feat: add extract query builders for both dialects

Add QueryFromObject helpers to build select queries from a table schema, with limits and geometry support.
This commit is contained in:
2026-04-09 16:00:00 -05:00
parent c5987fa7f8
commit 4f181269a4
5 changed files with 11 additions and 236 deletions

View File

@@ -1,11 +1,7 @@
package main
import (
"context"
"database/sql"
"fmt"
"sync"
"github.com/jackc/pgx/v5/pgxpool"
_ "github.com/microsoft/go-mssqldb"
@@ -21,43 +17,12 @@ func processMigrationJob(sourceDb *sql.DB, targetDb *pgxpool.Pool, job Migration
logColumnTypes(sourceColTypes, "Source col types")
logColumnTypes(targetColTypes, "Target col types")
chRowsExtract := make(chan []UnknownRowValues, QueueSize)
chRowsTransform := make(chan []UnknownRowValues)
mssqlContext := context.Background()
sourceQuery := buildExtractQueryMssql(job, sourceColTypes)
go func() {
if err := extractFromMssql(mssqlContext, job, sourceColTypes, ChunkSize, sourceDb, chRowsExtract); err != nil {
log.Error("Unexpected error extrating data from mssql: ", err)
}
close(chRowsExtract)
}()
log.Debug(sourceQuery)
go func() {
transformRowsMssql(sourceColTypes, chRowsExtract, chRowsTransform)
close(chRowsTransform)
}()
var wgFakeLoaders sync.WaitGroup
wgFakeLoaders.Go(func() {
fakeLoader(job, sourceColTypes, chRowsTransform)
})
chRowsExtractPostgres := make(chan []UnknownRowValues, QueueSize)
postgresContext := context.Background()
go func() {
if err := extractFromPostgres(postgresContext, job, sourceColTypes, ChunkSize, targetDb, chRowsExtractPostgres); err != nil {
log.Error("Unexpected error extrating data from postgres: ", err)
}
close(chRowsExtractPostgres)
}()
wgFakeLoaders.Go(func() {
fakeLoader(job, targetColTypes, chRowsExtractPostgres)
})
wgFakeLoaders.Wait()
targetQuery := buildExtractQueryPostgres(job, targetColTypes)
log.Debug(targetQuery)
}
func logColumnTypes(columnTypes []ColumnType, label string) {
@@ -67,45 +32,3 @@ func logColumnTypes(columnTypes []ColumnType, label string) {
log.Infof("%+v", col)
}
}
func transformRowsMssql(columns []ColumnType, in <-chan []UnknownRowValues, out chan<- []UnknownRowValues) {
for rows := range in {
log.Debugf("Chunk received, transforming...")
for _, rowValues := range rows {
for i, col := range columns {
value := rowValues[i]
if col.SystemType() == "uniqueidentifier" {
if b, ok := value.([]byte); ok {
rowValues[i] = mssqlUuidToBigEndian(b)
}
} else if col.SystemType() == "geometry" || col.SystemType() == "geography" {
if b, ok := value.([]byte); ok {
rowValues[i] = wkbToEwkbWithSrid(b, 4326)
}
}
}
}
out <- rows
}
}
func logSampleRow(job MigrationJob, columns []ColumnType, rowValues UnknownRowValues, tag string) {
log.Infof("[%s.%s] Sample row: (%s)", job.Schema, job.Table, tag)
for i, col := range columns {
log.Infof("%s (%T): %v", col.Name(), rowValues[i], rowValues[i])
}
}
func fakeLoader(job MigrationJob, columns []ColumnType, in <-chan []UnknownRowValues) {
for rows := range in {
log.Debugf("Chunk received, loading data into...")
for i, rowValues := range rows {
if i%100 == 0 {
logSampleRow(job, columns, rowValues, fmt.Sprintf("row %d", i))
}
}
}
}