From 43a74c5fa5e43ee563b1488634585cb0bf02ea24 Mon Sep 17 00:00:00 2001 From: bashonly Date: Fri, 5 Jul 2024 13:41:30 -0500 Subject: [PATCH 1/5] [core] Address gaps in allowed extensions Adds some extensions missing in 46521096433aceaa41b4caa845bed22ca6f377ce (from yt-dlp/yt-dlp#10362) Authored by: bashonly Co-authored by: dirkf --- youtube_dl/utils.py | 37 ++++++++++++++++++++++++++++++++++--- 1 file changed, 34 insertions(+), 3 deletions(-) diff --git a/youtube_dl/utils.py b/youtube_dl/utils.py index 3ec9d3811..ac1e78002 100644 --- a/youtube_dl/utils.py +++ b/youtube_dl/utils.py @@ -6604,27 +6604,53 @@ class _UnsafeExtensionError(Exception): ), # video MEDIA_EXTENSIONS.video, ( - 'avif', + 'asx', 'ismv', + 'm2t', 'm2ts', + 'm2v', 'm4s', 'mng', + 'mp2v', + 'mp4v', + 'mpe', 'mpeg', + 'mpeg1', + 'mpeg2', + 'mpeg4', + 'mxf', + 'ogm', 'qt', + 'rm', 'swf', 'ts', + 'vob', 'vp9', - 'wvm', ), # audio MEDIA_EXTENSIONS.audio, ( + '3ga', + 'ac3', + 'adts', + 'aif', + 'au', + 'dts', 'isma', + 'it', 'mid', + 'mod', 'mpga', + 'mp1', + 'mp2', + 'mp4a', + 'mpa', 'ra', + 'shn', + 'xm', ), # image MEDIA_EXTENSIONS.thumbnails, ( + 'avif', 'bmp', 'gif', 'ico', @@ -6634,6 +6660,7 @@ class _UnsafeExtensionError(Exception): 'jxl', 'svg', 'tif', + 'tiff', 'wbmp', ), # subtitle @@ -6641,10 +6668,15 @@ class _UnsafeExtensionError(Exception): 'dfxp', 'fs', 'ismt', + 'json3', 'sami', 'scc', + 'srv1', + 'srv2', + 'srv3', 'ssa', 'tt', + 'xml', ), # others MEDIA_EXTENSIONS.manifests, @@ -6658,7 +6690,6 @@ class _UnsafeExtensionError(Exception): # 'swp', # 'url', # 'webloc', - # 'xml', ))) def __init__(self, extension): From 756f6b45c7c9a77f0fa912f882afe7e04878a01b Mon Sep 17 00:00:00 2001 From: dirkf Date: Wed, 10 Jul 2024 17:58:38 +0100 Subject: [PATCH 2/5] [jsinterp] Re-align JSInterp and tests (esp.) with yt-dlp Thx: various yt-dlp authors --- test/test_jsinterp.py | 675 +++++++++++++++-------------------------- youtube_dl/jsinterp.py | 53 ++-- 2 files changed, 268 insertions(+), 460 deletions(-) diff --git a/test/test_jsinterp.py b/test/test_jsinterp.py index da8e98020..86d7f0d46 100644 --- a/test/test_jsinterp.py +++ b/test/test_jsinterp.py @@ -11,194 +11,146 @@ sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__)))) import math import re +from youtube_dl.compat import compat_str from youtube_dl.jsinterp import JS_Undefined, JSInterpreter +NaN = object() + class TestJSInterpreter(unittest.TestCase): + def _test(self, jsi_or_code, expected, func='f', args=()): + if isinstance(jsi_or_code, compat_str): + jsi_or_code = JSInterpreter(jsi_or_code) + got = jsi_or_code.call_function(func, *args) + if expected is NaN: + self.assertTrue(math.isnan(got), '{0} is not NaN'.format(got)) + else: + self.assertEqual(got, expected) + def test_basic(self): - jsi = JSInterpreter('function x(){;}') - self.assertEqual(jsi.call_function('x'), None) - self.assertEqual(repr(jsi.extract_function('x')), 'F') + jsi = JSInterpreter('function f(){;}') + self.assertEqual(repr(jsi.extract_function('f')), 'F') + self._test(jsi, None) - jsi = JSInterpreter('function x3(){return 42;}') - self.assertEqual(jsi.call_function('x3'), 42) - - jsi = JSInterpreter('function x3(){42}') - self.assertEqual(jsi.call_function('x3'), None) - - jsi = JSInterpreter('var x5 = function(){return 42;}') - self.assertEqual(jsi.call_function('x5'), 42) - - def test_calc(self): - jsi = JSInterpreter('function x4(a){return 2*a+1;}') - self.assertEqual(jsi.call_function('x4', 3), 7) + self._test('function f(){return 42;}', 42) + self._test('function f(){42}', None) + self._test('var f = function(){return 42;}', 42) def test_add(self): - jsi = JSInterpreter('function f(){return 42 + 7;}') - self.assertEqual(jsi.call_function('f'), 49) - jsi = JSInterpreter('function f(){return 42 + undefined;}') - self.assertTrue(math.isnan(jsi.call_function('f'))) - jsi = JSInterpreter('function f(){return 42 + null;}') - self.assertEqual(jsi.call_function('f'), 42) + self._test('function f(){return 42 + 7;}', 49) + self._test('function f(){return 42 + undefined;}', NaN) + self._test('function f(){return 42 + null;}', 42) def test_sub(self): - jsi = JSInterpreter('function f(){return 42 - 7;}') - self.assertEqual(jsi.call_function('f'), 35) - jsi = JSInterpreter('function f(){return 42 - undefined;}') - self.assertTrue(math.isnan(jsi.call_function('f'))) - jsi = JSInterpreter('function f(){return 42 - null;}') - self.assertEqual(jsi.call_function('f'), 42) + self._test('function f(){return 42 - 7;}', 35) + self._test('function f(){return 42 - undefined;}', NaN) + self._test('function f(){return 42 - null;}', 42) def test_mul(self): - jsi = JSInterpreter('function f(){return 42 * 7;}') - self.assertEqual(jsi.call_function('f'), 294) - jsi = JSInterpreter('function f(){return 42 * undefined;}') - self.assertTrue(math.isnan(jsi.call_function('f'))) - jsi = JSInterpreter('function f(){return 42 * null;}') - self.assertEqual(jsi.call_function('f'), 0) + self._test('function f(){return 42 * 7;}', 294) + self._test('function f(){return 42 * undefined;}', NaN) + self._test('function f(){return 42 * null;}', 0) def test_div(self): jsi = JSInterpreter('function f(a, b){return a / b;}') - self.assertTrue(math.isnan(jsi.call_function('f', 0, 0))) - self.assertTrue(math.isnan(jsi.call_function('f', JS_Undefined, 1))) - self.assertTrue(math.isinf(jsi.call_function('f', 2, 0))) - self.assertEqual(jsi.call_function('f', 0, 3), 0) + self._test(jsi, NaN, args=(0, 0)) + self._test(jsi, NaN, args=(JS_Undefined, 1)) + self._test(jsi, float('inf'), args=(2, 0)) + self._test(jsi, 0, args=(0, 3)) def test_mod(self): - jsi = JSInterpreter('function f(){return 42 % 7;}') - self.assertEqual(jsi.call_function('f'), 0) - jsi = JSInterpreter('function f(){return 42 % 0;}') - self.assertTrue(math.isnan(jsi.call_function('f'))) - jsi = JSInterpreter('function f(){return 42 % undefined;}') - self.assertTrue(math.isnan(jsi.call_function('f'))) + self._test('function f(){return 42 % 7;}', 0) + self._test('function f(){return 42 % 0;}', NaN) + self._test('function f(){return 42 % undefined;}', NaN) def test_exp(self): - jsi = JSInterpreter('function f(){return 42 ** 2;}') - self.assertEqual(jsi.call_function('f'), 1764) - jsi = JSInterpreter('function f(){return 42 ** undefined;}') - self.assertTrue(math.isnan(jsi.call_function('f'))) - jsi = JSInterpreter('function f(){return 42 ** null;}') - self.assertEqual(jsi.call_function('f'), 1) - jsi = JSInterpreter('function f(){return undefined ** 42;}') - self.assertTrue(math.isnan(jsi.call_function('f'))) + self._test('function f(){return 42 ** 2;}', 1764) + self._test('function f(){return 42 ** undefined;}', NaN) + self._test('function f(){return 42 ** null;}', 1) + self._test('function f(){return undefined ** 42;}', NaN) + + def test_calc(self): + self._test('function f(a){return 2*a+1;}', 7, args=[3]) def test_empty_return(self): - jsi = JSInterpreter('function f(){return; y()}') - self.assertEqual(jsi.call_function('f'), None) + self._test('function f(){return; y()}', None) def test_morespace(self): - jsi = JSInterpreter('function x (a) { return 2 * a + 1 ; }') - self.assertEqual(jsi.call_function('x', 3), 7) - - jsi = JSInterpreter('function f () { x = 2 ; return x; }') - self.assertEqual(jsi.call_function('f'), 2) + self._test('function f (a) { return 2 * a + 1 ; }', 7, args=[3]) + self._test('function f () { x = 2 ; return x; }', 2) def test_strange_chars(self): - jsi = JSInterpreter('function $_xY1 ($_axY1) { var $_axY2 = $_axY1 + 1; return $_axY2; }') - self.assertEqual(jsi.call_function('$_xY1', 20), 21) + self._test('function $_xY1 ($_axY1) { var $_axY2 = $_axY1 + 1; return $_axY2; }', + 21, args=[20], func='$_xY1') def test_operators(self): - jsi = JSInterpreter('function f(){return 1 << 5;}') - self.assertEqual(jsi.call_function('f'), 32) - - jsi = JSInterpreter('function f(){return 2 ** 5}') - self.assertEqual(jsi.call_function('f'), 32) - - jsi = JSInterpreter('function f(){return 19 & 21;}') - self.assertEqual(jsi.call_function('f'), 17) - - jsi = JSInterpreter('function f(){return 11 >> 2;}') - self.assertEqual(jsi.call_function('f'), 2) - - jsi = JSInterpreter('function f(){return []? 