aboutsummaryrefslogtreecommitdiff
diff options
context:
space:
mode:
-rw-r--r--test/test_utils.py2
-rw-r--r--youtube_dl/compat.py2240
-rw-r--r--youtube_dl/extractor/audiomack.py3
-rw-r--r--youtube_dl/extractor/dw.py2
-rw-r--r--youtube_dl/extractor/extractors.py3
-rw-r--r--youtube_dl/extractor/generic.py35
-rw-r--r--youtube_dl/extractor/godtv.py66
-rw-r--r--youtube_dl/extractor/lynda.py131
-rw-r--r--youtube_dl/extractor/nbc.py22
-rw-r--r--youtube_dl/extractor/openload.py7
-rw-r--r--youtube_dl/extractor/voicerepublic.py11
-rw-r--r--youtube_dl/extractor/vporn.py10
-rw-r--r--youtube_dl/extractor/vulture.py69
-rw-r--r--youtube_dl/extractor/wdr.py6
-rw-r--r--youtube_dl/extractor/weibo.py49
-rw-r--r--youtube_dl/extractor/xfileshare.py3
-rw-r--r--youtube_dl/extractor/xuite.py1
-rw-r--r--youtube_dl/extractor/yahoo.py2
-rw-r--r--youtube_dl/extractor/youtube.py5
-rw-r--r--youtube_dl/utils.py12
20 files changed, 2454 insertions, 225 deletions
diff --git a/test/test_utils.py b/test/test_utils.py
index feef80465..0e25de6b7 100644
--- a/test/test_utils.py
+++ b/test/test_utils.py
@@ -249,6 +249,8 @@ class TestUtil(unittest.TestCase):
self.assertEqual(unescapeHTML('/'), '/')
self.assertEqual(unescapeHTML('é'), 'é')
self.assertEqual(unescapeHTML('�'), '�')
+ # HTML5 entities
+ self.assertEqual(unescapeHTML('.''), '.\'')
def test_date_from_str(self):
self.assertEqual(date_from_str('yesterday'), date_from_str('now-1day'))
diff --git a/youtube_dl/compat.py b/youtube_dl/compat.py
index e3cab4dd0..67db1c7c6 100644
--- a/youtube_dl/compat.py
+++ b/youtube_dl/compat.py
@@ -63,6 +63,2244 @@ try:
except ImportError: # Python 2
import htmlentitydefs as compat_html_entities
+try: # Python >= 3.3
+ compat_html_entities_html5 = compat_html_entities.html5
+except AttributeError:
+ # Copied from CPython 3.5.1 html/entities.py
+ compat_html_entities_html5 = {
+ 'Aacute': '\xc1',
+ 'aacute': '\xe1',
+ 'Aacute;': '\xc1',
+ 'aacute;': '\xe1',
+ 'Abreve;': '\u0102',
+ 'abreve;': '\u0103',
+ 'ac;': '\u223e',
+ 'acd;': '\u223f',
+ 'acE;': '\u223e\u0333',
+ 'Acirc': '\xc2',
+ 'acirc': '\xe2',
+ 'Acirc;': '\xc2',
+ 'acirc;': '\xe2',
+ 'acute': '\xb4',
+ 'acute;': '\xb4',
+ 'Acy;': '\u0410',
+ 'acy;': '\u0430',
+ 'AElig': '\xc6',
+ 'aelig': '\xe6',
+ 'AElig;': '\xc6',
+ 'aelig;': '\xe6',
+ 'af;': '\u2061',
+ 'Afr;': '\U0001d504',
+ 'afr;': '\U0001d51e',
+ 'Agrave': '\xc0',
+ 'agrave': '\xe0',
+ 'Agrave;': '\xc0',
+ 'agrave;': '\xe0',
+ 'alefsym;': '\u2135',
+ 'aleph;': '\u2135',
+ 'Alpha;': '\u0391',
+ 'alpha;': '\u03b1',
+ 'Amacr;': '\u0100',
+ 'amacr;': '\u0101',
+ 'amalg;': '\u2a3f',
+ 'AMP': '&',
+ 'amp': '&',
+ 'AMP;': '&',
+ 'amp;': '&',
+ 'And;': '\u2a53',
+ 'and;': '\u2227',
+ 'andand;': '\u2a55',
+ 'andd;': '\u2a5c',
+ 'andslope;': '\u2a58',
+ 'andv;': '\u2a5a',
+ 'ang;': '\u2220',
+ 'ange;': '\u29a4',
+ 'angle;': '\u2220',
+ 'angmsd;': '\u2221',
+ 'angmsdaa;': '\u29a8',
+ 'angmsdab;': '\u29a9',
+ 'angmsdac;': '\u29aa',
+ 'angmsdad;': '\u29ab',
+ 'angmsdae;': '\u29ac',
+ 'angmsdaf;': '\u29ad',
+ 'angmsdag;': '\u29ae',
+ 'angmsdah;': '\u29af',
+ 'angrt;': '\u221f',
+ 'angrtvb;': '\u22be',
+ 'angrtvbd;': '\u299d',
+ 'angsph;': '\u2222',
+ 'angst;': '\xc5',
+ 'angzarr;': '\u237c',
+ 'Aogon;': '\u0104',
+ 'aogon;': '\u0105',
+ 'Aopf;': '\U0001d538',
+ 'aopf;': '\U0001d552',
+ 'ap;': '\u2248',
+ 'apacir;': '\u2a6f',
+ 'apE;': '\u2a70',
+ 'ape;': '\u224a',
+ 'apid;': '\u224b',
+ 'apos;': "'",
+ 'ApplyFunction;': '\u2061',
+ 'approx;': '\u2248',
+ 'approxeq;': '\u224a',
+ 'Aring': '\xc5',
+ 'aring': '\xe5',
+ 'Aring;': '\xc5',
+ 'aring;': '\xe5',
+ 'Ascr;': '\U0001d49c',
+ 'ascr;': '\U0001d4b6',
+ 'Assign;': '\u2254',
+ 'ast;': '*',
+ 'asymp;': '\u2248',
+ 'asympeq;': '\u224d',
+ 'Atilde': '\xc3',
+ 'atilde': '\xe3',
+ 'Atilde;': '\xc3',
+ 'atilde;': '\xe3',
+ 'Auml': '\xc4',
+ 'auml': '\xe4',
+ 'Auml;': '\xc4',
+ 'auml;': '\xe4',
+ 'awconint;': '\u2233',
+ 'awint;': '\u2a11',
+ 'backcong;': '\u224c',
+ 'backepsilon;': '\u03f6',
+ 'backprime;': '\u2035',
+ 'backsim;': '\u223d',
+ 'backsimeq;': '\u22cd',
+ 'Backslash;': '\u2216',
+ 'Barv;': '\u2ae7',
+ 'barvee;': '\u22bd',
+ 'Barwed;': '\u2306',
+ 'barwed;': '\u2305',
+ 'barwedge;': '\u2305',
+ 'bbrk;': '\u23b5',
+ 'bbrktbrk;': '\u23b6',
+ 'bcong;': '\u224c',
+ 'Bcy;': '\u0411',
+ 'bcy;': '\u0431',
+ 'bdquo;': '\u201e',
+ 'becaus;': '\u2235',
+ 'Because;': '\u2235',
+ 'because;': '\u2235',
+ 'bemptyv;': '\u29b0',
+ 'bepsi;': '\u03f6',
+ 'bernou;': '\u212c',
+ 'Bernoullis;': '\u212c',
+ 'Beta;': '\u0392',
+ 'beta;': '\u03b2',
+ 'beth;': '\u2136',
+ 'between;': '\u226c',
+ 'Bfr;': '\U0001d505',
+ 'bfr;': '\U0001d51f',
+ 'bigcap;': '\u22c2',
+ 'bigcirc;': '\u25ef',
+ 'bigcup;': '\u22c3',
+ 'bigodot;': '\u2a00',
+ 'bigoplus;': '\u2a01',
+ 'bigotimes;': '\u2a02',
+ 'bigsqcup;': '\u2a06',
+ 'bigstar;': '\u2605',
+ 'bigtriangledown;': '\u25bd',
+ 'bigtriangleup;': '\u25b3',
+ 'biguplus;': '\u2a04',
+ 'bigvee;': '\u22c1',
+ 'bigwedge;': '\u22c0',
+ 'bkarow;': '\u290d',
+ 'blacklozenge;': '\u29eb',
+ 'blacksquare;': '\u25aa',
+ 'blacktriangle;': '\u25b4',
+ 'blacktriangledown;': '\u25be',
+ 'blacktriangleleft;': '\u25c2',
+ 'blacktriangleright;': '\u25b8',
+ 'blank;': '\u2423',
+ 'blk12;': '\u2592',
+ 'blk14;': '\u2591',
+ 'blk34;': '\u2593',
+ 'block;': '\u2588',
+ 'bne;': '=\u20e5',
+ 'bnequiv;': '\u2261\u20e5',
+ 'bNot;': '\u2aed',
+ 'bnot;': '\u2310',
+ 'Bopf;': '\U0001d539',
+ 'bopf;': '\U0001d553',
+ 'bot;': '\u22a5',
+ 'bottom;': '\u22a5',
+ 'bowtie;': '\u22c8',
+ 'boxbox;': '\u29c9',
+ 'boxDL;': '\u2557',
+ 'boxDl;': '\u2556',
+ 'boxdL;': '\u2555',
+ 'boxdl;': '\u2510',
+ 'boxDR;': '\u2554',
+ 'boxDr;': '\u2553',
+ 'boxdR;': '\u2552',
+ 'boxdr;': '\u250c',
+ 'boxH;': '\u2550',
+ 'boxh;': '\u2500',
+ 'boxHD;': '\u2566',
+ 'boxHd;': '\u2564',
+ 'boxhD;': '\u2565',
+ 'boxhd;': '\u252c',
+ 'boxHU;': '\u2569',
+ 'boxHu;': '\u2567',
+ 'boxhU;': '\u2568',
+ 'boxhu;': '\u2534',
+ 'boxminus;': '\u229f',
+ 'boxplus;': '\u229e',
+ 'boxtimes;': '\u22a0',
+ 'boxUL;': '\u255d',
+ 'boxUl;': '\u255c',
+ 'boxuL;': '\u255b',
+ 'boxul;': '\u2518',
+ 'boxUR;': '\u255a',
+ 'boxUr;': '\u2559',
+ 'boxuR;': '\u2558',
+ 'boxur;': '\u2514',
+ 'boxV;': '\u2551',
+ 'boxv;': '\u2502',
+ 'boxVH;': '\u256c',
+ 'boxVh;': '\u256b',
+ 'boxvH;': '\u256a',
+ 'boxvh;': '\u253c',
+ 'boxVL;': '\u2563',
+ 'boxVl;': '\u2562',
+ 'boxvL;': '\u2561',
+ 'boxvl;': '\u2524',
+ 'boxVR;': '\u2560',
+ 'boxVr;': '\u255f',
+ 'boxvR;': '\u255e',
+ 'boxvr;': '\u251c',
+ 'bprime;': '\u2035',
+ 'Breve;': '\u02d8',
+ 'breve;': '\u02d8',
+ 'brvbar': '\xa6',
+ 'brvbar;': '\xa6',
+ 'Bscr;': '\u212c',
+ 'bscr;': '\U0001d4b7',
+ 'bsemi;': '\u204f',
+ 'bsim;': '\u223d',
+ 'bsime;': '\u22cd',
+ 'bsol;': '\\',
+ 'bsolb;': '\u29c5',
+ 'bsolhsub;': '\u27c8',
+ 'bull;': '\u2022',
+ 'bullet;': '\u2022',
+ 'bump;': '\u224e',
+ 'bumpE;': '\u2aae',
+ 'bumpe;': '\u224f',
+ 'Bumpeq;': '\u224e',
+ 'bumpeq;': '\u224f',
+ 'Cacute;': '\u0106',
+ 'cacute;': '\u0107',
+ 'Cap;': '\u22d2',
+ 'cap;': '\u2229',
+ 'capand;': '\u2a44',
+ 'capbrcup;': '\u2a49',
+ 'capcap;': '\u2a4b',
+ 'capcup;': '\u2a47',
+ 'capdot;': '\u2a40',
+ 'CapitalDifferentialD;': '\u2145',
+ 'caps;': '\u2229\ufe00',
+ 'caret;': '\u2041',
+ 'caron;': '\u02c7',
+ 'Cayleys;': '\u212d',
+ 'ccaps;': '\u2a4d',
+ 'Ccaron;': '\u010c',
+ 'ccaron;': '\u010d',
+ 'Ccedil': '\xc7',
+ 'ccedil': '\xe7',
+ 'Ccedil;': '\xc7',
+ 'ccedil;': '\xe7',
+ 'Ccirc;': '\u0108',
+ 'ccirc;': '\u0109',
+ 'Cconint;': '\u2230',
+ 'ccups;': '\u2a4c',
+ 'ccupssm;': '\u2a50',
+ 'Cdot;': '\u010a',
+ 'cdot;': '\u010b',
+ 'cedil': '\xb8',
+ 'cedil;': '\xb8',
+ 'Cedilla;': '\xb8',
+ 'cemptyv;': '\u29b2',
+ 'cent': '\xa2',
+ 'cent;': '\xa2',
+ 'CenterDot;': '\xb7',
+ 'centerdot;': '\xb7',
+ 'Cfr;': '\u212d',
+ 'cfr;': '\U0001d520',
+ 'CHcy;': '\u0427',
+ 'chcy;': '\u0447',
+ 'check;': '\u2713',
+ 'checkmark;': '\u2713',
+ 'Chi;': '\u03a7',
+ 'chi;': '\u03c7',
+ 'cir;': '\u25cb',
+ 'circ;': '\u02c6',
+ 'circeq;': '\u2257',
+ 'circlearrowleft;': '\u21ba',
+ 'circlearrowright;': '\u21bb',
+ 'circledast;': '\u229b',
+ 'circledcirc;': '\u229a',
+ 'circleddash;': '\u229d',
+ 'CircleDot;': '\u2299',
+ 'circledR;': '\xae',
+ 'circledS;': '\u24c8',
+ 'CircleMinus;': '\u2296',
+ 'CirclePlus;': '\u2295',
+ 'CircleTimes;': '\u2297',
+ 'cirE;': '\u29c3',
+ 'cire;': '\u2257',
+ 'cirfnint;': '\u2a10',
+ 'cirmid;': '\u2aef',
+ 'cirscir;': '\u29c2',
+ 'ClockwiseContourIntegral;': '\u2232',
+ 'CloseCurlyDoubleQuote;': '\u201d',
+ 'CloseCurlyQuote;': '\u2019',
+ 'clubs;': '\u2663',
+ 'clubsuit;': '\u2663',
+ 'Colon;': '\u2237',
+ 'colon;': ':',
+ 'Colone;': '\u2a74',
+ 'colone;': '\u2254',
+ 'coloneq;': '\u2254',
+ 'comma;': ',',
+ 'commat;': '@',
+ 'comp;': '\u2201',
+ 'compfn;': '\u2218',
+ 'complement;': '\u2201',
+ 'complexes;': '\u2102',
+ 'cong;': '\u2245',
+ 'congdot;': '\u2a6d',
+ 'Congruent;': '\u2261',
+ 'Conint;': '\u222f',
+ 'conint;': '\u222e',
+ 'ContourIntegral;': '\u222e',
+ 'Copf;': '\u2102',
+ 'copf;': '\U0001d554',
+ 'coprod;': '\u2210',
+ 'Coproduct;': '\u2210',
+ 'COPY': '\xa9',
+ 'copy': '\xa9',
+ 'COPY;': '\xa9',
+ 'copy;': '\xa9',
+ 'copysr;': '\u2117',
+ 'CounterClockwiseContourIntegral;': '\u2233',
+ 'crarr;': '\u21b5',
+ 'Cross;': '\u2a2f',
+ 'cross;': '\u2717',
+ 'Cscr;': '\U0001d49e',
+ 'cscr;': '\U0001d4b8',
+ 'csub;': '\u2acf',
