summaryrefslogtreecommitdiffstats
path: root/src
diff options
context:
space:
mode:
Diffstat (limited to 'src')
-rw-r--r--src/invidious/channels/community.cr2
-rw-r--r--src/invidious/helpers/json_filter.cr6
-rw-r--r--src/invidious/helpers/utils.cr2
-rw-r--r--src/invidious/routes/api/v1/videos.cr70
-rw-r--r--src/invidious/routing.cr2
-rw-r--r--src/invidious/videos/caption.cr66
-rw-r--r--src/invidious/yt_backend/extractors.cr16
7 files changed, 128 insertions, 36 deletions
diff --git a/src/invidious/channels/community.cr b/src/invidious/channels/community.cr
index 8e300288..76dff555 100644
--- a/src/invidious/channels/community.cr
+++ b/src/invidious/channels/community.cr
@@ -69,7 +69,7 @@ def fetch_channel_community(ucid, continuation, locale, format, thin_mode)
next if !post
content_html = post["contentText"]?.try { |t| parse_content(t) } || ""
- author = post["authorText"]?.try &.["simpleText"]? || ""
+ author = post["authorText"]["runs"]?.try &.[0]?.try &.["text"]? || ""
json.object do
json.field "author", author
diff --git a/src/invidious/helpers/json_filter.cr b/src/invidious/helpers/json_filter.cr
index b8e8f96d..3f4080ba 100644
--- a/src/invidious/helpers/json_filter.cr
+++ b/src/invidious/helpers/json_filter.cr
@@ -20,7 +20,7 @@ module JSONFilter
/^\(|\(\(|\/\(/
end
- def self.parse_fields(fields_text : String) : Nil
+ def self.parse_fields(fields_text : String, &) : Nil
if fields_text.empty?
raise FieldsParser::ParseError.new "Fields is empty"
end
@@ -42,7 +42,7 @@ module JSONFilter
parse_nest_groups(fields_text) { |nest_list| yield nest_list }
end
- def self.parse_single_nests(fields_text : String) : Nil
+ def self.parse_single_nests(fields_text : String, &) : Nil
single_nests = remove_nest_groups(fields_text)
if !single_nests.empty?
@@ -60,7 +60,7 @@ module JSONFilter
end
end
- def self.parse_nest_groups(fields_text : String) : Nil
+ def self.parse_nest_groups(fields_text : String, &) : Nil
nest_stack = [] of NamedTuple(group_name: String, closing_bracket_index: Int64)
bracket_pairs = get_bracket_pairs(fields_text, true)
diff --git a/src/invidious/helpers/utils.cr b/src/invidious/helpers/utils.cr
index 4448508c..500a2582 100644
--- a/src/invidious/helpers/utils.cr
+++ b/src/invidious/helpers/utils.cr
@@ -162,7 +162,7 @@ def number_with_separator(number)
end
def short_text_to_number(short_text : String) : Int64
- matches = /(?<number>\d+(\.\d+)?)\s?(?<suffix>[mMkKbB])?/.match(short_text)
+ matches = /(?<number>\d+(\.\d+)?)\s?(?<suffix>[mMkKbB]?)/.match(short_text)
number = matches.try &.["number"].to_f || 0.0
case matches.try &.["suffix"].downcase
diff --git a/src/invidious/routes/api/v1/videos.cr b/src/invidious/routes/api/v1/videos.cr
index 79f7bd3f..f312211e 100644
--- a/src/invidious/routes/api/v1/videos.cr
+++ b/src/invidious/routes/api/v1/videos.cr
@@ -93,45 +93,50 @@ module Invidious::Routes::API::V1::Videos
# as well as some other markup that makes it cumbersome, so we try to fix that here
if caption.name.includes? "auto-generated"