2+3: 4;}') - self.assertEqual(jsi.call_function('f'), 5) - - jsi = JSInterpreter('function f(){return 1 == 2}') - self.assertEqual(jsi.call_function('f'), False) - - jsi = JSInterpreter('function f(){return 0 && 1 || 2;}') - self.assertEqual(jsi.call_function('f'), 2) - - jsi = JSInterpreter('function f(){return 0 ?? 42;}') - self.assertEqual(jsi.call_function('f'), 0) - - jsi = JSInterpreter('function f(){return "life, the universe and everything" < 42;}') - self.assertFalse(jsi.call_function('f')) + self._test('function f(){return 1 << 5;}', 32) + self._test('function f(){return 2 ** 5}', 32) + self._test('function f(){return 19 & 21;}', 17) + self._test('function f(){return 11 >> 2;}', 2) + self._test('function f(){return []? 2+3: 4;}', 5) + self._test('function f(){return 1 == 2}', False) + self._test('function f(){return 0 && 1 || 2;}', 2) + self._test('function f(){return 0 ?? 42;}', 0) + self._test('function f(){return "life, the universe and everything" < 42;}', False) + # https://github.com/ytdl-org/youtube-dl/issues/32815 + self._test('function f(){return 0 - 7 * - 6;}', 42) def test_array_access(self): - jsi = JSInterpreter('function f(){var x = [1,2,3]; x[0] = 4; x[0] = 5; x[2.0] = 7; return x;}') - self.assertEqual(jsi.call_function('f'), [5, 2, 7]) + self._test('function f(){var x = [1,2,3]; x[0] = 4; x[0] = 5; x[2.0] = 7; return x;}', [5, 2, 7]) def test_parens(self): - jsi = JSInterpreter('function f(){return (1) + (2) * ((( (( (((((3)))))) )) ));}') - self.assertEqual(jsi.call_function('f'), 7) - - jsi = JSInterpreter('function f(){return (1 + 2) * 3;}') - self.assertEqual(jsi.call_function('f'), 9) + self._test('function f(){return (1) + (2) * ((( (( (((((3)))))) )) ));}', 7) + self._test('function f(){return (1 + 2) * 3;}', 9) def test_quotes(self): - jsi = JSInterpreter(r'function f(){return "a\"\\("}') - self.assertEqual(jsi.call_function('f'), r'a"\(') + self._test(r'function f(){return "a\"\\("}', r'a"\(') def test_assignments(self): - jsi = JSInterpreter('function f(){var x = 20; x = 30 + 1; return x;}') - self.assertEqual(jsi.call_function('f'), 31) - - jsi = JSInterpreter('function f(){var x = 20; x += 30 + 1; return x;}') - self.assertEqual(jsi.call_function('f'), 51) - - jsi = JSInterpreter('function f(){var x = 20; x -= 30 + 1; return x;}') - self.assertEqual(jsi.call_function('f'), -11) + self._test('function f(){var x = 20; x = 30 + 1; return x;}', 31) + self._test('function f(){var x = 20; x += 30 + 1; return x;}', 51) + self._test('function f(){var x = 20; x -= 30 + 1; return x;}', -11) + @unittest.skip('Not yet fully implemented') def test_comments(self): - 'Skipping: Not yet fully implemented' - return - jsi = JSInterpreter(''' - function x() { - var x = /* 1 + */ 2; - var y = /* 30 - * 40 */ 50; - return x + y; - } - ''') - self.assertEqual(jsi.call_function('x'), 52) + self._test(''' + function f() { + var x = /* 1 + */ 2; + var y = /* 30 + * 40 */ 50; + return x + y; + } + ''', 52) - jsi = JSInterpreter(''' - function f() { - var x = "/*"; - var y = 1 /* comment */ + 2; - return y; - } - ''') - self.assertEqual(jsi.call_function('f'), 3) + self._test(''' + function f() { + var x = "/*"; + var y = 1 /* comment */ + 2; + return y; + } + ''', 3) def test_precedence(self): - jsi = JSInterpreter(''' - function x() { - var a = [10, 20, 30, 40, 50]; - var b = 6; - a[0]=a[b%a.length]; - return a; - }''') - self.assertEqual(jsi.call_function('x'), [20, 20, 30, 40, 50]) + self._test(''' + function f() { + var a = [10, 20, 30, 40, 50]; + var b = 6; + a[0]=a[b%a.length]; + return a; + } + ''', [20, 20, 30, 40, 50]) def test_builtins(self): - jsi = JSInterpreter(''' - function x() { return NaN } - ''') - self.assertTrue(math.isnan(jsi.call_function('x'))) + self._test('function f() { return NaN }', NaN) def test_Date(self): - jsi = JSInterpreter(''' - function x(dt) { return new Date(dt) - 0; } - ''') - self.assertEqual(jsi.call_function('x', 'Wednesday 31 December 1969 18:01:26 MDT'), 86000) + self._test('function f() { return new Date("Wednesday 31 December 1969 18:01:26 MDT") - 0; }', 86000) + jsi = JSInterpreter('function f(dt) { return new Date(dt) - 0; }') # date format m/d/y - self.assertEqual(jsi.call_function('x', '12/31/1969 18:01:26 MDT'), 86000) - + self._test(jsi, 86000, args=['12/31/1969 18:01:26 MDT']) # epoch 0 - self.assertEqual(jsi.call_function('x', '1 January 1970 00:00:00 UTC'), 0) + self._test(jsi, 0, args=['1 January 1970 00:00:00 UTC']) def test_call(self): jsi = JSInterpreter(''' @@ -206,179 +158,115 @@ class TestJSInterpreter(unittest.TestCase): function y(a) { return x() + (a?a:0); } function z() { return y(3); } ''') - self.assertEqual(jsi.call_function('z'), 5) - self.assertEqual(jsi.call_function('y'), 2) + self._test(jsi, 5, func='z') + self._test(jsi, 2, func='y') def test_if(self): - jsi = JSInterpreter(''' - function x() { + self._test(''' + function f() { let a = 9; if (0==0) {a++} return a - }''') - self.assertEqual(jsi.call_function('x'), 10) + } + ''', 10) - jsi = JSInterpreter(''' - function x() { + self._test(''' + function f() { if (0==0) {return 10} - }''') - self.assertEqual(jsi.call_function('x'), 10) + } + ''', 10) - jsi = JSInterpreter(''' - function x() { + self._test(''' + function f() { if (0!=0) {return 1} else {return 10} - }''') - self.assertEqual(jsi.call_function('x'), 10) - - """ # Unsupported - jsi = JSInterpreter(''' - function x() { - if (0!=0) return 1; - else {return 10} - }''') - self.assertEqual(jsi.call_function('x'), 10) - """ + } + ''', 10) def test_elseif(self): - jsi = JSInterpreter(''' - function x() { - if (0!=0) {return 1} - else if (1==0) {return 2} - else {return 10} - }''') - self.assertEqual(jsi.call_function('x'), 10) - - """ # Unsupported - jsi = JSInterpreter(''' - function x() { - if (0!=0) return 1; - else if (1==0) {return 2} - else {return 10} - }''') - self.assertEqual(jsi.call_function('x'), 10) - # etc - """ + self._test(''' + function f() { + if (0!