+ 'csube;': '\u2ad1',
+ 'csup;': '\u2ad0',
+ 'csupe;': '\u2ad2',
+ 'ctdot;': '\u22ef',
+ 'cudarrl;': '\u2938',
+ 'cudarrr;': '\u2935',
+ 'cuepr;': '\u22de',
+ 'cuesc;': '\u22df',
+ 'cularr;': '\u21b6',
+ 'cularrp;': '\u293d',
+ 'Cup;': '\u22d3',
+ 'cup;': '\u222a',
+ 'cupbrcap;': '\u2a48',
+ 'CupCap;': '\u224d',
+ 'cupcap;': '\u2a46',
+ 'cupcup;': '\u2a4a',
+ 'cupdot;': '\u228d',
+ 'cupor;': '\u2a45',
+ 'cups;': '\u222a\ufe00',
+ 'curarr;': '\u21b7',
+ 'curarrm;': '\u293c',
+ 'curlyeqprec;': '\u22de',
+ 'curlyeqsucc;': '\u22df',
+ 'curlyvee;': '\u22ce',
+ 'curlywedge;': '\u22cf',
+ 'curren': '\xa4',
+ 'curren;': '\xa4',
+ 'curvearrowleft;': '\u21b6',
+ 'curvearrowright;': '\u21b7',
+ 'cuvee;': '\u22ce',
+ 'cuwed;': '\u22cf',
+ 'cwconint;': '\u2232',
+ 'cwint;': '\u2231',
+ 'cylcty;': '\u232d',
+ 'Dagger;': '\u2021',
+ 'dagger;': '\u2020',
+ 'daleth;': '\u2138',
+ 'Darr;': '\u21a1',
+ 'dArr;': '\u21d3',
+ 'darr;': '\u2193',
+ 'dash;': '\u2010',
+ 'Dashv;': '\u2ae4',
+ 'dashv;': '\u22a3',
+ 'dbkarow;': '\u290f',
+ 'dblac;': '\u02dd',
+ 'Dcaron;': '\u010e',
+ 'dcaron;': '\u010f',
+ 'Dcy;': '\u0414',
+ 'dcy;': '\u0434',
+ 'DD;': '\u2145',
+ 'dd;': '\u2146',
+ 'ddagger;': '\u2021',
+ 'ddarr;': '\u21ca',
+ 'DDotrahd;': '\u2911',
+ 'ddotseq;': '\u2a77',
+ 'deg': '\xb0',
+ 'deg;': '\xb0',
+ 'Del;': '\u2207',
+ 'Delta;': '\u0394',
+ 'delta;': '\u03b4',
+ 'demptyv;': '\u29b1',
+ 'dfisht;': '\u297f',
+ 'Dfr;': '\U0001d507',
+ 'dfr;': '\U0001d521',
+ 'dHar;': '\u2965',
+ 'dharl;': '\u21c3',
+ 'dharr;': '\u21c2',
+ 'DiacriticalAcute;': '\xb4',
+ 'DiacriticalDot;': '\u02d9',
+ 'DiacriticalDoubleAcute;': '\u02dd',
+ 'DiacriticalGrave;': '`',
+ 'DiacriticalTilde;': '\u02dc',
+ 'diam;': '\u22c4',
+ 'Diamond;': '\u22c4',
+ 'diamond;': '\u22c4',
+ 'diamondsuit;': '\u2666',
+ 'diams;': '\u2666',
+ 'die;': '\xa8',
+ 'DifferentialD;': '\u2146',
+ 'digamma;': '\u03dd',
+ 'disin;': '\u22f2',
+ 'div;': '\xf7',
+ 'divide': '\xf7',
+ 'divide;': '\xf7',
+ 'divideontimes;': '\u22c7',
+ 'divonx;': '\u22c7',
+ 'DJcy;': '\u0402',
+ 'djcy;': '\u0452',
+ 'dlcorn;': '\u231e',
+ 'dlcrop;': '\u230d',
+ 'dollar;': '$',
+ 'Dopf;': '\U0001d53b',
+ 'dopf;': '\U0001d555',
+ 'Dot;': '\xa8',
+ 'dot;': '\u02d9',
+ 'DotDot;': '\u20dc',
+ 'doteq;': '\u2250',
+ 'doteqdot;': '\u2251',
+ 'DotEqual;': '\u2250',
+ 'dotminus;': '\u2238',
+ 'dotplus;': '\u2214',
+ 'dotsquare;': '\u22a1',
+ 'doublebarwedge;': '\u2306',
+ 'DoubleContourIntegral;': '\u222f',
+ 'DoubleDot;': '\xa8',
+ 'DoubleDownArrow;': '\u21d3',
+ 'DoubleLeftArrow;': '\u21d0',
+ 'DoubleLeftRightArrow;': '\u21d4',
+ 'DoubleLeftTee;': '\u2ae4',
+ 'DoubleLongLeftArrow;': '\u27f8',
+ 'DoubleLongLeftRightArrow;': '\u27fa',
+ 'DoubleLongRightArrow;': '\u27f9',
+ 'DoubleRightArrow;': '\u21d2',
+ 'DoubleRightTee;': '\u22a8',
+ 'DoubleUpArrow;': '\u21d1',
+ 'DoubleUpDownArrow;': '\u21d5',
+ 'DoubleVerticalBar;': '\u2225',
+ 'DownArrow;': '\u2193',
+ 'Downarrow;': '\u21d3',
+ 'downarrow;': '\u2193',
+ 'DownArrowBar;': '\u2913',
+ 'DownArrowUpArrow;': '\u21f5',
+ 'DownBreve;': '\u0311',
+ 'downdownarrows;': '\u21ca',
+ 'downharpoonleft;': '\u21c3',
+ 'downharpoonright;': '\u21c2',
+ 'DownLeftRightVector;': '\u2950',
+ 'DownLeftTeeVector;': '\u295e',
+ 'DownLeftVector;': '\u21bd',
+ 'DownLeftVectorBar;': '\u2956',
+ 'DownRightTeeVector;': '\u295f',
+ 'DownRightVector;': '\u21c1',
+ 'DownRightVectorBar;': '\u2957',
+ 'DownTee;': '\u22a4',
+ 'DownTeeArrow;': '\u21a7',
+ 'drbkarow;': '\u2910',
+ 'drcorn;': '\u231f',
+ 'drcrop;': '\u230c',
+ 'Dscr;': '\U0001d49f',
+ 'dscr;': '\U0001d4b9',
+ 'DScy;': '\u0405',
+ 'dscy;': '\u0455',
+ 'dsol;': '\u29f6',
+ 'Dstrok;': '\u0110',
+ 'dstrok;': '\u0111',
+ 'dtdot;': '\u22f1',
+ 'dtri;': '\u25bf',
+ 'dtrif;': '\u25be',
+ 'duarr;': '\u21f5',
+ 'duhar;': '\u296f',
+ 'dwangle;': '\u29a6',
+ 'DZcy;': '\u040f',
+ 'dzcy;': '\u045f',
+ 'dzigrarr;': '\u27ff',
+ 'Eacute': '\xc9',
+ 'eacute': '\xe9',
+ 'Eacute;': '\xc9',
+ 'eacute;': '\xe9',
+ 'easter;': '\u2a6e',
+ 'Ecaron;': '\u011a',
+ 'ecaron;': '\u011b',
+ 'ecir;': '\u2256',
+ 'Ecirc': '\xca',
+ 'ecirc': '\xea',
+ 'Ecirc;': '\xca',
+ 'ecirc;': '\xea',
+ 'ecolon;': '\u2255',
+ 'Ecy;': '\u042d',
+ 'ecy;': '\u044d',
+ 'eDDot;': '\u2a77',
+ 'Edot;': '\u0116',
+ 'eDot;': '\u2251',
+ 'edot;': '\u0117',
+ 'ee;': '\u2147',
+ 'efDot;': '\u2252',
+ 'Efr;': '\U0001d508',
+ 'efr;': '\U0001d522',
+ 'eg;': '\u2a9a',
+ 'Egrave': '\xc8',
+ 'egrave': '\xe8',
+ 'Egrave;': '\xc8',
+ 'egrave;': '\xe8',
+ 'egs;': '\u2a96',
+ 'egsdot;': '\u2a98',
+ 'el;': '\u2a99',
+ 'Element;': '\u2208',
+ 'elinters;': '\u23e7',
+ 'ell;': '\u2113',
+ 'els;': '\u2a95',
+ 'elsdot;': '\u2a97',
+ 'Emacr;': '\u0112',
+ 'emacr;': '\u0113',
+ 'empty;': '\u2205',
+ 'emptyset;': '\u2205',
+ 'EmptySmallSquare;': '\u25fb',
+ 'emptyv;': '\u2205',
+ 'EmptyVerySmallSquare;': '\u25ab',
+ 'emsp13;': '\u2004',
+ 'emsp14;': '\u2005',
+ 'emsp;': '\u2003',
+ 'ENG;': '\u014a',
+ 'eng;': '\u014b',
+ 'ensp;': '\u2002',
+ 'Eogon;': '\u0118',
+ 'eogon;': '\u0119',
+ 'Eopf;': '\U0001d53c',
+ 'eopf;': '\U0001d556',
+ 'epar;': '\u22d5',
+ 'eparsl;': '\u29e3',
+ 'eplus;': '\u2a71',
+ 'epsi;': '\u03b5',
+ 'Epsilon;': '\u0395',
+ 'epsilon;': '\u03b5',
+ 'epsiv;': '\u03f5',
+ 'eqcirc;': '\u2256',
+ 'eqcolon;': '\u2255',
+ 'eqsim;': '\u2242',
+ 'eqslantgtr;': '\u2a96',
+ 'eqslantless;': '\u2a95',
+ 'Equal;': '\u2a75',
+ 'equals;': '=',
+ 'EqualTilde;': '\u2242',
+ 'equest;': '\u225f',
+ 'Equilibrium;': '\u21cc',
+ 'equiv;': '\u2261',
+ 'equivDD;': '\u2a78',
+ 'eqvparsl;': '\u29e5',
+ 'erarr;': '\u2971',
+ 'erDot;': '\u2253',
+ 'Escr;': '\u2130',
+ 'escr;': '\u212f',
+ 'esdot;': '\u2250',
+ 'Esim;': '\u2a73',
+ 'esim;': '\u2242',
+ 'Eta;': '\u0397',
+ 'eta;': '\u03b7',
+ 'ETH': '\xd0',
+ 'eth': '\xf0',
+ 'ETH;': '\xd0',
+ 'eth;': '\xf0',
+ 'Euml': '\xcb',
+ 'euml': '\xeb',
+ 'Euml;': '\xcb',
+ 'euml;': '\xeb',
+ 'euro;': '\u20ac',
+ 'excl;': '!',
+ 'exist;': '\u2203',
+ 'Exists;': '\u2203',
+ 'expectation;': '\u2130',
+ 'ExponentialE;': '\u2147',
+ 'exponentiale;': '\u2147',
+ 'fallingdotseq;': '\u2252',
+ 'Fcy;': '\u0424',
+ 'fcy;': '\u0444',
+ 'female;': '\u2640',
+ 'ffilig;': '\ufb03',
+ 'fflig;': '\ufb00',
+ 'ffllig;': '\ufb04',
+ 'Ffr;': '\U0001d509',
+ 'ffr;': '\U0001d523',
+ 'filig;': '\ufb01',
+ 'FilledSmallSquare;': '\u25fc',
+ 'FilledVerySmallSquare;': '\u25aa',
+ 'fjlig;': 'fj',
+ 'flat;': '\u266d',
+ 'fllig;': '\ufb02',
+ 'fltns;': '\u25b1',
+ 'fnof;': '\u0192',
+ 'Fopf;': '\U0001d53d',
+ 'fopf;': '\U0001d557',
+ 'ForAll;': '\u2200',
+ 'forall;': '\u2200',
+ 'fork;': '\u22d4',
+ 'forkv;': '\u2ad9',
+ 'Fouriertrf;': '\u2131',
+ 'fpartint;': '\u2a0d',
+ 'frac12': '\xbd',
+ 'frac12;': '\xbd',
+ 'frac13;': '\u2153',
+ 'frac14': '\xbc',
+ 'frac14;': '\xbc',
+ 'frac15;': '\u2155',
+ 'frac16;': '\u2159',
+ 'frac18;': '\u215b',
+ 'frac23;': '\u2154',
+ 'frac25;': '\u2156',
+ 'frac34': '\xbe',
+ 'frac34;': '\xbe',
+ 'frac35;': '\u2157',
+ 'frac38;': '\u215c',
+ 'frac45;': '\u2158',
+ 'frac56;': '\u215a',
+ 'frac58;': '\u215d',
+ 'frac78;': '\u215e',
+ 'frasl;': '\u2044',
+ 'frown;': '\u2322',
+ 'Fscr;': '\u2131',
+ 'fscr;': '\U0001d4bb',
+ 'gacute;': '\u01f5',
+ 'Gamma;': '\u0393',
+ 'gamma;': '\u03b3',
+ 'Gammad;': '\u03dc',
+ 'gammad;': '\u03dd',
+ 'gap;': '\u2a86',
+ 'Gbreve;': '\u011e',
+ 'gbreve;': '\u011f',
+ 'Gcedil;': '\u0122',
+ 'Gcirc;': '\u011c',
+ 'gcirc;': '\u011d',
+ 'Gcy;': '\u0413',
+ 'gcy;': '\u0433',
+ 'Gdot;': '\u0120',
+ 'gdot;': '\u0121',
+ 'gE;': '\u2267',
+ 'ge;': '\u2265',
+ 'gEl;': '\u2a8c',
+ 'gel;': '\u22db',
+ 'geq;': '\u2265',
+ 'geqq;': '\u2267',
+ 'geqslant;': '\u2a7e',
+ 'ges;': '\u2a7e',
+ 'gescc;': '\u2aa9',
+ 'gesdot;': '\u2a80',
+ 'gesdoto;': '\u2a82',
+ 'gesdotol;': '\u2a84',
+ 'gesl;': '\u22db\ufe00',
+ 'gesles;': '\u2a94',
+ 'Gfr;': '\U0001d50a',
+ 'gfr;': '\U0001d524',
+ 'Gg;': '\u22d9',
+ 'gg;': '\u226b',
+ 'ggg;': '\u22d9',
+ 'gimel;': '\u2137',
+ 'GJcy;': '\u0403',
+ 'gjcy;': '\u0453',
+ 'gl;': '\u2277',
+ 'gla;': '\u2aa5',
+ 'glE;': '\u2a92',
+ 'glj;': '\u2aa4',
+ 'gnap;': '\u2a8a',
+ 'gnapprox;': '\u2a8a',
+ 'gnE;': '\u2269',
+ 'gne;': '\u2a88',
+ 'gneq;': '\u2a88',
+ 'gneqq;': '\u2269',
+ 'gnsim;': '\u22e7',
+ 'Gopf;': '\U0001d53e',
+ 'gopf;': '\U0001d558',
+ 'grave;': '`',
+ 'GreaterEqual;': '\u2265',
+ 'GreaterEqualLess;': '\u22db',
+ 'GreaterFullEqual;': '\u2267',
+ 'GreaterGreater;': '\u2aa2',
+ 'GreaterLess;': '\u2277',
+ 'GreaterSlantEqual;': '\u2a7e',
+ 'GreaterTilde;': '\u2273',
+ 'Gscr;': '\U0001d4a2',
+ 'gscr;': '\u210a',
+ 'gsim;': '\u2273',
+ 'gsime;': '\u2a8e',
+ 'gsiml;': '\u2a90',
+ 'GT': '>',
+ 'gt': '>',
+ 'GT;': '>',
+ 'Gt;': '\u226b',
+ 'gt;': '>',
+ 'gtcc;': '\u2aa7',
+ 'gtcir;': '\u2a7a',
+ 'gtdot;': '\u22d7',
+ 'gtlPar;': '\u2995',
+ 'gtquest;': '\u2a7c',
+ 'gtrapprox;': '\u2a86',
+ 'gtrarr;': '\u2978',
+ 'gtrdot;': '\u22d7',
+ 'gtreqless;': '\u22db',
+ 'gtreqqless;': '\u2a8c',
+ 'gtrless;': '\u2277',
+ 'gtrsim;': '\u2273',
+ 'gvertneqq;': '\u2269\ufe00',
+ 'gvnE;': '\u2269\ufe00',
+ 'Hacek;': '\u02c7',
+ 'hairsp;': '\u200a',
+ 'half;': '\xbd',
+ 'hamilt;': '\u210b',
+ 'HARDcy;': '\u042a',
+ 'hardcy;': '\u044a',
+ 'hArr;': '\u21d4',
+ 'harr;': '\u2194',
+ 'harrcir;': '\u2948',
+ 'harrw;': '\u21ad',
+ 'Hat;': '^',
+ 'hbar;': '\u210f',
+ 'Hcirc;': '\u0124',
+ 'hcirc;': '\u0125',
+ 'hearts;': '\u2665',