caption_xml = YT_POOL.client &.get(url).body
- caption_xml = XML.parse(caption_xml)
- webvtt = String.build do |str|
- str << <<-END_VTT
- WEBVTT
- Kind: captions
- Language: #{tlang || caption.language_code}
+ if caption_xml.starts_with?("<?xml")
+ webvtt = caption.timedtext_to_vtt(caption_xml, tlang)
+ else
+ caption_xml = XML.parse(caption_xml)
+ webvtt = String.build do |str|
+ str << <<-END_VTT
+ WEBVTT
+ Kind: captions
+ Language: #{tlang || caption.language_code}
- END_VTT
- caption_nodes = caption_xml.xpath_nodes("//transcript/text")
- caption_nodes.each_with_index do |node, i|
- start_time = node["start"].to_f.seconds
- duration = node["dur"]?.try &.to_f.seconds
- duration ||= start_time
+ END_VTT
- if caption_nodes.size > i + 1
- end_time = caption_nodes[i + 1]["start"].to_f.seconds
- else
- end_time = start_time + duration
- end
+ caption_nodes = caption_xml.xpath_nodes("//transcript/text")
+ caption_nodes.each_with_index do |node, i|
+ start_time = node["start"].to_f.seconds
+ duration = node["dur"]?.try &.to_f.seconds
+ duration ||= start_time
- start_time = "#{start_time.hours.to_s.rjust(2, '0')}:#{start_time.minutes.to_s.rjust(2, '0')}:#{start_time.seconds.to_s.rjust(2, '0')}.#{start_time.milliseconds.to_s.rjust(3, '0')}"
- end_time = "#{end_time.hours.to_s.rjust(2, '0')}:#{end_time.minutes.to_s.rjust(2, '0')}:#{end_time.seconds.to_s.rjust(2, '0')}.#{end_time.milliseconds.to_s.rjust(3, '0')}"
+ if caption_nodes.size > i + 1
+ end_time = caption_nodes[i + 1]["start"].to_f.seconds
+ else
+ end_time = start_time + duration
+ end
- text = HTML.unescape(node.content)
- text = text.gsub(/<font color="#[a-fA-F0-9]{6}">/, "")
- text = text.gsub(/<\/font>/, "")
- if md = text.match(/(?<name>.*) : (?<text>.*)/)
- text = "<v #{md["name"]}>#{md["text"]}</v>"
- end
+ start_time = "#{start_time.hours.to_s.rjust(2, '0')}:#{start_time.minutes.to_s.rjust(2, '0')}:#{start_time.seconds.to_s.rjust(2, '0')}.#{start_time.milliseconds.to_s.rjust(3, '0')}"
+ end_time = "#{end_time.hours.to_s.rjust(2, '0')}:#{end_time.minutes.to_s.rjust(2, '0')}:#{end_time.seconds.to_s.rjust(2, '0')}.#{end_time.milliseconds.to_s.rjust(3, '0')}"
- str << <<-END_CUE
- #{start_time} --> #{end_time}
- #{text}
+ text = HTML.unescape(node.content)
+ text = text.gsub(/<font color="#[a-fA-F0-9]{6}">/, "")
+ text = text.gsub(/<\/font>/, "")
+ if md = text.match(/(?<name>.*) : (?<text>.*)/)
+ text = "<v #{md["name"]}>#{md["text"]}</v>"
+ end
+ str << <<-END_CUE
+ #{start_time} --> #{end_time}
+ #{text}
- END_CUE
+
+ END_CUE
+ end
end
end
else
@@ -141,7 +146,12 @@ module Invidious::Routes::API::V1::Videos
#
# See: https://github.com/iv-org/invidious/issues/2391
webvtt = YT_POOL.client &.get("#{url}&format=vtt").body
- .gsub(/([0-9:.]{12} --> [0-9:.]{12}).+/, "\\1")
+ if webvtt.starts_with?("<?xml")
+ webvtt = caption.timedtext_to_vtt(webvtt)
+ else
+ webvtt = YT_POOL.client &.get("#{url}&format=vtt").body
+ .gsub(/([0-9:.]{12} --> [0-9:.]{12}).+/, "\\1")
+ end
end
if title = env.params.query["title"]?
