browserSpider/program/spider/cli/data2html/main.go

package main

import (
	_ "embed"
	"encoding/json"
	"flag"
	"fmt"
	"html/template"
	"io/ioutil"
	"log"
	"os"
	"regexp"
	"strconv"
	"strings"
)

type (
	//Attach 附件
	Attach struct {
		Id         int
		Department string `json:"department"`
		Href       string `json:"attach_href"`
		Title      string `json:"info_title"`
		Raw        string
	}
	Attaches []*Attach
)

var (
	df  = flag.String("df", "./data.dat", "datafile 数据文件")
	of  = flag.String("of", "./data.html", "输出的html文件")
	tpl *template.Template
	//go:embed tpl.html
	html    string
	reg, _  = regexp.Compile("(\\d{4}年)")
	repeart = map[string]int{}
)

// readDataFile
func readDataFile() Attaches {
	ret := make(Attaches, 0, 0)
	bs, err := ioutil.ReadFile(*df)
	if err != nil {
		log.Fatal(err)
	}
	content := string(bs)
	index := 1
	for _, s := range strings.Split(content, "\n") {
		if len(s) == 0 {
			continue
		}
		var data string = s
		if strings.HasPrefix(s, ",") {
			data = s[1:]
		}
		var attach = new(Attach)
		err = json.Unmarshal([]byte(data), attach)
		if err == nil {
			ret = append(ret, attach)
			attach.Raw = s
			attach.Id = index
			index += 1
			//TODO 修改标题
			attach.Title = rename(attach.Title)
		}
	}
	return ret
}

// rename
func rename(text string) string {
	if reg.MatchString(text) {
		m := reg.FindAllString(text, -1)
		year := findMaxYear(m)
		text = reg.ReplaceAllString(text, "")
		ext := ""
		if strings.Contains(text, ".") {
			tmp := strings.Split(text, ".")
			ext = strings.ToLower(tmp[len(tmp)-1])
			if ext == "pdf" || ext == "doc" || ext == "docx" || ext == "xls" || ext == "xlsx" || ext == "zip" || ext == "rar" {
				text = text[:len(text)-4]
			}
		}
		key := fmt.Sprintf("%d_%s", year, text)
		if v, ok := repeart[key]; ok {
			repeart[key] = v + 1
			return fmt.Sprintf("%s_%d.%s", key, v+1, ext)
		} else {
			repeart[key] = 1
			return fmt.Sprintf("%s_%d.%s", key, 1, ext)
		}
	}
	return text
}

// findMaxYear
func findMaxYear(src []string) int {
	year := 2022
	for _, v := range src {
		y, err := strconv.Atoi(v[:4])
		if err == nil && y > year {
			year = y
		}
	}
	return year
}

// init
func init() {
	flag.Parse()
	var err error
	tpl, err = template.New("tpl").Parse(html)
	if err != nil {
		log.Fatal(err)
	}
}

// main
func main() {
	rs := readDataFile()
	fo, err := os.Create(*of)
	if err != nil {
		log.Fatal(err)
	}
	err = tpl.Execute(fo, rs)
	if err != nil {
		log.Fatal(err)
	}
	fo.Close()
}