2021-12-20 16:53:38 +04:00
|
|
|
|
// NNM-Club torrent files mass downloader
|
2019-09-28 19:59:02 +04:00
|
|
|
|
// Created for Uploaders group
|
2021-12-20 16:53:38 +04:00
|
|
|
|
// Copyright (c) 2012-2022 Vladimir "fat0troll" Hodakov
|
2019-09-28 19:59:02 +04:00
|
|
|
|
|
2021-12-20 16:53:38 +04:00
|
|
|
|
package fetcher
|
2019-09-28 19:59:02 +04:00
|
|
|
|
|
|
|
|
|
import (
|
2021-12-20 17:45:05 +04:00
|
|
|
|
"fmt"
|
2019-09-28 19:59:02 +04:00
|
|
|
|
"io"
|
|
|
|
|
"io/ioutil"
|
|
|
|
|
"net/http"
|
2019-09-28 20:20:23 +04:00
|
|
|
|
"time"
|
2019-09-28 19:59:02 +04:00
|
|
|
|
|
|
|
|
|
"golang.org/x/text/encoding/charmap"
|
|
|
|
|
"golang.org/x/text/transform"
|
|
|
|
|
)
|
|
|
|
|
|
2021-12-20 17:45:05 +04:00
|
|
|
|
const (
|
|
|
|
|
forumPageOffset = 50
|
|
|
|
|
)
|
|
|
|
|
|
|
|
|
|
func downloadAdditionalPages(forumID int) {
|
|
|
|
|
forumPage := 2
|
|
|
|
|
|
|
|
|
|
for {
|
2019-09-28 19:59:02 +04:00
|
|
|
|
dlog.Info().Int("номер страницы", forumPage).Msg("Скачивается ещё одна страница форума")
|
2020-02-22 18:33:33 +04:00
|
|
|
|
|
2021-12-20 17:45:05 +04:00
|
|
|
|
pageURL := fmt.Sprintf(
|
|
|
|
|
"https://%s/forum/viewforum.php?f=%d&start=%d#pagestart",
|
|
|
|
|
c.Config.URL, forumID, (forumPage-1)*forumPageOffset,
|
|
|
|
|
)
|
|
|
|
|
|
|
|
|
|
if uberDebug {
|
|
|
|
|
dlog.Debug().Str("URL страницы", pageURL).Msg("Сформирован URL для скачивания")
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
pageFile, err := downloadFile(pageURL, true)
|
2019-09-28 19:59:02 +04:00
|
|
|
|
if err != nil {
|
|
|
|
|
dlog.Fatal().Err(err).Msg("Не удалось загрузить страницу форума")
|
|
|
|
|
}
|
|
|
|
|
|
2021-12-20 17:45:05 +04:00
|
|
|
|
err = setQuerier(pageFile, "forumPage", forumPage)
|
|
|
|
|
if err != nil {
|
|
|
|
|
dlog.Info().Str("причина завершения", err.Error()).
|
|
|
|
|
Msg("Завершено скачивание страниц форума")
|
|
|
|
|
// Found empty page, leaving cycle
|
|
|
|
|
break
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
forumPage++
|
2019-09-28 19:59:02 +04:00
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2019-09-28 22:51:10 +04:00
|
|
|
|
func downloadFile(url string, transformEncoding bool) (string, error) {
|
2019-09-28 19:59:02 +04:00
|
|
|
|
req, err := http.NewRequest(http.MethodGet, url, nil)
|
|
|
|
|
if err != nil {
|
|
|
|
|
return "", err
|
|
|
|
|
}
|
2020-02-22 18:33:33 +04:00
|
|
|
|
|
2019-09-28 19:59:02 +04:00
|
|
|
|
for i := range c.Cookies {
|
|
|
|
|
req.AddCookie(c.Cookies[i])
|
|
|
|
|
}
|
|
|
|
|
|
2019-09-28 20:20:23 +04:00
|
|
|
|
var resp *http.Response
|
2020-02-22 18:33:33 +04:00
|
|
|
|
var retryCount int = 0
|
|
|
|
|
|
2019-09-28 20:20:23 +04:00
|
|
|
|
for {
|
|
|
|
|
if retryCount < 5 {
|
|
|
|
|
resp, err = dclient.Do(req)
|
|
|
|
|
if err != nil {
|
|
|
|
|
if uberDebug {
|
|
|
|
|
dlog.Debug().Err(err).Int("попытка", retryCount+1).Msg("Не удалось получить данные, пытаемся ещё раз")
|
|
|
|
|
}
|
2020-02-22 18:33:33 +04:00
|
|
|
|
|
2019-09-28 20:20:23 +04:00
|
|
|
|
time.Sleep(time.Second)
|
|
|
|
|
retryCount++
|
|
|
|
|
} else {
|
|
|
|
|
break
|
|
|
|
|
}
|
|
|
|
|
} else {
|
|
|
|
|
return "", err
|
|
|
|
|
}
|
2019-09-28 19:59:02 +04:00
|
|
|
|
}
|
2020-02-22 18:33:33 +04:00
|
|
|
|
|
2019-09-28 19:59:02 +04:00
|
|
|
|
defer resp.Body.Close()
|
|
|
|
|
|
|
|
|
|
tempF, err := ioutil.TempFile("", "massdl-*")
|
|
|
|
|
if err != nil {
|
|
|
|
|
return "", err
|
|
|
|
|
}
|
|
|
|
|
defer tempF.Close()
|
|
|
|
|
|
2019-09-28 22:51:10 +04:00
|
|
|
|
if transformEncoding {
|
|
|
|
|
respInUTF8 := transform.NewReader(resp.Body, charmap.Windows1251.NewDecoder())
|
2020-02-22 18:33:33 +04:00
|
|
|
|
|
2019-09-28 22:51:10 +04:00
|
|
|
|
_, err = io.Copy(tempF, respInUTF8)
|
|
|
|
|
if err != nil {
|
|
|
|
|
return "", err
|
|
|
|
|
}
|
|
|
|
|
} else {
|
|
|
|
|
_, err = io.Copy(tempF, resp.Body)
|
|
|
|
|
if err != nil {
|
|
|
|
|
return "", err
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if uberDebug {
|
|
|
|
|
dlog.Debug().Str("расположение", tempF.Name()).Msg("Загружен файл из сети")
|
2019-09-28 19:59:02 +04:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return tempF.Name(), nil
|
|
|
|
|
}
|