+			RetryCounter: 0,
+			Range: models.PartitionRange{
+				IsMinInclusive: true,
+			},
+		}
+
+		if err := rows.Scan(&partition.Range.Min, &partition.Range.Max); err != nil {
+			return nil, err
+		}
+
+		partitions = append(partitions, partition)
+	}
+
+	if err := rows.Err(); err != nil {
+		return nil, err
+	}
+
+	return partitions, nil
+}

internal/app/etl/table_analyzers/postgres.go

@@ -0,0 +1,174 @@
+package table_analyzers
+
+import (
+	"context"
+	"strings"
+	"time"
+
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/config"
+	dbwrapper "git.ksdemosapps.com/kylesoda/go-migrate/internal/app/db-wrapper"
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/etl"
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/models"
+)
+
+type PostgresTableAnalyzer struct {
+	db dbwrapper.DbWrapper
+}
+
+func NewPostgresTableAnalyzer(db dbwrapper.DbWrapper) etl.TableAnalyzer {
+	return &PostgresTableAnalyzer{db: db}
+}
+
+const postgresColumnMetadataQuery string = `
+SELECT 
+  c.column_name AS name,
+  c.data_type AS user_type,
+  c.udt_name AS system_type,
+  (CASE WHEN c.is_nullable = 'YES' THEN TRUE ELSE FALSE END) AS nullable,
+  COALESCE(c.character_maximum_length, -1) AS max_length,
+  COALESCE(c.numeric_precision, -1) AS precision,
+  COALESCE(c.numeric_scale, -1) AS scale
+FROM information_schema.columns c
+WHERE c.table_schema = $1 AND c.table_name = $2
+ORDER BY c.ordinal_position;`
+
+type rawColumnPostgres struct {
+	name       string
+	userType   string
+	systemType string
+	nullable   bool
+	maxLength  int64
+	precision  int64
+	scale      int64
+}
+
+func (ta *PostgresTableAnalyzer) systemTypeToUnifiedType(systemType string) string {
+	systemType = strings.ToLower(systemType)
+
+	if systemType == "varchar" || systemType == "char" || systemType == "nvarchar" || systemType == "nchar" || systemType == "text" || systemType == "ntext" {
+		return "STRING"
+	}
+
+	if systemType == "int" || systemType == "int4" || systemType == "integer" || systemType == "smallint" || systemType == "int2" || systemType == "bigint" || systemType == "int8" || systemType == "tinyint" {
+		return "INTEGER"
+	}
+
+	if systemType == "decimal" || systemType == "numeric" {
+		return "DECIMAL"
+	}
+
+	if systemType == "float" || systemType == "real" || systemType == "double precision" {
+		return "FLOAT"
+	}
+
+	if systemType == "bit" || systemType == "boolean" {
+		return "BOOLEAN"
+	}
+
+	if systemType == "date" {
+		return "DATE"
+	}
+	if systemType == "time" || systemType == "time without time zone" {
+		return "TIME"
+	}
+	if systemType == "datetime" || systemType == "datetime2" || systemType == "timestamp" || systemType == "timestamptz" || systemType == "timestamp with time zone" {
+		return "TIMESTAMP"
+	}
+
+	if systemType == "binary" || systemType == "varbinary" || systemType == "image" || systemType == "bytea" {
+		return "BINARY"
+	}
+
+	if systemType == "uniqueidentifier" || systemType == "uuid" {
+		return "UUID"
+	}
+
+	if systemType == "json" {
+		return "JSON"
+	}
+
+	if systemType == "geometry" || systemType == "geography" {
+		return "GEOMETRY"
+	}
+
+	return strings.ToUpper(systemType)
+}
+
+func (ta *PostgresTableAnalyzer) rawColumnToColumnType(rawColumn rawColumnPostgres) models.ColumnType {
+	const nullValue int64 = -1
+	stringTypes := map[string]bool{"varchar": true, "char": true, "text": true}
+	decimalTypes := map[string]bool{"decimal": true, "numeric": true}
+
+	if stringTypes[rawColumn.systemType] {
+		rawColumn.precision, rawColumn.scale = nullValue, nullValue
+	} else if decimalTypes[rawColumn.systemType] {
+		rawColumn.maxLength = nullValue
+	} else {
+		rawColumn.maxLength, rawColumn.precision, rawColumn.scale = nullValue, nullValue, nullValue
+	}
+
+	return models.NewColumnType(
+		rawColumn.name,
+		rawColumn.maxLength != nullValue,
+		rawColumn.precision != nullValue || rawColumn.scale != nullValue,
+		rawColumn.userType,
+		rawColumn.systemType,
+		ta.systemTypeToUnifiedType(rawColumn.systemType),
+		rawColumn.nullable,
+		rawColumn.maxLength,
+		rawColumn.precision,
+		rawColumn.scale,
+	)
+}
+
+func (ta *PostgresTableAnalyzer) QueryColumnTypes(
+	ctx context.Context,
+	tableInfo config.TableInfo,
+) ([]models.ColumnType, error) {
+	localCtx, cancel := context.WithTimeout(ctx, 1*time.Minute)
+	defer cancel()
+
+	rows, err := ta.db.Query(localCtx, postgresColumnMetadataQuery, tableInfo.Schema, tableInfo.Table)
+	if err != nil {
+		return nil, err
+	}
+	defer rows.Close()
+
+	var colTypes []models.ColumnType
+
+	for rows.Next() {
+		var column rawColumnPostgres
+
+		if err := rows.Scan(
+			&column.name,
+			&column.userType,
+			&column.systemType,
+			&column.nullable,
+			&column.maxLength,
+			&column.precision,
+			&column.scale,
+		); err != nil {
+			return nil, err
+		}
+
+		colTypes = append(colTypes, ta.rawColumnToColumnType(column))
+	}
+
+	return colTypes, nil
+}
+
+func (ta *PostgresTableAnalyzer) EstimateTotalRows(
+	ctx context.Context,
+	tableInfo config.TableInfo,
+) (int64, error) {
+	return 0, nil
+}
+
+func (ta *PostgresTableAnalyzer) CalculatePartitionRanges(
+	ctx context.Context,
+	tableInfo config.TableInfo,
+	partitionColumn string,
+	maxPartitions int64,
+) ([]models.Partition, error) {
+	return []models.Partition{}, nil
+}