+ 'heartsuit;': '\u2665',
+ 'hellip;': '\u2026',
+ 'hercon;': '\u22b9',
+ 'Hfr;': '\u210c',
+ 'hfr;': '\U0001d525',
+ 'HilbertSpace;': '\u210b',
+ 'hksearow;': '\u2925',
+ 'hkswarow;': '\u2926',
+ 'hoarr;': '\u21ff',
+ 'homtht;': '\u223b',
+ 'hookleftarrow;': '\u21a9',
+ 'hookrightarrow;': '\u21aa',
+ 'Hopf;': '\u210d',
+ 'hopf;': '\U0001d559',
+ 'horbar;': '\u2015',
+ 'HorizontalLine;': '\u2500',
+ 'Hscr;': '\u210b',
+ 'hscr;': '\U0001d4bd',
+ 'hslash;': '\u210f',
+ 'Hstrok;': '\u0126',
+ 'hstrok;': '\u0127',
+ 'HumpDownHump;': '\u224e',
+ 'HumpEqual;': '\u224f',
+ 'hybull;': '\u2043',
+ 'hyphen;': '\u2010',
+ 'Iacute': '\xcd',
+ 'iacute': '\xed',
+ 'Iacute;': '\xcd',
+ 'iacute;': '\xed',
+ 'ic;': '\u2063',
+ 'Icirc': '\xce',
+ 'icirc': '\xee',
+ 'Icirc;': '\xce',
+ 'icirc;': '\xee',
+ 'Icy;': '\u0418',
+ 'icy;': '\u0438',
+ 'Idot;': '\u0130',
+ 'IEcy;': '\u0415',
+ 'iecy;': '\u0435',
+ 'iexcl': '\xa1',
+ 'iexcl;': '\xa1',
+ 'iff;': '\u21d4',
+ 'Ifr;': '\u2111',
+ 'ifr;': '\U0001d526',
+ 'Igrave': '\xcc',
+ 'igrave': '\xec',
+ 'Igrave;': '\xcc',
+ 'igrave;': '\xec',
+ 'ii;': '\u2148',
+ 'iiiint;': '\u2a0c',
+ 'iiint;': '\u222d',
+ 'iinfin;': '\u29dc',
+ 'iiota;': '\u2129',
+ 'IJlig;': '\u0132',
+ 'ijlig;': '\u0133',
+ 'Im;': '\u2111',
+ 'Imacr;': '\u012a',
+ 'imacr;': '\u012b',
+ 'image;': '\u2111',
+ 'ImaginaryI;': '\u2148',
+ 'imagline;': '\u2110',
+ 'imagpart;': '\u2111',
+ 'imath;': '\u0131',
+ 'imof;': '\u22b7',
+ 'imped;': '\u01b5',
+ 'Implies;': '\u21d2',
+ 'in;': '\u2208',
+ 'incare;': '\u2105',
+ 'infin;': '\u221e',
+ 'infintie;': '\u29dd',
+ 'inodot;': '\u0131',
+ 'Int;': '\u222c',
+ 'int;': '\u222b',
+ 'intcal;': '\u22ba',
+ 'integers;': '\u2124',
+ 'Integral;': '\u222b',
+ 'intercal;': '\u22ba',
+ 'Intersection;': '\u22c2',
+ 'intlarhk;': '\u2a17',
+ 'intprod;': '\u2a3c',
+ 'InvisibleComma;': '\u2063',
+ 'InvisibleTimes;': '\u2062',
+ 'IOcy;': '\u0401',
+ 'iocy;': '\u0451',
+ 'Iogon;': '\u012e',
+ 'iogon;': '\u012f',
+ 'Iopf;': '\U0001d540',
+ 'iopf;': '\U0001d55a',
+ 'Iota;': '\u0399',
+ 'iota;': '\u03b9',
+ 'iprod;': '\u2a3c',
+ 'iquest': '\xbf',
+ 'iquest;': '\xbf',
+ 'Iscr;': '\u2110',
+ 'iscr;': '\U0001d4be',
+ 'isin;': '\u2208',
+ 'isindot;': '\u22f5',
+ 'isinE;': '\u22f9',
+ 'isins;': '\u22f4',
+ 'isinsv;': '\u22f3',
+ 'isinv;': '\u2208',
+ 'it;': '\u2062',
+ 'Itilde;': '\u0128',
+ 'itilde;': '\u0129',
+ 'Iukcy;': '\u0406',
+ 'iukcy;': '\u0456',
+ 'Iuml': '\xcf',
+ 'iuml': '\xef',
+ 'Iuml;': '\xcf',
+ 'iuml;': '\xef',
+ 'Jcirc;': '\u0134',
+ 'jcirc;': '\u0135',
+ 'Jcy;': '\u0419',
+ 'jcy;': '\u0439',
+ 'Jfr;': '\U0001d50d',
+ 'jfr;': '\U0001d527',
+ 'jmath;': '\u0237',
+ 'Jopf;': '\U0001d541',
+ 'jopf;': '\U0001d55b',
+ 'Jscr;': '\U0001d4a5',
+ 'jscr;': '\U0001d4bf',
+ 'Jsercy;': '\u0408',
+ 'jsercy;': '\u0458',
+ 'Jukcy;': '\u0404',
+ 'jukcy;': '\u0454',
+ 'Kappa;': '\u039a',
+ 'kappa;': '\u03ba',
+ 'kappav;': '\u03f0',
+ 'Kcedil;': '\u0136',
+ 'kcedil;': '\u0137',
+ 'Kcy;': '\u041a',
+ 'kcy;': '\u043a',
+ 'Kfr;': '\U0001d50e',
+ 'kfr;': '\U0001d528',
+ 'kgreen;': '\u0138',
+ 'KHcy;': '\u0425',
+ 'khcy;': '\u0445',
+ 'KJcy;': '\u040c',
+ 'kjcy;': '\u045c',
+ 'Kopf;': '\U0001d542',
+ 'kopf;': '\U0001d55c',
+ 'Kscr;': '\U0001d4a6',
+ 'kscr;': '\U0001d4c0',
+ 'lAarr;': '\u21da',
+ 'Lacute;': '\u0139',
+ 'lacute;': '\u013a',
+ 'laemptyv;': '\u29b4',
+ 'lagran;': '\u2112',
+ 'Lambda;': '\u039b',
+ 'lambda;': '\u03bb',
+ 'Lang;': '\u27ea',
+ 'lang;': '\u27e8',
+ 'langd;': '\u2991',
+ 'langle;': '\u27e8',
+ 'lap;': '\u2a85',
+ 'Laplacetrf;': '\u2112',
+ 'laquo': '\xab',
+ 'laquo;': '\xab',
+ 'Larr;': '\u219e',
+ 'lArr;': '\u21d0',
+ 'larr;': '\u2190',
+ 'larrb;': '\u21e4',
+ 'larrbfs;': '\u291f',
+ 'larrfs;': '\u291d',
+ 'larrhk;': '\u21a9',
+ 'larrlp;': '\u21ab',
+ 'larrpl;': '\u2939',
+ 'larrsim;': '\u2973',
+ 'larrtl;': '\u21a2',
+ 'lat;': '\u2aab',
+ 'lAtail;': '\u291b',
+ 'latail;': '\u2919',
+ 'late;': '\u2aad',
+ 'lates;': '\u2aad\ufe00',
+ 'lBarr;': '\u290e',
+ 'lbarr;': '\u290c',
+ 'lbbrk;': '\u2772',
+ 'lbrace;': '{',
+ 'lbrack;': '[',
+ 'lbrke;': '\u298b',
+ 'lbrksld;': '\u298f',
+ 'lbrkslu;': '\u298d',
+ 'Lcaron;': '\u013d',
+ 'lcaron;': '\u013e',
+ 'Lcedil;': '\u013b',
+ 'lcedil;': '\u013c',
+ 'lceil;': '\u2308',
+ 'lcub;': '{',
+ 'Lcy;': '\u041b',
+ 'lcy;': '\u043b',
+ 'ldca;': '\u2936',
+ 'ldquo;': '\u201c',
+ 'ldquor;': '\u201e',
+ 'ldrdhar;': '\u2967',
+ 'ldrushar;': '\u294b',
+ 'ldsh;': '\u21b2',
+ 'lE;': '\u2266',
+ 'le;': '\u2264',
+ 'LeftAngleBracket;': '\u27e8',
+ 'LeftArrow;': '\u2190',
+ 'Leftarrow;': '\u21d0',
+ 'leftarrow;': '\u2190',
+ 'LeftArrowBar;': '\u21e4',
+ 'LeftArrowRightArrow;': '\u21c6',
+ 'leftarrowtail;': '\u21a2',
+ 'LeftCeiling;': '\u2308',
+ 'LeftDoubleBracket;': '\u27e6',
+ 'LeftDownTeeVector;': '\u2961',
+ 'LeftDownVector;': '\u21c3',
+ 'LeftDownVectorBar;': '\u2959',
+ 'LeftFloor;': '\u230a',
+ 'leftharpoondown;': '\u21bd',
+ 'leftharpoonup;': '\u21bc',
+ 'leftleftarrows;': '\u21c7',
+ 'LeftRightArrow;': '\u2194',
+ 'Leftrightarrow;': '\u21d4',
+ 'leftrightarrow;': '\u2194',
+ 'leftrightarrows;': '\u21c6',
+ 'leftrightharpoons;': '\u21cb',
+ 'leftrightsquigarrow;': '\u21ad',
+ 'LeftRightVector;': '\u294e',
+ 'LeftTee;': '\u22a3',
+ 'LeftTeeArrow;': '\u21a4',
+ 'LeftTeeVector;': '\u295a',
+ 'leftthreetimes;': '\u22cb',
+ 'LeftTriangle;': '\u22b2',
+ 'LeftTriangleBar;': '\u29cf',
+ 'LeftTriangleEqual;': '\u22b4',
+ 'LeftUpDownVector;': '\u2951',
+ 'LeftUpTeeVector;': '\u2960',
+ 'LeftUpVector;': '\u21bf',
+ 'LeftUpVectorBar;': '\u2958',
+ 'LeftVector;': '\u21bc',
+ 'LeftVectorBar;': '\u2952',
+ 'lEg;': '\u2a8b',
+ 'leg;': '\u22da',
+ 'leq;': '\u2264',
+ 'leqq;': '\u2266',
+ 'leqslant;': '\u2a7d',
+ 'les;': '\u2a7d',
+ 'lescc;': '\u2aa8',
+ 'lesdot;': '\u2a7f',
+ 'lesdoto;': '\u2a81',
+ 'lesdotor;': '\u2a83',
+ 'lesg;': '\u22da\ufe00',
+ 'lesges;': '\u2a93',
+ 'lessapprox;': '\u2a85',
+ 'lessdot;': '\u22d6',
+ 'lesseqgtr;': '\u22da',
+ 'lesseqqgtr;': '\u2a8b',
+ 'LessEqualGreater;': '\u22da',
+ 'LessFullEqual;': '\u2266',
+ 'LessGreater;': '\u2276',
+ 'lessgtr;': '\u2276',
+ 'LessLess;': '\u2aa1',
+ 'lesssim;': '\u2272',
+ 'LessSlantEqual;': '\u2a7d',
+ 'LessTilde;': '\u2272',
+ 'lfisht;': '\u297c',
+ 'lfloor;': '\u230a',
+ 'Lfr;': '\U0001d50f',
+ 'lfr;': '\U0001d529',
+ 'lg;': '\u2276',
+ 'lgE;': '\u2a91',
+ 'lHar;': '\u2962',
+ 'lhard;': '\u21bd',
+ 'lharu;': '\u21bc',
+ 'lharul;': '\u296a',
+ 'lhblk;': '\u2584',
+ 'LJcy;': '\u0409',
+ 'ljcy;': '\u0459',
+ 'Ll;': '\u22d8',
+ 'll;': '\u226a',
+ 'llarr;': '\u21c7',
+ 'llcorner;': '\u231e',
+ 'Lleftarrow;': '\u21da',
+ 'llhard;': '\u296b',
+ 'lltri;': '\u25fa',
+ 'Lmidot;': '\u013f',
+ 'lmidot;': '\u0140',
+ 'lmoust;': '\u23b0',
+ 'lmoustache;': '\u23b0',
+ 'lnap;': '\u2a89',
+ 'lnapprox;': '\u2a89',
+ 'lnE;': '\u2268',
+ 'lne;': '\u2a87',
+ 'lneq;': '\u2a87',
+ 'lneqq;': '\u2268',
+ 'lnsim;': '\u22e6',
+ 'loang;': '\u27ec',
+ 'loarr;': '\u21fd',
+ 'lobrk;': '\u27e6',
+ 'LongLeftArrow;': '\u27f5',
+ 'Longleftarrow;': '\u27f8',
+ 'longleftarrow;': '\u27f5',
+ 'LongLeftRightArrow;': '\u27f7',
+ 'Longleftrightarrow;': '\u27fa',
+ 'longleftrightarrow;': '\u27f7',
+ 'longmapsto;': '\u27fc',
+ 'LongRightArrow;': '\u27f6',
+ 'Longrightarrow;': '\u27f9',
+ 'longrightarrow;': '\u27f6',
+ 'looparrowleft;': '\u21ab',
+ 'looparrowright;': '\u21ac',
+ 'lopar;': '\u2985',
+ 'Lopf;': '\U0001d543',
+ 'lopf;': '\U0001d55d',
+ 'loplus;': '\u2a2d',
+ 'lotimes;': '\u2a34',
+ 'lowast;': '\u2217',
+ 'lowbar;': '_',
+ 'LowerLeftArrow;': '\u2199',
+ 'LowerRightArrow;': '\u2198',
+ 'loz;': '\u25ca',
+ 'lozenge;': '\u25ca',
+ 'lozf;': '\u29eb',
+ 'lpar;': '(',
+ 'lparlt;': '\u2993',
+ 'lrarr;': '\u21c6',
+ 'lrcorner;': '\u231f',
+ 'lrhar;': '\u21cb',
+ 'lrhard;': '\u296d',
+ 'lrm;': '\u200e',
+ 'lrtri;': '\u22bf',
+ 'lsaquo;': '\u2039',
+ 'Lscr;': '\u2112',
+ 'lscr;': '\U0001d4c1',
+ 'Lsh;': '\u21b0',
+ 'lsh;': '\u21b0',
+ 'lsim;': '\u2272',
+ 'lsime;': '\u2a8d',
+ 'lsimg;': '\u2a8f',
+ 'lsqb;': '[',
+ 'lsquo;': '\u2018',
+ 'lsquor;': '\u201a',
+ 'Lstrok;': '\u0141',
+ 'lstrok;': '\u0142',
+ 'LT': '<',
+ 'lt': '<',
+ 'LT;': '<',
+ 'Lt;': '\u226a',
+ 'lt;': '<',
+ 'ltcc;': '\u2aa6',
+ 'ltcir;': '\u2a79',
+ 'ltdot;': '\u22d6',
+ 'lthree;': '\u22cb',
+ 'ltimes;': '\u22c9',
+ 'ltlarr;': '\u2976',
+ 'ltquest;': '\u2a7b',
+ 'ltri;': '\u25c3',
+ 'ltrie;': '\u22b4',
+ 'ltrif;': '\u25c2',
+ 'ltrPar;': '\u2996',
+ 'lurdshar;': '\u294a',
+ 'luruhar;': '\u2966',
+ 'lvertneqq;': '\u2268\ufe00',
+ 'lvnE;': '\u2268\ufe00',
+ 'macr': '\xaf',
+ 'macr;': '\xaf',
+ 'male;': '\u2642',
+ 'malt;': '\u2720',
+ 'maltese;': '\u2720',
+ 'Map;': '\u2905',
+ 'map;': '\u21a6',
+ 'mapsto;': '\u21a6',
+ 'mapstodown;': '\u21a7',
+ 'mapstoleft;': '\u21a4',
+ 'mapstoup;': '\u21a5',
+ 'marker;': '\u25ae',
+ 'mcomma;': '\u2a29',
+ 'Mcy;': '\u041c',
+ 'mcy;': '\u043c',
+ 'mdash;': '\u2014',
+ 'mDDot;': '\u223a',
+ 'measuredangle;': '\u2221',
+ 'MediumSpace;': '\u205f',
+ 'Mellintrf;': '\u2133',
+ 'Mfr;': '\U0001d510',
+ 'mfr;': '\U0001d52a',
+ 'mho;': '\u2127',
+ 'micro': '\xb5',
+ 'micro;': '\xb5',
+ 'mid;': '\u2223',
+ 'midast;': '*',
+ 'midcir;': '\u2af0',
+ 'middot': '\xb7',
+ 'middot;': '\xb7',
+ 'minus;': '\u2212',
+ 'minusb;': '\u229f',
+ 'minusd;': '\u2238',
+ 'minusdu;': '\u2a2a',
+ 'MinusPlus;': '\u2213',
+ 'mlcp;': '\u2adb',
+ 'mldr;': '\u2026',
