[2ch] support 'org' TLD (#8629)

This commit is contained in:
Mike Fährmann
2025-11-30 16:49:23 +01:00
parent f51544618e
commit 76765340eb
3 changed files with 27 additions and 10 deletions

View File

@@ -15,7 +15,7 @@ Consider all listed sites to potentially be NSFW.
<tbody valign="top">
<tr id="2ch" title="2ch">
<td>2ch</td>
<td>https://2ch.su/</td>
<td>https://2ch.org/</td>
<td>Boards, Threads</td>
<td></td>
</tr>

View File

@@ -4,28 +4,28 @@
# it under the terms of the GNU General Public License version 2 as
# published by the Free Software Foundation.
"""Extractors for https://2ch.su/"""
"""Extractors for https://2ch.org/"""
from .common import Extractor, Message
from .. import text, util
BASE_PATTERN = r"(?:https?://)?2ch\.(su|life|hk)"
BASE_PATTERN = r"(?:https?://)?2ch\.(org|su|life|hk)"
class _2chThreadExtractor(Extractor):
"""Extractor for 2ch threads"""
category = "2ch"
subcategory = "thread"
root = "https://2ch.su"
root = "https://2ch.org"
directory_fmt = ("{category}", "{board}", "{thread} {title}")
filename_fmt = "{tim}{filename:? //}.{extension}"
archive_fmt = "{board}_{thread}_{tim}"
pattern = rf"{BASE_PATTERN}/([^/?#]+)/res/(\d+)"
example = "https://2ch.su/a/res/12345.html"
example = "https://2ch.org/a/res/12345.html"
def __init__(self, match):
tld = match[1]
self.root = f"https://2ch.{'su' if tld == 'hk' else tld}"
self.root = f"https://2ch.{'org' if tld == 'hk' else tld}"
Extractor.__init__(self, match)
def items(self):
@@ -65,9 +65,9 @@ class _2chBoardExtractor(Extractor):
"""Extractor for 2ch boards"""
category = "2ch"
subcategory = "board"
root = "https://2ch.su"
root = "https://2ch.org"
pattern = rf"{BASE_PATTERN}/([^/?#]+)/?$"
example = "https://2ch.su/a/"
example = "https://2ch.org/a/"
def __init__(self, match):
tld = match[1]

View File

@@ -13,7 +13,7 @@ __tests__ = (
"#url" : "https://2ch.hk/a/res/6202876.html",
"#category": ("", "2ch", "thread"),
"#class" : _2ch._2chThreadExtractor,
"#pattern" : r"https://2ch\.su/a/src/6202876/\d+\.\w+",
"#pattern" : r"https://2ch\.org/a/src/6202876/\d+\.\w+",
"#count" : range(450, 1000),
"banned" : 0,
@@ -53,6 +53,16 @@ __tests__ = (
"width" : int,
},
{
"#url" : "https://2ch.org/a/res/6202876.html",
"#class" : _2ch._2chThreadExtractor,
},
{
"#url" : "https://2ch.su/a/res/6202876.html",
"#class" : _2ch._2chThreadExtractor,
},
{
"#url" : "https://2ch.life/a/res/6202876.html",
"#class" : _2ch._2chThreadExtractor,
@@ -63,10 +73,17 @@ __tests__ = (
{
"#url" : "https://2ch.hk/a/res/6202876.html",
"#class" : _2ch._2chThreadExtractor,
"#pattern" : r"https://2ch\.su/a/src/6202876/\d+\.\w+",
"#pattern" : r"https://2ch\.org/a/src/6202876/\d+\.\w+",
"#count" : range(450, 1000),
},
{
"#url" : "https://2ch.org/a/",
"#class" : _2ch._2chBoardExtractor,
"#pattern" : _2ch._2chThreadExtractor.pattern,
"#count" : range(200, 400),
},
{
"#url" : "https://2ch.su/a/",
"#class" : _2ch._2chBoardExtractor,