[bellazon] update (#8247)

- include 'filename' in default filename_fmt and archive_fmt as 'id'
  alone is not guaranteed to be unique, even in the same post
  https://www.bellazon.com/main/topic/3556-bipasha-basu
  /page/2/#findComment-2536060
- support 'inline' files
- ignore '/profile/' links
- do not increment 'num' on ignored files

gallery_dl/extractor/bellazon.py

@@ -20,13 +20,16 @@ class BellazonExtractor(Extractor):
     root = "https://www.bellazon.com/main"
     directory_fmt = ("{category}", "{thread[section]}",
                      "{thread[title]} ({thread[id]})")
-    filename_fmt = "{post[id]}_{num:>02}_{id}.{extension}"
+    filename_fmt = "{post[id]}_{num:>02}_{id}_{filename}.{extension}"
-    archive_fmt = "{post[id]}/{filename}"
+    archive_fmt = "{post[id]}/{id}_{filename}"
 
     def items(self):
         native = (f"{self.root}/", f"{self.root[6:]}/")
         extract_urls = text.re(
-            r'(?s)<((?:video .*?<source src|a [^>]*?href)="([^"]+).*?)</a>'
+            r'(?s)<('
+            r'(?:video .*?<source src|a [^>]*?href)="([^"]+).*?</a>'
+            r'|img [^>]*?src="([^"]+)"[^>]*>'
+            r')'
         ).findall
 
         if self.config("quoted", False):
@@ -44,9 +47,14 @@ class BellazonExtractor(Extractor):
             post["count"] = data["count"] = len(urls)
 
             yield Message.Directory, data
-            for data["num"], (info, url) in enumerate(urls, 1):
+            data["num"] = 0
-                url = text.unescape(url)
+            for info, url, url_img in urls:
+                url = text.unescape(url or url_img)
+
                 if url.startswith(native):
+                    if "/uploads/emoticons/" in url or "/profile/" in url:
+                        continue
+                    data["num"] += 1
                     if not (alt := text.extr(info, ' alt="', '"')) or (
                             alt.startswith("post-") and "_thumb." in alt):
                         name = url
@@ -60,13 +68,13 @@ class BellazonExtractor(Extractor):
                     elif "/core/interface/file/attachment.php" in url:
                         if not dc["id"]:
                             dc["id"] = url.rpartition("?id=")[2]
-                        if (pos := info.find(">")) >= 0 and \
+                        if name := text.extr(info, ">", "<").strip():
-                                (name := info[pos+1:].strip()):
                             text.nameext_from_url(name, dc)
 
                     if url[0] == "/":
                         url = f"https:{url}"
                     yield Message.Url, url, dc
+
                 else:
                     yield Message.Queue, url, data
 

test/results/bellazon.py

@@ -134,7 +134,7 @@ __tests__ = (
     "extension": "mp4",
     "filename" : r"re:^\d+$",
     "id"       : r"re:6361\d\d\d",
-    "num"      : range(3, 12),
+    "num"      : range(2, 11),
     "post"     : {
         "author_id"  : "101807",
         "author_slug": "rogerdanish",
@@ -190,6 +190,20 @@ __tests__ = (
     "id"       : "10919171",
 },
 
+{
+    "#url"     : "https://www.bellazon.com/main/topic/66334-charly-jordan/page/3/#findComment-4602714",
+    "#comment" : "'/profile/' link",
+    "#class"   : bellazon.BellazonPostExtractor,
+    "#count"   : 0,
+},
+
+{
+    "#url"     : "https://www.bellazon.com/main/topic/66334-charly-jordan/page/3/#findComment-4603172",
+    "#comment" : "'inline' image",
+    "#class"   : bellazon.BellazonPostExtractor,
+    "#results" : "https://www.bellazon.com/main/uploads/monthly_2018_04/30602369_1891291154222843_1650952189830496256_n.jpg.33e6ab78dd0e8723f790ad4f58f3761a.jpg",
+},
+
 {
     "#url"     : "https://www.bellazon.com/main/topic/57872-millie-brady/",
     "#class"   : bellazon.BellazonThreadExtractor,
@@ -244,7 +258,7 @@ __tests__ = (
     "#url"     : "https://www.bellazon.com/main/topic/1774-zhang-ziyi/",
     "#class"   : bellazon.BellazonThreadExtractor,
     "#range"   : "1-5",
-    "#options" : {"prder-posts": "asc"},
+    "#options" : {"order-posts": "asc"},
     "#results" : (
         "http://img292.echo.cx/my.php?image=4moon011rk.jpg",
         "http://img294.echo.cx/my.php?image=heroclip3jb.jpg",