+ 'mnplus;': '\u2213',
+ 'models;': '\u22a7',
+ 'Mopf;': '\U0001d544',
+ 'mopf;': '\U0001d55e',
+ 'mp;': '\u2213',
+ 'Mscr;': '\u2133',
+ 'mscr;': '\U0001d4c2',
+ 'mstpos;': '\u223e',
+ 'Mu;': '\u039c',
+ 'mu;': '\u03bc',
+ 'multimap;': '\u22b8',
+ 'mumap;': '\u22b8',
+ 'nabla;': '\u2207',
+ 'Nacute;': '\u0143',
+ 'nacute;': '\u0144',
+ 'nang;': '\u2220\u20d2',
+ 'nap;': '\u2249',
+ 'napE;': '\u2a70\u0338',
+ 'napid;': '\u224b\u0338',
+ 'napos;': '\u0149',
+ 'napprox;': '\u2249',
+ 'natur;': '\u266e',
+ 'natural;': '\u266e',
+ 'naturals;': '\u2115',
+ 'nbsp': '\xa0',
+ 'nbsp;': '\xa0',
+ 'nbump;': '\u224e\u0338',
+ 'nbumpe;': '\u224f\u0338',
+ 'ncap;': '\u2a43',
+ 'Ncaron;': '\u0147',
+ 'ncaron;': '\u0148',
+ 'Ncedil;': '\u0145',
+ 'ncedil;': '\u0146',
+ 'ncong;': '\u2247',
+ 'ncongdot;': '\u2a6d\u0338',
+ 'ncup;': '\u2a42',
+ 'Ncy;': '\u041d',
+ 'ncy;': '\u043d',
+ 'ndash;': '\u2013',
+ 'ne;': '\u2260',
+ 'nearhk;': '\u2924',
+ 'neArr;': '\u21d7',
+ 'nearr;': '\u2197',
+ 'nearrow;': '\u2197',
+ 'nedot;': '\u2250\u0338',
+ 'NegativeMediumSpace;': '\u200b',
+ 'NegativeThickSpace;': '\u200b',
+ 'NegativeThinSpace;': '\u200b',
+ 'NegativeVeryThinSpace;': '\u200b',
+ 'nequiv;': '\u2262',
+ 'nesear;': '\u2928',
+ 'nesim;': '\u2242\u0338',
+ 'NestedGreaterGreater;': '\u226b',
+ 'NestedLessLess;': '\u226a',
+ 'NewLine;': '\n',
+ 'nexist;': '\u2204',
+ 'nexists;': '\u2204',
+ 'Nfr;': '\U0001d511',
+ 'nfr;': '\U0001d52b',
+ 'ngE;': '\u2267\u0338',
+ 'nge;': '\u2271',
+ 'ngeq;': '\u2271',
+ 'ngeqq;': '\u2267\u0338',
+ 'ngeqslant;': '\u2a7e\u0338',
+ 'nges;': '\u2a7e\u0338',
+ 'nGg;': '\u22d9\u0338',
+ 'ngsim;': '\u2275',
+ 'nGt;': '\u226b\u20d2',
+ 'ngt;': '\u226f',
+ 'ngtr;': '\u226f',
+ 'nGtv;': '\u226b\u0338',
+ 'nhArr;': '\u21ce',
+ 'nharr;': '\u21ae',
+ 'nhpar;': '\u2af2',
+ 'ni;': '\u220b',
+ 'nis;': '\u22fc',
+ 'nisd;': '\u22fa',
+ 'niv;': '\u220b',
+ 'NJcy;': '\u040a',
+ 'njcy;': '\u045a',
+ 'nlArr;': '\u21cd',
+ 'nlarr;': '\u219a',
+ 'nldr;': '\u2025',
+ 'nlE;': '\u2266\u0338',
+ 'nle;': '\u2270',
+ 'nLeftarrow;': '\u21cd',
+ 'nleftarrow;': '\u219a',
+ 'nLeftrightarrow;': '\u21ce',
+ 'nleftrightarrow;': '\u21ae',
+ 'nleq;': '\u2270',
+ 'nleqq;': '\u2266\u0338',
+ 'nleqslant;': '\u2a7d\u0338',
+ 'nles;': '\u2a7d\u0338',
+ 'nless;': '\u226e',
+ 'nLl;': '\u22d8\u0338',
+ 'nlsim;': '\u2274',
+ 'nLt;': '\u226a\u20d2',
+ 'nlt;': '\u226e',
+ 'nltri;': '\u22ea',
+ 'nltrie;': '\u22ec',
+ 'nLtv;': '\u226a\u0338',
+ 'nmid;': '\u2224',
+ 'NoBreak;': '\u2060',
+ 'NonBreakingSpace;': '\xa0',
+ 'Nopf;': '\u2115',
+ 'nopf;': '\U0001d55f',
+ 'not': '\xac',
+ 'Not;': '\u2aec',
+ 'not;': '\xac',
+ 'NotCongruent;': '\u2262',
+ 'NotCupCap;': '\u226d',
+ 'NotDoubleVerticalBar;': '\u2226',
+ 'NotElement;': '\u2209',
+ 'NotEqual;': '\u2260',
+ 'NotEqualTilde;': '\u2242\u0338',
+ 'NotExists;': '\u2204',
+ 'NotGreater;': '\u226f',
+ 'NotGreaterEqual;': '\u2271',
+ 'NotGreaterFullEqual;': '\u2267\u0338',
+ 'NotGreaterGreater;': '\u226b\u0338',
+ 'NotGreaterLess;': '\u2279',
+ 'NotGreaterSlantEqual;': '\u2a7e\u0338',
+ 'NotGreaterTilde;': '\u2275',
+ 'NotHumpDownHump;': '\u224e\u0338',
+ 'NotHumpEqual;': '\u224f\u0338',
+ 'notin;': '\u2209',
+ 'notindot;': '\u22f5\u0338',
+ 'notinE;': '\u22f9\u0338',
+ 'notinva;': '\u2209',
+ 'notinvb;': '\u22f7',
+ 'notinvc;': '\u22f6',
+ 'NotLeftTriangle;': '\u22ea',
+ 'NotLeftTriangleBar;': '\u29cf\u0338',
+ 'NotLeftTriangleEqual;': '\u22ec',
+ 'NotLess;': '\u226e',
+ 'NotLessEqual;': '\u2270',
+ 'NotLessGreater;': '\u2278',
+ 'NotLessLess;': '\u226a\u0338',
+ 'NotLessSlantEqual;': '\u2a7d\u0338',
+ 'NotLessTilde;': '\u2274',
+ 'NotNestedGreaterGreater;': '\u2aa2\u0338',
+ 'NotNestedLessLess;': '\u2aa1\u0338',
+ 'notni;': '\u220c',
+ 'notniva;': '\u220c',
+ 'notnivb;': '\u22fe',
+ 'notnivc;': '\u22fd',
+ 'NotPrecedes;': '\u2280',
+ 'NotPrecedesEqual;': '\u2aaf\u0338',
+ 'NotPrecedesSlantEqual;': '\u22e0',
+ 'NotReverseElement;': '\u220c',
+ 'NotRightTriangle;': '\u22eb',
+ 'NotRightTriangleBar;': '\u29d0\u0338',
+ 'NotRightTriangleEqual;': '\u22ed',
+ 'NotSquareSubset;': '\u228f\u0338',
+ 'NotSquareSubsetEqual;': '\u22e2',
+ 'NotSquareSuperset;': '\u2290\u0338',
+ 'NotSquareSupersetEqual;': '\u22e3',
+ 'NotSubset;': '\u2282\u20d2',
+ 'NotSubsetEqual;': '\u2288',
+ 'NotSucceeds;': '\u2281',
+ 'NotSucceedsEqual;': '\u2ab0\u0338',
+ 'NotSucceedsSlantEqual;': '\u22e1',
+ 'NotSucceedsTilde;': '\u227f\u0338',
+ 'NotSuperset;': '\u2283\u20d2',
+ 'NotSupersetEqual;': '\u2289',
+ 'NotTilde;': '\u2241',
+ 'NotTildeEqual;': '\u2244',
+ 'NotTildeFullEqual;': '\u2247',
+ 'NotTildeTilde;': '\u2249',
+ 'NotVerticalBar;': '\u2224',
+ 'npar;': '\u2226',
+ 'nparallel;': '\u2226',
+ 'nparsl;': '\u2afd\u20e5',
+ 'npart;': '\u2202\u0338',
+ 'npolint;': '\u2a14',
+ 'npr;': '\u2280',
+ 'nprcue;': '\u22e0',
+ 'npre;': '\u2aaf\u0338',
+ 'nprec;': '\u2280',
+ 'npreceq;': '\u2aaf\u0338',
+ 'nrArr;': '\u21cf',
+ 'nrarr;': '\u219b',
+ 'nrarrc;': '\u2933\u0338',
+ 'nrarrw;': '\u219d\u0338',
+ 'nRightarrow;': '\u21cf',
+ 'nrightarrow;': '\u219b',
+ 'nrtri;': '\u22eb',
+ 'nrtrie;': '\u22ed',
+ 'nsc;': '\u2281',
+ 'nsccue;': '\u22e1',
+ 'nsce;': '\u2ab0\u0338',
+ 'Nscr;': '\U0001d4a9',
+ 'nscr;': '\U0001d4c3',
+ 'nshortmid;': '\u2224',
+ 'nshortparallel;': '\u2226',
+ 'nsim;': '\u2241',
+ 'nsime;': '\u2244',
+ 'nsimeq;': '\u2244',
+ 'nsmid;': '\u2224',
+ 'nspar;': '\u2226',
+ 'nsqsube;': '\u22e2',
+ 'nsqsupe;': '\u22e3',
+ 'nsub;': '\u2284',
+ 'nsubE;': '\u2ac5\u0338',
+ 'nsube;': '\u2288',
+ 'nsubset;': '\u2282\u20d2',
+ 'nsubseteq;': '\u2288',
+ 'nsubseteqq;': '\u2ac5\u0338',
+ 'nsucc;': '\u2281',
+ 'nsucceq;': '\u2ab0\u0338',
+ 'nsup;': '\u2285',
+ 'nsupE;': '\u2ac6\u0338',
+ 'nsupe;': '\u2289',
+ 'nsupset;': '\u2283\u20d2',
+ 'nsupseteq;': '\u2289',
+ 'nsupseteqq;': '\u2ac6\u0338',
+ 'ntgl;': '\u2279',
+ 'Ntilde': '\xd1',
+ 'ntilde': '\xf1',
+ 'Ntilde;': '\xd1',
+ 'ntilde;': '\xf1',
+ 'ntlg;': '\u2278',
+ 'ntriangleleft;': '\u22ea',
+ 'ntrianglelefteq;': '\u22ec',
+ 'ntriangleright;': '\u22eb',
+ 'ntrianglerighteq;': '\u22ed',
+ 'Nu;': '\u039d',
+ 'nu;': '\u03bd',
+ 'num;': '#',
+ 'numero;': '\u2116',
+ 'numsp;': '\u2007',
+ 'nvap;': '\u224d\u20d2',
+ 'nVDash;': '\u22af',
+ 'nVdash;': '\u22ae',
+ 'nvDash;': '\u22ad',
+ 'nvdash;': '\u22ac',
+ 'nvge;': '\u2265\u20d2',
+ 'nvgt;': '>\u20d2',
+ 'nvHarr;': '\u2904',
+ 'nvinfin;': '\u29de',
+ 'nvlArr;': '\u2902',
+ 'nvle;': '\u2264\u20d2',
+ 'nvlt;': '<\u20d2',
+ 'nvltrie;': '\u22b4\u20d2',
+ 'nvrArr;': '\u2903',
+ 'nvrtrie;': '\u22b5\u20d2',
+ 'nvsim;': '\u223c\u20d2',
+ 'nwarhk;': '\u2923',
+ 'nwArr;': '\u21d6',
+ 'nwarr;': '\u2196',
+ 'nwarrow;': '\u2196',
+ 'nwnear;': '\u2927',
+ 'Oacute': '\xd3',
+ 'oacute': '\xf3',
+ 'Oacute;': '\xd3',
+ 'oacute;': '\xf3',
+ 'oast;': '\u229b',
+ 'ocir;': '\u229a',
+ 'Ocirc': '\xd4',
+ 'ocirc': '\xf4',
+ 'Ocirc;': '\xd4',
+ 'ocirc;': '\xf4',
+ 'Ocy;': '\u041e',
+ 'ocy;': '\u043e',
+ 'odash;': '\u229d',
+ 'Odblac;': '\u0150',
+ 'odblac;': '\u0151',
+ 'odiv;': '\u2a38',
+ 'odot;': '\u2299',
+ 'odsold;': '\u29bc',
+ 'OElig;': '\u0152',
+ 'oelig;': '\u0153',
+ 'ofcir;': '\u29bf',
+ 'Ofr;': '\U0001d512',
+ 'ofr;': '\U0001d52c',
+ 'ogon;': '\u02db',
+ 'Ograve': '\xd2',
+ 'ograve': '\xf2',
+ 'Ograve;': '\xd2',
+ 'ograve;': '\xf2',
+ 'ogt;': '\u29c1',
+ 'ohbar;': '\u29b5',
+ 'ohm;': '\u03a9',
+ 'oint;': '\u222e',
+ 'olarr;': '\u21ba',
+ 'olcir;': '\u29be',
+ 'olcross;': '\u29bb',
+ 'oline;': '\u203e',
+ 'olt;': '\u29c0',
+ 'Omacr;': '\u014c',
+ 'omacr;': '\u014d',
+ 'Omega;': '\u03a9',
+ 'omega;': '\u03c9',
+ 'Omicron;': '\u039f',
+ 'omicron;': '\u03bf',
+ 'omid;': '\u29b6',
+ 'ominus;': '\u2296',
+ 'Oopf;': '\U0001d546',
+ 'oopf;': '\U0001d560',
+ 'opar;': '\u29b7',
+ 'OpenCurlyDoubleQuote;': '\u201c',
+ 'OpenCurlyQuote;': '\u2018',
+ 'operp;': '\u29b9',
+ 'oplus;': '\u2295',
+ 'Or;': '\u2a54',
+ 'or;': '\u2228',
+ 'orarr;': '\u21bb',
+ 'ord;': '\u2a5d',
+ 'order;': '\u2134',
+ 'orderof;': '\u2134',
+ 'ordf': '\xaa',
+ 'ordf;': '\xaa',
+ 'ordm': '\xba',
+ 'ordm;': '\xba',
+ 'origof;': '\u22b6',
+ 'oror;': '\u2a56',
+ 'orslope;': '\u2a57',
+ 'orv;': '\u2a5b',
+ 'oS;': '\u24c8',
+ 'Oscr;': '\U0001d4aa',
+ 'oscr;': '\u2134',
+ 'Oslash': '\xd8',
+ 'oslash': '\xf8',
+ 'Oslash;': '\xd8',
+ 'oslash;': '\xf8',
+ 'osol;': '\u2298',
+ 'Otilde': '\xd5',
+ 'otilde': '\xf5',
+ 'Otilde;': '\xd5',
+ 'otilde;': '\xf5',
+ 'Otimes;': '\u2a37',
+ 'otimes;': '\u2297',
+ 'otimesas;': '\u2a36',
+ 'Ouml': '\xd6',
+ 'ouml': '\xf6',
+ 'Ouml;': '\xd6',
+ 'ouml;': '\xf6',
+ 'ovbar;': '\u233d',
+ 'OverBar;': '\u203e',
+ 'OverBrace;': '\u23de',
+ 'OverBracket;': '\u23b4',
+ 'OverParenthesis;': '\u23dc',
+ 'par;': '\u2225',
+ 'para': '\xb6',
+ 'para;': '\xb6',
+ 'parallel;': '\u2225',
+ 'parsim;': '\u2af3',
+ 'parsl;': '\u2afd',
+ 'part;': '\u2202',
+ 'PartialD;': '\u2202',
+ 'Pcy;': '\u041f',
+ 'pcy;': '\u043f',
+ 'percnt;': '%',
+ 'period;': '.',
+ 'permil;': '\u2030',
+ 'perp;': '\u22a5',
+ 'pertenk;': '\u2031',
+ 'Pfr;': '\U0001d513',
+ 'pfr;': '\U0001d52d',
+ 'Phi;': '\u03a6',
+ 'phi;': '\u03c6',
+ 'phiv;': '\u03d5',
+ 'phmmat;': '\u2133',
+ 'phone;': '\u260e',
+ 'Pi;': '\u03a0',
+ 'pi;': '\u03c0',
+ 'pitchfork;': '\u22d4',
+ 'piv;': '\u03d6',
+ 'planck;': '\u210f',
+ 'planckh;': '\u210e',
+ 'plankv;': '\u210f',
+ 'plus;': '+',
+ 'plusacir;': '\u2a23',
+ 'plusb;': '\u229e',
