Videos: fix description text offset when emojis are present

This commit is contained in:
Samantaz Fox 2023-04-05 23:43:41 +02:00
parent 73d2ed6f77
commit b3c0afef02
No known key found for this signature in database
GPG Key ID: F42821059186176E
1 changed files with 47 additions and 24 deletions

View File

@ -46,37 +46,60 @@ def parse_command(command : JSON::Any?, string : String) : String?
return "(unknown YouTube desc command)" return "(unknown YouTube desc command)"
end end
def parse_description(desc : JSON::Any?) : String? private def copy_string(str : String::Builder, iter : Iterator, count : Int) : Int
if desc.nil? copied = 0
return "" while copied < count
cp = iter.next
break if cp.is_a?(Iterator::Stop)
str << cp.chr
# A codepoint from the SMP counts twice
copied += 1 if cp > 0xFFFF
copied += 1
end end
return copied
end
def parse_description(desc : JSON::Any?) : String?
return "" if desc.nil?
content = desc["content"].as_s content = desc["content"].as_s
if content.empty? return "" if content.empty?
return ""
end commands = desc["commandRuns"]?.try &.as_a
return content if commands.nil?
# Not everything is stored in UTF-8 on youtube's side. The SMP codepoints
# (0x10000 and above) are encoded as UTF-16 surrogate pairs, which are
# automatically decoded by the JSON parser. It means that we need to count
# copied byte in a special manner, preventing the use of regular string copy.
iter = content.each_codepoint
if commands = desc["commandRuns"]?.try &.as_a
description = String.build do |str|
index = 0 index = 0
return String.build do |str|
commands.each do |command| commands.each do |command|
start_index = command["startIndex"].as_i cmd_start = command["startIndex"].as_i
length = command["length"].as_i cmd_length = command["length"].as_i
if start_index > 0 && start_index - index > 0 # Copy the text chunk between this command and the previous if needed.
str << content[index...start_index] length = cmd_start - index
index = 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 end
str << parse_command(command, content[start_index, length]) str << parse_command(command, cmd_content)
index += length index += cmd_length
end
if index < content.size
str << content[index..content.size]
end
end
return description
end end
return content # 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 end