little-things/youtube-extract

#!/usr/bin/env python3
import re
import sys


if any(x in sys.argv for x in ['--help', '-h', '-?', 'help']):
	print('Usage: youtube-extract [massage|removenonyt]', file = sys.stderr)
	print(file = sys.stderr)
	print("In 'massage' mode (default), extracts any references to YouTube videos, channels, and playlists from the URLs on stdin and prints them on stdout.", file = sys.stderr)
	print('Lines that don\'t seem to contain references to such YouTube things are printed on stderr.', file = sys.stderr)
	print(file = sys.stderr)
	print("In 'removenonyt' mode, prints all URLs that look like they are or contain YouTube URLs on stdout.", file = sys.stderr)
	sys.exit(1)


# For all patterns: don't use \\? for an optional backslash (if needed) as it breaks the automatic pattern rewrite for parameters below. Use [\\]? instead.

# Channel/user URLs; the protocol and domain are stripped and replaced below.
channelPattern = re.compile('|'.join([
	r'//www\.youtube\.com/c/[^/?]+',
	r'//www\.youtube\.com/user/[^/?]+',
	r'//www\.youtube\.com/channel/UC[^/?]+',
	r'//www\.youtube\.com/(?!(c|user|channel|watch(_popup)?(\.php)?|embed|e|v|redirect)$)[^/?]+(?=/?$)',
  ]))

# Make sure that the last 11 chars of the match are always the video ID (because Python's re doesn't support \K).
# If necessary, use lookahead assertions to match further stuff after the video ID.
videoPattern = re.compile('|'.join([
	# Normal watch URL
	r'//www\.youtube\.com/watch(_popup)?(\.php)?/?\?(.*&)?v=[0-9A-Za-z_-]{11}',
	r'//www\.youtube\.com/watch/[0-9A-Za-z_-]{11}',
	# Embeds
	r'//www\.youtube\.com/e(mbed)?/(?!videoseries\?)[0-9A-Za-z_-]{11}',
	r'//www\.youtube\.com/embed/?\?(.*&)?v=[0-9A-Za-z_-]{11}',
	# Shortener
	r'//youtu\.be/[0-9A-Za-z_-]{11}',
	# Old (Flash) embeds
	r'//www\.youtube\.com/v/[0-9A-Za-z_-]{11}',
	# Redirects from links in video descriptions
	r'//www\.youtube\.com/redirect\?(.*&)?v=[0-9A-Za-z_-]{11}(?=&|$)',
	# Tracking and other crap
	r'//www\.youtube\.com/(ptracking|set_awesome)\?(.*&)?video_id=[0-9A-Za-z_-]{11}',
	r'//www\.youtube\.com/api/timedtext\?(.*&)?v=[0-9A-Za-z_-]{11}',
	r'//www\.youtube\.com/(my_videos_)?edit\?(.*&)?video_id=[0-9A-Za-z_-]{11}',
	# Generic v parameter on watch URLs including with percent encoding; this covers e.g. google.com/url?... or the oEmbed
	r'/watch\?(.*&)?v=[0-9A-Za-z_-]{11}',
  ]))

noisePattern = re.compile('|'.join([
	r'^//www\.youtube\.com/s/(desktop|player)/[0-9a-f]+/',
	r'^//www\.youtube\.com/s/gaming/emoji/',
	r'^//www\.youtube\.com/redirect\?event=channel_banner&',
	r'^//www\.youtube\.com/redirect\?(?=(.*&)?event=video_description(&|$))(?!(.*&)?v=)',
	r'^//www\.youtube\.com/yts/',
	r'^//www\.youtube\.com/img/',
	r'^//www\.youtube\.com/youtubei/',
	r'^//www\.youtube\.com/ads/',
	r'^//www\.youtube\.com/(player|iframe)_api\?',
	r'^//www\.youtube\.com/error(_204)?/?\?',
	r'^//www\.youtube\.com/(about|t|howyoutubeworks)[/?]',
	r'^//www\.youtube\.com/results/?\?',
	r'^//www\.youtube\.com/premium/?\?',
	r'^//www\.youtube\.com/?(\?|$)',
	r'^//www\.youtube\.com/embed/("|%22|' r"'|%27" r')(%20)?(\+|%3B)', # JS extraction stuff
	r'^//consent\.(youtube|google)\.com/',
  ]))


