1<?php
2
3namespace PhpOffice\PhpSpreadsheet\Helper;
4
5use DOMDocument;
6use DOMElement;
7use DOMNode;
8use DOMText;
9use PhpOffice\PhpSpreadsheet\RichText\RichText;
10use PhpOffice\PhpSpreadsheet\Style\Color;
11use PhpOffice\PhpSpreadsheet\Style\Font;
12
13class Html
14{
15    protected static $colourMap = [
16        'aliceblue' => 'f0f8ff',
17        'antiquewhite' => 'faebd7',
18        'antiquewhite1' => 'ffefdb',
19        'antiquewhite2' => 'eedfcc',
20        'antiquewhite3' => 'cdc0b0',
21        'antiquewhite4' => '8b8378',
22        'aqua' => '00ffff',
23        'aquamarine1' => '7fffd4',
24        'aquamarine2' => '76eec6',
25        'aquamarine4' => '458b74',
26        'azure1' => 'f0ffff',
27        'azure2' => 'e0eeee',
28        'azure3' => 'c1cdcd',
29        'azure4' => '838b8b',
30        'beige' => 'f5f5dc',
31        'bisque1' => 'ffe4c4',
32        'bisque2' => 'eed5b7',
33        'bisque3' => 'cdb79e',
34        'bisque4' => '8b7d6b',
35        'black' => '000000',
36        'blanchedalmond' => 'ffebcd',
37        'blue' => '0000ff',
38        'blue1' => '0000ff',
39        'blue2' => '0000ee',
40        'blue4' => '00008b',
41        'blueviolet' => '8a2be2',
42        'brown' => 'a52a2a',
43        'brown1' => 'ff4040',
44        'brown2' => 'ee3b3b',
45        'brown3' => 'cd3333',
46        'brown4' => '8b2323',
47        'burlywood' => 'deb887',
48        'burlywood1' => 'ffd39b',
49        'burlywood2' => 'eec591',
50        'burlywood3' => 'cdaa7d',
51        'burlywood4' => '8b7355',
52        'cadetblue' => '5f9ea0',
53        'cadetblue1' => '98f5ff',
54        'cadetblue2' => '8ee5ee',
55        'cadetblue3' => '7ac5cd',
56        'cadetblue4' => '53868b',
57        'chartreuse1' => '7fff00',
58        'chartreuse2' => '76ee00',
59        'chartreuse3' => '66cd00',
60        'chartreuse4' => '458b00',
61        'chocolate' => 'd2691e',
62        'chocolate1' => 'ff7f24',
63        'chocolate2' => 'ee7621',
64        'chocolate3' => 'cd661d',
65        'coral' => 'ff7f50',
66        'coral1' => 'ff7256',
67        'coral2' => 'ee6a50',
68        'coral3' => 'cd5b45',
69        'coral4' => '8b3e2f',
70        'cornflowerblue' => '6495ed',
71        'cornsilk1' => 'fff8dc',
72        'cornsilk2' => 'eee8cd',
73        'cornsilk3' => 'cdc8b1',
74        'cornsilk4' => '8b8878',
75        'cyan1' => '00ffff',
76        'cyan2' => '00eeee',
77        'cyan3' => '00cdcd',
78        'cyan4' => '008b8b',
79        'darkgoldenrod' => 'b8860b',
80        'darkgoldenrod1' => 'ffb90f',
81        'darkgoldenrod2' => 'eead0e',
82        'darkgoldenrod3' => 'cd950c',
83        'darkgoldenrod4' => '8b6508',
84        'darkgreen' => '006400',
85        'darkkhaki' => 'bdb76b',
86        'darkolivegreen' => '556b2f',
87        'darkolivegreen1' => 'caff70',
88        'darkolivegreen2' => 'bcee68',
89        'darkolivegreen3' => 'a2cd5a',
90        'darkolivegreen4' => '6e8b3d',
91        'darkorange' => 'ff8c00',
92        'darkorange1' => 'ff7f00',
93        'darkorange2' => 'ee7600',
94        'darkorange3' => 'cd6600',
95        'darkorange4' => '8b4500',
96        'darkorchid' => '9932cc',
97        'darkorchid1' => 'bf3eff',
98        'darkorchid2' => 'b23aee',
99        'darkorchid3' => '9a32cd',
100        'darkorchid4' => '68228b',
101        'darksalmon' => 'e9967a',
102        'darkseagreen' => '8fbc8f',
103        'darkseagreen1' => 'c1ffc1',
104        'darkseagreen2' => 'b4eeb4',
105        'darkseagreen3' => '9bcd9b',
106        'darkseagreen4' => '698b69',
107        'darkslateblue' => '483d8b',
108        'darkslategray' => '2f4f4f',
109        'darkslategray1' => '97ffff',
110        'darkslategray2' => '8deeee',
111        'darkslategray3' => '79cdcd',
112        'darkslategray4' => '528b8b',
