feat: unified the lanzou hosename and remove download client.

internal/driver/aliyun/common.go

@@ -17,7 +17,6 @@ var (
 
 type Aliyun struct {
 	client       *client.Client
-	downloader   *client.Client
 	tokenCache   *TokenResp
 	refreshToken string
 }
@@ -42,19 +41,8 @@ func New(c *client.Config, refreshToken string) (*Aliyun, error) {
 	// Set extra middleware for cleaning up the header.
 	cl.SetPreRequestHook(removeContentType)
 
-	c2, err := client.New(&client.Config{
-		HTTPS:      true,
-		UserAgent:  c.UserAgent,
-		Proxy:      c.Proxy,
-		ConfigRoot: c.ConfigRoot,
-	})
-	if err != nil {
-		return nil, err
-	}
-
 	return &Aliyun{
 		client:       cl,
-		downloader:   c2,
 		refreshToken: refreshToken,
 	}, nil
 }

internal/driver/aliyun/share.go

@@ -1,6 +1,10 @@
 package aliyun
 
-import "io"
+import (
+	"io"
+
+	"github.com/bookstairs/bookhunter/internal/log"
+)
 
 // AnonymousShare will try to access the share without the user information.
 func (ali *Aliyun) AnonymousShare(shareID string) (*ShareInfoResp, error) {
@@ -140,9 +144,12 @@ func (ali *Aliyun) DownloadURL(shareToken string, shareID string, fileID string)
 }
 
 func (ali *Aliyun) DownloadFile(downloadURL string) (io.ReadCloser, int64, error) {
-	resp, err := ali.downloader.R().
+	log.Debugf("Start to download file from aliyun drive: %s", downloadURL)
+
+	resp, err := ali.client.R().
 		SetDoNotParseResponse(true).
 		Get(downloadURL)
 	response := resp.RawResponse
+
 	return response.Body, response.ContentLength, err
 }

internal/driver/lanzou/common.go

@@ -0,0 +1,49 @@
+package lanzou
+
+import (
+	"io"
+
+	"github.com/bookstairs/bookhunter/internal/client"
+	"github.com/bookstairs/bookhunter/internal/log"
+)
+
+type Drive struct {
+	client *client.Client
+}
+
+// hostname is used to return a valid lanzou host.
+func hostname() string {
+	// We may add a roundrobin list in the future.
+	return "lanzoux.com"
+}
+
+func NewDrive(config *client.Config) (*Drive, error) {
+	cl, err := client.New(&client.Config{
+		HTTPS:      true,
+		Host:       hostname(),
+		UserAgent:  config.UserAgent,
+		Proxy:      config.Proxy,
+		ConfigRoot: config.ConfigRoot,
+	})
+
+	if err != nil {
+		return nil, err
+	}
+
+	cl.Client.
+		SetHeader("Accept-Language", "zh-CN,zh;q=0.9").
+		SetHeader("Referer", cl.BaseURL)
+
+	return &Drive{client: cl}, nil
+}
+
+func (l Drive) DownloadFile(downloadURL string) (io.ReadCloser, int64, error) {
+	log.Debugf("Start to download file from aliyun drive: %s", downloadURL)
+
+	resp, err := l.client.R().
+		SetDoNotParseResponse(true).
+		Get(downloadURL)
+	response := resp.RawResponse
+
+	return response.Body, response.ContentLength, err
+}

internal/driver/lanzou/response.go

@@ -0,0 +1,59 @@
+package lanzou
+
+import "regexp"
+
+var (
+	lxReg  = regexp.MustCompile(`'lx':(\d+),`)
+	fidReg = regexp.MustCompile(`'fid':(\d+),`)
+	uidReg = regexp.MustCompile(`'uid':'(\d+)',`)
+	repReg = regexp.MustCompile(`'rep':'(\d+)',`)
+	upReg  = regexp.MustCompile(`'up':(\d+),`)
+	lsReg  = regexp.MustCompile(`'ls':(\d+),`)
+	tVar   = regexp.MustCompile(`'t':(\S+),`)
+	kVar   = regexp.MustCompile(`'k':(\S+),`)
+
+	dirURLRe  = regexp.MustCompile(`(?m)https?://[a-zA-Z0-9-]*?\.?lanzou[a-z]\.com/(/s/)?b[a-zA-Z0-9]{7,}/?`)
+	fileURLRe = regexp.MustCompile(`(?m)https?://[a-zA-Z0-9-]*?\.?lanzou[a-z]\.com/(/s/)?i[a-zA-Z0-9]{7,}/?`)
+
+	find1Re      = regexp.MustCompile(`(?m)url\s:\s+'(.*?)',\n\t+data\s:\s+'(.*?)'\+pwd,`)
+	find2Re      = regexp.MustCompile(`(?m)<iframe.*?src="(/fn\?\w{10,})"\s.*>`)
+	find2TitleRe = regexp.MustCompile(`(?m)<title>(.*?)\s-\s蓝奏云</title>`)
+
+	htmlNoteRe = regexp.MustCompile(`(?m)<!--.+?-->|\s+//\s*.+`)
+	jsNoteRe   = regexp.MustCompile(`(?m)(.+?[,;])\s*//.+`)
+)
+
+type (
+	Response struct {
+		Code int64        `json:"code"`
+		Data ResponseData `json:"data"`
+		Msg  string       `json:"msg"`
+	}
+
+	ResponseData struct {
+		Name string `json:"name"`
+		URL  string `json:"url"`
+	}
+
+	Dom struct {
+		Zt  int         `json:"zt"`
+		Dom string      `json:"dom"`
+		URL string      `json:"url"`
+		Inf interface{} `json:"inf"`
+	}
+
+	FileList struct {
+		Zt   int    `json:"zt"`
+		Info string `json:"info"`
+		Text []struct {
+			Icon    string `json:"icon"`
+			T       int    `json:"t"`
+			ID      string `json:"id"`
+			NameAll string `json:"name_all"`
+			Size    string `json:"size"`
+			Time    string `json:"time"`
+			Duan    string `json:"duan"`
+			PIco    int    `json:"p_ico"`
+		} `json:"text"`
+	}
+)