From 3ada6a92348ea2133f2eb0b2387ce7fb490e2e67 Mon Sep 17 00:00:00 2001 From: Omar Roth Date: Sat, 4 Aug 2018 17:12:58 -0500 Subject: [PATCH] Add filters to '/api/v1/search' endpoint --- src/invidious.cr | 34 +++++++++++++- src/invidious/search.cr | 98 ++++++++++++++++++++++++++++++++++++++++- 2 files changed, 128 insertions(+), 4 deletions(-) diff --git a/src/invidious.cr b/src/invidious.cr index e82f0b9f..0a023345 100644 --- a/src/invidious.cr +++ b/src/invidious.cr @@ -2569,14 +2569,44 @@ get "/api/v1/search" do |env| page = env.params.query["page"]?.try &.to_i? page ||= 1 + sort_by = env.params.query["sort_by"]?.try &.downcase + sort_by ||= "relevance" + + date = env.params.query["date"]?.try &.downcase + date ||= "" + + duration = env.params.query["date"]?.try &.downcase + duration ||= "" + + features = env.params.query["features"]?.try &.split(",").map { |feature| feature.downcase } + features ||= [] of String + + # TODO: Support other content types + content_type = "video" + + begin + search_params = build_search_params(sort_by, date, content_type, duration, features) + rescue ex + env.response.content_type = "application/json" + next JSON.build do |json| + json.object do + json.field "error", ex.message + end + end + end + client = make_client(YT_URL) - html = client.get("/results?q=#{URI.escape(query)}&page=#{page}&sp=EgIQAVAU&disable_polymer=1").body + html = client.get("/results?q=#{URI.escape(query)}&page=#{page}&sp=#{search_params}&disable_polymer=1").body html = XML.parse_html(html) results = JSON.build do |json| json.array do html.xpath_nodes(%q(//ol[@class="item-section"]/li)).each do |node| - anchor = node.xpath_node(%q(.//h3[contains(@class,"yt-lockup-title")]/a)).not_nil! + anchor = node.xpath_node(%q(.//h3[contains(@class,"yt-lockup-title")]/a)) + if !anchor + next + end + if anchor["href"].starts_with? "https://www.googleadservices.com" next end diff --git a/src/invidious/search.cr b/src/invidious/search.cr index 034db789..7714de22 100644 --- a/src/invidious/search.cr +++ b/src/invidious/search.cr @@ -1,6 +1,6 @@ -def search(query, page = 1) +def search(query, page = 1, search_params = build_search_params(content_type: "video")) client = make_client(YT_URL) - html = client.get("/results?q=#{URI.escape(query)}&page=#{page}&sp=EgIQAVAU").body + html = client.get("/results?q=#{URI.escape(query)}&page=#{page}&sp=#{search_params}").body html = XML.parse_html(html) videos = [] of ChannelVideo @@ -28,3 +28,97 @@ def search(query, page = 1) return videos end + +def build_search_params(sort_by = "relevance", date : String = "", content_type : String = "", duration : String = "", features : Array(String) = [] of String) + head = "\x08" + head += case sort_by + when "relevance" + "\x00" + when "rating" + "\x01" + when "upload_date" + "\x02" + when "view_count" + "\x03" + else + raise "No sort #{sort_by}" + end + + body = "" + body += case date + when "hour" + "\x08\x01" + when "today" + "\x08\x02" + when "week" + "\x08\x03" + when "month" + "\x08\x04" + when "year" + "\x08\x05" + else + "" + end + + body += case content_type + when "video" + "\x10\x01" + when "channel" + "\x10\x02" + when "playlist" + "\x10\x03" + when "movie" + "\x10\x04" + when "show" + "\x10\x05" + else + "" + end + + body += case duration + when "short" + "\x18\x01" + when "long" + "\x18\x02" + else + "" + end + + features.each do |feature| + body += case feature + when "hd" + "\x20\x01" + when "subtitles" + "\x28\x01" + when "creative_commons" + "\x30\x01" + when "3d" + "\x38\x01" + when "live" + "\x40\x01" + when "purchased" + "\x48\x01" + when "4k" + "\x70\x01" + when "360" + "\x78\x01" + when "location" + "\xb8\x01\x01" + when "hdr" + "\xc8\x01\x01" + else + raise "Unknown feature #{feature}" + end + end + + if body.size > 0 + token = head + "\x12" + body.size.to_u8.unsafe_chr + body + else + token = head + end + + token = Base64.urlsafe_encode(token) + token = URI.escape(token) + + return token +end