internal/app/etl/transformers/mssql.go

@@ -0,0 +1,110 @@
+package transformers
+
+import (
+	"context"
+	"errors"
+	"sync"
+
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/custom_errors"
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/etl"
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/models"
+)
+
+type MssqlTransformer struct{}
+
+func NewMssqlTransformer() etl.Transformer {
+	return &MssqlTransformer{}
+}
+
+func computeTransformationPlan(columns []models.ColumnType) []etl.ColumnTransformPlan {
+	return []etl.ColumnTransformPlan{}
+}
+
+const processBatchCtxCheck = 4096
+
+func (mssqlTr *MssqlTransformer) ProcessBatch(
+	ctx context.Context,
+	batch *models.Batch,
+	transformationPlan []etl.ColumnTransformPlan,
+) error {
+	for i, rowValues := range batch.Rows {
+		if i%processBatchCtxCheck == 0 {
+			if err := ctx.Err(); err != nil {
+				return err
+			}
+		}
+
+		for _, task := range transformationPlan {
+			val := rowValues[task.Index]
+			if val == nil {
+				continue
+			}
+
+			transformed, err := task.Fn(val)
+			if err != nil {
+				return err
+			}
+			rowValues[task.Index] = transformed
+		}
+	}
+
+	return nil
+}
+
+func (mssqlTr *MssqlTransformer) Exec(
+	ctx context.Context,
+	columns []models.ColumnType,
+	chBatchesIn <-chan models.Batch,
+	chBatchesOut chan<- models.Batch,
+	chJobErrorsOut chan<- custom_errors.JobError,
+	wgActiveBatches *sync.WaitGroup,
+) {
+	transformationPlan := computeTransformationPlan(columns)
+
+	for {
+		if ctx.Err() != nil {
+			return
+		}
+
+		select {
+		case <-ctx.Done():
+			return
+
+		case batch, ok := <-chBatchesIn:
+			if !ok {
+				return
+			}
+
+			if len(transformationPlan) == 0 {
+				select {
+				case chBatchesOut <- batch:
+					wgActiveBatches.Add(1)
+					continue
+				case <-ctx.Done():
+					return
+				}
+			}
+
+			err := mssqlTr.ProcessBatch(ctx, &batch, transformationPlan)
+			if err != nil {
+				if errors.Is(err, ctx.Err()) {
+					return
+				}
+
+				select {
+				case chJobErrorsOut <- custom_errors.JobError{ShouldCancelJob: true, Msg: "Transformation failed", Prev: err}:
+				case <-ctx.Done():
+				}
+				return
+			}
+
+			select {
+			case chBatchesOut <- batch:
+			case <-ctx.Done():
+				return
+			}
+
+			wgActiveBatches.Add(1)
+		}
+	}
+}

