github/gallery-dl
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[bellazon] add initial support (#7480)

Browse Source
This commit is contained in:
Mike Fährmann
2025-09-12 17:21:34 +02:00
parent 9035b92230
commit 88bfc0991c
4 changed files with 385 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
docs/supportedsites.md
View File

@@ -133,6 +133,12 @@ Consider all listed sites to potentially be NSFW.
<td>Collections, Galleries, User Profiles</td>
<td></td>
</tr>
<tr id="bellazon" title="bellazon">
<td>Bellazon</td>
<td>https://www.bellazon.com/</td>
<td>Forums, Posts, Threads</td>
<td></td>
</tr>
<tr id="bilibili" title="bilibili">
<td>Bilibili</td>
<td>https://www.bilibili.com/</td>

1
gallery_dl/extractor/__init__.py
View File

@@ -31,6 +31,7 @@ modules = [
"batoto",
"bbc",
"behance",
"bellazon",
"bilibili",
"blogger",
"bluesky",

165
gallery_dl/extractor/bellazon.py Normal file
View File

@@ -0,0 +1,165 @@
# -*- coding: utf-8 -*-
# Copyright 2025 Mike Fährmann
#
# 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.
"""Extractors for https://www.bellazon.com/"""
from .common import Extractor, Message
from .. import text, exception
BASE_PATTERN = r"(?:https?://)?(?:www\.)?bellazon\.com/main"
class BellazonExtractor(Extractor):
"""Base class for bellazon extractors"""
category = "bellazon"
root = "https://www.bellazon.com/main"
directory_fmt = ("{category}", "{thread[section]}",
"{thread[title]} ({thread[id]})")
filename_fmt = "{post[id]}_{num:>02}_{id}.{extension}"
archive_fmt = "{post[id]}/{filename}"
def items(self):
extract_urls = text.re(r'<a ([^>]*?href="([^"]+)".*?)</a>').findall
native = f"{self.root}/"
for post in self.posts():
urls = extract_urls(post["content"])
data = {"post": post}
post["count"] = data["count"] = len(urls)
yield Message.Directory, data
for data["num"], (info, url) in enumerate(urls, 1):
url = text.unescape(url)
if url.startswith(native):
if not (alt := text.extr(info, ' alt="', '"')) or (
alt.startswith("post-") and "_thumb." in alt):
name = url
else:
name = text.unescape(alt)
dc = text.nameext_from_url(name, data.copy())
dc["id"] = text.extr(info, 'data-fileid="', '"')
if ext := text.extr(info, 'data-fileext="', '"'):
dc["extension"] = ext
yield Message.Url, url, dc
else:
yield Message.Queue, url, data
def _pagination(self, base, pnum=None):
base = f"{self.root}{base}"
if pnum is None:
url = f"{base}/"
pnum = 1
else:
url = f"{base}/page/{pnum}/"
pnum = None
while True:
page = self.request(url).text
yield page
if pnum is None or ' rel="next" ' not in page or text.extr(
page, " rel=\"next\" data-page='", "'") == str(pnum):
return
pnum += 1
url = f"{base}/page/{pnum}/"
def _parse_thread(self, page):
schema = self._extract_jsonld(page)
author = schema["author"]
stats = schema["interactionStatistic"]
url_t = schema["url"]
url_a = author["url"]
path = text.split_html(text.extr(
page, '<nav class="ipsBreadcrumb', "</nav>"))[2:-1]
thread = {
"url" : url_t,
"path" : path,
"title": schema["headline"],
"views": stats[0]["userInteractionCount"],
"posts": stats[1]["userInteractionCount"],
"date" : text.parse_datetime(schema["datePublished"]),
"date_updated": text.parse_datetime(schema["dateModified"]),
"description" : text.unescape(schema["text"]),
"section" : path[-2],
"author" : author["name"],
"author_url" : url_a,
}
thread["id"], _, thread["slug"] = \
url_t.rsplit("/", 2)[1].partition("-")
thread["author_id"], _, thread["author_slug"] = \
url_a.rsplit("/", 2)[1].partition("-")
return thread
def _parse_post(self, html):
extr = text.extract_from(html)
post = {
"id": extr('id="elComment_', '"'),
"author_url": extr(" href='", "'"),
"date": text.parse_datetime(extr("datetime='", "'")),
"content": extr("<!-- Post content -->", "\n\t\t</div>"),
}
if (pos := post["content"].find(">")) >= 0:
post["content"] = post["content"][pos+1:].strip()
post["author_id"], _, post["author_slug"] = \
post["author_url"].rsplit("/", 2)[1].partition("-")
return post
class BellazonPostExtractor(BellazonExtractor):
subcategory = "post"
pattern = (rf"{BASE_PATTERN}(/topic/\d+-[\w-]+(?:/page/\d+)?)"
rf"/?#findComment-(\d+)")
example = "https://www.bellazon.com/main/topic/123-SLUG/#findComment-12345"
def posts(self):
path, post_id = self.groups
page = self.request(f"{self.root}{path}").text
pos = page.find(f'id="elComment_{post_id}')
if pos < 0:
raise exception.NotFoundError("post")
html = text.extract(page, "<article ", "</article>", pos-100)[0]
self.kwdict["thread"] = self._parse_thread(page)
return (self._parse_post(html),)
class BellazonThreadExtractor(BellazonExtractor):
subcategory = "thread"
pattern = rf"{BASE_PATTERN}(/topic/\d+-[\w-]+)(?:/page/(\d+))?"
example = "https://www.bellazon.com/main/topic/123-SLUG/"
def posts(self):
for page in self._pagination(*self.groups):
if "thread" not in self.kwdict:
self.kwdict["thread"] = self._parse_thread(page)
for html in text.extract_iter(page, "<article ", "</article>"):
yield self._parse_post(html)
class BellazonForumExtractor(BellazonExtractor):
subcategory = "forum"
pattern = rf"{BASE_PATTERN}(/forum/\d+-[\w-]+)(?:/page/(\d+))?"
example = "https://www.bellazon.com/main/forum/123-SLUG/"
def items(self):
data = {"_extractor": BellazonThreadExtractor}
for page in self._pagination(*self.groups):
for row in text.extract_iter(
page, '<li data-ips-hook="topicRow"', "</"):
yield Message.Queue, text.extr(row, 'href="', '"'), data

