Refactor sources

This commit is contained in:
Albert Yi
2018-08-06 17:39:25 -07:00
parent 54363ffecf
commit 762dc3da24
71 changed files with 2340 additions and 2430 deletions

View File

@@ -1,182 +1,188 @@
module Sources
module Strategies
class NicoSeiga < Base
extend Memoist
def self.url_match?(url)
url =~ /^https?:\/\/(?:\w+\.)?nico(?:seiga|video)\.jp/
end
URL = %r!\Ahttps?://(?:\w+\.)?nico(?:seiga|video)\.jp!
DIRECT = %r!\Ahttps?://lohas\.nicoseiga\.jp/priv/[0-9a-f]+!
PAGE = %r!\Ahttps?://seiga\.nicovideo\.jp/seiga/im(\d+)!i
PROFILE = %r!\Ahttps?://seiga\.nicovideo\.jp/user/illust/(\d+)!i
def referer_url
if @referer_url =~ /seiga\.nicovideo\.jp\/seiga\/im\d+/ && @url =~ /http:\/\/lohas\.nicoseiga\.jp\/(?:priv|o)\//
@referer_url
else
@url
end
def self.match?(*urls)
urls.compact.any? { |x| x.match?(URL) }
end
def site_name
"Nico Seiga"
end
def unique_id
profile_url =~ /\/illust\/(\d+)/
"nicoseiga" + $1
end
def get
page = load_page
@artist_name, @profile_url = get_profile_from_api
@image_url = get_image_url_from_page(page)
@artist_commentary_title, @artist_commentary_desc = get_artist_commentary_from_api
# Log out before getting the tags.
# The reason for this is that if you're logged in and viewing a non-adult-rated work, the tags will be added with javascript after the page has loaded meaning we can't extract them easily.
# This does not apply if you're logged out (or if you're viewing an adult-rated work).
agent.cookie_jar.clear!
agent.get(normalized_url) do |page|
@tags = get_tags_from_page(page)
end
end
def normalized_for_artist_finder?
url =~ %r!https?://seiga\.nicovideo\.jp/user/illust/\d+/!i
end
def normalizable_for_artist_finder?
url =~ %r!https?://seiga\.nicovideo\.jp/seiga/im\d+!i
end
def normalize_for_artist_finder!
page = load_page
@illust_id = get_illust_id_from_url
@artist_name, @profile_url = get_profile_from_api
@profile_url + "/"
end
protected
def api_client
NicoSeigaApiClient.new(get_illust_id_from_url)
end
def get_illust_id_from_url
if normalized_url =~ %r!http://seiga.nicovideo.jp/seiga/im(\d+)!
$1.to_i
else
nil
end
end
def load_page
page = agent.get(normalized_url)
if page.search("a#link_btn_login").any?
# Session cache is invalid, clear it and log in normally.
Cache.delete("nico-seiga-session")
@agent = nil
page = agent.get(normalized_url)
def image_urls
if url =~ DIRECT
return [url]
end
page
end
def get_profile_from_api
return [api_client.moniker, "http://seiga.nicovideo.jp/user/illust/#{api_client.user_id}"]
end
def get_image_url_from_page(page)
link = page.search("a#illust_link")
if link.any?
image_url = "http://seiga.nicovideo.jp" + link[0]["href"]
page = agent.get(image_url) # need to follow this redirect while logged in or it won't work
if page.is_a?(Mechanize::Image)
return page.uri.to_s
return [page.uri.to_s]
end
images = page.search("div.illust_view_big").select {|x| x["data-src"] =~ /\/priv\//}
if images.any?
image_url = "http://lohas.nicoseiga.jp" + images[0]["data-src"]
end
else
image_url = nil
end
return image_url
end
def get_tags_from_page(page)
links = page.search("a.tag")
links.map do |node|
[node.text, "http://seiga.nicovideo.jp" + node.attr("href")]
end
end
def get_artist_commentary_from_api
[api_client.title, api_client.desc]
end
def normalized_url
@normalized_url ||= begin
if url =~ %r!\Ahttps?://lohas\.nicoseiga\.jp/o/[a-f0-9]+/\d+/(\d+)!
"http://seiga.nicovideo.jp/seiga/im#{$1}"
elsif url =~ %r{\Ahttps?://lohas\.nicoseiga\.jp/priv/(\d+)\?e=\d+&h=[a-f0-9]+}i
"http://seiga.nicovideo.jp/seiga/im#{$1}"
elsif url =~ %r{\Ahttps?://lohas\.nicoseiga\.jp/priv/[a-f0-9]+/\d+/(\d+)}i
"http://seiga.nicovideo.jp/seiga/im#{$1}"
elsif url =~ %r{\Ahttps?://lohas\.nicoseiga\.jp/priv/(\d+)}i
"http://seiga.nicovideo.jp/seiga/im#{$1}"
elsif url =~ %r{\Ahttps?://lohas\.nicoseiga\.jp//?thumb/(\d+)i?}i
"http://seiga.nicovideo.jp/seiga/im#{$1}"
