refactor: project code arch; feat: cancel operation anytime

This commit is contained in:
loveuer
2025-02-05 18:07:53 +08:00
parent b7b9c14e79
commit 4257cd3668
32 changed files with 1046 additions and 1628 deletions

81
internal/core/io.go Normal file
View File

@ -0,0 +1,81 @@
package core
import (
"context"
"encoding/json"
"fmt"
elastic7 "github.com/elastic/go-elasticsearch/v7"
"github.com/go-resty/resty/v2"
"github.com/loveuer/esgo2dump/internal/opt"
"github.com/loveuer/esgo2dump/internal/tool"
"github.com/loveuer/esgo2dump/internal/xfile"
"github.com/loveuer/esgo2dump/pkg/log"
"github.com/loveuer/esgo2dump/pkg/model"
"github.com/loveuer/esgo2dump/xes/es7"
"net/url"
"strings"
)
func NewIO(ctx context.Context, uri string, ioType model.IOType) (model.IO[map[string]any], error) {
type Version struct {
Name string
Version struct {
Number string `json:"number"`
} `json:"version"`
}
var (
err error
target *url.URL
rr *resty.Response
v Version
)
if target, err = url.Parse(uri); err != nil {
log.Debug("parse uri failed, type = %s, uri = %s, err = %s", ioType, uri, err.Error())
return xfile.NewClient(uri, ioType)
}
if err = tool.ValidScheme(target.Scheme); err != nil {
log.Debug("uri scheme check failed, type = %s, uri = %s", ioType, uri)
return xfile.NewClient(uri, ioType)
}
// elastic uri
index := strings.TrimPrefix(target.Path, "/")
if index == "" {
return nil, fmt.Errorf("uri invalid without index(path)")
}
log.Debug("%s uri es index = %s", ioType, index)
versionURL := fmt.Sprintf("%s://%s", target.Scheme, strings.Split(target.Host, ",")[0])
log.Debug("%s version url = %s", ioType, versionURL)
if rr, err = opt.HttpClient.R().Get(versionURL); err != nil {
log.Debug("get uri es version failed, type = %s, uri = %s, version_url = %s, err = %s", ioType, uri, versionURL, err.Error())
}
if err = json.Unmarshal(rr.Body(), &v); err != nil {
log.Debug("decode uri es version failed, type = %s, uri = %s, version_url = %s, err = %s", ioType, uri, versionURL, err.Error())
return nil, err
}
log.Debug("%s uri es version = %s", ioType, v.Version.Number)
mainVersion := strings.Split(v.Version.Number, ".")[0]
switch mainVersion {
case "8":
case "7":
var client *elastic7.Client
if client, err = es7.NewClient(ctx, uri); err != nil {
return nil, err
}
return es7.NewStreamer(ctx, client, index)
case "6":
default:
return nil, fmt.Errorf("es version not supported yet: %s", mainVersion)
}
return nil, nil
}

125
internal/core/run.data.go Normal file
View File

@ -0,0 +1,125 @@
package core
import (
"bufio"
"encoding/json"
"fmt"
"github.com/loveuer/esgo2dump/internal/opt"
"github.com/loveuer/esgo2dump/internal/tool"
"github.com/loveuer/esgo2dump/pkg/log"
"github.com/loveuer/esgo2dump/pkg/model"
"github.com/samber/lo"
"github.com/spf13/cobra"
"os"
"strings"
"sync"
)
func RunData(cmd *cobra.Command, input, output model.IO[map[string]any]) error {
var (
err error
// query chan
qc = make(chan map[string]any)
// error chan
ec = make(chan error)
// done chan
wc = &sync.WaitGroup{}
total = 0
)
wc.Add(1)
go func() {
var (
wroteCount = 0
items []map[string]any
)
defer wc.Done()
for query := range qc {
for {
limit := tool.CalculateLimit(opt.Cfg.Args.Limit, total, opt.Cfg.Args.Max)
log.Debug("one-step dump: arg.limit = %d, total = %d, arg.max = %d, calculate.limit = %d", opt.Cfg.Args.Limit, total, opt.Cfg.Args.Max, limit)
if limit == 0 {
break
}
if items, err = input.ReadData(
cmd.Context(),
limit,
query,
lo.Filter(strings.Split(opt.Cfg.Args.Field, ","), func(x string, _ int) bool { return x != "" }),
lo.Filter(strings.Split(opt.Cfg.Args.Sort, ","), func(x string, _ int) bool { return x != "" }),
); err != nil {
ec <- err
return
}
if len(items) == 0 {
break
}
if wroteCount, err = output.WriteData(cmd.Context(), items); err != nil {
ec <- err
return
}
total += wroteCount
if wroteCount != len(items) {
ec <- fmt.Errorf("got items %d, but wrote %d", len(items), wroteCount)
return
}
log.Info("Dump: dump data success = %d total = %d", wroteCount, total)
}
}
}()
switch {
case opt.Cfg.Args.QueryFile != "":
var (
// query file
qf *os.File
queryCount = 0
)
if qf, err = os.Open(opt.Cfg.Args.QueryFile); err != nil {
return err
}
scanner := bufio.NewScanner(qf)
for scanner.Scan() {
queryCount++
qm := make(map[string]any)
if err = json.Unmarshal(scanner.Bytes(), &qm); err != nil {
return err
}
qc <- qm
log.Debug("Dump: queries[%06d]", queryCount)
}
case opt.Cfg.Args.Query != "":
var (
qm = make(map[string]any)
)
if err = json.Unmarshal([]byte(opt.Cfg.Args.Query), &qm); err != nil {
return err
}
qc <- qm
default:
qc <- nil
}
// close query chan to stop trans_io_goroutine
close(qc)
wc.Wait()
log.Info("Dump: dump all data success, total = %d", total)
return nil
}

View File

@ -0,0 +1,19 @@
package core
import (
"github.com/loveuer/esgo2dump/pkg/model"
"github.com/spf13/cobra"
)
func RunMapping(cmd *cobra.Command, input model.IO[map[string]any], output model.IO[map[string]any]) error {
mapping, err := input.ReadMapping(cmd.Context())
if err != nil {
return err
}
if err = output.WriteMapping(cmd.Context(), mapping); err != nil {
return err
}
return nil
}

View File

@ -0,0 +1,19 @@
package core
import (
"github.com/loveuer/esgo2dump/pkg/model"
"github.com/spf13/cobra"
)
func RunSetting(cmd *cobra.Command, input model.IO[map[string]any], output model.IO[map[string]any]) error {
setting, err := input.ReadSetting(cmd.Context())
if err != nil {
return err
}
if err = output.WriteSetting(cmd.Context(), setting); err != nil {
return err
}
return nil
}