refactor: rename Batch to Partition across pipeline

Rename internal types and channels from Batch to Partition for consistency with the data model.
This commit is contained in:
2026-04-11 18:00:00 -05:00
parent 1f8689cdfc
commit 7f3d2b8cc4
28 changed files with 1250 additions and 781 deletions

View File

@@ -8,10 +8,8 @@ import (
)
type appConfig struct {
SourceDbUrl string
SourceDbType string
TargetDbUrl string
TargetDbType string
SourceDbUrl string
TargetDbUrl 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,
SourceDbUrl: sourceDbUrl,
TargetDbUrl: targetDbUrl,
}
}

View File

@@ -0,0 +1,125 @@
package config
import (
"fmt"
"os"
"gopkg.in/yaml.v3"
)
type RetryConfig struct {
Attempts int `yaml:"attempts"`
}
type JobConfig struct {
MaxExtractors int `yaml:"max_extractors"`
MaxLoaders int `yaml:"max_loaders"`
QueueSize int `yaml:"queue_size"`
ChunkSize int `yaml:"chunk_size"`
ChunksPerBatch int `yaml:"chunks_per_batch"`
RowsPerBatch int64
TruncateTarget bool `yaml:"truncate_target"`
TruncateMethod string `yaml:"truncate_method"`
Retry RetryConfig `yaml:"retry"`
}
type TableInfo struct {
Schema string `yaml:"schema"`
Table string `yaml:"table"`
}
type TargetTableInfo struct {
TableInfo `yaml:",inline"`
}
type SourceTableInfo struct {
TableInfo `yaml:",inline"`
PrimaryKey string `yaml:"primary_key"`
}
type Job struct {
Name string `yaml:"name"`
Enabled bool `yaml:"enabled"`
SourceTable SourceTableInfo `yaml:"source"`
TargetTable TargetTableInfo `yaml:"target"`
PreSQL []string `yaml:"pre_sql"`
PostSQL []string `yaml:"post_sql"`
JobConfig `yaml:",inline"`
}
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"`
}
func (c *MigrationConfig) UnmarshalYAML(value *yaml.Node) error {
var raw rawConfig
if err := value.Decode(&raw); err != nil {
return err
}
c.MaxParallelWorkers = raw.MaxParallelWorkers
c.Defaults = raw.Defaults
c.Defaults.RowsPerBatch = int64(raw.Defaults.ChunkSize * raw.Defaults.ChunksPerBatch)
for _, node := range raw.Jobs {
job := Job{
JobConfig: raw.Defaults,
}
if err := node.Decode(&job); err != nil {
return err
}
job.RowsPerBatch = int64(job.ChunkSize * job.ChunksPerBatch)
c.Jobs = append(c.Jobs, job)
}
return nil
}
const defaultConfigFileName string = "config.yaml"
func filenamesOrDefault(filenames []string) []string {
if len(filenames) == 0 {
return []string{defaultConfigFileName}
}
return filenames
}
func ReadMigrationConfig(filenames ...string) (MigrationConfig, error) {
filenames = filenamesOrDefault(filenames)
var data []byte
var err error
for _, filename := range filenames {
data, err = os.ReadFile(filename)
if err != nil {
continue
}
break
}
if err != nil {
return MigrationConfig{}, fmt.Errorf("Error reading config file: %v", err)
}
var config MigrationConfig
if err := yaml.Unmarshal(data, &config); err != nil {
return MigrationConfig{}, fmt.Errorf("Error parsing config file: %v", err)
}
return config, nil
}