cgimport/importer/importer.go

package importer

import (
	"encoding/json"
	"fmt"
	"io"
	"os"
	"path/filepath"
	"strings"
	"sync"
	"sync/atomic"
	"time"

	"git.wecise.com/wecise/cgimport/odbc"
	"git.wecise.com/wecise/cgimport/reader"
	"git.wecise.com/wecise/cgimport/schema"
	"git.wecise.com/wecise/util/filewalker"
	"git.wecise.com/wecise/util/merrs"
	"git.wecise.com/wecise/util/rc"
)

var mcfg = odbc.Config
var logger = odbc.Logger

type Importer struct {
	datapath     string
	parallel     int
	reload       bool
	importrc     *rc.RoutinesController
	odbcimporter *ODBCImporter
}

func ImportDir(datapath string, parallel int, reload bool) (totalfilescount, totalrecordscount int64, totalusetime time.Duration, filescount, recordscount int64, usetime time.Duration, err error) {
	importer := &Importer{
		datapath:     datapath,
		parallel:     parallel,
		reload:       reload,
		importrc:     rc.NewRoutinesController("", 100),
		odbcimporter: NewODBCImporter(),
	}
	return importer.Import()
}

func (importer *Importer) Import() (totalfilescount, totalrecordscount int64, totalusetime time.Duration, filescount, recordscount int64, usetime time.Duration, err error) {
	var cgirc = rc.NewRoutinesController("", importer.parallel)
	var wg sync.WaitGroup
	fw, e := filewalker.NewFileWalker([]string{importer.datapath}, ".*")
	if e != nil {
		err = e
		return
	}
	cgistatus := NewCGIStatus()
	if odbc.DevPhase&odbc.DP_PROCESSCONTINUE != 0 && !importer.reload {
		err = cgistatus.Load()
		if err != nil {
			return
		}
	} else {
		// reload
		// 清除已有类
		err = importer.odbcimporter.reload()
		if err != nil {
			return
		}
	}
	// 建类
	err = importer.odbcimporter.ReviseClassStruct()
	if err != nil {
		return
	}
	totalfilescount = int64(len(cgistatus.ImportStatus))
	for _, v := range cgistatus.ImportStatus {
		totalrecordscount += v.RecordsCount
	}
	totalusetime = cgistatus.TotalUseTime
	st := time.Now().Add(-totalusetime)
	cst := time.Now()
	// 遍历文件目录
	e = fw.List(func(basedir string, fpath string) bool {
		if err != nil {
			return false
		}
		filename := filepath.Join(basedir, fpath)
		wg.Add(1)
		cgirc.ConcurCall(1,
			func() {
				defer wg.Done()
				cgistatus.mutex.RLock()
				importstatus := cgistatus.ImportStatus[filename]
				cgistatus.mutex.RUnlock()
				if importstatus != nil {
					return
				}
				records, e := importer.ImportFile(filename)
				if e != nil {
					err = e
					return
				}
				atomic.AddInt64(&filescount, 1)
				atomic.AddInt64(&recordscount, records)
				atomic.AddInt64(&totalfilescount, 1)
				atomic.AddInt64(&totalrecordscount, records)
				usetime = time.Since(cst)
				totalusetime = time.Since(st)
				cgistatus.mutex.Lock()
				cgistatus.ImportStatus[filename] = &ImportStatus{RecordsCount: records}
				cgistatus.TotalUseTime = totalusetime
				cgistatus.mutex.Unlock()
				cgistatus.Save()
			},
		)
		return true
	})
	wg.Wait()
	if e != nil {
		if os.IsNotExist(e) {
			err = merrs.NewError(`directory "`+importer.datapath+`" not exist specified by "datapath"`, e)
		} else {
			err = merrs.NewError(e)
		}
		return
	}
	cgistatus.WaitSaveDone()
	importer.alldone()
	return
}

func (importer *Importer) ImportFile(filepath string) (blockcount int64, err error) {
	f, e := os.Open(filepath)
	if e != nil {
		return blockcount, merrs.NewError(e, merrs.SSMaps{{"filename": filepath}})
	}
	defer f.Close()
	return importer.importReader(filepath, f)
}

