aaaaa

haha.go

package main

import (
	"fmt"
	"io/ioutil"
	"github.com/anaskhan96/soup"
	"strings"
	"regexp"
	"os"
)

func main() {
	b, err := ioutil.ReadFile("file.html") 
	if err != nil {
		fmt.Print(err)
	}

	str := string(b)
	
	r := regexp.MustCompile(`https:[^\s"><]*\bdetail\/[0-9]*`)
	matches := r.FindAllString(str, -1)

	file, err := os.Create("links.json")
	if err != nil {
		fmt.Println(err)
		return
	}

	file.WriteString("{\n")
	file.WriteString("\t" + `"ads": [` + "\n")
	
	for index, each := range matches {

		if (index % 2) == 1 {
			fmt.Println(index -1)
			file.WriteString("\t\t{\n")
			resp, err := soup.Get(each)
			if err != nil {
				continue
			}
			doc := soup.HTMLParse(resp)
			if doc.Error != nil {
				continue
			}

			itemsTemp := doc.Find("div", "class", "detail_part3")
			if itemsTemp.Error != nil {
				fmt.Println(itemsTemp.Error)
				continue
			}

			items := doc.Find("div", "class", "detail_part3").FindAll("li")
			if len(items) < 1 {
				continue
			}
			file.WriteString("\t\t\t" + `"link": ` + `"` + each + `"` + ", \n")
			file.WriteString("\t\t\t" + `"title": ` + `"` + doc.Find("div", "class", "title").Find("h1").Text() + `"` + ", \n")
			for itemIndex, item := range items {

				hasUsefulSpan := false
				if item.Find("span", "class", "name").Text() == "گروه آگهی :: " {

					catsAndSubCats := item.Find("span", "class", "val").FindAll("a")
					file.WriteString("\t\t\t" + `"category": ` + `"` + catsAndSubCats[0].Text() + `"` + ", \n")
					if len(catsAndSubCats) > 1 {
						file.WriteString("\t\t\t" + `"subcategory": ` + `"` + catsAndSubCats[1].Text() + `"`)
					}
					hasUsefulSpan = true

				} else if item.Find("span", "class", "name").Text() == "نام و نام خانوادگی :: " {

					file.WriteString("\t\t\t" + `"name": ` + `"` + item.Find("span", "class", "val").Text() + `"`)
					hasUsefulSpan = true

				} else if item.Find("span", "class", "name").Text() == "تلفن تماس :: " {

					file.WriteString("\t\t\t" + `"phone": ` + `"` + item.Find("span", "class", "val").Text() + `"`)
					hasUsefulSpan = true

				} else if item.Find("span", "class", "name").Text() == "لینک :: " {

					file.WriteString("\t\t\t" + `"website": ` + `"` + item.Find("span", "class", "val").Find("a").Text() + `"`)
					hasUsefulSpan = true

				} else if item.Find("span", "class", "name").Text() == "موقعیت :: " {

					file.WriteString("\t\t\t" + `"location": ` + `"` + item.Find("span", "class", "val").Text() + `"`)
					hasUsefulSpan = true

				} else if item.Find("span", "class", "name").Text() == "نشانی :: " {

					file.WriteString("\t\t\t" + `"address": ` + `"` + strings.Replace(item.Find("span", "class", "val").Text(), "\n", " ", -1) + `"`)
					hasUsefulSpan = true
				}
				if (len(items) - 1) == itemIndex {
					file.WriteString("\n")
				} else {
					if hasUsefulSpan == true {
						file.WriteString(",\n")
					}
				}
			}

			if (len(matches) - 1) == index {
				file.WriteString("\t\t}\n")
			} else {
				file.WriteString("\t\t},\n")
			}

		}
	}

	file.WriteString("\t]\n")
	file.WriteString("}")

	err = file.Close()
	if err != nil {
		fmt.Println(err)
		return
	}

}