[extractor:ehentai] updated data extraction

gan-of-culture · Dec 13, 2023 · 65c1fce · 65c1fce
1 parent 62d4f75
commit 65c1fce
Show file tree

Hide file tree

Showing 2 changed files with 5 additions and 9 deletions.
diff --git a/extractors/ehentai/ehentai.go b/extractors/ehentai/ehentai.go
@@ -16,7 +16,7 @@ const site = "https://e-hentai.org/"
 
 var reNumbOfPages = regexp.MustCompile(`([0-9]+) pages`)
 var reIMGURLs = regexp.MustCompile(`https://e-hentai.org/s[^"]+-[0-9]+`)
-var reFileInfo = regexp.MustCompile(`<div>[^.]+\.([^::]+):: ([^::]+) :: ([^.]+.[0-9]+) ([A-Za-z]{2,3})`)
+var reFileInfo = regexp.MustCompile(`<div>[^.]+\.(\w+)\s*:: ([^::]+) :: ([^.]+.[0-9]+) ([A-Za-z]{2,3})`)
 var reSourceURL = regexp.MustCompile(`<img id="img" src="([^"]+)`)
 
 type extractor struct{}
@@ -69,10 +69,6 @@ func parseURL(URL string) []string {
 }
 
 func extractData(URL string) ([]*static.Data, error) {
-	if !strings.Contains(URL, "?nw=session") {
-		URL = URL + "?nw=session"
-	}
-
 	htmlString, err := request.Get(URL)
 	if err != nil {
 		return nil, err

diff --git a/extractors/htdoujin/htdoujin_test.go b/extractors/htdoujin/htdoujin_test.go
@@ -14,7 +14,7 @@ func TestParseURL(t *testing.T) {
 		URL  string
 		Want int
 	}{
-		/*{
+		{
 			Name: "Single Gallery HentaiEra",
 			URL:  "https://hentaiera.com/gallery/150354/",
 			Want: 1,
@@ -26,7 +26,7 @@ func TestParseURL(t *testing.T) {
 			Name: "Tag HentaiEra",
 			URL:  "https://hentaiera.com/tag/ahegao/",
 			Want: 25,
-		},*/{
+		}, {
 			Name: "Single Gallery HentaiEnvy",
 			URL:  "https://hentaienvy.com/gallery/808735/",
 			Want: 1,
@@ -91,15 +91,15 @@ func TestExtract(t *testing.T) {
 		Name string
 		Args test.Args
 	}{
-		/*{
+		{
 			Name: "Single Gallery HentaiEra",
 			Args: test.Args{
 				URL:     "https://hentaiera.com/gallery/610929/",
 				Title:   "Senran Princess G",
 				Quality: "",
 				Size:    0,
 			},
-		},*/
+		},
 		{
 			Name: "Single Gallery HentaiEnvy",
 			Args: test.Args{