func (importer *Importer) importReader(filename string, buf io.Reader) (blockcount int64, err error) {
	var filetype schema.FileType
	switch {
	case strings.Contains(filename, "_L1_"):
		filetype = schema.FT_LEVEL1
	case strings.Contains(filename, "_L2_"):
		filetype = schema.FT_LEVEL2
	case strings.Contains(filename, "_L3_"):
		filetype = schema.FT_LEVEL3
	case strings.Contains(filename, "_L4_"):
		filetype = schema.FT_LEVEL4
	case strings.Contains(filename, "_L5_"):
		filetype = schema.FT_LEVEL5
	case strings.Contains(filename, "_L6_"):
		filetype = schema.FT_LEVEL6
	case strings.Contains(filename, "_L7_"):
		filetype = schema.FT_LEVEL7
	case strings.Contains(filename, "_L8_"):
		filetype = schema.FT_LEVEL8
	case strings.Contains(filename, "MASTER"):
		filetype = schema.FT_MASTER
	case strings.Contains(filename, "EDGE"):
		filetype = schema.FT_EDGE
	default:
		err = merrs.NewError("filename does not conform to the agreed format " + filename)
		return
	}
	br, e := reader.NewBlockReader(filename, filetype, buf)
	if e != nil {
		return blockcount, merrs.NewError(e, merrs.SSMaps{{"filename": filename}})
	}
	var wg sync.WaitGroup
	defer importer.done()
	defer wg.Wait()
	for {
		if err != nil {
			break
		}
		block, line, linecount, e := br.ReadBlock()
		if e != nil {
			return blockcount, merrs.NewError(e, merrs.SSMaps{{"filename": filename}, {"linecount": fmt.Sprint(linecount)}, {"line": line}})
		}
		if block == nil {
			return
		}
		wg.Add(1)
		e = importer.importrc.ConcurCall(1, func() {
			defer wg.Done()
			e = importer.importRecord(block, line, filename, filetype, linecount)
			if e != nil {
				err = merrs.NewError(e, merrs.SSMaps{{"filename": filename}, {"linecount": fmt.Sprint(linecount)}, {"line": line}})
				return
			}
			atomic.AddInt64(&blockcount, 1)
		})
		if e != nil {
			return blockcount, merrs.NewError(e, merrs.SSMaps{{"filename": filename}, {"linecount": fmt.Sprint(linecount)}, {"line": line}})
		}
	}
	return
}

func (importer *Importer) importRecord(record map[string]any, line string, filename string, filetype schema.FileType, linecount int) (err error) {
	if odbc.LogDebug {
		bs, e := json.MarshalIndent(record, "", "  ")
		if e != nil {
			return merrs.NewError(e)
		}
		logger.Debug(fmt.Sprint("import ", filename, "[", linecount, "]:", string(bs)))
	}
	var classname string
	switch filetype {
	case schema.FT_EDGE:
		err = importer.odbcimporter.InsertEdge(record)
		if err != nil {
			err = merrs.NewError(err, merrs.SSMaps{{"filename": filename}, {"linecount": fmt.Sprint(linecount)}, {"line": line}})
			return
		}
	default:
		classname = string(filetype)
		err = importer.odbcimporter.InsertData(classname, record)
		if err != nil {
			err = merrs.NewError(err, merrs.SSMaps{{"filename": filename}, {"linecount": fmt.Sprint(linecount)}, {"line": line}})
			return
		}
	}
	return
}

func (importer *Importer) alldone() {
	importer.odbcimporter.alldone()
}

func (importer *Importer) done() {
	importer.odbcimporter.done()
}

func Check() {
	client := odbc.ODBClient
	if client == nil {
		return
	}
	{
		mql := "select id,uniqueid,tags,contain,day,vtime from level1 where uniqueid='E2E:OTR0002L'"
		r, e := client.Query(mql).Do()
		if e != nil {
			panic(merrs.NewError(e))
		}
		bs, _ := json.MarshalIndent(r.Data, "", "  ")
		fmt.Println(string(bs))
	}
}