=0) {return 1} + else if (1==0) {return 2} + else {return 10} + } + ''', 10) def test_for_loop(self): - # function x() { a=0; for (i=0; i-10; i++) {a++} a } - jsi = JSInterpreter(''' - function x() { a=0; for (i=0; i-10; i++) {a++} return a } - ''') - self.assertEqual(jsi.call_function('x'), 10) + self._test('function f() { a=0; for (i=0; i-10; i++) {a++} return a }', 10) def test_while_loop(self): - # function x() { a=0; while (a<10) {a++} a } - jsi = JSInterpreter(''' - function x() { a=0; while (a<10) {a++} return a } - ''') - self.assertEqual(jsi.call_function('x'), 10) + self._test('function f() { a=0; while (a<10) {a++} return a }', 10) def test_switch(self): jsi = JSInterpreter(''' - function x(f) { switch(f){ - case 1:f+=1; - case 2:f+=2; - case 3:f+=3;break; - case 4:f+=4; - default:f=0; - } return f } + function f(x) { switch(x){ + case 1:x+=1; + case 2:x+=2; + case 3:x+=3;break; + case 4:x+=4; + default:x=0; + } return x } ''') - self.assertEqual(jsi.call_function('x', 1), 7) - self.assertEqual(jsi.call_function('x', 3), 6) - self.assertEqual(jsi.call_function('x', 5), 0) + self._test(jsi, 7, args=[1]) + self._test(jsi, 6, args=[3]) + self._test(jsi, 0, args=[5]) def test_switch_default(self): jsi = JSInterpreter(''' - function x(f) { switch(f){ - case 2: f+=2; - default: f-=1; - case 5: - case 6: f+=6; - case 0: break; - case 1: f+=1; - } return f } + function f(x) { switch(x){ + case 2: x+=2; + default: x-=1; + case 5: + case 6: x+=6; + case 0: break; + case 1: x+=1; + } return x } ''') - self.assertEqual(jsi.call_function('x', 1), 2) - self.assertEqual(jsi.call_function('x', 5), 11) - self.assertEqual(jsi.call_function('x', 9), 14) + self._test(jsi, 2, args=[1]) + self._test(jsi, 11, args=[5]) + self._test(jsi, 14, args=[9]) def test_try(self): - jsi = JSInterpreter(''' - function x() { try{return 10} catch(e){return 5} } - ''') - self.assertEqual(jsi.call_function('x'), 10) + self._test('function f() { try{return 10} catch(e){return 5} }', 10) def test_catch(self): - jsi = JSInterpreter(''' - function x() { try{throw 10} catch(e){return 5} } - ''') - self.assertEqual(jsi.call_function('x'), 5) + self._test('function f() { try{throw 10} catch(e){return 5} }', 5) def test_finally(self): - jsi = JSInterpreter(''' - function x() { try{throw 10} finally {return 42} } - ''') - self.assertEqual(jsi.call_function('x'), 42) - jsi = JSInterpreter(''' - function x() { try{throw 10} catch(e){return 5} finally {return 42} } - ''') - self.assertEqual(jsi.call_function('x'), 42) + self._test('function f() { try{throw 10} finally {return 42} }', 42) + self._test('function f() { try{throw 10} catch(e){return 5} finally {return 42} }', 42) def test_nested_try(self): - jsi = JSInterpreter(''' - function x() {try { - try{throw 10} finally {throw 42} + self._test(''' + function f() {try { + try{throw 10} finally {throw 42} } catch(e){return 5} } - ''') - self.assertEqual(jsi.call_function('x'), 5) + ''', 5) def test_for_loop_continue(self): - jsi = JSInterpreter(''' - function x() { a=0; for (i=0; i-10; i++) { continue; a++ } return a } - ''') - self.assertEqual(jsi.call_function('x'), 0) + self._test('function f() { a=0; for (i=0; i-10; i++) { continue; a++ } return a }', 0) def test_for_loop_break(self): - jsi = JSInterpreter(''' - function x() { a=0; for (i=0; i-10; i++) { break; a++ } return a } - ''') - self.assertEqual(jsi.call_function('x'), 0) + self._test('function f() { a=0; for (i=0; i-10; i++) { break; a++ } return a }', 0) def test_for_loop_try(self): - jsi = JSInterpreter(''' - function x() { - for (i=0; i-10; i++) { try { if (i == 5) throw i} catch {return 10} finally {break} }; - return 42 } - ''') - self.assertEqual(jsi.call_function('x'), 42) + self._test(''' + function f() { + for (i=0; i-10; i++) { try { if (i == 5) throw i} catch {return 10} finally {break} }; + return 42 } + ''', 42) def test_literal_list(self): - jsi = JSInterpreter(''' - function x() { return [1, 2, "asdf", [5, 6, 7]][3] } - ''') - self.assertEqual(jsi.call_function('x'), [5, 6, 7]) + self._test('function f() { return [1, 2, "asdf", [5, 6, 7]][3] }', [5, 6, 7]) def test_comma(self): - jsi = JSInterpreter(''' - function x() { a=5; a -= 1, a+=3; return a } - ''') - self.assertEqual(jsi.call_function('x'), 7) - jsi = JSInterpreter(''' - function x() { a=5; return (a -= 1, a+=3, a); } - ''') - self.assertEqual(jsi.call_function('x'), 7) - - jsi = JSInterpreter(''' - function x() { return (l=[0,1,2,3], function(a, b){return a+b})((l[1], l[2]), l[3]) } - ''') - self.assertEqual(jsi.call_function('x'), 5) + self._test('function f() { a=5; a -= 1, a+=3; return a }', 7) + self._test('function f() { a=5; return (a -= 1, a+=3, a); }', 7) + self._test('function f() { return (l=[0,1,2,3], function(a, b){return a+b})((l[1], l[2]), l[3]) }', 5) def test_void(self): - jsi = JSInterpreter(''' - function x() { return void 42; } - ''') - self.assertEqual(jsi.call_function('x'), None) + self._test('function f() { return void 42; }', None) def test_return_function(self): jsi = JSInterpreter(''' @@ -387,110 +275,60 @@ class TestJSInterpreter(unittest.TestCase): self.assertEqual(jsi.call_function('x')([]), 1) def test_null(self): - jsi = JSInterpreter(''' - function x() { return null; } - ''') - self.assertIs(jsi.call_function('x'), None) - - jsi = JSInterpreter(''' - function x() { return [null > 0, null < 0, null == 0, null === 0]; } - ''') - self.assertEqual(jsi.call_function('x'), [False, False, False, False]) - - jsi = JSInterpreter(''' - function x() { return [null >= 0, null <= 0]; } - ''') - self.assertEqual(jsi.call_function('x'), [True, True]) + self._test('function f() { return null; }', None) + self._test('function f() { return [null > 0, null < 0, null == 0, null === 0]; }', + [False, False, False, False]) + self._test('function f() { return [null >= 0, null <= 0]; }', [True, True]) def test_undefined(self): - jsi = JSInterpreter(''' - function x() { return undefined === undefined; } - ''') - self.assertTrue(jsi.call_function('x')) + self._test('function f() { return undefined === undefined; }', True) + self._test('function f() { return undefined; }', JS_Undefined) + self._test('function f() {return undefined ?? 42; }', 42) + self._test('function f() { let v; return v; }', JS_Undefined) + self._test('function f() { let v; return v**0; }', 1) + self._test('function f() { let v; return [v>42, v<=42, v&&42, 42&&v]; }', + [False, False, JS_Undefined, JS_Undefined]) + + self._test(''' + function f() { return [ + undefined === undefined, + undefined == undefined, + undefined == null + ]; } + ''', [True] * 3) + self._test(''' + function f() { return [ + undefined < undefined, + undefined > undefined, + undefined === 0, + undefined == 0, + undefined < 0, + undefined > 0, + undefined >= 0, + undefined <= 0, + undefined > null, + undefined < null, + undefined === null + ]; } + ''', [False] * 11) jsi = JSInterpreter(''' - function x() { return undefined; } - ''') - self.assertIs(jsi.call_function('x'), JS_Undefined) - - jsi = JSInterpreter(''' - function x() { let v; return v; } - ''') - self.assertIs(jsi.call_function('x'), JS_Undefined) - - jsi = JSInterpreter(''' - function x() { return [undefined === undefined, undefined == undefined, undefined < undefined, undefined > undefined]; } - ''') - self.assertEqual(jsi.call_function('x'), [True, True, False, False]) - - jsi = JSInterpreter(''' - function x() { return [undefined === 0, undefined == 0, undefined < 0, undefined > 0]; } - ''') - self.assertEqual(jsi.call_function('x'), [False, False, False, False]) - - jsi = JSInterpreter(''' - function x() { return [undefined >= 0, undefined <= 0]; } - ''') - self.assertEqual(jsi.call_function('x'), [False, False]) - - jsi = JSInterpreter(''' - function x() { return [undefined > null, undefined < null, undefined == null, undefined === null]; } - ''') - self.assertEqual(jsi.call_function('x'), [False, False, True, False]) - - jsi = JSInterpreter(''' - function x() { return [undefined === null, undefined == null, undefined < null, undefined > null]; } - ''') - self.assertEqual(jsi.call_function('x'), [False, True, False, False]) - - jsi = JSInterpreter(''' - function x() { let v; return [42+v, v+42, v**42, 42**v, 0**v]; } + function x() { let v; return [42+v, v+42, v**42, 42**v, 0**v]; } ''') for y in jsi.call_function('x'): self.assertTrue(math.isnan(y)) - jsi = JSInterpreter(''' - function x() { let v; return v**0; } - ''') - self.assertEqual(jsi.call_function('x'), 1) - - jsi = JSInterpreter(''' - function x() { let v; return [v>42, v<=42, v&&42, 42&&v]; } - ''') - self.assertEqual(jsi.call_function('x'), [False, False, JS_Undefined, JS_Undefined]) - - jsi = JSInterpreter('function x(){return undefined ?? 42; }') - self.assertEqual(jsi.call_function('x'), 42) - def test_object(self): - jsi = JSInterpreter(''' - function x() { return {}; } - ''') - self.assertEqual(jsi.call_function('x'), {}) - - jsi = JSInterpreter(''' - function x() { let a = {m1: 42, m2: 0 }; return [a["m1"], a.m2]; } - ''') - self.assertEqual(jsi.call_function('x'), [42, 0]) - - jsi = JSInterpreter(''' - function x() { let a; return a?.qq; } - ''') - self.assertIs(jsi.call_function('x'), JS_Undefined) - - jsi = JSInterpreter(''' - function x() { let a = {m1: 42, m2: 0 }; return a?.qq; } - ''') - self.assertIs(jsi.call_function('x'), JS_Undefined) + self._test('function f() { return {}; }', {}) + self._test('function f() { let a = {m1: 42, m2: 0 }; return [a["m1"], a.m2]; }', [42, 0]) + self._test('function f() { let a; return a?.qq; }', JS_Undefined) + self._test('function f() { let a = {m1: 42, m2: 0 }; return a?.qq; }', JS_Undefined) def test_regex(self): - jsi = JSInterpreter(''' - function x() { let a=/,,[/,913,/](,)}/; } - ''') - self.assertIs(jsi.call_function('x'), None) + self._test('function f() { let a=/,,[/,913,/](,)}/; }', None) jsi = JSInterpreter(''' - function x() { let a=/,,[/,913,/](,)}/; "".replace(a, ""); return a; } + function x() { let a=/,,[/,913,/](,)}/; "".replace(a, ""); return a; } ''') attrs = set(('findall', 'finditer', 'match', 'scanner', 'search', 'split', 'sub', 'subn')) @@ -500,94 +338,65 @@ class TestJSInterpreter(unittest.TestCase): self.assertSetEqual(set(dir(jsi.call_function('x'))) & attrs, attrs) jsi = JSInterpreter(''' - function x() { let a=/,,[/,913,/](,)}/i; return a; } + function x() { let a=/,,[/,913,/](,)}/i; return a; } ''') self.assertEqual(jsi.call_function('x').flags & ~re.U, re.I) - jsi = JSInterpreter(r''' - function x() { let a="data-name".replace("data-", ""); return a } - ''') - self.assertEqual(jsi.call_function('x'), 'name') + jsi = JSInterpreter(r'function f() { let a=/,][}",],()}(\[)/; return a; }') + self.assertEqual(jsi.call_function('f').pattern, r',][}",],()}(\[)') - jsi = JSInterpreter(r''' - function x() { let a="data-name".replace(new RegExp("^.+-"), ""); return a; } - ''') - self.assertEqual(jsi.call_function('x'), 'name') + jsi = JSInterpreter(r'function f() { let a=[/[)\\]/]; return a[0]; }') + self.assertEqual(jsi.call_function('f').pattern, r'[)\\]') - jsi = JSInterpreter(r''' - function x() { let a="data-name".replace(/^.+-/, ""); return a; } - ''') - self.assertEqual(jsi.call_function('x'), 'name') - - jsi = JSInterpreter(r''' - function x() { let a="data-name".replace(/a/g, "o"); return a; } - ''') - self.assertEqual(jsi.call_function('x'), 'doto-nome') - - jsi = JSInterpreter(r''' - function x() { let a="data-name".replaceAll("a", "o"); return a; } - ''') - self.assertEqual(jsi.call_function('x'), 'doto-nome') - - jsi = JSInterpreter(r''' - function x() { let a=[/[)\\]/]; return a[0]; } - ''') - self.assertEqual(jsi.call_function('x').pattern, r'[)\\]') - - """ # fails - jsi = JSInterpreter(r''' - function x() { let a=100; a/=/[0-9]+/.exec('divide by 20 today')[0]; } - ''') - self.assertEqual(jsi.call_function('x'), 5) - """ + def test_replace(self): + self._test('function f() { let a="data-name".replace("data-", ""); return a }', + 'name') + self._test('function f() { let a="data-name".replace(new RegExp("^.+-"), ""); return a; }', + 'name') + self._test('function f() { let a="data-name".replace(/^.+-/, ""); return a; }', + 'name') + self._test('function f() { let a="data-name".replace(/a/g, "o"); return a; }', + 'doto-nome') + self._test('function f() { let a="data-name".replaceAll("a", "o"); return a; }', + 'doto-nome') def test_char_code_at(self): - jsi = JSInterpreter('function x(i){return "test".charCodeAt(i)}') - self.assertEqual(jsi.call_function('x', 0), 116) - self.assertEqual(jsi.call_function('x', 1), 101) - self.assertEqual(jsi.call_function('x', 2), 115) - self.assertEqual(jsi.call_function('x', 3), 116) - self.assertEqual(jsi.call_function('x', 4), None) - self.assertEqual(jsi.call_function('x', 'not_a_number'), 116) + jsi = JSInterpreter('function f(i){return "test".charCodeAt(i)}') + self._test(jsi, 116, args=[0]) + self._test(jsi, 101, args=[1]) + self._test(jsi, 115, args=[2]) + self._test(jsi, 116, args=[3]) + self._test(jsi, None, args=[4]) + self._test(jsi, 116, args=['not_a_number']) def test_bitwise_operators_overflow(self): - jsi = JSInterpreter('function x(){return -524999584 << 5}') - self.assertEqual(jsi.call_function('x'), 379882496) + self._test('function f(){return -524999584 << 5}', 379882496) + self._test('function f(){return 1236566549 << 5}', 915423904) - jsi = JSInterpreter('function x(){return 1236566549 << 5}') - self.assertEqual(jsi.call_function('x'), 915423904) + def test_bitwise_operators_typecast(self): + # madness + self._test('function f(){return null << 5}', 0) + self._test('function f(){return undefined >> 5}', 0) + self._test('function f(){return 42 << NaN}', 42) + self._test('function f(){return 42 << Infinity}', 42) - def test_bitwise_operators_madness(self): - jsi = JSInterpreter('function x(){return null << 5}') - self.assertEqual(jsi.call_function('x'), 0) - - jsi = JSInterpreter('function x(){return undefined >> 5}') - self.assertEqual(jsi.call_function('x'), 0) - - jsi = JSInterpreter('function x(){return 42 << NaN}') - self.assertEqual(jsi.call_function('x'), 42) - - jsi = JSInterpreter('function x(){return 42 << Infinity}') - self.assertEqual(jsi.call_function('x'), 42) + def test_negative(self): + self._test('function f(){return 2 * -2.0 ;}', -4) + self._test('function f(){return 2 - - -2 ;}', 0) + self._test('function f(){return 2 - - - -2 ;}', 4) + self._test('function f(){return 2 - + + - -2;}', 0) + self._test('function f(){return 