Let `--match-filter` reject entries early

[yt-dlp.git] / test / test_utils.py

diff --git a/test/test_utils.py b/test/test_utils.py
index cf541de4ad8b22fa71e334cdeb571ed23b1a03de..dedc598f7b9f37b9c99a834f8ceb98ea689032fd 100644 (file)
--- a/test/test_utils.py
+++ b/test/test_utils.py
@@ -1,4 +1,4 @@
-#!/usr/bin/env python
+#!/usr/bin/env python3
 # coding: utf-8
 
 from __future__ import unicode_literals
@@ -12,6 +12,7 @@
 
 # Various small unit tests
 import io
+import itertools
 import json
 import xml.etree.ElementTree
 
@@ -108,6 +109,7 @@
     cli_bool_option,
     parse_codecs,
     iri_to_uri,
+    LazyList,
 )
 from yt_dlp.compat import (
     compat_chr,
@@ -126,6 +128,7 @@ def test_timeconvert(self):
         self.assertTrue(timeconvert('bougrg') is None)
 
     def test_sanitize_filename(self):
+        self.assertEqual(sanitize_filename(''), '')
         self.assertEqual(sanitize_filename('abc'), 'abc')
         self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
 
@@ -1051,6 +1054,9 @@ def test_js_to_json_edgecases(self):
         on = js_to_json('{ "040": "040" }')
         self.assertEqual(json.loads(on), {'040': '040'})
 
+        on = js_to_json('[1,//{},\n2]')
+        self.assertEqual(json.loads(on), [1, 2])
+
     def test_js_to_json_malformed(self):
         self.assertEqual(js_to_json('42a1'), '42"a1"')
         self.assertEqual(js_to_json('42a-1'), '42"a"-1')
@@ -1201,12 +1207,26 @@ def test_render_table(self):
             '9999 51')
 
     def test_match_str(self):
-        self.assertRaises(ValueError, match_str, 'xy>foobar', {})
+        # 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', {}))
@@ -1214,10 +1234,25 @@ def test_match_str(self):
         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'}))
         self.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'}))
+        self.assertTrue(match_str('y^=foo', {'y': 'foobar42'}))
+        self.assertFalse(match_str('y!^=foo', {'y': 'foobar42'}))
+        self.assertFalse(match_str('y^=bar', {'y': 'foobar42'}))
+        self.assertTrue(match_str('y!^=bar', {'y': 'foobar42'}))
+        self.assertRaises(ValueError, match_str, 'x^=42', {'x': 42})
+        self.assertTrue(match_str('y*=bar', {'y': 'foobar42'}))
+        self.assertFalse(match_str('y!*=bar', {'y': 'foobar42'}))
+        self.assertFalse(match_str('y*=baz', {'y': 'foobar42'}))
+        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'}))
@@ -1230,18 +1265,35 @@ def test_match_str(self):
         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(match_str(
+            r"!is_live & like_count>?100 & description~='(?i)\bcats \& dogs\b'",
+            {'description': 'Raining Cats & Dogs'}))
+
+        # Incomplete
+        self.assertFalse(match_str('id!=foo', {'id': 'foo'}, True))
+        self.assertTrue(match_str('x', {'id': 'foo'}, True))
+        self.assertTrue(match_str('!x', {'id': 'foo'}, True))
+        self.assertFalse(match_str('x', {'id': 'foo'}, False))
 
     def test_parse_dfxp_time_expr(self):
         self.assertEqual(parse_dfxp_time_expr(None), None)
@@ -1524,6 +1576,51 @@ def test_clean_podcast_url(self):
         self.assertEqual(clean_podcast_url('https://www.podtrac.com/pts/redirect.mp3/chtbl.com/track/5899E/traffic.megaphone.fm/HSW7835899191.mp3'), 'https://traffic.megaphone.fm/HSW7835899191.mp3')
         self.assertEqual(clean_podcast_url('https://play.podtrac.com/npr-344098539/edge1.pod.npr.org/anon.npr-podcasts/podcast/npr/waitwait/2020/10/20201003_waitwait_wwdtmpodcast201003-015621a5-f035-4eca-a9a1-7c118d90bc3c.mp3'), 'https://edge1.pod.npr.org/anon.npr-podcasts/podcast/npr/waitwait/2020/10/20201003_waitwait_wwdtmpodcast201003-015621a5-f035-4eca-a9a1-7c118d90bc3c.mp3')
 
+    def test_LazyList(self):
+        it = list(range(10))
+
+        self.assertEqual(list(LazyList(it)), it)
+        self.assertEqual(LazyList(it).exhaust(), it)
+        self.assertEqual(LazyList(it)[5], it[5])
+
+        self.assertEqual(LazyList(it)[5:], it[5:])
+        self.assertEqual(LazyList(it)[:5], it[:5])
+        self.assertEqual(LazyList(it)[::2], it[::2])
+        self.assertEqual(LazyList(it)[1::2], it[1::2])
+        self.assertEqual(LazyList(it)[5::-1], it[5::-1])
+        self.assertEqual(LazyList(it)[6:2:-2], it[6:2:-2])
+        self.assertEqual(LazyList(it)[::-1], it[::-1])
+
+        self.assertTrue(LazyList(it))
+        self.assertFalse(LazyList(range(0)))
+        self.assertEqual(len(LazyList(it)), len(it))
+        self.assertEqual(repr(LazyList(it)), repr(it))
+        self.assertEqual(str(LazyList(it)), str(it))
+
+        self.assertEqual(list(LazyList(it).reverse()), it[::-1])
+        self.assertEqual(list(LazyList(it).reverse()[1:3:7]), it[::-1][1:3:7])
+        self.assertEqual(list(LazyList(it).reverse()[::-1]), it)
+
+    def test_LazyList_laziness(self):
+
+        def test(ll, idx, val, cache):
+            self.assertEqual(ll[idx], val)
+            self.assertEqual(getattr(ll, '_LazyList__cache'), list(cache))
+
+        ll = LazyList(range(10))
+        test(ll, 0, 0, range(1))
+        test(ll, 5, 5, range(6))
+        test(ll, -3, 7, range(10))
+
+        ll = LazyList(range(10)).reverse()
+        test(ll, -1, 0, range(1))
+        test(ll, 3, 6, range(10))
+
+        ll = LazyList(itertools.count())
+        test(ll, 10, 10, range(11))
+        ll.reverse()
+        test(ll, -15, 14, range(15))
+
 
 if __name__ == '__main__':
     unittest.main()