+ 'pluscir;': '\u2a22',
+ 'plusdo;': '\u2214',
+ 'plusdu;': '\u2a25',
+ 'pluse;': '\u2a72',
+ 'PlusMinus;': '\xb1',
+ 'plusmn': '\xb1',
+ 'plusmn;': '\xb1',
+ 'plussim;': '\u2a26',
+ 'plustwo;': '\u2a27',
+ 'pm;': '\xb1',
+ 'Poincareplane;': '\u210c',
+ 'pointint;': '\u2a15',
+ 'Popf;': '\u2119',
+ 'popf;': '\U0001d561',
+ 'pound': '\xa3',
+ 'pound;': '\xa3',
+ 'Pr;': '\u2abb',
+ 'pr;': '\u227a',
+ 'prap;': '\u2ab7',
+ 'prcue;': '\u227c',
+ 'prE;': '\u2ab3',
+ 'pre;': '\u2aaf',
+ 'prec;': '\u227a',
+ 'precapprox;': '\u2ab7',
+ 'preccurlyeq;': '\u227c',
+ 'Precedes;': '\u227a',
+ 'PrecedesEqual;': '\u2aaf',
+ 'PrecedesSlantEqual;': '\u227c',
+ 'PrecedesTilde;': '\u227e',
+ 'preceq;': '\u2aaf',
+ 'precnapprox;': '\u2ab9',
+ 'precneqq;': '\u2ab5',
+ 'precnsim;': '\u22e8',
+ 'precsim;': '\u227e',
+ 'Prime;': '\u2033',
+ 'prime;': '\u2032',
+ 'primes;': '\u2119',
+ 'prnap;': '\u2ab9',
+ 'prnE;': '\u2ab5',
+ 'prnsim;': '\u22e8',
+ 'prod;': '\u220f',
+ 'Product;': '\u220f',
+ 'profalar;': '\u232e',
+ 'profline;': '\u2312',
+ 'profsurf;': '\u2313',
+ 'prop;': '\u221d',
+ 'Proportion;': '\u2237',
+ 'Proportional;': '\u221d',
+ 'propto;': '\u221d',
+ 'prsim;': '\u227e',
+ 'prurel;': '\u22b0',
+ 'Pscr;': '\U0001d4ab',
+ 'pscr;': '\U0001d4c5',
+ 'Psi;': '\u03a8',
+ 'psi;': '\u03c8',
+ 'puncsp;': '\u2008',
+ 'Qfr;': '\U0001d514',
+ 'qfr;': '\U0001d52e',
+ 'qint;': '\u2a0c',
+ 'Qopf;': '\u211a',
+ 'qopf;': '\U0001d562',
+ 'qprime;': '\u2057',
+ 'Qscr;': '\U0001d4ac',
+ 'qscr;': '\U0001d4c6',
+ 'quaternions;': '\u210d',
+ 'quatint;': '\u2a16',
+ 'quest;': '?',
+ 'questeq;': '\u225f',
+ 'QUOT': '"',
+ 'quot': '"',
+ 'QUOT;': '"',
+ 'quot;': '"',
+ 'rAarr;': '\u21db',
+ 'race;': '\u223d\u0331',
+ 'Racute;': '\u0154',
+ 'racute;': '\u0155',
+ 'radic;': '\u221a',
+ 'raemptyv;': '\u29b3',
+ 'Rang;': '\u27eb',
+ 'rang;': '\u27e9',
+ 'rangd;': '\u2992',
+ 'range;': '\u29a5',
+ 'rangle;': '\u27e9',
+ 'raquo': '\xbb',
+ 'raquo;': '\xbb',
+ 'Rarr;': '\u21a0',
+ 'rArr;': '\u21d2',
+ 'rarr;': '\u2192',
+ 'rarrap;': '\u2975',
+ 'rarrb;': '\u21e5',
+ 'rarrbfs;': '\u2920',
+ 'rarrc;': '\u2933',
+ 'rarrfs;': '\u291e',
+ 'rarrhk;': '\u21aa',
+ 'rarrlp;': '\u21ac',
+ 'rarrpl;': '\u2945',
+ 'rarrsim;': '\u2974',
+ 'Rarrtl;': '\u2916',
+ 'rarrtl;': '\u21a3',
+ 'rarrw;': '\u219d',
+ 'rAtail;': '\u291c',
+ 'ratail;': '\u291a',
+ 'ratio;': '\u2236',
+ 'rationals;': '\u211a',
+ 'RBarr;': '\u2910',
+ 'rBarr;': '\u290f',
+ 'rbarr;': '\u290d',
+ 'rbbrk;': '\u2773',
+ 'rbrace;': '}',
+ 'rbrack;': ']',
+ 'rbrke;': '\u298c',
+ 'rbrksld;': '\u298e',
+ 'rbrkslu;': '\u2990',
+ 'Rcaron;': '\u0158',
+ 'rcaron;': '\u0159',
+ 'Rcedil;': '\u0156',
+ 'rcedil;': '\u0157',
+ 'rceil;': '\u2309',
+ 'rcub;': '}',
+ 'Rcy;': '\u0420',
+ 'rcy;': '\u0440',
+ 'rdca;': '\u2937',
+ 'rdldhar;': '\u2969',
+ 'rdquo;': '\u201d',
+ 'rdquor;': '\u201d',
+ 'rdsh;': '\u21b3',
+ 'Re;': '\u211c',
+ 'real;': '\u211c',
+ 'realine;': '\u211b',
+ 'realpart;': '\u211c',
+ 'reals;': '\u211d',
+ 'rect;': '\u25ad',
+ 'REG': '\xae',
+ 'reg': '\xae',
+ 'REG;': '\xae',
+ 'reg;': '\xae',
+ 'ReverseElement;': '\u220b',
+ 'ReverseEquilibrium;': '\u21cb',
+ 'ReverseUpEquilibrium;': '\u296f',
+ 'rfisht;': '\u297d',
+ 'rfloor;': '\u230b',
+ 'Rfr;': '\u211c',
+ 'rfr;': '\U0001d52f',
+ 'rHar;': '\u2964',
+ 'rhard;': '\u21c1',
+ 'rharu;': '\u21c0',
+ 'rharul;': '\u296c',
+ 'Rho;': '\u03a1',
+ 'rho;': '\u03c1',
+ 'rhov;': '\u03f1',
+ 'RightAngleBracket;': '\u27e9',
+ 'RightArrow;': '\u2192',
+ 'Rightarrow;': '\u21d2',
+ 'rightarrow;': '\u2192',
+ 'RightArrowBar;': '\u21e5',
+ 'RightArrowLeftArrow;': '\u21c4',
+ 'rightarrowtail;': '\u21a3',
+ 'RightCeiling;': '\u2309',
+ 'RightDoubleBracket;': '\u27e7',
+ 'RightDownTeeVector;': '\u295d',
+ 'RightDownVector;': '\u21c2',
+ 'RightDownVectorBar;': '\u2955',
+ 'RightFloor;': '\u230b',
+ 'rightharpoondown;': '\u21c1',
+ 'rightharpoonup;': '\u21c0',
+ 'rightleftarrows;': '\u21c4',
+ 'rightleftharpoons;': '\u21cc',
+ 'rightrightarrows;': '\u21c9',
+ 'rightsquigarrow;': '\u219d',
+ 'RightTee;': '\u22a2',
+ 'RightTeeArrow;': '\u21a6',
+ 'RightTeeVector;': '\u295b',
+ 'rightthreetimes;': '\u22cc',
+ 'RightTriangle;': '\u22b3',
+ 'RightTriangleBar;': '\u29d0',
+ 'RightTriangleEqual;': '\u22b5',
+ 'RightUpDownVector;': '\u294f',
+ 'RightUpTeeVector;': '\u295c',
+ 'RightUpVector;': '\u21be',
+ 'RightUpVectorBar;': '\u2954',
+ 'RightVector;': '\u21c0',
+ 'RightVectorBar;': '\u2953',
+ 'ring;': '\u02da',
+ 'risingdotseq;': '\u2253',
+ 'rlarr;': '\u21c4',
+ 'rlhar;': '\u21cc',
+ 'rlm;': '\u200f',
+ 'rmoust;': '\u23b1',
+ 'rmoustache;': '\u23b1',
+ 'rnmid;': '\u2aee',
+ 'roang;': '\u27ed',
+ 'roarr;': '\u21fe',
+ 'robrk;': '\u27e7',
+ 'ropar;': '\u2986',
+ 'Ropf;': '\u211d',
+ 'ropf;': '\U0001d563',
+ 'roplus;': '\u2a2e',
+ 'rotimes;': '\u2a35',
+ 'RoundImplies;': '\u2970',
+ 'rpar;': ')',
+ 'rpargt;': '\u2994',
+ 'rppolint;': '\u2a12',
+ 'rrarr;': '\u21c9',
+ 'Rrightarrow;': '\u21db',
+ 'rsaquo;': '\u203a',
+ 'Rscr;': '\u211b',
+ 'rscr;': '\U0001d4c7',
+ 'Rsh;': '\u21b1',
+ 'rsh;': '\u21b1',
+ 'rsqb;': ']',
+ 'rsquo;': '\u2019',
+ 'rsquor;': '\u2019',
+ 'rthree;': '\u22cc',
+ 'rtimes;': '\u22ca',
+ 'rtri;': '\u25b9',
+ 'rtrie;': '\u22b5',
+ 'rtrif;': '\u25b8',
+ 'rtriltri;': '\u29ce',
+ 'RuleDelayed;': '\u29f4',
+ 'ruluhar;': '\u2968',
+ 'rx;': '\u211e',
+ 'Sacute;': '\u015a',
+ 'sacute;': '\u015b',
+ 'sbquo;': '\u201a',
+ 'Sc;': '\u2abc',
+ 'sc;': '\u227b',
+ 'scap;': '\u2ab8',
+ 'Scaron;': '\u0160',
+ 'scaron;': '\u0161',
+ 'sccue;': '\u227d',
+ 'scE;': '\u2ab4',
+ 'sce;': '\u2ab0',
+ 'Scedil;': '\u015e',
+ 'scedil;': '\u015f',
+ 'Scirc;': '\u015c',
+ 'scirc;': '\u015d',
+ 'scnap;': '\u2aba',
+ 'scnE;': '\u2ab6',
+ 'scnsim;': '\u22e9',
+ 'scpolint;': '\u2a13',
+ 'scsim;': '\u227f',
+ 'Scy;': '\u0421',
+ 'scy;': '\u0441',
+ 'sdot;': '\u22c5',
+ 'sdotb;': '\u22a1',
+ 'sdote;': '\u2a66',
+ 'searhk;': '\u2925',
+ 'seArr;': '\u21d8',
+ 'searr;': '\u2198',
+ 'searrow;': '\u2198',
+ 'sect': '\xa7',
+ 'sect;': '\xa7',
+ 'semi;': ';',
+ 'seswar;': '\u2929',
+ 'setminus;': '\u2216',
+ 'setmn;': '\u2216',
+ 'sext;': '\u2736',
+ 'Sfr;': '\U0001d516',
+ 'sfr;': '\U0001d530',
+ 'sfrown;': '\u2322',
+ 'sharp;': '\u266f',
+ 'SHCHcy;': '\u0429',
+ 'shchcy;': '\u0449',
+ 'SHcy;': '\u0428',
+ 'shcy;': '\u0448',
+ 'ShortDownArrow;': '\u2193',
+ 'ShortLeftArrow;': '\u2190',
+ 'shortmid;': '\u2223',
+ 'shortparallel;': '\u2225',
+ 'ShortRightArrow;': '\u2192',
+ 'ShortUpArrow;': '\u2191',
+ 'shy': '\xad',
+ 'shy;': '\xad',
+ 'Sigma;': '\u03a3',
+ 'sigma;': '\u03c3',
+ 'sigmaf;': '\u03c2',
+ 'sigmav;': '\u03c2',
+ 'sim;': '\u223c',
+ 'simdot;': '\u2a6a',
+ 'sime;': '\u2243',
+ 'simeq;': '\u2243',
+ 'simg;': '\u2a9e',
+ 'simgE;': '\u2aa0',
+ 'siml;': '\u2a9d',
+ 'simlE;': '\u2a9f',
+ 'simne;': '\u2246',
+ 'simplus;': '\u2a24',
+ 'simrarr;': '\u2972',
+ 'slarr;': '\u2190',
+ 'SmallCircle;': '\u2218',
+ 'smallsetminus;': '\u2216',
+ 'smashp;': '\u2a33',
+ 'smeparsl;': '\u29e4',
+ 'smid;': '\u2223',
+ 'smile;': '\u2323',
+ 'smt;': '\u2aaa',
+ 'smte;': '\u2aac',
+ 'smtes;': '\u2aac\ufe00',
+ 'SOFTcy;': '\u042c',
+ 'softcy;': '\u044c',
+ 'sol;': '/',
+ 'solb;': '\u29c4',
+ 'solbar;': '\u233f',
+ 'Sopf;': '\U0001d54a',
+ 'sopf;': '\U0001d564',
+ 'spades;': '\u2660',
+ 'spadesuit;': '\u2660',
+ 'spar;': '\u2225',
+ 'sqcap;': '\u2293',
+ 'sqcaps;': '\u2293\ufe00',
+ 'sqcup;': '\u2294',
+ 'sqcups;': '\u2294\ufe00',
+ 'Sqrt;': '\u221a',
+ 'sqsub;': '\u228f',
+ 'sqsube;': '\u2291',
+ 'sqsubset;': '\u228f',
+ 'sqsubseteq;': '\u2291',
+ 'sqsup;': '\u2290',
+ 'sqsupe;': '\u2292',
+ 'sqsupset;': '\u2290',
+ 'sqsupseteq;': '\u2292',
+ 'squ;': '\u25a1',
+ 'Square;': '\u25a1',
+ 'square;': '\u25a1',
+ 'SquareIntersection;': '\u2293',
+ 'SquareSubset;': '\u228f',
+ 'SquareSubsetEqual;': '\u2291',
+ 'SquareSuperset;': '\u2290',
+ 'SquareSupersetEqual;': '\u2292',
+ 'SquareUnion;': '\u2294',
+ 'squarf;': '\u25aa',
+ 'squf;': '\u25aa',
+ 'srarr;': '\u2192',
+ 'Sscr;': '\U0001d4ae',
+ 'sscr;': '\U0001d4c8',
+ 'ssetmn;': '\u2216',
+ 'ssmile;': '\u2323',
+ 'sstarf;': '\u22c6',
+ 'Star;': '\u22c6',
+ 'star;': '\u2606',
+ 'starf;': '\u2605',
+ 'straightepsilon;': '\u03f5',
+ 'straightphi;': '\u03d5',
+ 'strns;': '\xaf',
+ 'Sub;': '\u22d0',
+ 'sub;': '\u2282',
+ 'subdot;': '\u2abd',
+ 'subE;': '\u2ac5',
+ 'sube;': '\u2286',
+ 'subedot;': '\u2ac3',
+ 'submult;': '\u2ac1',
+ 'subnE;': '\u2acb',
+ 'subne;': '\u228a',
+ 'subplus;': '\u2abf',
+ 'subrarr;': '\u2979',
+ 'Subset;': '\u22d0',
+ 'subset;': '\u2282',
+ 'subseteq;': '\u2286',
+ 'subseteqq;': '\u2ac5',
+ 'SubsetEqual;': '\u2286',
+ 'subsetneq;': '\u228a',
+ 'subsetneqq;': '\u2acb',
+ 'subsim;': '\u2ac7',
+ 'subsub;': '\u2ad5',
+ 'subsup;': '\u2ad3',
+ 'succ;': '\u227b',
+ 'succapprox;': '\u2ab8',
+ 'succcurlyeq;': '\u227d',
+ 'Succeeds;': '\u227b',
+ 'SucceedsEqual;': '\u2ab0',
+ 'SucceedsSlantEqual;': '\u227d',
+ 'SucceedsTilde;': '\u227f',
+ 'succeq;': '\u2ab0',
+ 'succnapprox;': '\u2aba',
+ 'succneqq;': '\u2ab6',
+ 'succnsim;': '\u22e9',
+ 'succsim;': '\u227f',
+ 'SuchThat;': '\u220b',
+ 'Sum;': '\u2211',
+ 'sum;': '\u2211',
+ 'sung;': '\u266a',
+ 'sup1': '\xb9',
+ 'sup1;': '\xb9',
+ 'sup2': '\xb2',
+ 'sup2;': '\xb2',
+ 'sup3': '\xb3',
+ 'sup3;': '\xb3',
+ 'Sup;': '\u22d1',
+ 'sup;': '\u2283',
+ 'supdot;': '\u2abe',
+ 'supdsub;': '\u2ad8',