213
test/results/bellazon.py Normal file
View File

@@ -0,0 +1,213 @@
# -*- 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 bellazon
__tests__ = (
{
"#url" : "https://www.bellazon.com/main/topic/57872-millie-brady/#findComment-4351049",
"#class" : bellazon.BellazonPostExtractor,
"#results" : (
"https://www.bellazon.com/main/uploads/monthly_2017_06/595482b77fd89_millieb280617BZNImage101.jpg.10b91b9141b374e657a1a4c3d0c96b64.jpg",
"https://www.bellazon.com/main/uploads/monthly_2017_06/595482c3f2fa0_millieb280617BZNImage102.jpg.1b706048fc525151775cf4b7c734b283.jpg",
"https://www.bellazon.com/main/uploads/monthly_2017_06/595482cdc66ad_millieb280617BZNImage103.jpg.6fa2226a314d0f0e9f9426e7f90f4808.jpg",
"https://www.bellazon.com/main/uploads/monthly_2017_06/595482dac786c_millieb280617BZNImage104.jpg.e579be6b585cef90b965d4d09969a66a.jpg",
"https://www.bellazon.com/main/uploads/monthly_2017_06/595482e772acd_millieb280617BZNImage105.jpg.428df8a841957b48452a6a6ab64ddacb.jpg",
),
"id" : r"re:55\d+",
"filename" : str,
"extension": "jpg",
"count" : 5,
"num" : range(1, 5),
"post" : {
"author_id" : "72476",
"author_slug": "shepherd",
"author_url" : "https://www.bellazon.com/main/profile/72476-shepherd/",
"count" : 5,
"date" : "dt:2017-06-29 04:32:43",
"id" : "4351049",
"content" : """\
<p>
\tSerpentine Galleries Summer Party, London, Jun 28 '17
</p>
<p>
\t 
</p>
<p>
\t<a class="ipsAttachLink ipsAttachLink_image" href="https://www.bellazon.com/main/uploads/monthly_2017_06/595482b77fd89_millieb280617BZNImage101.jpg.10b91b9141b374e657a1a4c3d0c96b64.jpg" data-fileid="5550073" rel=""><img alt="millieb280617BZNImage101.jpg" class="ipsImage ipsImage_thumbnailed" data-fileid="5550073" src="https://www.bellazon.com/main/uploads/monthly_2017_06/595482b7c4730_millieb280617BZNImage101.thumb.jpg.5b5240deead09ec5546a6bbf68aff724.jpg" data-ratio="66.56" loading="lazy"></a> <a class="ipsAttachLink ipsAttachLink_image" href="https://www.bellazon.com/main/uploads/monthly_2017_06/595482c3f2fa0_millieb280617BZNImage102.jpg.1b706048fc525151775cf4b7c734b283.jpg" data-fileid="5550074" rel=""><img alt="millieb280617BZNImage102.jpg" class="ipsImage ipsImage_thumbnailed" data-fileid="5550074" src="https://www.bellazon.com/main/uploads/monthly_2017_06/595482c4529af_millieb280617BZNImage102.thumb.jpg.1b9f9ec5f002eaaaa80a174d1a7853d0.jpg" data-ratio="150" loading="lazy"></a> <a class="ipsAttachLink ipsAttachLink_image" href="https://www.bellazon.com/main/uploads/monthly_2017_06/595482cdc66ad_millieb280617BZNImage103.jpg.6fa2226a314d0f0e9f9426e7f90f4808.jpg" data-fileid="5550075" rel=""><img alt="millieb280617BZNImage103.jpg" class="ipsImage ipsImage_thumbnailed" data-fileid="5550075" src="https://www.bellazon.com/main/uploads/monthly_2017_06/595482ce268f7_millieb280617BZNImage103.thumb.jpg.580d38335424d6fa65bd5d476625864b.jpg" data-ratio="150.23" loading="lazy"></a>
</p>
<p>
