fix: 修复hcomic的漫画下载不了的问题

ystyle · Jun 30, 2020 · 41837e5 · 41837e5
1 parent 610c937
commit 41837e5
Show file tree

Hide file tree

Showing 4 changed files with 30 additions and 4 deletions.
diff --git a/model/hcomic.go b/model/hcomic.go
@@ -29,6 +29,7 @@ type HcomicSection struct {
 	Index      int    // 顺序
 	Title      string // 标题
 	Url        string // 地址
+	BackupUrl  string // 备选地址
 	ImgFile    string // 图片缓存路径
 	HtmlFile   string // html 缓存路径
 	InnerHtml  string // html 内部目录
@@ -51,16 +52,25 @@ func (hcomic *HcomicInfo) SetDefault() {
 	hcomic.ZipFile = path.Join(config.StoreDir, "hcomic", fmt.Sprintf("%s.zip", hcomic.ID))
 }
 
-func (hcomic *HcomicInfo) AddSection(title, url string) {
+func (hcomic *HcomicInfo) AddSection(title, url, backupURL string) {
 	index := len(hcomic.Sections)
 	imageFilename := path.Base(url)
 	hcomic.Sections = append(hcomic.Sections, &HcomicSection{
 		Index:      index,
 		Title:      title,
 		Url:        url,
+		BackupUrl:  backupURL,
 		ImgFile:    path.Join(hcomic.ScaledImagesDir, imageFilename),
 		HtmlFile:   path.Join(hcomic.HtmlDir, fmt.Sprintf("Page-%d.html", index)),
 		InnerHtml:  fmt.Sprintf("html/Page-%d.html", index),
 		InnerImage: fmt.Sprintf("scaled-images/%s", imageFilename),
 	})
 }
+
+func (hs *HcomicSection) ResetToBackupURL() {
+	// 重置文件名， 在预览图和高清图格式不一致时用
+	imageFilename := path.Base(hs.BackupUrl)
+	ScaledImagesDir := path.Dir(hs.ImgFile)
+	hs.ImgFile = path.Join(ScaledImagesDir, imageFilename)
+	hs.InnerImage = fmt.Sprintf("scaled-images/%s", imageFilename)
+}
diff --git a/services/hcomic.go b/services/hcomic.go
@@ -14,6 +14,7 @@ import (
 	"io/ioutil"
 	"os"
 	"path"
+	"strings"
 	"sync"
 	"time"
 )
@@ -110,7 +111,11 @@ func Submit(client *core.WsClient, message core.Message) {
 
 func download(wg *sync.WaitGroup, client *core.WsClient, section *model.HcomicSection) {
 	defer wg.Done()
-	_ = web.Download(section.Url, section.ImgFile)
+	err := web.Download(section.Url, section.ImgFile)
+	if err != nil && section.BackupUrl != "" {
+		section.ResetToBackupURL()
+		_ = web.Download(section.BackupUrl, section.ImgFile)
+	}
 	client.WsSend <- core.NewMessage("info", fmt.Sprintf("下载完成: %s", section.Url))
 }
 

diff --git a/util/hcomic/images.go b/util/hcomic/images.go
@@ -6,6 +6,7 @@ import (
 	"github.com/ystyle/kas/util/web"
 	"net/url"
 	"path"
+	"regexp"
 	"strings"
 )
 
@@ -14,6 +15,7 @@ func GetAllImages(book *model.HcomicInfo) error {
 	if err != nil {
 		return err
 	}
+	reg, _ := regexp.Compile(`this.src='(.*)';`)
 	meta := html.Find("meta[name=\"applicable-device\"]")
 	if attr, has := meta.Attr("content"); has && attr == "pc,mobile" {
 		if book.BookName == "" {
@@ -23,7 +25,13 @@ func GetAllImages(book *model.HcomicInfo) error {
 		for i := range imgs.Nodes {
 			img := imgs.Eq(i)
 			src, _ := img.Attr("data-src")
-			book.AddSection(fmt.Sprintf("#%d", i+1), GetHDImage(src))
+			jstext, _ := img.Attr("onerror")
+			group := reg.FindStringSubmatch(jstext)
+			backurl := ""
+			if len(group) > 0 {
+				backurl = group[1]
+			}
+			book.AddSection(fmt.Sprintf("#%d", i+1), GetHDImage(src), GetHDImage(backurl))
 		}
 	} else {
 		lis := html.Find(".img_list li")
@@ -35,7 +43,7 @@ func GetAllImages(book *model.HcomicInfo) error {
 			url, _ := li.Find("img").First().Attr("src")
 			title := li.Find("label").Text()
 			if url != "" {
-				book.AddSection(title, GetHDImage(url))
+				book.AddSection(title, GetHDImage(url), "")
 			}
 		}
 	}

diff --git a/util/web/http.go b/util/web/http.go
@@ -59,6 +59,9 @@ func Download(url string, dir string) error {
 	if err != nil {
 		return fmt.Errorf("目标网站无法连接: %w", err)
 	}
+	if res.StatusCode >= 400 {
+		return fmt.Errorf("目标网站不存在: %w", err)
+	}
 	f, err := os.Create(dir)
 	if err != nil {
 		return fmt.Errorf("服务器错误: 无法创建文件")