+ 'supE;': '\u2ac6',
+ 'supe;': '\u2287',
+ 'supedot;': '\u2ac4',
+ 'Superset;': '\u2283',
+ 'SupersetEqual;': '\u2287',
+ 'suphsol;': '\u27c9',
+ 'suphsub;': '\u2ad7',
+ 'suplarr;': '\u297b',
+ 'supmult;': '\u2ac2',
+ 'supnE;': '\u2acc',
+ 'supne;': '\u228b',
+ 'supplus;': '\u2ac0',
+ 'Supset;': '\u22d1',
+ 'supset;': '\u2283',
+ 'supseteq;': '\u2287',
+ 'supseteqq;': '\u2ac6',
+ 'supsetneq;': '\u228b',
+ 'supsetneqq;': '\u2acc',
+ 'supsim;': '\u2ac8',
+ 'supsub;': '\u2ad4',
+ 'supsup;': '\u2ad6',
+ 'swarhk;': '\u2926',
+ 'swArr;': '\u21d9',
+ 'swarr;': '\u2199',
+ 'swarrow;': '\u2199',
+ 'swnwar;': '\u292a',
+ 'szlig': '\xdf',
+ 'szlig;': '\xdf',
+ 'Tab;': '\t',
+ 'target;': '\u2316',
+ 'Tau;': '\u03a4',
+ 'tau;': '\u03c4',
+ 'tbrk;': '\u23b4',
+ 'Tcaron;': '\u0164',
+ 'tcaron;': '\u0165',
+ 'Tcedil;': '\u0162',
+ 'tcedil;': '\u0163',
+ 'Tcy;': '\u0422',
+ 'tcy;': '\u0442',
+ 'tdot;': '\u20db',
+ 'telrec;': '\u2315',
+ 'Tfr;': '\U0001d517',
+ 'tfr;': '\U0001d531',
+ 'there4;': '\u2234',
+ 'Therefore;': '\u2234',
+ 'therefore;': '\u2234',
+ 'Theta;': '\u0398',
+ 'theta;': '\u03b8',
+ 'thetasym;': '\u03d1',
+ 'thetav;': '\u03d1',
+ 'thickapprox;': '\u2248',
+ 'thicksim;': '\u223c',
+ 'ThickSpace;': '\u205f\u200a',
+ 'thinsp;': '\u2009',
+ 'ThinSpace;': '\u2009',
+ 'thkap;': '\u2248',
+ 'thksim;': '\u223c',
+ 'THORN': '\xde',
+ 'thorn': '\xfe',
+ 'THORN;': '\xde',
+ 'thorn;': '\xfe',
+ 'Tilde;': '\u223c',
+ 'tilde;': '\u02dc',
+ 'TildeEqual;': '\u2243',
+ 'TildeFullEqual;': '\u2245',
+ 'TildeTilde;': '\u2248',
+ 'times': '\xd7',
+ 'times;': '\xd7',
+ 'timesb;': '\u22a0',
+ 'timesbar;': '\u2a31',
+ 'timesd;': '\u2a30',
+ 'tint;': '\u222d',
+ 'toea;': '\u2928',
+ 'top;': '\u22a4',
+ 'topbot;': '\u2336',
+ 'topcir;': '\u2af1',
+ 'Topf;': '\U0001d54b',
+ 'topf;': '\U0001d565',
+ 'topfork;': '\u2ada',
+ 'tosa;': '\u2929',
+ 'tprime;': '\u2034',
+ 'TRADE;': '\u2122',
+ 'trade;': '\u2122',
+ 'triangle;': '\u25b5',
+ 'triangledown;': '\u25bf',
+ 'triangleleft;': '\u25c3',
+ 'trianglelefteq;': '\u22b4',
+ 'triangleq;': '\u225c',
+ 'triangleright;': '\u25b9',
+ 'trianglerighteq;': '\u22b5',
+ 'tridot;': '\u25ec',
+ 'trie;': '\u225c',
+ 'triminus;': '\u2a3a',
+ 'TripleDot;': '\u20db',
+ 'triplus;': '\u2a39',
+ 'trisb;': '\u29cd',
+ 'tritime;': '\u2a3b',
+ 'trpezium;': '\u23e2',
+ 'Tscr;': '\U0001d4af',
+ 'tscr;': '\U0001d4c9',
+ 'TScy;': '\u0426',
+ 'tscy;': '\u0446',
+ 'TSHcy;': '\u040b',
+ 'tshcy;': '\u045b',
+ 'Tstrok;': '\u0166',
+ 'tstrok;': '\u0167',
+ 'twixt;': '\u226c',
+ 'twoheadleftarrow;': '\u219e',
+ 'twoheadrightarrow;': '\u21a0',
+ 'Uacute': '\xda',
+ 'uacute': '\xfa',
+ 'Uacute;': '\xda',
+ 'uacute;': '\xfa',
+ 'Uarr;': '\u219f',
+ 'uArr;': '\u21d1',
+ 'uarr;': '\u2191',
+ 'Uarrocir;': '\u2949',
+ 'Ubrcy;': '\u040e',
+ 'ubrcy;': '\u045e',
+ 'Ubreve;': '\u016c',
+ 'ubreve;': '\u016d',
+ 'Ucirc': '\xdb',
+ 'ucirc': '\xfb',
+ 'Ucirc;': '\xdb',
+ 'ucirc;': '\xfb',
+ 'Ucy;': '\u0423',
+ 'ucy;': '\u0443',
+ 'udarr;': '\u21c5',
+ 'Udblac;': '\u0170',
+ 'udblac;': '\u0171',
+ 'udhar;': '\u296e',
+ 'ufisht;': '\u297e',
+ 'Ufr;': '\U0001d518',
+ 'ufr;': '\U0001d532',
+ 'Ugrave': '\xd9',
+ 'ugrave': '\xf9',
+ 'Ugrave;': '\xd9',
+ 'ugrave;': '\xf9',
+ 'uHar;': '\u2963',
+ 'uharl;': '\u21bf',
+ 'uharr;': '\u21be',
+ 'uhblk;': '\u2580',
+ 'ulcorn;': '\u231c',
+ 'ulcorner;': '\u231c',
+ 'ulcrop;': '\u230f',
+ 'ultri;': '\u25f8',
+ 'Umacr;': '\u016a',
+ 'umacr;': '\u016b',
+ 'uml': '\xa8',
+ 'uml;': '\xa8',
+ 'UnderBar;': '_',
+ 'UnderBrace;': '\u23df',
+ 'UnderBracket;': '\u23b5',
+ 'UnderParenthesis;': '\u23dd',
+ 'Union;': '\u22c3',
+ 'UnionPlus;': '\u228e',
+ 'Uogon;': '\u0172',
+ 'uogon;': '\u0173',
+ 'Uopf;': '\U0001d54c',
+ 'uopf;': '\U0001d566',
+ 'UpArrow;': '\u2191',
+ 'Uparrow;': '\u21d1',
+ 'uparrow;': '\u2191',
+ 'UpArrowBar;': '\u2912',
+ 'UpArrowDownArrow;': '\u21c5',
+ 'UpDownArrow;': '\u2195',
+ 'Updownarrow;': '\u21d5',
+ 'updownarrow;': '\u2195',
+ 'UpEquilibrium;': '\u296e',
+ 'upharpoonleft;': '\u21bf',
+ 'upharpoonright;': '\u21be',
+ 'uplus;': '\u228e',
+ 'UpperLeftArrow;': '\u2196',
+ 'UpperRightArrow;': '\u2197',
+ 'Upsi;': '\u03d2',
+ 'upsi;': '\u03c5',
+ 'upsih;': '\u03d2',
+ 'Upsilon;': '\u03a5',
+ 'upsilon;': '\u03c5',
+ 'UpTee;': '\u22a5',
+ 'UpTeeArrow;': '\u21a5',
+ 'upuparrows;': '\u21c8',
+ 'urcorn;': '\u231d',
+ 'urcorner;': '\u231d',
+ 'urcrop;': '\u230e',
+ 'Uring;': '\u016e',
+ 'uring;': '\u016f',
+ 'urtri;': '\u25f9',
+ 'Uscr;': '\U0001d4b0',
+ 'uscr;': '\U0001d4ca',
+ 'utdot;': '\u22f0',
+ 'Utilde;': '\u0168',
+ 'utilde;': '\u0169',
+ 'utri;': '\u25b5',
+ 'utrif;': '\u25b4',
+ 'uuarr;': '\u21c8',
+ 'Uuml': '\xdc',
+ 'uuml': '\xfc',
+ 'Uuml;': '\xdc',
+ 'uuml;': '\xfc',
+ 'uwangle;': '\u29a7',
+ 'vangrt;': '\u299c',
+ 'varepsilon;': '\u03f5',
+ 'varkappa;': '\u03f0',
+ 'varnothing;': '\u2205',
+ 'varphi;': '\u03d5',
+ 'varpi;': '\u03d6',
+ 'varpropto;': '\u221d',
+ 'vArr;': '\u21d5',
+ 'varr;': '\u2195',
+ 'varrho;': '\u03f1',
+ 'varsigma;': '\u03c2',
+ 'varsubsetneq;': '\u228a\ufe00',
+ 'varsubsetneqq;': '\u2acb\ufe00',
+ 'varsupsetneq;': '\u228b\ufe00',
+ 'varsupsetneqq;': '\u2acc\ufe00',
+ 'vartheta;': '\u03d1',
+ 'vartriangleleft;': '\u22b2',
+ 'vartriangleright;': '\u22b3',
+ 'Vbar;': '\u2aeb',
+ 'vBar;': '\u2ae8',
+ 'vBarv;': '\u2ae9',
+ 'Vcy;': '\u0412',
+ 'vcy;': '\u0432',
+ 'VDash;': '\u22ab',
+ 'Vdash;': '\u22a9',
+ 'vDash;': '\u22a8',
+ 'vdash;': '\u22a2',
+ 'Vdashl;': '\u2ae6',
+ 'Vee;': '\u22c1',
+ 'vee;': '\u2228',
+ 'veebar;': '\u22bb',
+ 'veeeq;': '\u225a',
+ 'vellip;': '\u22ee',
+ 'Verbar;': '\u2016',
+ 'verbar;': '|',
+ 'Vert;': '\u2016',
+ 'vert;': '|',
+ 'VerticalBar;': '\u2223',
+ 'VerticalLine;': '|',
+ 'VerticalSeparator;': '\u2758',
+ 'VerticalTilde;': '\u2240',
+ 'VeryThinSpace;': '\u200a',
+ 'Vfr;': '\U0001d519',
+ 'vfr;': '\U0001d533',
+ 'vltri;': '\u22b2',
+ 'vnsub;': '\u2282\u20d2',
+ 'vnsup;': '\u2283\u20d2',
+ 'Vopf;': '\U0001d54d',
+ 'vopf;': '\U0001d567',
+ 'vprop;': '\u221d',
+ 'vrtri;': '\u22b3',
+ 'Vscr;': '\U0001d4b1',
+ 'vscr;': '\U0001d4cb',
+ 'vsubnE;': '\u2acb\ufe00',
+ 'vsubne;': '\u228a\ufe00',
+ 'vsupnE;': '\u2acc\ufe00',
+ 'vsupne;': '\u228b\ufe00',
+ 'Vvdash;': '\u22aa',
+ 'vzigzag;': '\u299a',
+ 'Wcirc;': '\u0174',
+ 'wcirc;': '\u0175',
+ 'wedbar;': '\u2a5f',
+ 'Wedge;': '\u22c0',
+ 'wedge;': '\u2227',
+ 'wedgeq;': '\u2259',
+ 'weierp;': '\u2118',
+ 'Wfr;': '\U0001d51a',
+ 'wfr;': '\U0001d534',
+ 'Wopf;': '\U0001d54e',
+ 'wopf;': '\U0001d568',
+ 'wp;': '\u2118',
+ 'wr;': '\u2240',
+ 'wreath;': '\u2240',
+ 'Wscr;': '\U0001d4b2',
+ 'wscr;': '\U0001d4cc',
+ 'xcap;': '\u22c2',
+ 'xcirc;': '\u25ef',
+ 'xcup;': '\u22c3',
+ 'xdtri;': '\u25bd',
+ 'Xfr;': '\U0001d51b',
+ 'xfr;': '\U0001d535',
+ 'xhArr;': '\u27fa',
+ 'xharr;': '\u27f7',
+ 'Xi;': '\u039e',
+ 'xi;': '\u03be',
+ 'xlArr;': '\u27f8',
+ 'xlarr;': '\u27f5',
+ 'xmap;': '\u27fc',
+ 'xnis;': '\u22fb',
+ 'xodot;': '\u2a00',
+ 'Xopf;': '\U0001d54f',
+ 'xopf;': '\U0001d569',
+ 'xoplus;': '\u2a01',
+ 'xotime;': '\u2a02',
+ 'xrArr;': '\u27f9',
+ 'xrarr;': '\u27f6',
+ 'Xscr;': '\U0001d4b3',
+ 'xscr;': '\U0001d4cd',
+ 'xsqcup;': '\u2a06',
+ 'xuplus;': '\u2a04',
+ 'xutri;': '\u25b3',
+ 'xvee;': '\u22c1',
+ 'xwedge;': '\u22c0',
+ 'Yacute': '\xdd',
+ 'yacute': '\xfd',
+ 'Yacute;': '\xdd',
+ 'yacute;': '\xfd',
+ 'YAcy;': '\u042f',
+ 'yacy;': '\u044f',
+ 'Ycirc;': '\u0176',
+ 'ycirc;': '\u0177',
+ 'Ycy;': '\u042b',
+ 'ycy;': '\u044b',
+ 'yen': '\xa5',
+ 'yen;': '\xa5',
+ 'Yfr;': '\U0001d51c',
+ 'yfr;': '\U0001d536',
+ 'YIcy;': '\u0407',
+ 'yicy;': '\u0457',
+ 'Yopf;': '\U0001d550',
+ 'yopf;': '\U0001d56a',
+ 'Yscr;': '\U0001d4b4',
+ 'yscr;': '\U0001d4ce',
+ 'YUcy;': '\u042e',
+ 'yucy;': '\u044e',
+ 'yuml': '\xff',
+ 'Yuml;': '\u0178',
+ 'yuml;': '\xff',
+ 'Zacute;': '\u0179',
+ 'zacute;': '\u017a',
+ 'Zcaron;': '\u017d',
+ 'zcaron;': '\u017e',
+ 'Zcy;': '\u0417',
+ 'zcy;': '\u0437',
+ 'Zdot;': '\u017b',
+ 'zdot;': '\u017c',
+ 'zeetrf;': '\u2128',
+ 'ZeroWidthSpace;': '\u200b',
+ 'Zeta;': '\u0396',
+ 'zeta;': '\u03b6',
+ 'Zfr;': '\u2128',
+ 'zfr;': '\U0001d537',
+ 'ZHcy;': '\u0416',
+ 'zhcy;': '\u0436',
+ 'zigrarr;': '\u21dd',
+ 'Zopf;': '\u2124',
+ 'zopf;': '\U0001d56b',
+ 'Zscr;': '\U0001d4b5',
+ 'zscr;': '\U0001d4cf',
+ 'zwj;': '\u200d',
+ 'zwnj;': '\u200c',
+ }
+
try:
import http.client as compat_http_client
except ImportError: # Python 2
@@ -83,7 +2321,6 @@ try:
except ImportError: # Python 2
from HTMLParser import HTMLParser as compat_HTMLParser
-
try:
from subprocess import DEVNULL
compat_subprocess_get_DEVNULL = lambda: DEVNULL
@@ -626,6 +2863,7 @@ __all__ = [
'compat_getenv',
'compat_getpass',
'compat_html_entities',
+ 'compat_html_entities_html5',
'compat_http_client',
'compat_http_server',
'compat_input',
diff --git a/youtube_dl/extractor/audiomack.py b/youtube_dl/extractor/audiomack.py
index a52d26cec..f3bd4d444 100644
--- a/youtube_dl/extractor/audiomack.py
+++ b/youtube_dl/extractor/audiomack.py
@@ -6,6 +6,7 @@ import time
from .common import InfoExtractor
from .soundcloud import SoundcloudIE
+from ..compat import compat_str
from ..utils import (
ExtractorError,
url_basename,
@@ -136,7 +137,7 @@ class AudiomackAlbumIE(InfoExtractor):
result[resultkey] = api_response[apikey]
song_id = url_basename(api_response['url']).rpartition('.')[0]
result['entries'].append({
