From 4b5de77bdb7765df4797bf068592926285ba709a Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Sergey=20M=E2=80=A4?= Date: Mon, 6 Mar 2017 03:57:46 +0700 Subject: [PATCH] [utils] Process bytestrings in urljoin (closes #12369) --- test/test_utils.py | 3 +++ youtube_dl/utils.py | 7 ++++++- 2 files changed, 9 insertions(+), 1 deletion(-) diff --git a/test/test_utils.py b/test/test_utils.py index aefd94518..173c49514 100644 --- a/test/test_utils.py +++ b/test/test_utils.py @@ -455,6 +455,9 @@ class TestUtil(unittest.TestCase): def test_urljoin(self): self.assertEqual(urljoin('http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt') + self.assertEqual(urljoin(b'http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt') + self.assertEqual(urljoin('http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt') + self.assertEqual(urljoin(b'http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt') self.assertEqual(urljoin('//foo.de/', '/a/b/c.txt'), '//foo.de/a/b/c.txt') self.assertEqual(urljoin('http://foo.de/', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt') self.assertEqual(urljoin('http://foo.de', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt') diff --git a/youtube_dl/utils.py b/youtube_dl/utils.py index 8738aa249..d293c7498 100644 --- a/youtube_dl/utils.py +++ b/youtube_dl/utils.py @@ -1748,11 +1748,16 @@ def base_url(url): def urljoin(base, path): + if isinstance(path, bytes): + path = path.decode('utf-8') if not isinstance(path, compat_str) or not path: return None if re.match(r'^(?:https?:)?//', path): return path - if not isinstance(base, compat_str) or not re.match(r'^(?:https?:)?//', base): + if isinstance(base, bytes): + base = base.decode('utf-8') + if not isinstance(base, compat_str) or not re.match( + r'^(?:https?:)?//', base): return None return compat_urlparse.urljoin(base, path)