2 + - + - -2;}', 0) def test_32066(self): - jsi = JSInterpreter("function x(){return Math.pow(3, 5) + new Date('1970-01-01T08:01:42.000+08:00') / 1000 * -239 - -24205;}") - self.assertEqual(jsi.call_function('x'), 70) + self._test( + "function f(){return Math.pow(3, 5) + new Date('1970-01-01T08:01:42.000+08:00') / 1000 * -239 - -24205;}", + 70) - def test_unary_operators(self): - jsi = JSInterpreter('function f(){return 2 - - - 2;}') - self.assertEqual(jsi.call_function('f'), 0) - jsi = JSInterpreter('function f(){return 2 + - + - - 2;}') - self.assertEqual(jsi.call_function('f'), 0) - # https://github.com/ytdl-org/youtube-dl/issues/32815 - jsi = JSInterpreter('function f(){return 0 - 7 * - 6;}') - self.assertEqual(jsi.call_function('f'), 42) - - """ # fails so far + @unittest.skip('Not yet working') def test_packed(self): - jsi = JSInterpreter('''function x(p,a,c,k,e,d){while(c--)if(k[c])p=p.replace(new RegExp('\\b'+c.toString(a)+'\\b','g'),k[c]);return p}''') - self.assertEqual(jsi.call_function('x', '''h 7=g("1j");7.7h({7g:[{33:"w://7f-7e-7d-7c.v.7b/7a/79/78/77/76.74?t=73&s=2s&e=72&f=2t&71=70.0.0.1&6z=6y&6x=6w"}],6v:"w://32.v.u/6u.31",16:"r%",15:"r%",6t:"6s",6r:"",6q:"l",6p:"l",6o:"6n",6m:\'6l\',6k:"6j",9:[{33:"/2u?b=6i&n=50&6h=w://32.v.u/6g.31",6f:"6e"}],1y:{6d:1,6c:\'#6b\',6a:\'#69\',68:"67",66:30,65:r,},"64":{63:"%62 2m%m%61%5z%5y%5x.u%5w%5v%5u.2y%22 2k%m%1o%22 5t%m%1o%22 5s%m%1o%22 2j%m%5r%22 16%m%5q%22 15%m%5p%22 5o%2z%5n%5m%2z",5l:"w://v.u/d/1k/5k.2y",5j:[]},\'5i\':{"5h":"5g"},5f:"5e",5d:"w://v.u",5c:{},5b:l,1x:[0.25,0.50,0.75,1,1.25,1.5,2]});h 1m,1n,5a;h 59=0,58=0;h 7=g("1j");h 2x=0,57=0,56=0;$.55({54:{\'53-52\':\'2i-51\'}});7.j(\'4z\',6(x){c(5>0&&x.1l>=5&&1n!=1){1n=1;$(\'q.4y\').4x(\'4w\')}});7.j(\'13\',6(x){2x=x.1l});7.j(\'2g\',6(x){2w(x)});7.j(\'4v\',6(){$(\'q.2v\').4u()});6 2w(x){$(\'q.2v\').4t();c(1m)19;1m=1;17=0;c(4s.4r===l){17=1}$.4q(\'/2u?b=4p&2l=1k&4o=2t-4n-4m-2s-4l&4k=&4j=&4i=&17=\'+17,6(2r){$(\'#4h\').4g(2r)});$(\'.3-8-4f-4e:4d("4c")\').2h(6(e){2q();g().4b(0);g().4a(l)});6 2q(){h $14=$("").2p({1l:"49",16:"r%",15:"r%",48:0,2n:0,2o:47,46:"45(10%, 10%, 10%, 0.4)","44-43":"42"});$("<41 />").2p({16:"60%",15:"60%",2o:40,"3z-2n":"3y"}).3x({\'2m\':\'/?b=3w&2l=1k\',\'2k\':\'0\',\'2j\':\'2i\'}).2f($14);$14.2h(6(){$(3v).3u();g().2g()});$14.2f($(\'#1j\'))}g().13(0);}6 3t(){h 9=7.1b(2e);2d.2c(9);c(9.n>1){1r(i=0;i<9.n;i++){c(9[i].1a==2e){2d.2c(\'!!=\'+i);7.1p(i)}}}}7.j(\'3s\',6(){g().1h("/2a/3r.29","3q 10 28",6(){g().13(g().27()+10)},"2b");$("q[26=2b]").23().21(\'.3-20-1z\');g().1h("/2a/3p.29","3o 10 28",6(){h 12=g().27()-10;c(12<0)12=0;g().13(12)},"24");$("q[26=24]").23().21(\'.3-20-1z\');});6 1i(){}7.j(\'3n\',6(){1i()});7.j(\'3m\',6(){1i()});7.j("k",6(y){h 9=7.1b();c(9.n<2)19;$(\'.3-8-3l-3k\').3j(6(){$(\'#3-8-a-k\').1e(\'3-8-a-z\');$(\'.3-a-k\').p(\'o-1f\',\'11\')});7.1h("/3i/3h.3g","3f 3e",6(){$(\'.3-1w\').3d(\'3-8-1v\');$(\'.3-8-1y, .3-8-1x\').p(\'o-1g\',\'11\');c($(\'.3-1w\').3c(\'3-8-1v\')){$(\'.3-a-k\').p(\'o-1g\',\'l\');$(\'.3-a-k\').p(\'o-1f\',\'l\');$(\'.3-8-a\').1e(\'3-8-a-z\');$(\'.3-8-a:1u\').3b(\'3-8-a-z\')}3a{$(\'.3-a-k\').p(\'o-1g\',\'11\');$(\'.3-a-k\').p(\'o-1f\',\'11\');$(\'.3-8-a:1u\').1e(\'3-8-a-z\')}},"39");7.j("38",6(y){1d.37(\'1c\',y.9[y.36].1a)});c(1d.1t(\'1c\')){35("1s(1d.1t(\'1c\'));",34)}});h 18;6 1s(1q){h 9=7.1b();c(9.n>1){1r(i=0;i<9.n;i++){c(9[i].1a==1q){c(i==18){19}18=i;7.1p(i)}}}}',36,270,'|||jw|||function|player|settings|tracks|submenu||if||||jwplayer|var||on|audioTracks|true|3D|length|aria|attr|div|100|||sx|filemoon|https||event|active||false|tt|seek|dd|height|width|adb|current_audio|return|name|getAudioTracks|default_audio|localStorage|removeClass|expanded|checked|addButton|callMeMaybe|vplayer|0fxcyc2ajhp1|position|vvplay|vvad|220|setCurrentAudioTrack|audio_name|for|audio_set|getItem|last|open|controls|playbackRates|captions|rewind|icon|insertAfter||detach|ff00||button|getPosition|sec|png|player8|ff11|log|console|track_name|appendTo|play|click|no|scrolling|frameborder|file_code|src|top|zIndex|css|showCCform|data|1662367683|383371|dl|video_ad|doPlay|prevt|mp4|3E||jpg|thumbs|file|300|setTimeout|currentTrack|setItem|audioTrackChanged|dualSound|else|addClass|hasClass|toggleClass|Track|Audio|svg|dualy|images|mousedown|buttons|topbar|playAttemptFailed|beforePlay|Rewind|fr|Forward|ff|ready|set_audio_track|remove|this|upload_srt|prop|50px|margin|1000001|iframe|center|align|text|rgba|background|1000000|left|absolute|pause|setCurrentCaptions|Upload|contains|item|content|html|fviews|referer|prem|embed|3e57249ef633e0d03bf76ceb8d8a4b65|216|83|hash|view|get|TokenZir|window|hide|show|complete|slow|fadeIn|video_ad_fadein|time||cache|Cache|Content|headers|ajaxSetup|v2done|tott|vastdone2|vastdone1|vvbefore|playbackRateControls|cast|aboutlink|FileMoon|abouttext|UHD|1870|qualityLabels|sites|GNOME_POWER|link|2Fiframe|3C|allowfullscreen|22360|22640|22no|marginheight|marginwidth|2FGNOME_POWER|2F0fxcyc2ajhp1|2Fe|2Ffilemoon|2F|3A||22https|3Ciframe|code|sharing|fontOpacity|backgroundOpacity|Tahoma|fontFamily|303030|backgroundColor|FFFFFF|color|userFontScale|thumbnails|kind|0fxcyc2ajhp10000|url|get_slides|start|startparam|none|preload|html5|primary|hlshtml|androidhls|duration|uniform|stretching|0fxcyc2ajhp1_xt|image|2048|sp|6871|asn|127|srv|43200|_g3XlBcu2lmD9oDexD2NLWSmah2Nu3XcDrl93m9PwXY|m3u8||master|0fxcyc2ajhp1_x|00076|01|hls2|to|s01|delivery|storage|moon|sources|setup'''.split('|'))) - """ + self._test( + '''function f(p,a,c,k,e,d){while(c--)if(k[c])p=p.replace(new RegExp('\\b'+c.toString(a)+'\\b','g'),k[c]);return p}''', + '''h 7=g("1j");7.7h({7g:[{33:"w://7f-7e-7d-7c.v.7b/7a/79/78/77/76.74?t=73&s=2s&e=72&f=2t&71=70.0.0.1&6z=6y&6x=6w"}],6v:"w://32.v.u/6u.31",16:"r%",15:"r%",6t:"6s",6r:"",6q:"l",6p:"l",6o:"6n",6m:\'6l\',6k:"6j",9:[{33:"/2u?b=6i&n=50&6h=w://32.v.u/6g.31",6f:"6e"}],1y:{6d:1,6c:\'#6b\',6a:\'#69\',68:"67",66:30,65:r,},"64":{63:"%62 2m%m%61%5z%5y%5x.u%5w%5v%5u.2y%22 2k%m%1o%22 5t%m%1o%22 5s%m%1o%22 2j%m%5r%22 16%m%5q%22 15%m%5p%22 5o%2z%5n%5m%2z",5l:"w://v.u/d/1k/5k.2y",5j:[]},\'5i\':{"5h":"5g"},5f:"5e",5d:"w://v.u",5c:{},5b:l,1x:[0.25,0.50,0.75,1,1.25,1.5,2]});h 1m,1n,5a;h 59=0,58=0;h 7=g("1j");h 2x=0,57=0,56=0;$.55({54:{\'53-52\':\'2i-51\'}});7.j(\'4z\',6(x){c(5>0&&x.1l>=5&&1n!=1){1n=1;$(\'q.4y\').4x(\'4w\')}});7.j(\'13\',6(x){2x=x.1l});7.j(\'2g\',6(x){2w(x)});7.j(\'4v\',6(){$(\'q.2v\').4u()});6 2w(x){$(\'q.2v\').4t();c(1m)19;1m=1;17=0;c(4s.4r===l){17=1}$.4q(\'/2u?b=4p&2l=1k&4o=2t-4n-4m-2s-4l&4k=&4j=&4i=&17=\'+17,6(2r){$(\'#4h\').4g(2r)});$(\'.3-8-4f-4e:4d("4c")\').2h(6(e){2q();g().4b(0);g().4a(l)});6 2q(){h $14=$("").2p({1l:"49",16:"r%",15:"r%",48:0,2n:0,2o:47,46:"45(10%, 10%, 10%, 0.4)","44-43":"42"});$("<41 />").2p({16:"60%",15:"60%",2o:40,"3z-2n":"3y"}).3x({\'2m\':\'/?b=3w&2l=1k\',\'2k\':\'0\',\'2j\':\'2i\'}).2f($14);$14.2h(6(){$(3v).3u();g().2g()});$14.2f($(\'#1j\'))}g().13(0);}6 3t(){h 9=7.1b(2e);2d.2c(9);c(9.n>1){1r(i=0;i<9.n;i++){c(9[i].1a==2e){2d.2c(\'!!