internal/app/etl/transformers/types.go

@@ -0,0 +1 @@
+package transformers

internal/app/etl/transformers/utils.go

@@ -1,4 +1,4 @@
-package main
+package transformers
 
 import (
 	"encoding/binary"

internal/app/etl/types.go

@@ -0,0 +1,98 @@
+package etl
+
+import (
+	"context"
+	"sync"
+
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/config"
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/custom_errors"
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/models"
+)
+
+type Extractor interface {
+	ProcessPartition(
+		ctx context.Context,
+		tableInfo config.SourceTableInfo,
+		columns []models.ColumnType,
+		batchSize int,
+		partition models.Partition,
+		indexPrimaryKey int,
+		chBatchesOut chan<- models.Batch,
+	) (int, error)
+
+	Exec(
+		ctx context.Context,
+		tableInfo config.SourceTableInfo,
+		columns []models.ColumnType,
+		batchSize int,
+		chPartitionsIn <-chan models.Partition,
+		chBatchesOut chan<- models.Batch,
+		chErrorsOut chan<- custom_errors.ExtractorError,
+		chJobErrorsOut chan<- custom_errors.JobError,
+		wgActivePartitions *sync.WaitGroup,
+		rowsRead *int64,
+	)
+}
+
+type TransformerFunc func(any) (any, error)
+
+type ColumnTransformPlan struct {
+	Index int
+	Fn    TransformerFunc
+}
+
+type Transformer interface {
+	ProcessBatch(
+		ctx context.Context,
+		batch *models.Batch,
+		transformationPlan []ColumnTransformPlan,
+	) error
+
+	Exec(
+		ctx context.Context,
+		columns []models.ColumnType,
+		chBatchesIn <-chan models.Batch,
+		chBactchesOut chan<- models.Batch,
+		chJobErrorsOut chan<- custom_errors.JobError,
+		wgActiveBatches *sync.WaitGroup,
+	)
+}
+
+type Loader interface {
+	ProcessBatch(
+		ctx context.Context,
+		tableInfo config.TargetTableInfo,
+		colNames []string,
+		batch models.Batch,
+	) (int, error)
+
+	Exec(
+		ctx context.Context,
+		tableInfo config.TargetTableInfo,
+		columns []models.ColumnType,
+		chBatchesIn <-chan models.Batch,
+		chErrorsOut chan<- custom_errors.LoaderError,
+		chJobErrorsOut chan<- custom_errors.JobError,
+		wgActiveBatches *sync.WaitGroup,
+		rowsLoaded *int64,
+	)
+}
+
+type TableAnalyzer interface {
+	QueryColumnTypes(
+		ctx context.Context,
+		tableInfo config.TableInfo,
+	) ([]models.ColumnType, error)
+
+	EstimateTotalRows(
+		ctx context.Context,
+		tableInfo config.TableInfo,
+	) (int64, error)
+
+	CalculatePartitionRanges(
+		ctx context.Context,
+		tableInfo config.TableInfo,
+		partitionColumn string,
+		maxPartitions int64,
+	) ([]models.Partition, error)
+}

