From 9974d4613c7d1c632ed4c42dbcea8def6228a611 Mon Sep 17 00:00:00 2001
From: JustAnotherArchivist <JustAnotherArchivist@users.noreply.github.com>
Date: Thu, 26 Nov 2020 03:48:59 +0000
Subject: [PATCH] Stop trying to rewrite patterns for percent encoding

---
 youtube-extract | 16 ++++------------
 1 file changed, 4 insertions(+), 12 deletions(-)

diff --git a/youtube-extract b/youtube-extract
index e96f2da..ed39ce0 100755
--- a/youtube-extract
+++ b/youtube-extract
@@ -103,15 +103,7 @@ matchers = [
 
 # Compile pattern and generate one for parameters if desired
 for e in matchers:
-	pattern, paramSearch, f = e
-	e[0] = re.compile(pattern)
-	if paramSearch:
-		p2 = pattern.replace('//', '/{1,2}').replace('/', '(/|%2F)').replace(r'\?', r'(\?|%3F)')
-		p2 = re.sub(r'(?<!\(\?):', '(:|%3A)', p2)
-		p2 = re.sub(r'(?<!\(\?)=', '(=|%3D)', p2)
-		e[1] = re.compile(p2.replace('&', '(&|%26)'))
-	else:
-		e[1] = None
+	e[0] = re.compile(e[0])
 
 # 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)/')
@@ -121,11 +113,11 @@ for origLine in sys.stdin:
 	line = re.sub(r'^https?://', '//', origLine)
 	line = domainPattern.sub('/www.youtube.com/', line)
 	hadMatches = False
-	for pattern1, pattern2, f in matchers:
+	for pattern, paramSearch, f in matchers:
 		results = set()
-		for m, encoded in itertools.chain(((x, False) for x in pattern1.finditer(line)), ((x, True) for x in pattern2.finditer(line)) if pattern2 else ()):
+		for m in itertools.chain((x for x in pattern.finditer(line)), (x for x in pattern.finditer(percentdecode(line))) if paramSearch else ()):
 			hadMatches = True
-			r = f(m if not encoded else [percentdecode(x) if x else x for x in itertools.chain((m[0],), m.groups())])
+			r = f(m)
 			if r in results:
 				continue
 			results.add(r)