=\'+i);7.1p(i)}}}}7.j(\'3s\',6(){g().1h("/2a/3r.29","3q 10 28",6(){g().13(g().27()+10)},"2b");$("q[26=2b]").23().21(\'.3-20-1z\');g().1h("/2a/3p.29","3o 10 28",6(){h 12=g().27()-10;c(12<0)12=0;g().13(12)},"24");$("q[26=24]").23().21(\'.3-20-1z\');});6 1i(){}7.j(\'3n\',6(){1i()});7.j(\'3m\',6(){1i()});7.j("k",6(y){h 9=7.1b();c(9.n<2)19;$(\'.3-8-3l-3k\').3j(6(){$(\'#3-8-a-k\').1e(\'3-8-a-z\');$(\'.3-a-k\').p(\'o-1f\',\'11\')});7.1h("/3i/3h.3g","3f 3e",6(){$(\'.3-1w\').3d(\'3-8-1v\');$(\'.3-8-1y, .3-8-1x\').p(\'o-1g\',\'11\');c($(\'.3-1w\').3c(\'3-8-1v\')){$(\'.3-a-k\').p(\'o-1g\',\'l\');$(\'.3-a-k\').p(\'o-1f\',\'l\');$(\'.3-8-a\').1e(\'3-8-a-z\');$(\'.3-8-a:1u\').3b(\'3-8-a-z\')}3a{$(\'.3-a-k\').p(\'o-1g\',\'11\');$(\'.3-a-k\').p(\'o-1f\',\'11\');$(\'.3-8-a:1u\').1e(\'3-8-a-z\')}},"39");7.j("38",6(y){1d.37(\'1c\',y.9[y.36].1a)});c(1d.1t(\'1c\')){35("1s(1d.1t(\'1c\'));",34)}});h 18;6 1s(1q){h 9=7.1b();c(9.n>1){1r(i=0;i<9.n;i++){c(9[i].1a==1q){c(i==18){19}18=i;7.1p(i)}}}}',36,270,'|||jw|||function|player|settings|tracks|submenu||if||||jwplayer|var||on|audioTracks|true|3D|length|aria|attr|div|100|||sx|filemoon|https||event|active||false|tt|seek|dd|height|width|adb|current_audio|return|name|getAudioTracks|default_audio|localStorage|removeClass|expanded|checked|addButton|callMeMaybe|vplayer|0fxcyc2ajhp1|position|vvplay|vvad|220|setCurrentAudioTrack|audio_name|for|audio_set|getItem|last|open|controls|playbackRates|captions|rewind|icon|insertAfter||detach|ff00||button|getPosition|sec|png|player8|ff11|log|console|track_name|appendTo|play|click|no|scrolling|frameborder|file_code|src|top|zIndex|css|showCCform|data|1662367683|383371|dl|video_ad|doPlay|prevt|mp4|3E||jpg|thumbs|file|300|setTimeout|currentTrack|setItem|audioTrackChanged|dualSound|else|addClass|hasClass|toggleClass|Track|Audio|svg|dualy|images|mousedown|buttons|topbar|playAttemptFailed|beforePlay|Rewind|fr|Forward|ff|ready|set_audio_track|remove|this|upload_srt|prop|50px|margin|1000001|iframe|center|align|text|rgba|background|1000000|left|absolute|pause|setCurrentCaptions|Upload|contains|item|content|html|fviews|referer|prem|embed|3e57249ef633e0d03bf76ceb8d8a4b65|216|83|hash|view|get|TokenZir|window|hide|show|complete|slow|fadeIn|video_ad_fadein|time||cache|Cache|Content|headers|ajaxSetup|v2done|tott|vastdone2|vastdone1|vvbefore|playbackRateControls|cast|aboutlink|FileMoon|abouttext|UHD|1870|qualityLabels|sites|GNOME_POWER|link|2Fiframe|3C|allowfullscreen|22360|22640|22no|marginheight|marginwidth|2FGNOME_POWER|2F0fxcyc2ajhp1|2Fe|2Ffilemoon|2F|3A||22https|3Ciframe|code|sharing|fontOpacity|backgroundOpacity|Tahoma|fontFamily|303030|backgroundColor|FFFFFF|color|userFontScale|thumbnails|kind|0fxcyc2ajhp10000|url|get_slides|start|startparam|none|preload|html5|primary|hlshtml|androidhls|duration|uniform|stretching|0fxcyc2ajhp1_xt|image|2048|sp|6871|asn|127|srv|43200|_g3XlBcu2lmD9oDexD2NLWSmah2Nu3XcDrl93m9PwXY|m3u8||master|0fxcyc2ajhp1_x|00076|01|hls2|to|s01|delivery|storage|moon|sources|setup'''.split('|')) if __name__ == '__main__': diff --git a/youtube_dl/jsinterp.py b/youtube_dl/jsinterp.py index 02adf6678..799497acb 100644 --- a/youtube_dl/jsinterp.py +++ b/youtube_dl/jsinterp.py @@ -20,7 +20,9 @@ from .compat import ( compat_basestring, compat_chr, compat_collections_chain_map as ChainMap, + compat_filter as filter, compat_itertools_zip_longest as zip_longest, + compat_map as map, compat_str, ) @@ -252,7 +254,7 @@ class Debugger(object): cls.write('=> Raises:', e, '<-|', stmt, level=allow_recursion) raise if cls.ENABLED and stmt.strip(): - if should_ret or not repr(ret) == stmt: + if should_ret or repr(ret) != stmt: cls.write(['->', '=>'][should_ret], repr(ret), '<-|', stmt, level=allow_recursion) return ret, should_ret return interpret_statement @@ -365,6 +367,8 @@ class JSInterpreter(object): start, splits, pos, delim_len = 0, 0, 0, len(delim) - 1 in_quote, escaping, after_op, in_regex_char_group = None, False, True, False skipping = 0 + if skip_delims: + skip_delims = variadic(skip_delims) for idx, char in enumerate(expr): paren_delta = 0 if not in_quote: @@ -391,7 +395,7 @@ class JSInterpreter(object): continue elif pos == 0 and skip_delims: here = expr[idx:] - for s in variadic(skip_delims): + for s in skip_delims: if here.startswith(s) and s: skipping = len(s) - 1 break @@ -412,7 +416,6 @@ class JSInterpreter(object): if delim is None: delim = expr and _MATCHING_PARENS[expr[0]] separated = list(cls._separate(expr, delim, 1)) - if len(separated) < 2: raise cls.Exception('No terminating paren {delim} in {expr!r:.5500}'.format(**locals())) return separated[0][1:].strip(), separated[1].strip() @@ -487,6 +490,7 @@ class JSInterpreter(object): # fails on (eg) if (...) stmt1; else stmt2; sub_statements = list(self._separate(stmt, ';')) or [''] expr = stmt = sub_statements.pop().strip() + for sub_stmt in sub_statements: ret, should_return = self.interpret_statement(sub_stmt, local_vars, allow_recursion) if should_return: @@ -626,8 +630,7 @@ class JSInterpreter(object): if m.group('err'): catch_vars[m.group('err')] = err.error if isinstance(err, JS_Throw) else err catch_vars = local_vars.new_child(m=catch_vars) - err = None - pending = self.interpret_statement(sub_expr, catch_vars, allow_recursion) + err, pending = None, self.interpret_statement(sub_expr, catch_vars, allow_recursion) m = self._FINALLY_RE.match(expr) if m: @@ -877,12 +880,12 @@ class JSInterpreter(object): self.interpret_expression(v, local_vars, allow_recursion) for v in self._separate(arg_str)] - if obj == compat_str: + if obj is compat_str: if member == 'fromCharCode': assertion(argvals, 'takes one or more arguments') return ''.join(map(compat_chr, argvals)) raise self.Exception('Unsupported string method ' + member, expr=expr) - elif obj == float: + elif obj is float: if member == 'pow': assertion(len(argvals) == 2, 'takes two arguments') return argvals[0] ** argvals[1] @@ -907,12 +910,12 @@ class JSInterpreter(object): elif member == 'splice': assertion(isinstance(obj, list), 'must be applied on a list') assertion(argvals, 'takes one or more arguments') - index, howMany = map(int, (argvals + [len(obj)])[:2]) + index, how_many = map(int, (argvals + [len(obj)])[:2]) if index < 0: index += len(obj) add_items = argvals[2:] res = [] - for i in range(index, min(index + howMany, len(obj))): + for _ in range(index, min(index + how_many, len(obj))): res.append(obj.pop(index)) for i, item in enumerate(add_items): obj.insert(index + i, item) @@ -1002,28 +1005,25 @@ class JSInterpreter(object): def extract_object(self, objname): _FUNC_NAME_RE = r'''(?:[a-zA-Z$0-9]+|"[a-zA-Z$0-9]+"|'[a-zA-Z$0-9]+')''' obj = {} - fields = None - for obj_m in re.finditer( + fields = next(filter(None, ( + obj_m.group('fields') for obj_m in re.finditer( r'''(?xs) {0}\s*\.\s*{1}|{1}\s*=\s*\{{\s* (?P({2}\s*:\s*function\s*\(.*?\)\s*\{{.*?}}(?:,\s*)?)