internal/app/models/colum-type.go

@@ -42,3 +42,29 @@ func (c *ColumnType) Nullable() bool {
 func (c *ColumnType) Type() string {
 	return c.unifiedType
 }
+
+func NewColumnType(
+	name string,
+	hasMaxLength bool,
+	hasPrecisionScale bool,
+	userType string,
+	systemType string,
+	unifiedType string,
+	nullable bool,
+	maxLength int64,
+	precision int64,
+	scale int64,
+) ColumnType {
+	return ColumnType{
+		name,
+		hasMaxLength,
+		hasPrecisionScale,
+		userType,
+		systemType,
+		unifiedType,
+		nullable,
+		maxLength,
+		precision,
+		scale,
+	}
+}

internal/app/models/main.go

@@ -1,22 +1,41 @@
 package models
 
-import "github.com/google/uuid"
+import (
+	"time"
+
+	"github.com/google/uuid"
+)
 
 type UnknownRowValues = []any
 
-type Chunk struct {
+type Batch struct {
 	Id           uuid.UUID
-	BatchId      uuid.UUID
+	PartitionId  uuid.UUID
-	Data         []UnknownRowValues
+	Rows         []UnknownRowValues
 	RetryCounter int
 }
 
-type Batch struct {
+type PartitionRange struct {
+	Min            int64
+	Max            int64
+	IsMinInclusive bool
+	IsMaxInclusive bool
+}
+
+type Partition struct {
 	Id           uuid.UUID
 	ParentId     uuid.UUID
-	LowerLimit            int64
+	Range        PartitionRange
-	UpperLimit            int64
+	HasRange     bool
-	IsLowerLimitInclusive bool
-	ShouldUseRange        bool
 	RetryCounter int
 }
+
+type JobResult struct {
+	JobName    string
+	StartTime  time.Time
+	Duration   time.Duration
+	RowsRead   int64
+	RowsLoaded int64
+	RowsFailed int64
+	Error      error
+}

scripts/pg-info-test/main.go

@@ -8,13 +8,32 @@ import (
 	"time"
 
 	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/config"
-	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/db"
 	"github.com/jackc/pgx/v5"
 	"github.com/jackc/pgx/v5/pgtype"
 	"github.com/jackc/pgx/v5/pgxpool"
 	log "github.com/sirupsen/logrus"
 )
 
+func Connect(ctx context.Context, dbURL string) (*pgxpool.Pool, error) {
+	pool, err := pgxpool.New(ctx, dbURL)
+	if err != nil {
+		return nil, fmt.Errorf("unable to connect to database: %w", err)
+	}
+
+	if err := pool.Ping(ctx); err != nil {
+		pool.Close()
+		return nil, fmt.Errorf("unable to ping database: %w", err)
+	}
+
+	return pool, nil
+}
+
+func Close(pool *pgxpool.Pool) {
+	if pool != nil {
+		pool.Close()
+	}
+}
+
 func main() {
 	log.SetFormatter(&log.TextFormatter{
 		FullTimestamp:   true,
@@ -27,8 +46,8 @@ func main() {
 	ctxSource, cancel := context.WithTimeout(context.Background(), 20*time.Second)
 	defer cancel()
 
-	sourcePool, err := db.Connect(ctxSource, config.App.SourceDbUrl)
+	sourcePool, err := Connect(ctxSource, config.App.SourceDbUrl)
-	defer db.Close(sourcePool)
+	defer Close(sourcePool)
 	if err != nil {
 		log.Fatal(err)
 	}
@@ -37,8 +56,8 @@ func main() {
 	ctxTarget, cancel := context.WithTimeout(context.Background(), 20*time.Second)
 	defer cancel()
 
-	targetPool, err := db.Connect(ctxTarget, config.App.TargetDbUrl)
+	targetPool, err := Connect(ctxTarget, config.App.TargetDbUrl)
-	defer db.Close(targetPool)
+	defer Close(targetPool)
 	if err != nil {
 		log.Fatal(err)
 	}