From b962dacf7be6b606c23a0293bd5b64f6b714e1e6 Mon Sep 17 00:00:00 2001
From: Tangent <rose@tangentfox.com>
Date: Tue, 20 Feb 2024 01:23:37 -0700
Subject: [PATCH] subtitle downloader is better

---
 download-subtitles.lua | 49 +++++++++++++++++++++++++++++++++++-------
 1 file changed, 41 insertions(+), 8 deletions(-)

diff --git a/download-subtitles.lua b/download-subtitles.lua
index e775a76..e3c5e4b 100644
--- a/download-subtitles.lua
+++ b/download-subtitles.lua
@@ -1,12 +1,45 @@
-local source = "URL"
+#!/usr/bin/env luajit
 
-os.execute("yt-dlp --flat-playlist --print-to-file \"%(id)s\" download-list.txt --skip-download \"" .. source .. "\"")
-os.execute("yt-dlp --flat-playlist --print-to-file \"%(id)s=%(title)s\" VIDEO-TITLES.txt --skip-download \"" .. source .. "\"")
+local source = arg[1]
+assert(source and #source > 0, "Specify a URL!")
+
+local function make_safe(file_name)
+  -- return file_name:gsub("[%w%s%_%-%,%.%[%]%(%)%'%+]", "") -- this is the literal opposite of what I needed, oops
+  -- return file_name:gsub("[%\"%:%\\%!%@%#%$%%%^%&%*%=%{%}%|%;%<%>%?%/]", "") -- this should handle everything but I'm not certain! :D
+
+  file_name = file_name:gsub("[%\"%:%\\%!%@%#%$%%%^%*%=%{%}%|%;%<%>%?%/]", "") -- everything except the &
+  file_name = file_name:gsub(" %&", ",")   -- replacing & with a comma works for 99% of things
+  file_name = file_name:gsub("%&", ",")    -- replacing & with a comma works for 99% of things
+  file_name = file_name:gsub("[%s+]", " ") -- more than one space in succession should be a single space
+  return file_name
+end
+
+local function download_subtitle(id, file_name)
+  os.execute("sleep 6") -- up-front in case chromium isn't ready / delay from previous download to prevent rate limit issues
+  os.execute("chromium \"https://www.downloadyoutubesubtitles.com/?u=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3D" .. id .. "\"")
+  os.execute("sleep 12")
+  os.execute("curl \"https://www.downloadyoutubesubtitles.com/get2.php?i=" .. id .. "&format=txt&hl=en&a=\" > \"./subtitles/" .. (file_name or id) .. ".txt\"")
+end
+
+os.execute("pwsh -command chromium") -- open browser first, to make sure it is ready by the time we start using it
+
+os.execute("yt-dlp --flat-playlist --print-to-file \"%(id)s=%(title)s\" videos.txt --skip-download \"" .. source .. "\"")
 os.execute("mkdir subtitles")
 
-for line in io.lines("download-list.txt") do
-  os.execute("chromium \"https://www.downloadyoutubesubtitles.com/?u=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3D" .. line .. "\"")
-  os.execute("sleep 12")
-  os.execute("curl \"https://www.downloadyoutubesubtitles.com/get2.php?i=" .. line .. "&format=txt&hl=en&a=\" > \"./subtitles/" .. line .. ".txt\"")
-  os.execute("sleep 6")
+local videos = {}
+
+local file = io.open("videos.txt", "r")
+assert(file, "videos.txt doesn't exist.. somehow")
+for line in file:lines() do
+  local i, j = line:find("%=")
+  videos[#videos + 1] = {
+    id = line:sub(1, i - 1),
+    title = line:sub(j + 1),
+  }
+end
+file:close()
+
+for index, video in ipairs(videos) do
+  print(index, #videos, video.title)
+  download_subtitle(video.id, make_safe(video.title))
 end