spider_creater/bind4comm.go

// 绑定公共接口
package main

import (
	"container/list"
	"encoding/json"
	"fmt"
	"github.com/bmaupin/go-epub"
	"github.com/wailsapp/wails/v2/pkg/runtime"
	"github.com/xuri/excelize/v2"
	qu "jygit.jydev.jianyu360.cn/data_processing/common_utils"
	"log"
	"os"
	"os/exec"
	sysruntim "runtime"
	be "spider_creator/backend"
	"strconv"
	"strings"
)

var FileType = map[string]runtime.FileFilter{
	"epub": runtime.FileFilter{Pattern: "*.epub", DisplayName: "epub file *.epub"},
	"xlsx": runtime.FileFilter{Pattern: "*.xlsx", DisplayName: "excel file *.xlsx"},
	"json": runtime.FileFilter{Pattern: "*.json", DisplayName: "json file *.json"},
}

// Greet returns a greeting for the given name
func (a *App) Dispatch(event string, data interface{}) error {
	runtime.EventsEmit(a.ctx, event, data)
	return nil
}

// SelectSaveFilePath
func (a *App) SelectSaveFilePath(defaultDirectory, defaultFileName, defaulFileType string) string {
	qu.Debug("导出文件位置：", defaultDirectory, defaultFileName, defaulFileType)
	path, err := runtime.SaveFileDialog(a.ctx, runtime.SaveDialogOptions{Filters: []runtime.FileFilter{
		FileType[defaulFileType],
	},
		DefaultFilename:  defaultFileName,
		DefaultDirectory: defaultDirectory,
	})
	if err != nil {
		qu.Debug(err.Error())
		return ""
	}
	return path
}

// SelectOpenFilePath
func (a *App) SelectOpenFilePath() string {
	path, err := runtime.OpenFileDialog(a.ctx, runtime.OpenDialogOptions{Filters: []runtime.FileFilter{
		{Pattern: "*.xlsx", DisplayName: "excel file *.xlsx"},
	}})
	if err != nil {
		qu.Debug(err.Error())
		return ""
	}
	return path
}

// RunExportExcelFile 数据集导出到excel文件中
func (a *App) RunExportExcelFile(filepath, code string, currentResult *list.List) error {
	qu.Debug("filepath---", filepath)
	f := excelize.NewFile()
	defer f.Close()
	f.SetCellStr("Sheet1", "A1", "站点")
	f.SetCellStr("Sheet1", "B1", "栏目")
	f.SetCellStr("Sheet1", "C1", "爬虫")
	//写入数据
	f.SetCellStr("Sheet1", "D1", "标题")
	f.SetCellStr("Sheet1", "E1", "链接")
	f.SetCellStr("Sheet1", "F1", "发布单位")
	f.SetCellStr("Sheet1", "G1", "发布时间")
	f.SetCellStr("Sheet1", "H1", "正文")
	f.SetCellStr("Sheet1", "I1", "附件")
	i := 0
	for el := currentResult.Front(); el != nil; el = el.Next() {
		r, _ := el.Value.(*be.ResultItem)
		//写入站点信息
		iStr := strconv.Itoa(i + 2)
		f.SetCellStr("Sheet1", "A"+iStr, r.Site)
		f.SetCellStr("Sheet1", "B"+iStr, r.Channel)
		f.SetCellStr("Sheet1", "C"+iStr, code)
		//写入数据
		f.SetCellStr("Sheet1", "D"+iStr, r.Title)
		f.SetCellStr("Sheet1", "E"+iStr, r.Href)
		f.SetCellStr("Sheet1", "F"+iStr, r.PublishUnit)
		f.SetCellStr("Sheet1", "G"+iStr, r.ListPubTime)
		f.SetCellStr("Sheet1", "H"+iStr, r.Content)
		f.SetCellStr("Sheet1", "I"+iStr, "")
		if len(r.AttachLinks) > 0 {
			bs, err := json.Marshal(r.AttachLinks)
			if err == nil {
				f.SetCellStr("Sheet1", "I"+iStr, string(bs))
			}
		}
		i += 1
	}
	err := f.SaveAs(filepath)
	if err != nil {
		return err
	}
	return nil
}

func (a *App) RunExportJsonFile(filepath, code string, currentResult *list.List) error {
	qu.Debug("filepath---", filepath)
	var result []map[string]interface{}
	for el := currentResult.Front(); el != nil; el = el.Next() {
		r, _ := el.Value.(*be.ResultItem)
		rmap := map[string]interface{}{
			"site":        r.Site,
			"channel":     r.Channel,
			"code":        code,
			"title":       r.Title,
			"href":        r.Href,
			"publishdept": r.PublishUnit,
			"publishtime": r.ListPubTime,
			"detail":      r.Content,
			"attachLinks": r.AttachLinks,
		}
		result = append(result, rmap)
	}
	jsonData, err := json.MarshalIndent(result, "", "	")
	if err != nil {
		return err
	}
	fo, err := os.Create(filepath)
	if err != nil {
		return err
	}
	defer fo.Close()
	if _, err := fo.Write(jsonData); err != nil {
		return fmt.Errorf("failed to write data to file: %w", err)
	}
	return nil
}

// RunExportEpubFile 导出epub文件
func (a *App) RunExportEpubFile(bookname, filepath string, currentResult *list.List) error {
	qu.Debug("filepath---", filepath)
	output := epub.NewEpub(bookname)
	output.SetTitle(bookname)
	output.SetDescription(bookname)
	output.SetAuthor("unknow")
	i := 1
	for el := currentResult.Front(); el != nil; el = el.Next() {
		art, _ := el.Value.(*be.ResultItem)
		body := "<h2>" + art.Title + "</h2><p>" + strings.Join(strings.Split(art.Content, "\n"), "</p><p>") + "</p>"
		output.AddSection(body, art.Title, fmt.Sprintf("%06d.xhtml", i+1), "")
		i += 1
	}
	fo, err := os.Create(filepath)
	if err != nil {
		a.Dispatch("debug_event", err.Error())
	}
	output.WriteTo(fo)
	fo.Close()
	return nil
}

// 杀死所有chrome进程
func (a *App) KillAllChrome() string {
	killChrome()
	return "ok"
}

// 杀死chrome进程
func killChrome() {
	// 根据操作系统选择不同的命令
	var cmd *exec.Cmd
	qu.Debug("电脑系统：", sysruntim.GOOS)
	if sysruntim.GOOS == "windows" {
		// 在Windows上使用taskkill命令
		cmd = exec.Command("taskkill", "/F", "/IM", "chrome.exe")
	} else {
		// 在类Unix系统上使用pkill命令
		cmd = exec.Command("pkill", "-f", "chrome")
	}
	// 执行命令
	err := cmd.Run()
	if err != nil {
		log.Println("Error killing process:", err)
		return
	}
	log.Println("Chrome process killed successfully.")
}