- 'id': api_response.get('id', song_id),
+ 'id': compat_str(api_response.get('id', song_id)),
'uploader': api_response.get('artist'),
'title': api_response.get('title', song_id),
'url': api_response['url'],
diff --git a/youtube_dl/extractor/dw.py b/youtube_dl/extractor/dw.py
index 0f0f0b8d3..d740652f1 100644
--- a/youtube_dl/extractor/dw.py
+++ b/youtube_dl/extractor/dw.py
@@ -35,6 +35,7 @@ class DWIE(InfoExtractor):
'upload_date': '20160311',
}
}, {
+ # DW documentaries, only last for one or two weeks
'url': 'http://www.dw.com/en/documentaries-welcome-to-the-90s-2016-05-21/e-19220158-9798',
'md5': '56b6214ef463bfb9a3b71aeb886f3cf1',
'info_dict': {
@@ -44,6 +45,7 @@ class DWIE(InfoExtractor):
'description': 'Welcome to the 90s - The Golden Decade of Hip Hop',
'upload_date': '20160521',
},
+ 'skip': 'Video removed',
}]
def _real_extract(self, url):
diff --git a/youtube_dl/extractor/extractors.py b/youtube_dl/extractor/extractors.py
index aa98782a5..38708294a 100644
--- a/youtube_dl/extractor/extractors.py
+++ b/youtube_dl/extractor/extractors.py
@@ -292,6 +292,7 @@ from .globo import (
GloboArticleIE,
)
from .godtube import GodTubeIE
+from .godtv import GodTVIE
from .goldenmoustache import GoldenMoustacheIE
from .golem import GolemIE
from .googledrive import GoogleDriveIE
@@ -957,7 +958,6 @@ from .vporn import VpornIE
from .vrt import VRTIE
from .vube import VubeIE
from .vuclip import VuClipIE
-from .vulture import VultureIE
from .walla import WallaIE
from .washingtonpost import (
WashingtonPostIE,
@@ -973,7 +973,6 @@ from .webofstories import (
WebOfStoriesIE,
WebOfStoriesPlaylistIE,
)
-from .weibo import WeiboIE
from .weiqitv import WeiqiTVIE
from .wimp import WimpIE
from .wistia import WistiaIE
diff --git a/youtube_dl/extractor/generic.py b/youtube_dl/extractor/generic.py
index 90575ab0e..ef18ce3dc 100644
--- a/youtube_dl/extractor/generic.py
+++ b/youtube_dl/extractor/generic.py
@@ -627,13 +627,13 @@ class GenericIE(InfoExtractor):
},
# MTVSercices embed
{
- 'url': 'http://www.gametrailers.com/news-post/76093/north-america-europe-is-getting-that-mario-kart-8-mercedes-dlc-too',
- 'md5': '35727f82f58c76d996fc188f9755b0d5',
+ 'url': 'http://www.vulture.com/2016/06/new-key-peele-sketches-released.html',
+ 'md5': 'ca1aef97695ef2c1d6973256a57e5252',
'info_dict': {
- 'id': '0306a69b-8adf-4fb5-aace-75f8e8cbfca9',
+ 'id': '769f7ec0-0692-4d62-9b45-0d88074bffc1',
'ext': 'mp4',
- 'title': 'Review',
- 'description': 'Mario\'s life in the fast lane has never looked so good.',
+ 'title': 'Key and Peele|October 10, 2012|2|203|Liam Neesons - Uncensored',
+ 'description': 'Two valets share their love for movie star Liam Neesons.',
},
},
# YouTube embed via <data-embed-url="">
@@ -1032,6 +1032,17 @@ class GenericIE(InfoExtractor):
'timestamp': 1389118457,
},
},
+ # NBC News embed
+ {
+ 'url': 'http://www.vulture.com/2016/06/letterman-couldnt-care-less-about-late-night.html',
+ 'md5': '1aa589c675898ae6d37a17913cf68d66',
+ 'info_dict': {
+ 'id': '701714499682',
+ 'ext': 'mp4',
+ 'title': 'PREVIEW: On Assignment: David Letterman',
+ 'description': 'A preview of Tom Brokaw\'s interview with David Letterman as part of the On Assignment series powered by Dateline. Airs Sunday June 12 at 7/6c.',
+ },
+ },
# UDN embed
{
'url': 'https://video.udn.com/news/300346',
@@ -1846,14 +1857,6 @@ class GenericIE(InfoExtractor):
url = unescapeHTML(mobj.group('url'))
return self.url_result(url)
- # Look for embedded vulture.com player
- mobj = re.search(
- r'<iframe src="(?P<url>https?://video\.vulture\.com/[^"]+)"',
- webpage)
- if mobj is not None:
- url = unescapeHTML(mobj.group('url'))
- return self.url_result(url, ie='Vulture')
-
# Look for embedded mtvservices player
mtvservices_url = MTVServicesEmbeddedIE._extract_url(webpage)
if mtvservices_url:
@@ -1966,6 +1969,12 @@ class GenericIE(InfoExtractor):
if nbc_sports_url:
return self.url_result(nbc_sports_url, 'NBCSportsVPlayer')
+ # Look for NBC News embeds
+ nbc_news_embed_url = re.search(
+ r'<iframe[^>]+src=(["\'])(?P<url>(?:https?:)?//www\.nbcnews\.com/widget/video-embed/[^"\']+)\1', webpage)
+ if nbc_news_embed_url:
+ return self.url_result(nbc_news_embed_url.group('url'), 'NBCNews')
+
# Look for Google Drive embeds
google_drive_url = GoogleDriveIE._extract_url(webpage)
if google_drive_url:
diff --git a/youtube_dl/extractor/godtv.py b/youtube_dl/extractor/godtv.py
new file mode 100644
index 000000000..c5d3b4e6a
--- /dev/null
+++ b/youtube_dl/extractor/godtv.py
@@ -0,0 +1,66 @@
+from __future__ import unicode_literals
+
+from .common import InfoExtractor
+from .ooyala import OoyalaIE
+from ..utils import js_to_json
+
+
+class GodTVIE(InfoExtractor):
+ _VALID_URL = r'https?://(?:www\.)?god\.tv(?:/[^/]+)*/(?P<id>[^/?#&]+)'
+ _TESTS = [{
+ 'url': 'http://god.tv/jesus-image/video/jesus-conference-2016/randy-needham',
+ 'info_dict': {
+ 'id': 'lpd3g2MzE6D1g8zFAKz8AGpxWcpu6o_3',
+ 'ext': 'mp4',
+ 'title': 'Randy Needham',
+ 'duration': 3615.08,
+ },
+ 'params': {
+ 'skip_download': True,
+ }
+ }, {
+ 'url': 'http://god.tv/playlist/bible-study',
+ 'info_dict': {
+ 'id': 'bible-study',
+ },
+ 'playlist_mincount': 37,
+ }, {
+ 'url': 'http://god.tv/node/15097',
+ 'only_matching': True,
+ }, {
+ 'url': 'http://god.tv/live/africa',
+ 'only_matching': True,
+ }, {
+ 'url': 'http://god.tv/liveevents',
+ 'only_matching': True,
+ }]
+
+ def _real_extract(self, url):
+ display_id = self._match_id(url)
+
+ webpage = self._download_webpage(url, display_id)
+
+ settings = self._parse_json(
+ self._search_regex(
+ r'jQuery\.extend\(Drupal\.settings\s*,\s*({.+?})\);',
+ webpage, 'settings', default='{}'),
+ display_id, transform_source=js_to_json, fatal=False)
+
+ ooyala_id = None
+
+ if settings:
+ playlist = settings.get('playlist')
+ if playlist and isinstance(playlist, list):
+ entries = [
+ OoyalaIE._build_url_result(video['content_id'])
+ for video in playlist if video.get('content_id')]
+ if entries:
+ return self.playlist_result(entries, display_id)
+ ooyala_id = settings.get('ooyala', {}).get('content_id')
+
+ if not ooyala_id:
+ ooyala_id = self._search_regex(
+ r'["\']content_id["\']\s*:\s*(["\'])(?P<id>[\w-]+)\1',
+ webpage, 'ooyala id', group='id')
+
+ return OoyalaIE._build_url_result(ooyala_id)
diff --git a/youtube_dl/extractor/lynda.py b/youtube_dl/extractor/lynda.py
index c1bca5678..5b458d9bc 100644
--- a/youtube_dl/extractor/lynda.py
+++ b/youtube_dl/extractor/lynda.py
@@ -1,93 +1,94 @@
from __future__ import unicode_literals
import re
-import json
from .common import InfoExtractor
-from ..compat import compat_str
+from ..compat import (
+ compat_HTTPError,
+ compat_str,
+ compat_urlparse,
+)
from ..utils import (
ExtractorError,
- clean_html,
int_or_none,
- sanitized_Request,
urlencode_postdata,
)
class LyndaBaseIE(InfoExtractor):
- _LOGIN_URL = 'https://www.lynda.com/login/login.aspx'
+ _SIGNIN_URL = 'https://www.lynda.com/signin'
+ _PASSWORD_URL = 'https://www.lynda.com/signin/password'
+ _USER_URL = 'https://www.lynda.com/signin/user'
_ACCOUNT_CREDENTIALS_HINT = 'Use --username and --password options to provide lynda.com account credentials.'
_NETRC_MACHINE = 'lynda'
def _real_initialize(self):
self._login()
+ @staticmethod
+ def _check_error(json_string, key_or_keys):
+ keys = [key_or_keys] if isinstance(key_or_keys, compat_str) else key_or_keys
+ for key in keys:
+ error = json_string.get(key)
+ if error:
+ raise ExtractorError('Unable to login: %s' % error, expected=True)
+
+ def _login_step(self, form_html, fallback_action_url, extra_form_data, note, referrer_url):
+ action_url = self._search_regex(
+ r'<form[^>]+action=(["\'])(?P<url>.+?)\1', form_html,
+ 'post url', default=fallback_action_url, group='url')
+
+ if not action_url.startswith('http'):
+ action_url = compat_urlparse.urljoin(self._SIGNIN_URL, action_url)
+
+ form_data = self._hidden_inputs(form_html)
+ form_data.update(extra_form_data)
+
+ try:
+ response = self._download_json(
+ action_url, None, note,
+ data=urlencode_postdata(form_data),
+ headers={
+ 'Referer': referrer_url,
+ 'X-Requested-With': 'XMLHttpRequest',
+ })
+ except ExtractorError as e:
+ if isinstance(e.cause, compat_HTTPError) and e.cause.code == 500:
+ response = self._parse_json(e.cause.read().decode('utf-8'), None)
+ self._check_error(response, ('email', 'password'))
+ raise
+
+ self._check_error(response, 'ErrorMessage')
+
+ return response, action_url
+
def _login(self):
username, password = self._get_login_info()
if username is None:
return
- login_form = {
- 'username': username,
- 'password': password,
- 'remember': 'false',
- 'stayPut': 'false'
- }
- request = sanitized_Request(
- self._LOGIN_URL, urlencode_postdata(login_form))
- login_page = self._download_webpage(
- request, None, 'Logging in as %s' % username)
-
- # Not (yet) logged in
- m = re.search(r'loginResultJson\s*=\s*\'(?P<json>[^\']+)\';', login_page)
- if m is not None:
- response = m.group('json')
- response_json = json.loads(response)
- state = response_json['state']
-
- if state == 'notlogged':
- raise ExtractorError(
- 'Unable to login, incorrect username and/or password',
- expected=True)
-
- # This is when we get popup:
- # > You're already logged in to lynda.com on two devices.
- # > If you log in here, we'll log you out of another device.
- # So, we need to confirm this.
