From a047eeb6d2cbbee944e5088f7cba111746f4bb0b Mon Sep 17 00:00:00 2001
From: pukkandan <pukkandan.ytdlp@gmail.com>
Date: Thu, 5 Aug 2021 03:01:23 +0530
Subject: Add regex to `--match-filter`

This does not fully deprecate `--match-title`/`--reject-title`
since `--match-filter` is only checked after the extraction is complete,
while `--match-title` can often be checked from the flat playlist.

Fixes: https://github.com/ytdl-org/youtube-dl/issues/9092, https://github.com/ytdl-org/youtube-dl/issues/23035
---
 test/test_utils.py | 54 ++++++++++++++++++++++++++++++++++++++++++------------
 1 file changed, 42 insertions(+), 12 deletions(-)

(limited to 'test/test_utils.py')

diff --git a/test/test_utils.py b/test/test_utils.py
index 5ac5dedc9..aef59e491 100644
--- a/test/test_utils.py
+++ b/test/test_utils.py
@@ -1207,11 +1207,26 @@ ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
             '9999 51')
 
     def test_match_str(self):
+        # Unary
         self.assertFalse(match_str('xy', {'x': 1200}))
         self.assertTrue(match_str('!xy', {'x': 1200}))
         self.assertTrue(match_str('x', {'x': 1200}))
         self.assertFalse(match_str('!x', {'x': 1200}))
         self.assertTrue(match_str('x', {'x': 0}))
+        self.assertTrue(match_str('is_live', {'is_live': True}))
+        self.assertFalse(match_str('is_live', {'is_live': False}))
+        self.assertFalse(match_str('is_live', {'is_live': None}))
+        self.assertFalse(match_str('is_live', {}))
+        self.assertFalse(match_str('!is_live', {'is_live': True}))
+        self.assertTrue(match_str('!is_live', {'is_live': False}))
+        self.assertTrue(match_str('!is_live', {'is_live': None}))
+        self.assertTrue(match_str('!is_live', {}))
+        self.assertTrue(match_str('title', {'title': 'abc'}))
+        self.assertTrue(match_str('title', {'title': ''}))
+        self.assertFalse(match_str('!title', {'title': 'abc'}))
+        self.assertFalse(match_str('!title', {'title': ''}))
+
+        # Numeric
         self.assertFalse(match_str('x>0', {'x': 0}))
         self.assertFalse(match_str('x>0', {}))
         self.assertTrue(match_str('x>?0', {}))
@@ -1219,6 +1234,8 @@ ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
         self.assertFalse(match_str('x>2K', {'x': 1200}))
         self.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200}))
         self.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200}))
+
+        # String
         self.assertFalse(match_str('y=a212', {'y': 'foobar42'}))
         self.assertTrue(match_str('y=foobar42', {'y': 'foobar42'}))
         self.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'}))
@@ -1234,6 +1251,8 @@ ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
         self.assertTrue(match_str('y!*=baz', {'y': 'foobar42'}))
         self.assertTrue(match_str('y$=42', {'y': 'foobar42'}))
         self.assertFalse(match_str('y$=43', {'y': 'foobar42'}))
+
+        # And
         self.assertFalse(match_str(
             'like_count > 100 & dislike_count <? 50 & description',
             {'like_count': 90, 'description': 'foo'}))
@@ -1246,18 +1265,29 @@ ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
         self.assertFalse(match_str(
             'like_count > 100 & dislike_count <? 50 & description',
             {'like_count': 190, 'dislike_count': 10}))
-        self.assertTrue(match_str('is_live', {'is_live': True}))
-        self.assertFalse(match_str('is_live', {'is_live': False}))
-        self.assertFalse(match_str('is_live', {'is_live': None}))
-        self.assertFalse(match_str('is_live', {}))
-        self.assertFalse(match_str('!is_live', {'is_live': True}))
-        self.assertTrue(match_str('!is_live', {'is_live': False}))
-        self.assertTrue(match_str('!is_live', {'is_live': None}))
-        self.assertTrue(match_str('!is_live', {}))
-        self.assertTrue(match_str('title', {'title': 'abc'}))
-        self.assertTrue(match_str('title', {'title': ''}))
-        self.assertFalse(match_str('!title', {'title': 'abc'}))
-        self.assertFalse(match_str('!title', {'title': ''}))
+
+        # Regex
+        self.assertTrue(match_str(r'x~=\bbar', {'x': 'foo bar'}))
+        self.assertFalse(match_str(r'x~=\bbar.+', {'x': 'foo bar'}))
+        self.assertFalse(match_str(r'x~=^FOO', {'x': 'foo bar'}))
+        self.assertTrue(match_str(r'x~=(?i)^FOO', {'x': 'foo bar'}))
+
+        # Quotes
+        self.assertTrue(match_str(r'x^="foo"', {'x': 'foo "bar"'}))
+        self.assertFalse(match_str(r'x^="foo  "', {'x': 'foo "bar"'}))
+        self.assertFalse(match_str(r'x$="bar"', {'x': 'foo "bar"'}))
+        self.assertTrue(match_str(r'x$=" \"bar\""', {'x': 'foo "bar"'}))
+
+        # Escaping &
+        self.assertFalse(match_str(r'x=foo & bar', {'x': 'foo & bar'}))
+        self.assertTrue(match_str(r'x=foo \& bar', {'x': 'foo & bar'}))
+        self.assertTrue(match_str(r'x=foo \& bar & x^=foo', {'x': 'foo & bar'}))
+        self.assertTrue(match_str(r'x="foo \& bar" & x^=foo', {'x': 'foo & bar'}))
+
+        # Example from docs
+        self.assertTrue(
+            r'!is_live & like_count>?100 & description~=\'(?i)\bcats \& dogs\b\'',
+            {'description': 'Raining Cats & Dogs'})
 
     def test_parse_dfxp_time_expr(self):
         self.assertEqual(parse_dfxp_time_expr(None), None)
-- 
cgit v1.2.3