[hentaicosplays] update domains (#6578)

inherit from BaseExtractor to make differentiating between sites easier

docs/supportedsites.md

@@ -325,24 +325,12 @@ Consider all listed sites to potentially be NSFW.
     <td>Archive, Individual Posts, Home Feed, Search Results</td>
     <td></td>
 </tr>
-<tr>
-    <td>Hentai Cosplay</td>
-    <td>https://hentai-cosplay-xxx.com/</td>
-    <td>Galleries</td>
-    <td></td>
-</tr>
 <tr>
     <td>Hentai Foundry</td>
     <td>https://www.hentai-foundry.com/</td>
     <td>Favorites, individual Images, Pictures, Popular Images, Recent Images, Scraps, Stories, Tag Searches, User Profiles</td>
     <td></td>
 </tr>
-<tr>
-    <td>Hentai Image</td>
-    <td>https://hentai-img.com/</td>
-    <td>Galleries</td>
-    <td></td>
-</tr>
 <tr>
     <td>Hentai2Read</td>
     <td>https://hentai2read.com/</td>
@@ -751,12 +739,6 @@ Consider all listed sites to potentially be NSFW.
     <td>Posts Images, Search Results, User Profiles</td>
     <td></td>
 </tr>
-<tr>
-    <td>Porn Image</td>
-    <td>https://porn-images-xxx.com/</td>
-    <td>Galleries</td>
-    <td></td>
-</tr>
 <tr>
     <td>Pornhub</td>
     <td>https://www.pornhub.com/</td>
@@ -1268,6 +1250,28 @@ Consider all listed sites to potentially be NSFW.
     <td></td>
 </tr>
 
+<tr>
+    <td colspan="4"><strong>Hentai Cosplay Instances</strong></td>
+</tr>
+<tr>
+    <td>Hentai Cosplay</td>
+    <td>https://hentai-cosplay-xxx.com/</td>
+    <td>Galleries</td>
+    <td></td>
+</tr>
+<tr>
+    <td>Hentai Image</td>
+    <td>https://hentai-img-xxx.com/</td>
+    <td>Galleries</td>
+    <td></td>
+</tr>
+<tr>
+    <td>Porn Image</td>
+    <td>https://porn-image.com/</td>
+    <td>Galleries</td>
+    <td></td>
+</tr>
+
 <tr>
     <td colspan="4"><strong>jschan Imageboards</strong></td>
 </tr>

gallery_dl/extractor/hentaicosplays.py

@@ -5,31 +5,46 @@
 # published by the Free Software Foundation.
 
 """Extractors for https://hentai-cosplay-xxx.com/
-(also works for hentai-img.com and porn-images-xxx.com)"""
+(also works for hentai-img-xxx.com and porn-image.com)"""
 
-from .common import GalleryExtractor
+from .common import BaseExtractor, GalleryExtractor
 from .. import text
 
 
-class HentaicosplaysGalleryExtractor(GalleryExtractor):
+class HentaicosplaysExtractor(BaseExtractor):
+    basecategory = "hentaicosplays"
+
+
+BASE_PATTERN = HentaicosplaysExtractor.update({
+    "hentaicosplay": {
+        "root": "https://hentai-cosplay-xxx.com",
+        "pattern": r"(?:\w\w\.)?hentai-cosplays?(?:-xxx)?\.com",
+    },
+    "hentaiimg": {
+        "root": "https://hentai-img-xxx.com",
+        "pattern": r"(?:\w\w\.)?hentai-img(?:-xxx)?\.com",
+    },
+    "pornimage": {
+        "root": "https://porn-image.com",
+        "pattern": r"(?:\w\w\.)?porn-images?(?:-xxx)?\.com",
+    },
+})
+
+
+class HentaicosplaysGalleryExtractor(
+        HentaicosplaysExtractor, GalleryExtractor):
     """Extractor for image galleries from
-    hentai-cosplay-xxx.com, hentai-img.com, and porn-images-xxx.com"""
+    hentai-cosplay-xxx.com, hentai-img-xxx.com, and porn-image.com"""
-    category = "hentaicosplays"
     directory_fmt = ("{site}", "{title}")
     filename_fmt = "{filename}.{extension}"
     archive_fmt = "{title}_{filename}"
-    pattern = r"((?:https?://)?(?:\w{2}\.)?" \
+    pattern = BASE_PATTERN + r"/(?:image|story)/([\w-]+)"
-              r"(hentai-cosplay(?:s|-xxx)|hentai-img|porn-images-xxx)\.com)/" \
-              r"(?:image|story)/([\w-]+)"
     example = "https://hentai-cosplay-xxx.com/image/TITLE/"
 
     def __init__(self, match):
-        root, self.site, self.slug = match.groups()
+        BaseExtractor.__init__(self, match)
-        self.root = text.ensure_http_scheme(root)
+        self.slug = self.groups[-1]
-        if self.root == "https://hentai-cosplays.com":
+        self.gallery_url = "{}/story/{}/".format(self.root, self.slug)
-            self.root = "https://hentai-cosplay-xxx.com"
-        url = "{}/story/{}/".format(self.root, self.slug)
-        GalleryExtractor.__init__(self, match, url)
 
     def _init(self):
         self.session.headers["Referer"] = self.gallery_url
@@ -39,7 +54,7 @@ class HentaicosplaysGalleryExtractor(GalleryExtractor):
         return {
             "title": text.unescape(title.rpartition(" Story Viewer - ")[0]),
             "slug" : self.slug,
-            "site" : self.site,
+            "site" : self.root.partition("://")[2].rpartition(".")[0],
         }
 
     def images(self, page):