\t<a class="ipsAttachLink ipsAttachLink_image" href="https://www.bellazon.com/main/uploads/monthly_2017_06/595482dac786c_millieb280617BZNImage104.jpg.e579be6b585cef90b965d4d09969a66a.jpg" data-fileid="5550076" rel=""><img alt="millieb280617BZNImage104.jpg" class="ipsImage ipsImage_thumbnailed" data-fileid="5550076" src="https://www.bellazon.com/main/uploads/monthly_2017_06/595482db10e03_millieb280617BZNImage104.thumb.jpg.958eba72b585110a4b8c08f1efd9cfc8.jpg" title="" data-ratio="150.26" loading="lazy"></a> <a class="ipsAttachLink ipsAttachLink_image" href="https://www.bellazon.com/main/uploads/monthly_2017_06/595482e772acd_millieb280617BZNImage105.jpg.428df8a841957b48452a6a6ab64ddacb.jpg" data-fileid="5550077" rel=""><img alt="millieb280617BZNImage105.jpg" class="ipsImage ipsImage_thumbnailed" data-fileid="5550077" src="https://www.bellazon.com/main/uploads/monthly_2017_06/595482e7e6bc1_millieb280617BZNImage105.thumb.jpg.1e5ce2b85f7ceed7446d7f13caa9ce2b.jpg" data-ratio="150.22" loading="lazy"></a>
</p>\
""",
},
"thread" : {
"author" : "Shepherd",
"author_id" : "72476",
"author_slug" : "shepherd",
"author_url" : "https://www.bellazon.com/main/profile/72476-shepherd/",
"date" : "dt:2015-06-20 21:34:31",
"date_updated": "dt:2017-06-29 04:32:43",
"description" : "Previously featured in the popular TV series, Mr Selfridge, emerging British born actress Millie Brady is set for huge success. \nMillie has just been confirmed as the lead role in ‘The Clan of the Cave Bear’ which will begin filming in May 2015. The drama pilot is from Imagine TV, Allison Shearmur Productions, Fox 21 TV and Lionsgate TV. Millie is also due to appear in the eagerly awaited black comedy, 'Pride and Prejudice and Zombies', staring alongside Matt Smith, Sally Philiips, Douglas Booth, Lily james and Sam Riley. She is currently filming 'Knights of the Roundtable: King Arthur' directed by Guy Ritchie. \n  \n  \nFarfetch, Jun 2015 \nLinda Brownlee photos \n  \n \n",
"id" : "57872",
"posts" : 1,
"section" : "Actresses",
"slug" : "millie-brady",
"title" : "Millie Brady",
"url" : "https://www.bellazon.com/main/topic/57872-millie-brady/",
"views" : range(3_800, 5_000),
"path" : [
"Females",
"Actresses",
"Millie Brady",
],
},
},
{
"#url" : "https://www.bellazon.com/main/topic/3556-bipasha-basu/#findComment-2134610",
"#class" : bellazon.BellazonPostExtractor,
"#results" : "https://www.bellazon.com/main/uploads/monthly_04_2010/post-35864-1270985307.jpg",
"id" : "1002749",
"filename" : "post-35864-1270985307",
"extension": "jpg",
"count" : 1,
"num" : 1,
"post" : {
"author_id" : "35864",
"author_slug": "egluze",
"author_url" : "https://www.bellazon.com/main/profile/35864-egluze/",
"count" : 1,
"date" : "dt:2010-04-11 11:28:43",
"id" : "2134610",
"content" : """\
<p><strong>Marie Claire India April 2010</strong></p>
<p><a class="ipsAttachLink ipsAttachLink_image" href="https://www.bellazon.com/main/uploads/monthly_04_2010/post-35864-1270985307.jpg" rel="external nofollow"><img class="ipsImage ipsImage_thumbnailed" src="https://www.bellazon.com/main/uploads/monthly_04_2010/post-35864-1270985307_thumb.jpg" data-fileid="1002749" alt="post-35864-1270985307_thumb.jpg" data-ratio="133.67" loading="lazy"></a></p>\