*) }}\s*; '''.format(_NAME_RE, re.escape(objname), _FUNC_NAME_RE), - self.code): - fields = obj_m.group('fields') - if fields: - break - else: + self.code))), None) + if not fields: raise self.Exception('Could not find object ' + objname) # Currently, it only supports function definitions - fields_m = re.finditer( - r'''(?x) - (?P%s)\s*:\s*function\s*\((?P(?:%s|,)*)\){(?P[^}]+)} - ''' % (_FUNC_NAME_RE, _NAME_RE), - fields) - for f in fields_m: + for f in re.finditer( + r'''(?x) + (?P%s)\s*:\s*function\s*\((?P(?:%s|,)*)\){(?P[^}]+)} + ''' % (_FUNC_NAME_RE, _NAME_RE), + fields): argnames = self.build_arglist(f.group('args')) - obj[remove_quotes(f.group('key'))] = self.build_function(argnames, f.group('code')) + name = remove_quotes(f.group('key')) + obj[name] = function_with_repr(self.build_function(argnames, f.group('code')), 'F<{0}>'.format(name)) return obj @@ -1058,7 +1058,7 @@ class JSInterpreter(object): def extract_function(self, funcname): return function_with_repr( self.extract_function_from_code(*self.extract_function_code(funcname)), - 'F<%s>' % (funcname, )) + 'F<%s>' % (funcname,)) def extract_function_from_code(self, argnames, code, *global_stack): local_vars = {} @@ -1067,7 +1067,7 @@ class JSInterpreter(object): if mobj is None: break start, body_start = mobj.span() - body, remaining = self._separate_at_paren(code[body_start - 1:], '}') + body, remaining = self._separate_at_paren(code[body_start - 1:]) name = self._named_object(local_vars, self.extract_function_from_code( [x.strip() for x in mobj.group('args').split(',')], body, local_vars, *global_stack)) @@ -1095,8 +1095,7 @@ class JSInterpreter(object): argnames = tuple(argnames) def resf(args, kwargs={}, allow_recursion=100): - global_stack[0].update( - zip_longest(argnames, args, fillvalue=None)) + global_stack[0].update(zip_longest(argnames, args, fillvalue=None)) global_stack[0].update(kwargs) var_stack = LocalNameSpace(*global_stack) ret, should_abort = self.interpret_statement(code.replace('\n', ' '), var_stack, allow_recursion - 1) From 76ac69917ec76ba663da843795f46916831e6da9 Mon Sep 17 00:00:00 2001 From: dirkf Date: Wed, 10 Jul 2024 18:02:11 +0100 Subject: [PATCH 3/5] [jsinterp] Further improve expression parsing (fix fd8242e) Passes tests from yt-dlp --- youtube_dl/jsinterp.py | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-) diff --git a/youtube_dl/jsinterp.py b/youtube_dl/jsinterp.py index 799497acb..a2074a91e 100644 --- a/youtube_dl/jsinterp.py +++ b/youtube_dl/jsinterp.py @@ -804,16 +804,19 @@ class JSInterpreter(object): if op in ('+', '-'): # simplify/adjust consecutive instances of these operators undone = 0 - while len(separated) > 1 and not separated[-1].strip(): + separated = [s.strip() for s in separated] + while len(separated) > 1 and not separated[-1]: undone += 1 separated.pop() if op == '-' and undone % 2 != 0: right_expr = op + right_expr elif op == '+': - while len(separated) > 1 and separated[-1].strip() in self.OP_CHARS: + while len(separated) > 1 and set(separated[-1]) <= self.OP_CHARS: + right_expr = separated.pop() + right_expr + if separated[-1][-1:] in self.OP_CHARS: right_expr = separated.pop() + right_expr # hanging op at end of left => unary + (strip) or - (push right) - left_val = separated[-1] + left_val = separated[-1] if separated else '' for dm_op in ('*', '%', '/', '**'): bodmas = tuple(self._separate(left_val, dm_op, skip_delims=skip_delim)) if len(bodmas) > 1 and not bodmas[-1].strip(): From d35ce6ce95aac9b98c5f8272824a82e4623b777a Mon Sep 17 00:00:00 2001 From: dirkf Date: Wed, 10 Jul 2024 18:07:20 +0100 Subject: [PATCH 4/5] [jsinterp] Support functionality for player `b22ef6e7` * support `prototype` for call() and apply() (yt-dlp/yt-dlp#10392, thx Grub4k) * map JS `Array` to `list` --- test/test_jsinterp.py | 27 +++++++++++++++++++++++++++ youtube_dl/jsinterp.py | 24 +++++++++++++++++++++--- 2 files changed, 48 insertions(+), 3 deletions(-) diff --git a/test/test_jsinterp.py b/test/test_jsinterp.py index 86d7f0d46..104e766be 100644 --- a/test/test_jsinterp.py +++ b/test/test_jsinterp.py @@ -398,6 +398,33 @@ class TestJSInterpreter(unittest.TestCase): '''function f(p,a,c,k,e,d){while(c--)if(k[c])p=p.replace(new RegExp('\\b'+c.toString(a)+'\\b','g'),k[c]);return p}''', '''h 7=g("1j");7.7h({7g:[{33:"w://7f-7e-7d-7c.v.7b/7a/79/78/77/76.74?t=73&s=2s&e=72&f=2t&71=70.0.0.1&6z=6y&6x=6w"}],6v:"w://32.v.u/6u.31",16:"r%",15:"r%",6t:"6s",6r:"",6q:"l",6p:"l",6o:"6n",6m:\'6l\',6k:"6j",9:[{33:"/2u?b=6i&n=50&6h=w://32.v.u/6g.31",6f:"6e"}],1y:{6d:1,6c:\'#6b\',6a:\'#69\',68:"67",66:30,65:r,},"64":{63:"%62 2m%m%61%5z%5y%5x.u%5w%5v%5u.2y%22 2k%m%1o%22 5t%m%1o%22 5s%m%1o%22 2j%m%5r%22 16%m%5q%22 15%m%5p%22 5o%2z%5n%5m%2z",5l:"w://v.u/d/1k/5k.2y",5j:[]},\'5i\':{"5h":"5g"},5f:"5e",5d:"w://v.u",5c:{},5b:l,1x:[0.25,0.50,0.75,1,1.25,1.5,2]});h 1m,1n,5a;h 59=0,58=0;h 7=g("1j");h 2x=0,57=0,56=0;$.55({54:{\'53-52\':\'2i-51\'}});7.j(\'4z\',6(x){c(5>0&&x.1l>=5&&1n!=1){1n=1;$(\'q.4y\').4x(\'4w\')}});7.j(\'13\',6(x){2x=x.1l});7.j(\'2g\',6(x){2w(x)});7.j(\'4v\',6(){$(\'q.2v\').4u()});6 2w(x){$(\'q.2v\').4t();c(1m)19;1m=1;17=0;c(4s.4r===l){17=1}$.4q(\'/2u?b=4p&2l=1k&4o=2t-4n-4m-2s-4l&4k=&4j=&4i=&17=\'+17,6(2r){$(\'#4h\').4g(2r)});$(\'.3-8-4f-4e:4d("4c")\').2h(6(e){2q();g().4b(0);g().4a(l)});6 2q(){h $14=$("").2p({1l:"49",16:"r%",15:"r%",48:0,2n:0,2o:47,46:"45(10%, 10%, 10%, 0.4)","44-43":"42"});$("<41 />").2p({16:"60%",15:"60%",2o:40,"3z-2n":"3y"}).3x({\'2m\':\'/?b=3w&2l=1k\',\'2k\':\'0\',\'2j\':\'2i\'}).2f($14);$14.2h(6(){$(3v).3u();g().2g()});$14.2f($(\'#1j\'))}g().13(0);}6 3t(){h 9=7.1b(2e);2d.2c(9);c(9.n>1){1r(i=0;i<9.n;i++){c(9[i].1a==2e){2d.2c(\'!!=\'+i);7.1p(i)}}}}7.j(\'3s\',6(){g().1h("/2a/3r.29","3q 10 28",6(){g().13(g().27()+10)},"2b");$("q[26=2b]").23().21(\'.3-20-1z\');g().1h("/2a/3p.29","3o 10 28",6(){h 12=g().27()-10;c(12<0)12=0;g().13(12)},"24");$("q[26=24]").23().21(\'.3-20-1z\');});6 1i(){}7.j(\'3n\',6(){1i()});7.j(\'3m\',6(){1i()});7.j("k",6(y){h 9=7.1b();c(9.n<2)19;$(\'.3-8-3l-3k\').3j(6(){$(\'#3-8-a-k\').1e(\'3-8-a-z\');$(\'.3-a-k\').p(\'o-1f\',\'11\')});7.1h("/3i/3h.3g","3f 3e",6(){$(\'.3-1w\').3d(\'3-8-1v\');$(\'.3-8-1y, .3-8-1x\').p(\'o-1g\',\'11\');c($(\'.3-1w\').3c(\'3-8-1v\')){$(\'.3-a-k\').p(\'o-1g\',\'l\');$(\'.3-a-k\').p(\'o-1f\',\'l\');$(\'.3-8-a\').1e(\'3-8-a-z\');$(\'.3-8-a:1u\').3b(\'3-8-a-z\')}3a{$(\'.3-a-k\').p(\'o-1g\',\'11\');$(\'.3-a-k\').p(\'o-1f\',\'11\');$(\'.3-8-a:1u\').1e(\'3-8-a-z\')}},"39");7.j("38",6(y){1d.37(\'1c\',y.9[y.36].1a)});c(1d.1t(\'1c\')){35("1s(1d.1t(\'1c\'));",34)}});h 18;6 1s(1q){h 9=7.1b();c(9.n>1){1r(i=0;i<9.n;i++){c(9[i].1a==1q){c(i==18){19}18=i;7.1p(i)}}}}',36,270,'|||jw|||function|player|settings|tracks|submenu||if||||jwplayer|var||on|audioTracks|true|3D|length|aria|attr|div|100|||sx|filemoon|https||event|active||false|tt|seek|dd|height|width|adb|current_audio|return|name|getAudioTracks|default_audio|localStorage|removeClass|expanded|checked|addButton|callMeMaybe|vplayer|0fxcyc2ajhp1|position|vvplay|vvad|220|setCurrentAudioTrack|audio_name|for|audio_set|getItem|last|open|controls|playbackRates|captions|rewind|icon|insertAfter||detach|ff00||button|getPosition|sec|png|player8|ff11|log|console|track_name|appendTo|play|click|no|scrolling|frameborder|file_code|src|top|zIndex|css|showCCform|data|1662367683|383371|dl|video_ad|doPlay|prevt|mp4|3E||jpg|thumbs|file|300|setTimeout|currentTrack|setItem|audioTrackChanged|dualSound|else|addClass|hasClass|toggleClass|Track|Audio|svg|dualy|images|mousedown|buttons|topbar|playAttemptFailed|beforePlay|Rewind|fr|Forward|ff|ready|set_audio_track|remove|this|upload_srt|prop|50px|margin|1000001|iframe|center|align|text|rgba|background|1000000|left|absolute|pause|setCurrentCaptions|Upload|contains|item|content|html|fviews|referer|prem|embed|3e57249ef633e0d03bf76ceb8d8a4b65|216|83|hash|view|get|TokenZir|window|hide|show|complete|slow|fadeIn|video_ad_fadein|time||cache|Cache|Content|headers|ajaxSetup|v2done|tott|vastdone2|vastdone1|vvbefore|playbackRateControls|cast|aboutlink|FileMoon|abouttext|UHD|1870|qualityLabels|sites|GNOME_POWER|link|2Fiframe|3C|allowfullscreen|22360|22640|22no|marginheight|marginwidth|2FGNOME_POWER|2F0fxcyc2ajhp1|2Fe|2Ffilemoon|2F|3A||22https|3Ciframe|code|sharing|fontOpacity|backgroundOpacity|Tahoma|fontFamily|303030|backgroundColor|FFFFFF|color|userFontScale|thumbnails|kind|0fxcyc2ajhp10000|url|get_slides|start|startparam|none|preload|html5|primary|hlshtml|androidhls|duration|uniform|stretching|0fxcyc2ajhp1_xt|image|2048|sp|6871|asn|127|srv|43200|_g3XlBcu2lmD9oDexD2NLWSmah2Nu3XcDrl93m9PwXY|m3u8||master|0fxcyc2ajhp1_x|00076|01|hls2|to|s01|delivery|storage|moon|sources|setup'''.split('|')) + def