jaculabilis/danbooru
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

tumblr: fix _640 images not being recognized (#3944).

Browse Source 
Fixes _640 images not being matched by the IMAGE regex and therefore not
being rewritten to the largest size.
This commit is contained in:
evazion
2018-10-09 11:57:23 -05:00
parent d874c68419
commit 184a5ebf3e
2 changed files with 24 additions and 12 deletions
Download Patch File Download Diff File Expand all files Collapse all files

22
app/logical/sources/strategies/tumblr.rb
View File

@@ -1,12 +1,13 @@
module Sources::Strategies
class Tumblr < Base
SIZES = %w[1280 640 540 500h 500 400 250 100]
BASE_URL = %r!\Ahttps?://(?:[^/]+\.)*tumblr\.com!i
DOMAIN = %r{(data|(\d+\.)?media)\.tumblr\.com}
MD5 = %r{(?<md5>[0-9a-f]{32})}i
FILENAME = %r{(?<filename>(tumblr_(inline_)?)?[a-z0-9]+(_r[0-9]+)?)}i
SIZES = %r{(?:250|400|500|500h|540|1280|raw)}i
EXT = %r{(?<ext>\w+)}
IMAGE = %r!\Ahttps?://#{DOMAIN}/(?<dir>#{MD5}/)?#{FILENAME}_#{SIZES}\.#{EXT}\z!i
IMAGE = %r!\Ahttps?://#{DOMAIN}/(?<dir>#{MD5}/)?#{FILENAME}_(?<size>\w+)\.#{EXT}\z!i
POST = %r!\Ahttps?://(?<blog_name>[^.]+)\.tumblr\.com/(?:post|image)/(?<post_id>\d+)!i
def self.enabled?
@@ -132,19 +133,16 @@ module Sources::Strategies
#
# http://media.tumblr.com/tumblr_m24kbxqKAX1rszquso1_1280.jpg
# => https://media.tumblr.com/tumblr_m24kbxqKAX1rszquso1_1280.jpg
def find_largest(x)
if x =~ IMAGE
sizes = [1280, 640, 540, "500h", 500, 400, 250]
candidates = sizes.map do |size|
"https://media.tumblr.com/#{$~[:dir]}#{$~[:filename]}_#{size}.#{$~[:ext]}"
end
def find_largest(url, sizes: SIZES)
return url unless url =~ IMAGE
return candidates.find do |candidate|
http_exists?(candidate, headers)
end
candidates = sizes.map do |size|
"https://media.tumblr.com/#{$~[:dir]}#{$~[:filename]}_#{size}.#{$~[:ext]}"
end
return x
candidates.find do |candidate|
http_exists?(candidate, headers)
end
end
def inline_images

14
test/unit/sources/tumblr_test.rb
View File

@@ -209,5 +209,19 @@ module Sources
assert_operator(data, :<, site.to_h)
end
end
context "A download for a 'http://*.media.tumblr.com/$hash/tumblr_$id_$size.png' image" do
should "find the largest image" do
%w[100 250 400 500 500h 540 640 1280].each do |size|
page = "https://natsuki-teru.tumblr.com/post/178728919271"
image = "https://66.media.tumblr.com/b9395771b2d0435fe4efee926a5a7d9c/tumblr_pg2wu1L9DM1trd056o2_#{size}.png"
full = "https://media.tumblr.com/b9395771b2d0435fe4efee926a5a7d9c/tumblr_pg2wu1L9DM1trd056o2_1280.png"
site = Sources::Strategies.find(image, page)
assert_equal(full, site.image_url)
assert_equal(full, site.image_urls.second)
end
end
end
end
end