elsif url =~ %r{/seiga/im\d+}
url
else
nil
return ["http://lohas.nicoseiga.jp" + images[0]["data-src"]]
end
end
raise "image url not found for (#{url}, #{referer_url})"
end
def page_url
[url, referer_url].each do |x|
if x =~ %r!\Ahttps?://lohas\.nicoseiga\.jp/o/[a-f0-9]+/\d+/(\d+)!
return "http://seiga.nicovideo.jp/seiga/im#{$1}"
end
if x =~ %r{\Ahttps?://lohas\.nicoseiga\.jp/priv/(\d+)\?e=\d+&h=[a-f0-9]+}i
return "http://seiga.nicovideo.jp/seiga/im#{$1}"
end
if x =~ %r{\Ahttps?://lohas\.nicoseiga\.jp/priv/[a-f0-9]+/\d+/(\d+)}i
return "http://seiga.nicovideo.jp/seiga/im#{$1}"
end
if x =~ %r{\Ahttps?://lohas\.nicoseiga\.jp/priv/(\d+)}i
return "http://seiga.nicovideo.jp/seiga/im#{$1}"
end
if x =~ %r{\Ahttps?://lohas\.nicoseiga\.jp//?thumb/(\d+)i?}i
return "http://seiga.nicovideo.jp/seiga/im#{$1}"
end
if x =~ %r{/seiga/im\d+}
return x
end
end
return super
end
def profile_url
if url =~ PROFILE
return url
end
"http://seiga.nicovideo.jp/user/illust/#{api_client.user_id}"
end
def artist_name
api_client.moniker
end
def artist_commentary_title
api_client.title
end
def artist_commentary_desc
api_client.desc
end
def headers
super.merge(
"Referer" => "https://seiga.nicovideo.jp"
)
end
def normalized_for_artist_finder?
url =~ PROFILE
end
def normalizable_for_artist_finder?
url =~ PAGE || url =~ PROFILE
end
def normalize_for_artist_finder
"#{profile_url}/"
end
def unique_id
"nicoseiga#{api_client.user_id}"
end
def tags
string = page.at("meta[name=keywords]").try(:[], "content") || ""
string.split(/,/).map do |name|
[name, "https://seiga.nicovideo.jp/tag/#{CGI.escape(name)}"]
end
end
memoize :tags
public
def api_client
NicoSeigaApiClient.new(illust_id)
end
memoize :api_client
def illust_id
if page_url =~ PAGE
return $1.to_i
end
return nil
end
def page
doc = agent.get(page_url)
if doc.search("a#link_btn_login").any?
# Session cache is invalid, clear it and log in normally.
Cache.delete("nico-seiga-session")
doc = agent.get(page_url)
end
doc
end
memoize :page
def agent
@agent ||= begin
mech = Mechanize.new
mech.redirect_ok = false
mech.keep_alive = false
mech = Mechanize.new
mech.redirect_ok = false
mech.keep_alive = false
session = Cache.get("nico-seiga-session")
if session
cookie = Mechanize::Cookie.new("user_session", session)
cookie.domain = ".nicovideo.jp"
cookie.path = "/"
mech.cookie_jar.add(cookie)
else
mech.get("https://account.nicovideo.jp/login") do |page|
page.form_with(:id => "login_form") do |form|
form["mail_tel"] = Danbooru.config.nico_seiga_login
form["password"] = Danbooru.config.nico_seiga_password
end.click_button
end
session = mech.cookie_jar.cookies.select{|c| c.name == "user_session"}.first
if session
Cache.put("nico-seiga-session", session.value, 1.month)
else
raise "Session not found"
end
end
# This cookie needs to be set to allow viewing of adult works
cookie = Mechanize::Cookie.new("skip_fetish_warning", "1")
cookie.domain = "seiga.nicovideo.jp"
session = Cache.get("nico-seiga-session")
if session
cookie = Mechanize::Cookie.new("user_session", session)
cookie.domain = ".nicovideo.jp"
cookie.path = "/"
mech.cookie_jar.add(cookie)
mech.redirect_ok = true
mech
else
mech.get("https://account.nicovideo.jp/login") do |page|
page.form_with(:id => "login_form") do |form|
form["mail_tel"] = Danbooru.config.nico_seiga_login
form["password"] = Danbooru.config.nico_seiga_password
end.click_button
end
session = mech.cookie_jar.cookies.select{|c| c.name == "user_session"}.first
if session
Cache.put("nico-seiga-session", session.value, 1.month)
else
raise "Session not found"
end
end
end
memoize :api_client
# This cookie needs to be set to allow viewing of adult works
cookie = Mechanize::Cookie.new("skip_fetish_warning", "1")
cookie.domain = "seiga.nicovideo.jp"
cookie.path = "/"
mech.cookie_jar.add(cookie)
mech.redirect_ok = true
mech
end
memoize :agent
end
end
end