test_join(self): + test_input = list('test') + tests = [ + 'function f(a, b){return a.join(b)}', + 'function f(a, b){return Array.prototype.join.call(a, b)}', + 'function f(a, b){return Array.prototype.join.apply(a, [b])}', + ] + for test in tests: + jsi = JSInterpreter(test) + self._test(jsi, 'test', args=[test_input, '']) + self._test(jsi, 't-e-s-t', args=[test_input, '-']) + self._test(jsi, '', args=[[], '-']) + + def test_split(self): + test_result = list('test') + tests = [ + 'function f(a, b){return a.split(b)}', + 'function f(a, b){return String.prototype.split.call(a, b)}', + 'function f(a, b){return String.prototype.split.apply(a, [b])}', + ] + for test in tests: + jsi = JSInterpreter(test) + self._test(jsi, test_result, args=['test', '']) + self._test(jsi, test_result, args=['t-e-s-t', '-']) + self._test(jsi, [''], args=['', '-']) + self._test(jsi, [], args=['', '']) + if __name__ == '__main__': unittest.main() diff --git a/youtube_dl/jsinterp.py b/youtube_dl/jsinterp.py index a2074a91e..949f77775 100644 --- a/youtube_dl/jsinterp.py +++ b/youtube_dl/jsinterp.py @@ -850,7 +850,7 @@ class JSInterpreter(object): memb = member raise self.Exception('{memb} {msg}'.format(**locals()), expr=expr) - def eval_method(): + def eval_method(variable, member): if (variable, member) == ('console', 'debug'): if Debugger.ENABLED: Debugger.write(self.interpret_expression('[{}]'.format(arg_str), local_vars, allow_recursion)) @@ -858,6 +858,7 @@ class JSInterpreter(object): types = { 'String': compat_str, 'Math': float, + 'Array': list, } obj = local_vars.get(variable) if obj in (JS_Undefined, None): @@ -883,6 +884,23 @@ class JSInterpreter(object): self.interpret_expression(v, local_vars, allow_recursion) for v in self._separate(arg_str)] + # Fixup prototype call + if isinstance(obj, type): + new_member, rest = member.partition('.')[0::2] + if new_member == 'prototype': + new_member, func_prototype = rest.partition('.')[0::2] + assertion(argvals, 'takes one or more arguments') + assertion(isinstance(argvals[0], obj), 'must bind to type {0}'.format(obj)) + if func_prototype == 'call': + obj = argvals.pop(0) + elif func_prototype == 'apply': + assertion(len(argvals) == 2, 'takes two arguments') + obj, argvals = argvals + assertion(isinstance(argvals, list), 'second argument must be a list') + else: + raise self.Exception('Unsupported Function method ' + func_prototype, expr) + member = new_member + if obj is compat_str: if member == 'fromCharCode': assertion(argvals, 'takes one or more arguments') @@ -976,11 +994,11 @@ class JSInterpreter(object): if remaining: ret, should_abort = self.interpret_statement( - self._named_object(local_vars, eval_method()) + remaining, + self._named_object(local_vars, eval_method(variable, member)) + remaining, local_vars, allow_recursion) return ret, should_return or should_abort else: - return eval_method(), should_return + return eval_method(variable, member), should_return elif md.get('function'): fname = m.group('fname') From 16f5bbc464602773e61eeafef51d1dbc47987bb4 Mon Sep 17 00:00:00 2001 From: dirkf Date: Wed, 10 Jul 2024 18:20:59 +0100 Subject: [PATCH 5/5] [YouTube] Fix nsig processing for player `b22ef6e7` * improve extraction of function name (like yt-dlp/yt-dlp#10390) * always use JSInterp to extract function code (yt-dlp/yt-dlp#10396, thx seproDev, pukkandan) --- test/test_youtube_signature.py | 4 ++++ youtube_dl/extractor/youtube.py | 24 ++++++++++-------------- 2 files changed, 14 insertions(+), 14 deletions(-) diff --git a/test/test_youtube_signature.py b/test/test_youtube_signature.py index cafba7a5c..cc18d0f7b 100644 --- a/test/test_youtube_signature.py +++ b/test/test_youtube_signature.py @@ -162,6 +162,10 @@ _NSIG_TESTS = [ 'https://www.youtube.com/s/player/590f65a6/player_ias.vflset/en_US/base.js', '1tm7-g_A9zsI8_Lay_', 'xI4Vem4Put_rOg', ), + ( + 'https://www.youtube.com/s/player/b22ef6e7/player_ias.vflset/en_US/base.js', + 'b6HcntHGkvBLk_FRf', 'kNPW6A7FyP2l8A', + ), ] diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py index 90c16e172..2e31a8979 100644 --- a/youtube_dl/extractor/youtube.py +++ b/youtube_dl/extractor/youtube.py @@ -1636,7 +1636,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor): try: jsi, player_id, func_code = self._extract_n_function_code(video_id, player_url) except ExtractorError as e: - raise ExtractorError('Unable to extract nsig jsi, player_id, func_codefunction code', cause=e) + raise ExtractorError('Unable to extract nsig function code', cause=e) if self.get_param('youtube_print_sig_code'): self.to_screen('Extracted nsig function from {0}:\n{1}\n'.format( player_id, func_code[1])) @@ -1658,8 +1658,14 @@ class YoutubeIE(YoutubeBaseInfoExtractor): def _extract_n_function_name(self, jscode): func_name, idx = self._search_regex( - r'\.get\("n"\)\)&&\(b=(?P[a-zA-Z_$][\w$]*)(?:\[(?P\d+)\])?\([\w$]+\)', - jscode, 'Initial JS player n function name', group=('nfunc', 'idx')) + # new: (b=String.fromCharCode(110),c=a.get(b))&&c=nfunc[idx](c) + # old: .get("n"))&&(b=nfunc[idx](b) + # older: .get("n"))&&(b=nfunc(b) + r'''(?x) + (?:\(\s*(?P[a-z])\s*=\s*String\s*\.\s*fromCharCode\s*\(\s*110\s*\)\s*,(?P[a-z])\s*=\s*[a-z]\s*)? + \.\s*get\s*\(\s*(?(b)(?P=b)|"n")(?:\s*\)){2}\s*&&\s*\(\s*(?(c)(?P=c)|b)\s*=\s* + (?P[a-zA-Z_$][\w$]*)(?:\s*\[(?P\d+)\])?\s*\(\s*[\w$]+\s*\) + ''', jscode, 'Initial JS player n function name', group=('nfunc', 'idx')) if not idx: return func_name @@ -1679,17 +1685,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor): func_name = self._extract_n_function_name(jscode) - # For redundancy - func_code = self._search_regex( - r'''(?xs)%s\s*=\s*function\s*\((?P[\w$]+)\)\s* - # NB: The end of the regex is intentionally kept strict - {(?P.+?}\s*return\ [\w$]+.join\(""\))};''' % func_name, - jscode, 'nsig function', group=('var', 'code'), default=None) - if func_code: - func_code = ([func_code[0]], func_code[1]) - else: - self.write_debug('Extracting nsig function with jsinterp') - func_code = jsi.extract_function_code(func_name) + func_code = jsi.extract_function_code(func_name) self.cache.store('youtube-nsig', player_id, func_code) return jsi, player_id, func_code