matchers = [
	# (pattern, paramSearch, function(match) -> output str or None); returning None stops further processing
	[channelPattern, True, lambda m: 'https://www.youtube.com/' + m[0].split('/', 3)[-1]],
	[videoPattern, True, lambda m: f'https://www.youtube.com/watch?v={m[0][-11:]}'],
	[re.compile(r'//www\.youtube\.com/(playlist|embed(/videoseries|/\+lastest|/playlist)?/?)\?(.*&)?list=UU[0-9A-Za-z_-]+'), True, lambda m: f'https://www.youtube.com/channel/UC{m[0].rsplit("=", 1)[1][2:]}'],
	[re.compile(r'//www\.youtube\.com/(playlist|embed(/videoseries|/\+lastest|/playlist)?/?)\?(.*&)?list=(PL|FL|RD|OL)[0-9A-Za-z_-]+'), True, lambda m: f'https://www.youtube.com/playlist?list={m[0].rsplit("=", 1)[1]}'],
	[re.compile(r'//www\.youtube\.com/embed/?\?(?=(.*&)?listType=user_uploads(&|$))(.*&)?list=[^&]+'), True, lambda m: f'https://www.youtube.com/user/{m[0].rsplit("=", 1)[1]}'],
	[re.compile(r'//www\.youtube\.com/rss/user/[^/]+'), True, lambda m: f'https://www.youtube.com/user/{m[0].rsplit("/", 1)[1]}'],
	[re.compile(r'//www\.youtube\.com/feeds/videos\.xml\?(.*&)?channel_id=UC[0-9A-Za-z_-]+'), True, lambda m: f'https://www.youtube.com/channel/{m[0].rsplit("=", 1)[1]}'],
	[re.compile(r'//www\.youtube\.com(/view_play_list\?p=|/playlist\?(.*&)?list=)[0-9A-F]{16}(?=(&|$))'), True, lambda m: f'https://www.youtube.com/playlist?list=PL{m[0].rsplit("=", 1)[1]}'],
	[re.compile(r'^//i\.ytimg\.com/vi/([0-9A-Za-z_-]{11})/'), False, lambda m: f'https://www.youtube.com/watch?v={m[1]}'],
	[noisePattern, False, lambda m: None],
]

# Compile second pattern for parameters if needed
for e in matchers:
	pattern, paramSearch, f = e
	if paramSearch:
		p2 = pattern.pattern.replace('//', '/{1,2}').replace('/', '(/|%2F)').replace(':', '(:|%3A)').replace(r'\?', r'(\?|%3F)')
		p2 = re.sub(r'(?<!\(\?)=', '(=|%3D)', p2)
		e[1] = re.compile(p2.replace('&', '(&|%26)'), pattern.flags)
	else:
		e[1] = None

# Only one slash before so it still matches inside URLs when slashes were collapsed.
domainPattern = re.compile(r'/(www\.)?youtube\.(com|de|fr|co\.uk|it|es|at|pt|gr|hu|ro|pl|dk|no|se|fi|ee|lt|lv|ru|by|cz|sk|si|rs|hr|ca)/')

for origLine in sys.stdin:
	origLine = origLine.strip()
	line = re.sub(r'^https?://', '//', origLine)
	line = domainPattern.sub('/www.youtube.com/', line)
	if sys.argv[1] == 'massage':
		hadMatches = False
		for pattern1, pattern2, f in matchers:
			patterns = [pattern1]
			if pattern2:
				patterns.append(pattern2)
			results = set()
			for pattern in patterns:
				m = pattern.search(line)
				if m:
					hadMatches = True
					r = f(m)
					if r in results:
						continue
					results.add(r)
					if r is None:
						break
					print(r)
			if None in results:
				break
		if not hadMatches:
			print(origLine, file = sys.stderr)
	elif sys.argv[1] == 'removenonyt':
		if any(x in line for x in ('/www.youtube.com/', '/youtu.be/', '%2Fwww.youtube.com%2F', '%2Fyoutu.be%2F')):
			print(origLine)