scripts/supportedsites.py

@@ -59,7 +59,7 @@ CATEGORY_MAP = {
     "hatenablog"     : "HatenaBlog",
     "hbrowse"        : "HBrowse",
     "hentai2read"    : "Hentai2Read",
-    "hentaicosplays" : "Hentai Cosplay",
+    "hentaicosplay"  : "Hentai Cosplay",
     "hentaifoundry"  : "Hentai Foundry",
     "hentaifox"      : "HentaiFox",
     "hentaihand"     : "HentaiHand",
@@ -112,7 +112,7 @@ CATEGORY_MAP = {
     "photovogue"     : "PhotoVogue",
     "pidgiwiki"      : "PidgiWiki",
     "pixeldrain"     : "pixeldrain",
-    "pornimagesxxx"  : "Porn Image",
+    "pornimage"      : "Porn Image",
     "pornpics"       : "PornPics.com",
     "pornreactor"    : "PornReactor",
     "readcomiconline": "Read Comic Online",
@@ -366,6 +366,7 @@ BASE_MAP = {
     "foolslide"   : "FoOlSlide Instances",
     "gelbooru_v01": "Gelbooru Beta 0.1.11",
     "gelbooru_v02": "Gelbooru Beta 0.2",
+    "hentaicosplays": "Hentai Cosplay Instances",
     "jschan"      : "jschan Imageboards",
     "lolisafe"    : "lolisafe and chibisafe",
     "lynxchan"    : "LynxChan Imageboards",
@@ -561,13 +562,6 @@ def build_extractor_list():
     default["coomerparty"] = default["kemonoparty"]
     domains["coomerparty"] = domains["kemonoparty"].replace("kemono", "coomer")
 
-    # add hentai-cosplays sister sites (hentai-img, porn-images-xxx)
-    default["hentaiimg"] = default["hentaicosplays"]
-    domains["hentaiimg"] = "https://hentai-img.com/"
-
-    default["pornimagesxxx"] = default["hentaicosplays"]
-    domains["pornimagesxxx"] = "https://porn-images-xxx.com/"
-
     # add manga4life.com
     default["mangalife"] = default["mangasee"]
     domains["mangalife"] = "https://manga4life.com/"

test/results/hentaicosplays.py

@@ -10,7 +10,7 @@ from gallery_dl.extractor import hentaicosplays
 __tests__ = (
 {
     "#url"     : "https://hentai-cosplay-xxx.com/image/---devilism--tide-kurihara-/",
-    "#category": ("", "hentaicosplays", "gallery"),
+    "#category": ("hentaicosplays", "hentaicosplay", "gallery"),
     "#class"   : hentaicosplays.HentaicosplaysGalleryExtractor,
     "#pattern" : r"https://static\d?\.hentai-cosplay-xxx\.com/upload/\d+/\d+/\d+/\d+\.jpg$",
 
@@ -22,36 +22,48 @@ __tests__ = (
 
 {
     "#url"     : "https://hentai-cosplays.com/image/---devilism--tide-kurihara-/",
-    "#category": ("", "hentaicosplays", "gallery"),
+    "#category": ("hentaicosplays", "hentaicosplay", "gallery"),
     "#class"   : hentaicosplays.HentaicosplaysGalleryExtractor,
     "#pattern" : r"https://static\d?\.hentai-cosplay-xxx\.com/upload/\d+/\d+/\d+/\d+\.jpg$",
 
     "count": 18,
-    "site" : "hentai-cosplays",
+    "site" : "hentai-cosplay-xxx",
     "slug" : "---devilism--tide-kurihara-",
     "title": "艦 こ れ-devilism の tide Kurihara 憂",
 },
 
 {
-    "#url"     : "https://fr.porn-images-xxx.com/image/enako-enako-24/",
+    "#url"     : "https://fr.porn-image.com/image/enako-enako-24/",
-    "#category": ("", "hentaicosplays", "gallery"),
+    "#category": ("hentaicosplays", "pornimage", "gallery"),
     "#class"   : hentaicosplays.HentaicosplaysGalleryExtractor,
-    "#pattern" : r"https://static\d?.porn-images-xxx.com/upload/\d+/\d+/\d+/\d+.jpg$",
+    "#pattern" : r"https://static\d?.porn-image.com/upload/\d+/\d+/\d+/\d+.jpg$",
 
     "count": 11,
-    "site" : "porn-images-xxx",
+    "site" : "porn-image",
+    "title": str,
+},
+
+{
+    "#url"     : "https://fr.porn-images-xxx.com/image/enako-enako-24/",
+    "#category": ("hentaicosplays", "pornimage", "gallery"),
+    "#class"   : hentaicosplays.HentaicosplaysGalleryExtractor,
+},
+
+{
+    "#url"     : "https://ja.hentai-img-xxx.com/image/hollow-cora-502/",
+    "#category": ("hentaicosplays", "hentaiimg", "gallery"),
+    "#class"   : hentaicosplays.HentaicosplaysGalleryExtractor,
+    "#pattern" : r"https://static\d?.hentai-img-xxx.com/upload/\d+/\d+/\d+/\d+.jpg$",
+
+    "count": 2,
+    "site" : "hentai-img-xxx",
     "title": str,
 },
 
 {
     "#url"     : "https://ja.hentai-img.com/image/hollow-cora-502/",
-    "#category": ("", "hentaicosplays", "gallery"),
+    "#category": ("hentaicosplays", "hentaiimg", "gallery"),
     "#class"   : hentaicosplays.HentaicosplaysGalleryExtractor,
-    "#pattern" : r"https://static\d?.hentai-img.com/upload/\d+/\d+/\d+/\d+.jpg$",
-
-    "count": 2,
-    "site" : "hentai-img",
-    "title": str,
 },
 
 )