mirror of
https://github.com/iv-org/invidious.git
synced 2026-09-07 01:22:49 -05:00
fix: use UTF-16 code units for description link offsets
YouTube provides startIndex and length values in UTF-16 code units, but the code was iterating Unicode codepoints. This mismatch caused links to be misaligned when the description contained emoji or other supplementary characters before a link. Fixes #5821
This commit is contained in:
parent
986471320d
commit
05e1caf208
@ -1,76 +0,0 @@
|
|||||||
require "json"
|
|
||||||
require "uri"
|
|
||||||
|
|
||||||
private def copy_string(str : String::Builder, iter : Iterator, count : Int) : Int
|
|
||||||
copied = 0
|
|
||||||
while copied < count
|
|
||||||
cp = iter.next
|
|
||||||
break if cp.is_a?(Iterator::Stop)
|
|
||||||
|
|
||||||
if cp == 0x26 # Ampersand (&)
|
|
||||||
str << "&"
|
|
||||||
elsif cp == 0x27 # Single quote (')
|
|
||||||
str << "'"
|
|
||||||
elsif cp == 0x22 # Double quote (")
|
|
||||||
str << """
|
|
||||||
elsif cp == 0x3C # Less-than (<)
|
|
||||||
str << "<"
|
|
||||||
elsif cp == 0x3E # Greater than (>)
|
|
||||||
str << ">"
|
|
||||||
else
|
|
||||||
str << cp.chr
|
|
||||||
end
|
|
||||||
|
|
||||||
copied += 1
|
|
||||||
end
|
|
||||||
|
|
||||||
return copied
|
|
||||||
end
|
|
||||||
|
|
||||||
def parse_description(desc, video_id : String) : String?
|
|
||||||
return "" if desc.nil?
|
|
||||||
|
|
||||||
content = desc["content"].as_s
|
|
||||||
return "" if content.empty?
|
|
||||||
|
|
||||||
commands = desc["commandRuns"]?.try &.as_a
|
|
||||||
if commands.nil?
|
|
||||||
# Slightly faster than HTML.escape, as we're only doing one pass on
|
|
||||||
# the string instead of five for the standard library
|
|
||||||
return String.build do |str|
|
|
||||||
copy_string(str, content.each_codepoint, content.size)
|
|
||||||
end
|
|
||||||
end
|
|
||||||
|
|
||||||
iter = content.each_codepoint
|
|
||||||
|
|
||||||
index = 0
|
|
||||||
|
|
||||||
return String.build do |str|
|
|
||||||
commands.each do |command|
|
|
||||||
cmd_start = command["startIndex"].as_i
|
|
||||||
cmd_length = command["length"].as_i
|
|
||||||
|
|
||||||
# Copy the text chunk between this command and the previous if needed.
|
|
||||||
length = cmd_start - index
|
|
||||||
index += copy_string(str, iter, length)
|
|
||||||
|
|
||||||
# We need to copy the command's text using the iterator
|
|
||||||
# and the special function defined above.
|
|
||||||
cmd_content = String.build(cmd_length) do |str2|
|
|
||||||
copy_string(str2, iter, cmd_length)
|
|
||||||
end
|
|
||||||
|
|
||||||
link = cmd_content
|
|
||||||
if on_tap = command.dig?("onTap", "innertubeCommand")
|
|
||||||
link = parse_link_endpoint(on_tap, cmd_content, video_id)
|
|
||||||
end
|
|
||||||
str << link
|
|
||||||
index += cmd_length
|
|
||||||
end
|
|
||||||
|
|
||||||
# Copy the end of the string (past the last command).
|
|
||||||
remaining_length = content.size - index
|
|
||||||
copy_string(str, iter, remaining_length) if remaining_length > 0
|
|
||||||
end
|
|
||||||
end
|
|
||||||
Loading…
x
Reference in New Issue
Block a user