diff --git a/src/invidious/routing.cr b/src/invidious/routing.cr
index 491022a5..157e6de7 100644
--- a/src/invidious/routing.cr
+++ b/src/invidious/routing.cr
@@ -132,6 +132,8 @@ module Invidious::Routing
get "/c/:user#{path}", Routes::Channels, :brand_redirect
# /user/linustechtips | Not always the same as /c/
get "/user/:user#{path}", Routes::Channels, :brand_redirect
+ # /@LinusTechTips | Handle
+ get "/@:user#{path}", Routes::Channels, :brand_redirect
# /attribution_link?a=anything&u=/channel/UCZYTClx2T1of7BRZ86-8fow
get "/attribution_link#{path}", Routes::Channels, :brand_redirect
# /profile?user=linustechtips
diff --git a/src/invidious/videos/caption.cr b/src/invidious/videos/caption.cr
index 4642c1a7..13f81a31 100644
--- a/src/invidious/videos/caption.cr
+++ b/src/invidious/videos/caption.cr
@@ -31,6 +31,72 @@ module Invidious::Videos
return captions_list
end
+ def timedtext_to_vtt(timedtext : String, tlang = nil) : String
+ # In the future, we could just directly work with the url. This is more of a POC
+ cues = [] of XML::Node
+ tree = XML.parse(timedtext)
+ tree = tree.children.first
+
+ tree.children.each do |item|
+ if item.name == "body"
+ item.children.each do |cue|
+ if cue.name == "p" && !(cue.children.size == 1 && cue.children[0].content == "\n")
+ cues << cue
+ end
+ end
+ break
+ end
+ end
+ result = String.build do |result|
+ result << <<-END_VTT
+ WEBVTT
+ Kind: captions
+ Language: #{tlang || @language_code}
+
+
+ END_VTT
+
+ result << "\n\n"
+
+ cues.each_with_index do |node, i|
+ start_time = node["t"].to_f.milliseconds
+
+ duration = node["d"]?.try &.to_f.milliseconds
+
+ duration ||= start_time
+
+ if cues.size > i + 1
+ end_time = cues[i + 1]["t"].to_f.milliseconds
+ else
+ end_time = start_time + duration
+ end
+
+ # start_time
+ result << start_time.hours.to_s.rjust(2, '0')
+ result << ':' << start_time.minutes.to_s.rjust(2, '0')
+ result << ':' << start_time.seconds.to_s.rjust(2, '0')
+ result << '.' << start_time.milliseconds.to_s.rjust(3, '0')
+
+ result << " --> "
+
+ # end_time
+ result << end_time.hours.to_s.rjust(2, '0')
+ result << ':' << end_time.minutes.to_s.rjust(2, '0')
+ result << ':' << end_time.seconds.to_s.rjust(2, '0')
+ result << '.' << end_time.milliseconds.to_s.rjust(3, '0')
+
+ result << "\n"
+
+ node.children.each do |s|
+ result << s.content
+ end
+ result << "\n"
+ result << "\n"
+ end
+ end
+ return result
+ end
+
# List of all caption languages available on Youtube.
LANGUAGES = {
"",
diff --git a/src/invidious/yt_backend/extractors.cr b/src/invidious/yt_backend/extractors.cr
index 65d107b2..b14ad7b9 100644
--- a/src/invidious/yt_backend/extractors.cr
+++ b/src/invidious/yt_backend/extractors.cr
@@ -172,7 +172,17 @@ private module Parsers
# When public subscriber count is disabled, the subscriberCountText isn't sent by InnerTube.
# Always simpleText
# TODO change default value to nil
+
subscriber_count = item_contents.dig?("subscriberCountText", "simpleText")
+
+ # Since youtube added channel handles, `VideoCountText` holds the number of
+ # subscribers and `subscriberCountText` holds the handle, except when the
+ # channel doesn't have a handle (e.g: some topic music channels).
+ # See https://github.com/iv-org/invidious/issues/3394#issuecomment-1321261688
+ if !subscriber_count || !subscriber_count.as_s.includes? " subscriber"
+ subscriber_count = item_contents.dig?("videoCountText", "simpleText")
+ end
+ subscriber_count = subscriber_count
.try { |s| short_text_to_number(s.as_s.split(" ")[0]).to_i32 } || 0
# Auto-generated channels doesn't have videoCountText
@@ -682,7 +692,11 @@ module HelperExtractors
# Returns a 0 when it's unable to do so
def self.get_video_count(container : JSON::Any) : Int32
if box = container["videoCountText"]?
- return extract_text(box).try &.gsub(/\D/, "").to_i || 0
+ if (extracted_text = extract_text(box)) && !extracted_text.includes? " subscriber"
+ return extracted_text.gsub(/\D/, "").to_i
+ else
+ return 0
+ end
elsif box = container["videoCount"]?
return box.as_s.to_i
else
generated by cgit v1.2.3 (git 2.39.1) at 2025-12-17 16:50:26 +0000