- if state == 'conflicted':
- confirm_form = {
- 'username': '',
- 'password': '',
- 'resolve': 'true',
- 'remember': 'false',
- 'stayPut': 'false',
- }
- request = sanitized_Request(
- self._LOGIN_URL, urlencode_postdata(confirm_form))
- login_page = self._download_webpage(
- request, None,
- 'Confirming log in and log out from another device')
-
- if all(not re.search(p, login_page) for p in ('isLoggedIn\s*:\s*true', r'logout\.aspx', r'>Log out<')):
- if 'login error' in login_page:
- mobj = re.search(
- r'(?s)<h1[^>]+class="topmost">(?P<title>[^<]+)</h1>\s*<div>(?P<description>.+?)</div>',
- login_page)
- if mobj:
- raise ExtractorError(
- 'lynda returned error: %s - %s'
- % (mobj.group('title'), clean_html(mobj.group('description'))),
- expected=True)
- raise ExtractorError('Unable to log in')
-
- def _logout(self):
- username, _ = self._get_login_info()
- if username is None:
+ # Step 1: download signin page
+ signin_page = self._download_webpage(
+ self._SIGNIN_URL, None, 'Downloading signin page')
+
+ # Already logged in
+ if any(re.search(p, signin_page) for p in (
+ 'isLoggedIn\s*:\s*true', r'logout\.aspx', r'>Log out<')):
return
- self._download_webpage(
- 'http://www.lynda.com/ajax/logout.aspx', None,
- 'Logging out', 'Unable to log out', fatal=False)
+ # Step 2: submit email
+ signin_form = self._search_regex(
+ r'(?s)(<form[^>]+data-form-name=["\']signin["\'][^>]*>.+?</form>)',
+ signin_page, 'signin form')
+ signin_page, signin_url = self._login_step(
+ signin_form, self._PASSWORD_URL, {'email': username},
+ 'Submitting email', self._SIGNIN_URL)
+
+ # Step 3: submit password
+ password_form = signin_page['body']
+ self._login_step(
+ password_form, self._USER_URL, {'email': username, 'password': password},
+ 'Submitting password', signin_url)
class LyndaIE(LyndaBaseIE):
@@ -212,8 +213,6 @@ class LyndaCourseIE(LyndaBaseIE):
'http://www.lynda.com/ajax/player?courseId=%s&type=course' % course_id,
course_id, 'Downloading course JSON')
- self._logout()
-
if course.get('Status') == 'NotFound':
raise ExtractorError(
'Course %s does not exist' % course_id, expected=True)
diff --git a/youtube_dl/extractor/nbc.py b/youtube_dl/extractor/nbc.py
index f27c7f139..6b7da1149 100644
--- a/youtube_dl/extractor/nbc.py
+++ b/youtube_dl/extractor/nbc.py
@@ -266,6 +266,11 @@ class NBCNewsIE(ThePlatformIE):
'url': 'http://www.nbcnews.com/watch/dateline/full-episode--deadly-betrayal-386250819952',
'only_matching': True,
},
+ {
+ # From http://www.vulture.com/2016/06/letterman-couldnt-care-less-about-late-night.html
+ 'url': 'http://www.nbcnews.com/widget/video-embed/701714499682',
+ 'only_matching': True,
+ },
]
def _real_extract(self, url):
@@ -289,18 +294,17 @@ class NBCNewsIE(ThePlatformIE):
webpage = self._download_webpage(url, display_id)
info = None
bootstrap_json = self._search_regex(
- r'(?m)var\s+(?:bootstrapJson|playlistData)\s*=\s*({.+});?\s*$',
+ [r'(?m)(?:var\s+(?:bootstrapJson|playlistData)|NEWS\.videoObj)\s*=\s*({.+});?\s*$',
+ r'videoObj\s*:\s*({.+})', r'data-video="([^"]+)"'],
webpage, 'bootstrap json', default=None)
- if bootstrap_json:
- bootstrap = self._parse_json(bootstrap_json, display_id)
+ bootstrap = self._parse_json(
+ bootstrap_json, display_id, transform_source=unescapeHTML)
+ if 'results' in bootstrap:
info = bootstrap['results'][0]['video']
+ elif 'video' in bootstrap:
+ info = bootstrap['video']
else:
- player_instance_json = self._search_regex(
- r'videoObj\s*:\s*({.+})', webpage, 'player instance', default=None)
- if not player_instance_json:
- player_instance_json = self._html_search_regex(
- r'data-video="([^"]+)"', webpage, 'video json')
- info = self._parse_json(player_instance_json, display_id)
+ info = bootstrap
video_id = info['mpxId']
title = info['title']
diff --git a/youtube_dl/extractor/openload.py b/youtube_dl/extractor/openload.py
index 5049b870e..6415b8fdc 100644
--- a/youtube_dl/extractor/openload.py
+++ b/youtube_dl/extractor/openload.py
@@ -14,7 +14,7 @@ from ..utils import (
class OpenloadIE(InfoExtractor):
- _VALID_URL = r'https://openload.(?:co|io)/(?:f|embed)/(?P<id>[a-zA-Z0-9-]+)'
+ _VALID_URL = r'https://openload.(?:co|io)/(?:f|embed)/(?P<id>[a-zA-Z0-9-_]+)'
_TESTS = [{
'url': 'https://openload.co/f/kUEfGclsU9o',
@@ -32,6 +32,9 @@ class OpenloadIE(InfoExtractor):
'url': 'https://openload.io/f/ZAn6oz-VZGE/',
'only_matching': True,
}, {
+ 'url': 'https://openload.co/f/_-ztPaZtMhM/',
+ 'only_matching': True,
+ }, {
# unavailable via https://openload.co/f/Sxz5sADo82g/, different layout
# for title and ext
'url': 'https://openload.co/embed/Sxz5sADo82g/',
@@ -100,7 +103,7 @@ class OpenloadIE(InfoExtractor):
raise ExtractorError('File not found', expected=True)
code = self._search_regex(
- r'</video>\s*</div>\s*<script[^>]+>([^<]+)</script>',
+ r'</video>\s*</div>\s*<script[^>]+>[^>]+</script>\s*<script[^>]+>([^<]+)</script>',
webpage, 'JS code')
decoded = self.openload_decode(code)
diff --git a/youtube_dl/extractor/voicerepublic.py b/youtube_dl/extractor/voicerepublic.py
index 93d15a556..4f1a99a89 100644
--- a/youtube_dl/extractor/voicerepublic.py
+++ b/youtube_dl/extractor/voicerepublic.py
@@ -3,7 +3,10 @@ from __future__ import unicode_literals
import re
from .common import InfoExtractor
-from ..compat import compat_urlparse
+from ..compat import (
+ compat_str,
+ compat_urlparse,
+)
from ..utils import (
ExtractorError,
determine_ext,
@@ -16,13 +19,13 @@ class VoiceRepublicIE(InfoExtractor):
_VALID_URL = r'https?://voicerepublic\.com/(?:talks|embed)/(?P<id>[0-9a-z-]+)'
_TESTS = [{
'url': 'http://voicerepublic.com/talks/watching-the-watchers-building-a-sousveillance-state',
- 'md5': '0554a24d1657915aa8e8f84e15dc9353',
+ 'md5': 'b9174d651323f17783000876347116e3',
'info_dict': {
'id': '2296',
'display_id': 'watching-the-watchers-building-a-sousveillance-state',
'ext': 'm4a',
'title': 'Watching the Watchers: Building a Sousveillance State',
- 'description': 'md5:715ba964958afa2398df615809cfecb1',
+ 'description': 'Secret surveillance programs have metadata too. The people and companies that operate secret surveillance programs can be surveilled.',
'thumbnail': 're:^https?://.*\.(?:png|jpg)$',
'duration': 1800,
'view_count': int,
@@ -52,7 +55,7 @@ class VoiceRepublicIE(InfoExtractor):
if data:
title = data['title']
description = data.get('teaser')
- talk_id = data.get('talk_id') or display_id
+ talk_id = compat_str(data.get('talk_id') or display_id)
talk = data['talk']
duration = int_or_none(talk.get('duration'))
formats = [{
diff --git a/youtube_dl/extractor/vporn.py b/youtube_dl/extractor/vporn.py
index 92c90e517..1557a0e04 100644
--- a/youtube_dl/extractor/vporn.py
+++ b/youtube_dl/extractor/vporn.py
@@ -4,6 +4,7 @@ import re
from .common import InfoExtractor
from ..utils import (
+ ExtractorError,
parse_duration,
str_to_int,
)
@@ -27,7 +28,8 @@ class VpornIE(InfoExtractor):
'duration': 393,
'age_limit': 18,
'view_count': int,
- }
+ },
+ 'skip': 'video removed',
},
{
'url': 'http://www.vporn.com/female/hana-shower/523564/',
@@ -40,7 +42,7 @@ class VpornIE(InfoExtractor):
'description': 'Hana showers at the bathroom.',
'thumbnail': 're:^https?://.*\.jpg$',
'uploader': 'Hmmmmm',
- 'categories': ['Big Boobs', 'Erotic', 'Teen', 'Female'],
+ 'categories': ['Big Boobs', 'Erotic', 'Teen', 'Female', '720p'],
'duration': 588,
'age_limit': 18,
'view_count': int,
@@ -55,6 +57,10 @@ class VpornIE(InfoExtractor):
webpage = self._download_webpage(url, display_id)
+ errmsg = 'This video has been deleted due to Copyright Infringement or by the account owner!'
+ if errmsg in webpage:
+ raise ExtractorError('%s said: %s' % (self.IE_NAME, errmsg), expected=True)
+
title = self._html_search_regex(
r'videoname\s*=\s*\'([^\']+)\'', webpage, 'title').strip()
description = self._html_search_regex(
diff --git a/youtube_dl/extractor/vulture.py b/youtube_dl/extractor/vulture.py
deleted file mode 100644
index faa167e65..000000000
--- a/youtube_dl/extractor/vulture.py
+++ /dev/null
@@ -1,69 +0,0 @@
-from __future__ import unicode_literals
-
-import json
-import os.path
-import re
-
-from .common import InfoExtractor
-from ..utils import (
- int_or_none,
- parse_iso8601,
-)
-
-
-class VultureIE(InfoExtractor):
- IE_NAME = 'vulture.com'
- _VALID_URL = r'https?://video\.vulture\.com/video/(?P<display_id>[^/]+)/'
- _TEST = {
- 'url': 'http://video.vulture.com/video/Mindy-Kaling-s-Harvard-Speech/player?layout=compact&read_more=1',
- 'md5': '8d997845642a2b5152820f7257871bc8',
- 'info_dict': {
- 'id': '6GHRQL3RV7MSD1H4',
- 'ext': 'mp4',
- 'title': 'kaling-speech-2-MAGNIFY STANDARD CONTAINER REVISED',
- 'uploader_id': 'Sarah',
- 'thumbnail': 're:^http://.*\.jpg$',
- 'timestamp': 1401288564,
- 'upload_date': '20140528',
- 'description': 'Uplifting and witty, as predicted.',
- 'duration': 1015,
- }
- }
-
- def _real_extract(self, url):
- mobj = re.match(self._VALID_URL, url)
- display_id = mobj.group('display_id')
-
- webpage = self._download_webpage(url, display_id)
- query_string = self._search_regex(
- r"queryString\s*=\s*'([^']+)'", webpage, 'query string')
- video_id = self._search_regex(
- r'content=([^&]+)', query_string, 'video ID')
- query_url = 'http://video.vulture.com/embed/player/container/1000/1000/?%s' % query_string
-
- query_webpage = self._download_webpage(
- query_url, display_id, note='Downloading query page')
- params_json = self._search_regex(
- r'(?sm)new MagnifyEmbeddablePlayer\({.*?contentItem:\s*(\{.*?\})\n?,\n',
- query_webpage,
- 'player params')
- params = json.loads(params_json)
-
- upload_timestamp = parse_iso8601(params['posted'].replace(' ', 'T'))
- uploader_id = params.get('user', {}).get('handle')
-
- media_item = params['media_item']
- title = os.path.splitext(media_item['title'])[0]
- duration = int_or_none(media_item.get('duration_seconds'))
-
- return {
- 'id': video_id,
- 'display_id': display_id,
- 'url': media_item['pipeline_xid'],
- 'title': title,
- 'timestamp': upload_timestamp,
- 'thumbnail': params.get('thumbnail_url'),
- 'uploader_id': uploader_id,
- 'description': params.get('description'),
- 'duration': duration,
- }
diff --git a/youtube_dl/extractor/wdr.py b/youtube_dl/extractor/wdr.py
index a9238cbeb..6b83a2a04 100644
--- a/youtube_dl/extractor/wdr.py
+++ b/youtube_dl/extractor/wdr.py
@@ -34,7 +34,8 @@ class WDRIE(InfoExtractor):
'description': 'md5:87be8ff14d8dfd7a7ee46f0299b52318',
'is_live': False,
'subtitles': {'de': [{
- 'url': 'http://ondemand-ww.wdr.de/medp/fsk0/105/1058683/1058683_12220974.xml'
+ 'url': 'http://ondemand-ww.wdr.de/medp/fsk0/105/1058683/1058683_12220974.xml',
+ 'ext': 'ttml',
}]},
},
},
@@ -190,7 +191,8 @@ class WDRIE(InfoExtractor):
caption_url = metadata_media_resource.get('captionURL')
if caption_url:
subtitles['de'] = [{
- 'url': caption_url
+ 'url': caption_url,
+ 'ext': 'ttml',
}]
title = metadata_tracker_data.get('trackerClipTitle')
diff --git a/youtube_dl/extractor/weibo.py b/youtube_dl/extractor/weibo.py
deleted file mode 100644
index 20bb039d3..000000000
--- a/youtube_dl/extractor/weibo.py
+++ /dev/null
@@ -1,49 +0,0 @@
-# coding: utf-8
-from __future__ import unicode_literals
-
-import re
-
-from .common import InfoExtractor
-
-
-class WeiboIE(InfoExtractor):
- """
- The videos in Weibo come from different sites, this IE just finds the link
- to the external video and returns it.
- """
- _VALID_URL = r'https?://video\.weibo\.com/v/weishipin/t_(?P<id>.+?)\.htm'
-
- _TEST = {
- 'url': 'http://video.weibo.com/v/weishipin/t_zjUw2kZ.htm',
- 'info_dict': {
- 'id': '98322879',
- 'ext': 'flv',
- 'title': '魔声耳机最新广告“All Eyes On Us”',
- },
- 'params': {
- 'skip_download': True,
- },
- 'add_ie': ['Sina'],
- }
-
- # Additional example videos from different sites
- # Youku: http://video.weibo.com/v/weishipin/t_zQGDWQ8.htm
- # 56.com: http://video.weibo.com/v/weishipin/t_zQ44HxN.htm
-
- def _real_extract(self, url):
- mobj = re.match(self._VALID_URL, url, flags=re.VERBOSE)
- video_id = mobj.group('id')
- info_url = 'http://video.weibo.com/?s=v&a=play_list&format=json&mix_video_id=t_%s' % video_id
- info = self._download_json(info_url, video_id)
-
- videos_urls = map(lambda v: v['play_page_url'], info['result']['data'])
- # Prefer sina video since they have thumbnails
- videos_urls = sorted(videos_urls, key=lambda u: 'video.sina.com' in u)
- player_url = videos_urls[-1]
- m_sina = re.match(r'https?://video\.sina\.com\.cn/v/b/(\d+)-\d+\.html',
- player_url)
- if m_sina is not None:
- self.to_screen('Sina video detected')
- sina_id = m_sina.group(1)
- player_url = 'http://you.video.sina.com.cn/swf/quotePlayer.swf?vid=%s' % sina_id
- return self.url_result(player_url)
diff --git a/youtube_dl/extractor/xfileshare.py b/youtube_dl/extractor/xfileshare.py
index 769003735..ee4d04c20 100644
--- a/youtube_dl/extractor/xfileshare.py
+++ b/youtube_dl/extractor/xfileshare.py
@@ -62,7 +62,8 @@ class XFileShareIE(InfoExtractor):
'ext': 'mp4',
'title': 'youtube-dl test video \'äBaW_jenozKc.mp4.mp4',
'thumbnail': 're:http://.*\.jpg',
- }
+ },
+ 'skip': 'Video removed',
}, {
'url': 'http://vidto.me/ku5glz52nqe1.html',
'info_dict': {
diff --git a/youtube_dl/extractor/xuite.py b/youtube_dl/extractor/xuite.py
index 2466410fa..0be8932ad 100644
--- a/youtube_dl/extractor/xuite.py
+++ b/youtube_dl/extractor/xuite.py
@@ -66,6 +66,7 @@ class XuiteIE(InfoExtractor):
'uploader_id': '242127761',
'categories': ['電玩動漫'],
},
+ 'skip': 'Video removed',
}, {
'url': 'http://vlog.xuite.net/play/S1dDUjdyLTMyOTc3NjcuZmx2/%E5%AD%AB%E7%87%95%E5%A7%BF-%E7%9C%BC%E6%B7%9A%E6%88%90%E8%A9%A9',
'only_matching': True,
diff --git a/youtube_dl/extractor/yahoo.py b/youtube_dl/extractor/yahoo.py
index b376f2b93..927a964a4 100644
--- a/youtube_dl/extractor/yahoo.py
+++ b/youtube_dl/extractor/yahoo.py
@@ -343,7 +343,7 @@ class YahooIE(InfoExtractor):
webpage, 'region', fatal=False, default='US')
data = compat_urllib_parse_urlencode({
'protocol': 'http',
- 'region': region,
+ 'region': region.upper(),
})
query_url = (
'https://video.media.yql.yahoo.com/v1/video/sapi/streams/'
diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py
index 6c9f77d95..00dd602ff 100644
--- a/youtube_dl/extractor/youtube.py
+++ b/youtube_dl/extractor/youtube.py
@@ -1988,7 +1988,7 @@ class YoutubeChannelIE(YoutubePlaylistBaseInfoExtractor):
class YoutubeUserIE(YoutubeChannelIE):
IE_DESC = 'YouTube.com user videos (URL or "ytuser" keyword)'
- _VALID_URL = r'(?:(?:https?://(?:\w+\.)?youtube\.com/(?:user/)?(?!(?:attribution_link|watch|results)(?:$|[^a-z_A-Z0-9-])))|ytuser:)(?!feed/)(?P<id>[A-Za-z0-9_-]+)'
+ _VALID_URL = r'(?:(?:https?://(?:\w+\.)?youtube\.com/(?:user/|c/)?(?!(?:attribution_link|watch|results)(?:$|[^a-z_A-Z0-9-])))|ytuser:)(?!feed/)(?P<id>[A-Za-z0-9_-]+)'
_TEMPLATE_URL = 'https://www.youtube.com/user/%s/videos'
IE_NAME = 'youtube:user'
@@ -2001,6 +2001,9 @@ class YoutubeUserIE(YoutubeChannelIE):
}, {
'url': 'ytuser:phihag',
'only_matching': True,
+ }, {
+ 'url': 'https://www.youtube.com/c/gametrailers',
+ 'only_matching': True,
}]
@classmethod
diff --git a/youtube_dl/utils.py b/youtube_dl/utils.py
index 229de4b39..f77ab8650 100644
--- a/youtube_dl/utils.py
+++ b/youtube_dl/utils.py
@@ -39,6 +39,7 @@ from .compat import (
compat_chr,
compat_etree_fromstring,
compat_html_entities,
+ compat_html_entities_html5,
compat_http_client,
compat_kwargs,
compat_parse_qs,
@@ -456,12 +457,19 @@ def orderedSet(iterable):
return res
-def _htmlentity_transform(entity):
+def _htmlentity_transform(entity_with_semicolon):
"""Transforms an HTML entity to a character."""
+ entity = entity_with_semicolon[:-1]
+
# Known non-numeric HTML entity
if entity in compat_html_entities.name2codepoint:
return compat_chr(compat_html_entities.name2codepoint[entity])
+ # TODO: HTML5 allows entities without a semicolon. For example,
+ # '&Eacuteric' should be decoded as 'Éric'.
+ if entity_with_semicolon in compat_html_entities_html5:
+ return compat_html_entities_html5[entity_with_semicolon]
+
mobj = re.match(r'#(x[0-9a-fA-F]+|[0-9]+)', entity)
if mobj is not None:
numstr = mobj.group(1)
@@ -486,7 +494,7 @@ def unescapeHTML(s):
assert type(s) == compat_str
return re.sub(
- r'&([^;]+);', lambda m: _htmlentity_transform(m.group(1)), s)
+ r'&([^;]+;)', lambda m: _htmlentity_transform(m.group(1)), s)
def get_subprocess_encoding():