merge #4945: {shimmie2[ support 'rule34hentai.net' (#861, #4789)

docs/supportedsites.md

@@ -1371,6 +1371,12 @@ Consider all listed sites to potentially be NSFW.
     <td>Posts, Tag Searches</td>
     <td></td>
 </tr>
+<tr>
+    <td>Rule34Hentai</td>
+    <td>https://rule34hentai.net/</td>
+    <td>Posts, Tag Searches</td>
+    <td></td>
+</tr>
 
 <tr>
     <td colspan="4"><strong>szurubooru Instances</strong></td>

gallery_dl/extractor/shimmie2.py

@@ -35,6 +35,8 @@ def _init(self):
 
         if self.category == "giantessbooru":
             self.posts = self._posts_giantessbooru
+        elif self.category == "rule34hentai":
+            self.posts = self._posts_rule34hentai
 
     def items(self):
         data = self.metadata()
@@ -85,6 +87,10 @@ def posts(self):
         "pattern": r"booru\.cavemanon\.xyz",
         "file_url": "{0}/index.php?q=image/{2}.{4}",
     },
+    "rule34hentai": {
+        "root": "https://rule34hentai.net",
+        "pattern": r"rule34hentai\.net",
+    },
 }
 
 BASE_PATTERN = Shimmie2Extractor.update(INSTANCES) + r"/(?:index\.php\?q=/?)?"
@@ -187,6 +193,56 @@ def _posts_giantessbooru(self):
             if not extr('/{}">{}<'.format(pnum, pnum), ">"):
                 return
 
+    def _posts_rule34hentai(self):
+        pnum = text.parse_int(self.page, 1)
+        file_url_fmt = self.file_url_fmt.format
+
+        init = True
+        mime = ""
+
+        while True:
+            url = "{}/post/list/{}/{}".format(self.root, self.tags, pnum)
+            page = self.request(url).text
+            extr = text.extract_from(page)
+
+            if init:
+                init = False
+                has_mime = ("data-mime=\"" in page)
+                has_pid = ("data-post-id=\"" in page)
+
+            while True:
+                if has_mime:
+                    mime = extr("data-mime=\"", "\"")
+                if has_pid:
+                    pid = extr("data-post-id=\"", "\"")
+                else:
+                    pid = extr("href='/post/view/", "?")
+
+                if not pid:
+                    break
+
+                tags, dimensions, size, ext = extr(
+                    "title=\"", "\"").split(" // ")
+                width, _, height = dimensions.partition("x")
+                md5 = extr("/_thumbs/", "/")
+
+                yield {
+                    "file_url": file_url_fmt(
+                        self.root, md5, pid, text.quote(tags),
+                        mime.rpartition("/")[2] if mime else "jpg"),
+                    "id": pid,
+                    "md5": md5,
+                    "tags": tags,
+                    "width": width,
+                    "height": height,
+                    "size": text.parse_bytes(size[:-1]),
+                }
+
+            pnum += 1
+            if not extr(">Next<", ">"):
+                if not extr("/{}'>{}<".format(pnum, pnum), ">"):
+                    return
+
 
 class Shimmie2PostExtractor(Shimmie2Extractor):
     """Extractor for single shimmie2 posts"""
@@ -234,3 +290,25 @@ def _posts_giantessbooru(self):
             "height"  : 0,
             "size"    : 0,
         },)
+
+    def _posts_rule34hentai(self):
+        url = "{}/post/view/{}".format(self.root, self.post_id)
+        extr = text.extract_from(self.request(url).text)
+
+        post = {
+            "id"      : self.post_id,
+            "tags"    : extr(": ", "<").partition(" - ")[0].rstrip(")"),
+            "md5"     : extr("/_thumbs/", "/"),
+            "file_url": self.root + (
+                extr('id="main_image" src="', '"') or
+                extr('<source src="', '"')).lstrip("."),
+            "width"   : extr("data-width=", " ").strip("\"'"),
+            "height"  : extr("data-height=", ">").partition(
+                " ")[0].strip("\"'"),
+            "size"    : 0,
+        }
+
+        if not post["md5"]:
+            post["md5"] = text.extr(post["file_url"], "/_images/", "/")
+
+        return (post,)

scripts/supportedsites.py

@@ -106,6 +106,7 @@
     "redgifs"        : "RedGIFs",
     "rozenarcana"    : "Rozen Arcana",
     "rule34"         : "Rule 34",
+    "rule34hentai"   : "Rule34Hentai",
     "rule34us"       : "Rule 34",
     "sankaku"        : "Sankaku Channel",
     "sankakucomplex" : "Sankaku Complex",

test/results/rule34hentai.py

@@ -0,0 +1,51 @@
+# -*- coding: utf-8 -*-
+
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License version 2 as
+# published by the Free Software Foundation.
+
+from gallery_dl.extractor import shimmie2
+
+
+__tests__ = (
+{
+    "#url"     : "https://rule34hentai.net/post/list/mizuki_kotora/1",
+    "#category": ("shimmie2", "rule34hentai", "tag"),
+    "#class"   : shimmie2.Shimmie2TagExtractor,
+    "#urls"    : (
+        "https://rule34hentai.net/_images/7f3a411263d0f6de936e47ae8f9d35fb/332%20-%20Darkstalkers%20Felicia%20mizuki_kotora.jpeg",
+        "https://rule34hentai.net/_images/1a8eca7c04f8bf325bc993c5751a91c4/264%20-%20Darkstalkers%20Felicia%20mizuki_kotora.jpeg",
+        "https://rule34hentai.net/_images/09511511c4c9e9e1f9b795e059a60832/259%20-%20Darkstalkers%20Felicia%20mizuki_kotora.jpeg",
+    ),
+
+    "extension"  : "jpeg",
+    "file_url"   : r"re:https://rule34hentai.net/_images/.+\.jpeg",
+    "filename"   : r"re:\d+ - \w+",
+    "height"     : range(496, 875),
+    "id"         : range(259, 332),
+    "md5"        : r"re:^[0-9a-f]{32}$",
+    "search_tags": "mizuki_kotora",
+    "size"       : int,
+    "tags"       : str,
+    "width"      : range(500, 850),
+},
+
+{
+    "#url"     : "https://rule34hentai.net/post/view/264",
+    "#category": ("shimmie2", "rule34hentai", "post"),
+    "#class"   : shimmie2.Shimmie2PostExtractor,
+    "#urls"        : "https://rule34hentai.net/_images/1a8eca7c04f8bf325bc993c5751a91c4/264%20-%20Darkstalkers%20Felicia%20mizuki_kotora.jpg",
+    "#sha1_content": "6c23780bb78673cbff1bca9accb77ea11ec734f3",
+
+    "extension": "jpg",
+    "file_url" : "https://rule34hentai.net/_images/1a8eca7c04f8bf325bc993c5751a91c4/264%20-%20Darkstalkers%20Felicia%20mizuki_kotora.jpg",
+    "filename" : "264 - Darkstalkers Felicia mizuki_kotora",
+    "height"   : 875,
+    "id"       : 264,
+    "md5"      : "1a8eca7c04f8bf325bc993c5751a91c4",
+    "size"     : 0,
+    "tags"     : "Darkstalkers Felicia mizuki_kotora",
+    "width"    : 657,
+},
+
+)