""",
},
"thread" : {
"author" : "SaBrIaNa",
"author_id" : "1324",
"author_slug" : "sabriana",
"author_url" : "https://www.bellazon.com/main/profile/1324-sabriana/",
"date" : "dt:2005-12-26 20:31:33",
"date_updated": "dt:2017-06-17 05:19:09",
"description" : str,
"id" : "3556",
"posts" : 44,
"section" : "Actresses",
"slug" : "bipasha-basu",
"title" : "Bipasha Basu",
"url" : "https://www.bellazon.com/main/topic/3556-bipasha-basu/",
"views" : range(20_000, 50_000),
"path" : [
"Females",
"Actresses",
"Bipasha Basu",
],
},
},
{
"#url" : "https://www.bellazon.com/main/topic/57872-millie-brady/",
"#class" : bellazon.BellazonThreadExtractor,
"#pattern" : r"https://www\.bellazon\.com/main/uploads/monthly_\d+_\d+/.+\.jpg",
"#count" : 13,
"id" : r"re:\d+",
"filename" : str,
"extension": "jpg",
"count" : {5, 8},
"num" : range(1, 8),
"post" : {
"id" : {"3721257", "4351049"},
"count" : {5, 8},
"author_id": "72476",
"date" : "type:datetime",
},
"thread" : {
"id" : "57872",
"title" : "Millie Brady",
"author" : "Shepherd",
"author_id": "72476",
"date" : "dt:2015-06-20 21:34:31",
},
},
{
"#url" : "https://www.bellazon.com/main/topic/3556-bipasha-basu/",
"#class" : bellazon.BellazonThreadExtractor,
"#pattern" : r"https?://(www\.bellazon\.com/main/uploads/.+\.\w+|www\.[^.]+\.(com|ru)|img\d+.imagevenue.com|imagesion.com)",
"#count" : 247,
"count" : range(0, 30),
"num" : range(0, 30),
"post" : {
"id" : r"re:\d+",
"author_id": r"re:\d+",
"count" : range(0, 30),
"date" : "type:datetime",
},
"thread" : {
"id" : "3556",
"title" : "Bipasha Basu",
"author" : "SaBrIaNa",
"author_id" : "1324",
"date" : "dt:2005-12-26 20:31:33",
"date_updated": "dt:2017-06-17 05:19:09",
},
},
{
"#url" : "https://www.bellazon.com/main/topic/1774-zhang-ziyi/",
"#class" : bellazon.BellazonThreadExtractor,
"#range" : "1-5",
"#results" : (
"http://img292.echo.cx/my.php?image=4moon011rk.jpg",
"http://img294.echo.cx/my.php?image=heroclip3jb.jpg",
"http://img294.echo.cx/my.php?image=heroclip29ut.jpg",
"http://img294.echo.cx/my.php?image=heroclip35lp.jpg",
"http://img36.echo.cx/my.php?image=895welzz4514nv.jpg",
),
"thread": {
"author" : "Hiro",
"author_id" : "26",
"author_slug" : "hiro",
"author_url" : "https://www.bellazon.com/main/profile/26-hiro/",
"date" : "dt:2005-06-08 03:02:03",
"date_updated": "dt:2023-07-09 07:33:19",
"description" : str,
"id" : "1774",
"posts" : 480,
"section" : "Actresses",
"slug" : "zhang-ziyi",
"title" : "Zhang Ziyi",
"url" : "https://www.bellazon.com/main/topic/1774-zhang-ziyi/",
"views" : int,
"path" : [
"Females",
"Actresses",
"Zhang Ziyi",
],
},
},
{
"#url" : "https://www.bellazon.com/main/forum/3-actresses/",
"#class" : bellazon.BellazonForumExtractor,
"#pattern" : bellazon.BellazonThreadExtractor.pattern,
"#range" : "1-100",
"#count" : 100,
},
)