[utils] Rework URL path munging for ., .. components

* move processing to YoutubeDLHandler
* also process `Location` header for redirect
* use tests from https://github.com/yt-dlp/yt-dlp/pull/7662

2 files changed, 42 insertions, 1 deletions

diff --git a/test/test_http.py b/test/test_http.py
index 793bea359..485c4c6fc 100644
--- a/test/test_http.py
+++ b/test/test_http.py
@@ -180,6 +180,12 @@ class HTTPTestRequestHandler(compat_http_server.BaseHTTPRequestHandler):
             respond()
         elif self.path == '/%c7%9f':
             respond()
+        elif self.path == '/redirect_dotsegments':
+            self.send_response(301)
+            # redirect to /headers but with dot segments before
+            self.send_header('Location', '/a/b/./../../headers')
+            self.send_header('Content-Length', '0')
+            self.end_headers()
         elif self.path.startswith('/redirect_'):
             self._redirect()
         elif self.path.startswith('/method'):
@@ -489,6 +495,14 @@ class TestHTTP(unittest.TestCase):
             self.assertEqual(res.headers.get('Content-Encoding'), 'unsupported')
             self.assertEqual(res.read(), b'raw')
 
+    def test_remove_dot_segments(self):
+        with FakeYDL() as ydl:
+            res = ydl.urlopen(sanitized_Request(self._test_url('a/b/./../../headers')))
+            self.assertEqual(compat_urllib_parse.urlparse(res.geturl()).path, '/headers')
+
+            res = ydl.urlopen(sanitized_Request(self._test_url('redirect_dotsegments')))
+            self.assertEqual(compat_urllib_parse.urlparse(res.geturl()).path, '/headers')
+
 
 def _build_proxy_handler(name):
     class HTTPTestRequestHandler(compat_http_server.BaseHTTPRequestHandler):
diff --git a/test/test_utils.py b/test/test_utils.py
index e83977f29..fdae1f744 100644
--- a/test/test_utils.py
+++ b/test/test_utils.py
@@ -64,6 +64,7 @@ from youtube_dl.utils import (
     parse_age_limit,
     parse_duration,
     parse_filesize,
+    parse_codecs,
     parse_count,
     parse_iso8601,
     parse_resolution,
@@ -114,7 +115,7 @@ from youtube_dl.utils import (
     cli_option,
     cli_valueless_option,
     cli_bool_option,
-    parse_codecs,
+    YoutubeDLHandler,
 )
 from youtube_dl.compat import (
     compat_chr,
@@ -905,6 +906,32 @@ class TestUtil(unittest.TestCase):
         )
         self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
 
+    def test_remove_dot_segments(self):
+
+        def remove_dot_segments(p):
+            q = '' if p.startswith('/') else '/'
+            p = 'http://example.com' + q + p
+            p = compat_urlparse.urlsplit(YoutubeDLHandler._fix_path(p)).path
+            return p[1:] if q else p
+
+        self.assertEqual(remove_dot_segments('/a/b/c/./../../g'), '/a/g')
+        self.assertEqual(remove_dot_segments('mid/content=5/../6'), 'mid/6')
+        self.assertEqual(remove_dot_segments('/ad/../cd'), '/cd')
+        self.assertEqual(remove_dot_segments('/ad/../cd/'), '/cd/')
+        self.assertEqual(remove_dot_segments('/..'), '/')
+        self.assertEqual(remove_dot_segments('/./'), '/')
+        self.assertEqual(remove_dot_segments('/./a'), '/a')
+        self.assertEqual(remove_dot_segments('/abc/./.././d/././e/.././f/./../../ghi'), '/ghi')
+        self.assertEqual(remove_dot_segments('/'), '/')
+        self.assertEqual(remove_dot_segments('/t'), '/t')
+        self.assertEqual(remove_dot_segments('t'), 't')
+        self.assertEqual(remove_dot_segments(''), '')
+        self.assertEqual(remove_dot_segments('/../a/b/c'), '/a/b/c')
+        self.assertEqual(remove_dot_segments('../a'), 'a')
+        self.assertEqual(remove_dot_segments('./a'), 'a')
+        self.assertEqual(remove_dot_segments('.'), '')
+        self.assertEqual(remove_dot_segments('////'), '////')
+
     def test_js_to_json_vars_strings(self):
         self.assertDictEqual(
             json.loads(js_to_json(