From 289d6b65d2027bf0423b3c95b00438ececc133dc Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Mike=20F=C3=A4hrmann?= Date: Mon, 16 Oct 2017 21:21:19 +0200 Subject: [PATCH] [danbooru] extend and improve URL regex - add support for danbooru mirrors: - hijiribe.donmai.us - sonohara.donmai.us - todo: actually use these domains instead of redirecting everything to danbooru itself - improve handling of query string parameters --- gallery_dl/extractor/danbooru.py | 24 +++++++++++++++--------- 1 file changed, 15 insertions(+), 9 deletions(-) diff --git a/gallery_dl/extractor/danbooru.py b/gallery_dl/extractor/danbooru.py index 9e44d789..5e5ac206 100644 --- a/gallery_dl/extractor/danbooru.py +++ b/gallery_dl/extractor/danbooru.py @@ -19,16 +19,21 @@ class DanbooruExtractor(booru.JSONBooruExtractor): class DanbooruTagExtractor(DanbooruExtractor, booru.BooruTagExtractor): """Extractor for images from danbooru based on search-tags""" - pattern = [r"(?:https?://)?danbooru\.donmai\.us/posts" - r"\?(?:utf8=%E2%9C%93&)?tags=([^&]+)"] - test = [("https://danbooru.donmai.us/posts?tags=bonocho", { - "content": "b196fb9f1668109d7774a0a82efea3ffdda07746", - })] + pattern = [r"(?:https?://)?(?:danbooru|hijiribe|sonohara)\.donmai\.us" + r"/posts\?(?:[^&#]*&)*tags=([^&#]+)"] + test = [ + ("https://danbooru.donmai.us/posts?tags=bonocho", { + "content": "b196fb9f1668109d7774a0a82efea3ffdda07746", + }), + ("https://hijiribe.donmai.us/posts?tags=bonocho", None), + ("https://sonohara.donmai.us/posts?tags=bonocho", None), + ] class DanbooruPoolExtractor(DanbooruExtractor, booru.BooruPoolExtractor): """Extractor for image-pools from danbooru""" - pattern = [r"(?:https?://)?danbooru\.donmai\.us/pools/(\d+)"] + pattern = [r"(?:https?://)?(?:danbooru|hijiribe|sonohara)\.donmai\.us" + r"/pools/(\d+)"] test = [("https://danbooru.donmai.us/pools/7659", { "content": "b16bab12bea5f7ea9e0a836bf8045f280e113d99", })] @@ -36,7 +41,8 @@ class DanbooruPoolExtractor(DanbooruExtractor, booru.BooruPoolExtractor): class DanbooruPostExtractor(DanbooruExtractor, booru.BooruPostExtractor): """Extractor for single images from danbooru""" - pattern = [r"(?:https?://)?danbooru\.donmai\.us/posts/(\d+)"] + pattern = [r"(?:https?://)?(?:danbooru|hijiribe|sonohara)\.donmai\.us" + r"/posts/(\d+)"] test = [("https://danbooru.donmai.us/posts/294929", { "content": "5e255713cbf0a8e0801dc423563c34d896bb9229", })] @@ -44,8 +50,8 @@ class DanbooruPostExtractor(DanbooruExtractor, booru.BooruPostExtractor): class DanbooruPopularExtractor(DanbooruExtractor, booru.BooruPopularExtractor): """Extractor for popular images from danbooru""" - pattern = [r"(?:https?://)?danbooru\.donmai\.us/" - r"explore/posts/popular()(?:\?([^#]*))?"] + pattern = [r"(?:https?://)?(?:danbooru|hijiribe|sonohara)\.donmai\.us" + r"/explore/posts/popular()(?:\?([^#]*))?"] test = [ ("https://danbooru.donmai.us/explore/posts/popular", None), (("https://danbooru.donmai.us/explore/posts/popular"