113        'darkturquoise' => '00ced1',
114        'darkviolet' => '9400d3',
115        'deeppink1' => 'ff1493',
116        'deeppink2' => 'ee1289',
117        'deeppink3' => 'cd1076',
118        'deeppink4' => '8b0a50',
119        'deepskyblue1' => '00bfff',
120        'deepskyblue2' => '00b2ee',
121        'deepskyblue3' => '009acd',
122        'deepskyblue4' => '00688b',
123        'dimgray' => '696969',
124        'dodgerblue1' => '1e90ff',
125        'dodgerblue2' => '1c86ee',
126        'dodgerblue3' => '1874cd',
127        'dodgerblue4' => '104e8b',
128        'firebrick' => 'b22222',
129        'firebrick1' => 'ff3030',
130        'firebrick2' => 'ee2c2c',
131        'firebrick3' => 'cd2626',
132        'firebrick4' => '8b1a1a',
133        'floralwhite' => 'fffaf0',
134        'forestgreen' => '228b22',
135        'fuchsia' => 'ff00ff',
136        'gainsboro' => 'dcdcdc',
137        'ghostwhite' => 'f8f8ff',
138        'gold1' => 'ffd700',
139        'gold2' => 'eec900',
140        'gold3' => 'cdad00',
141        'gold4' => '8b7500',
142        'goldenrod' => 'daa520',
143        'goldenrod1' => 'ffc125',
144        'goldenrod2' => 'eeb422',
145        'goldenrod3' => 'cd9b1d',
146        'goldenrod4' => '8b6914',
147        'gray' => 'bebebe',
148        'gray1' => '030303',
149        'gray10' => '1a1a1a',
150        'gray11' => '1c1c1c',
151        'gray12' => '1f1f1f',
152        'gray13' => '212121',
153        'gray14' => '242424',
154        'gray15' => '262626',
155        'gray16' => '292929',
156        'gray17' => '2b2b2b',
157        'gray18' => '2e2e2e',
158        'gray19' => '303030',
159        'gray2' => '050505',
160        'gray20' => '333333',
161        'gray21' => '363636',
162        'gray22' => '383838',
163        'gray23' => '3b3b3b',
164        'gray24' => '3d3d3d',
165        'gray25' => '404040',
166        'gray26' => '424242',
167        'gray27' => '454545',
168        'gray28' => '474747',
169        'gray29' => '4a4a4a',
170        'gray3' => '080808',
171        'gray30' => '4d4d4d',
172        'gray31' => '4f4f4f',
173        'gray32' => '525252',
174        'gray33' => '545454',
175        'gray34' => '575757',
176        'gray35' => '595959',
177        'gray36' => '5c5c5c',
178        'gray37' => '5e5e5e',
179        'gray38' => '616161',
180        'gray39' => '636363',
181        'gray4' => '0a0a0a',
182        'gray40' => '666666',
183        'gray41' => '696969',
184        'gray42' => '6b6b6b',
185        'gray43' => '6e6e6e',
186        'gray44' => '707070',
187        'gray45' => '737373',
188        'gray46' => '757575',
189        'gray47' => '787878',
190        'gray48' => '7a7a7a',
191        'gray49' => '7d7d7d',
192        'gray5' => '0d0d0d',
193        'gray50' => '7f7f7f',
194        'gray51' => '828282',
195        'gray52' => '858585',
196        'gray53' => '878787',
197        'gray54' => '8a8a8a',
198        'gray55' => '8c8c8c',
199        'gray56' => '8f8f8f',
200        'gray57' => '919191',
201        'gray58' => '949494',
202        'gray59' => '969696',
203        'gray6' => '0f0f0f',
204        'gray60' => '999999',
205        'gray61' => '9c9c9c',
206        'gray62' => '9e9e9e',
207        'gray63' => 'a1a1a1',
208        'gray64' => 'a3a3a3',
209        'gray65' => 'a6a6a6',
210        'gray66' => 'a8a8a8',
211        'gray67' => 'ababab',
212        'gray68' => 'adadad',
213        'gray69' => 'b0b0b0',
214        'gray7' => '121212',
215        'gray70' => 'b3b3b3',
216        'gray71' => 'b5b5b5',
217        'gray72' => 'b8b8b8',
218        'gray73' => 'bababa',
219        'gray74' => 'bdbdbd',
220        'gray75' => 'bfbfbf',
221        'gray76' => 'c2c2c2',
222        'gray77' => 'c4c4c4',
223        'gray78' => 'c7c7c7',
224        'gray79' => 'c9c9c9',
225        'gray8' => '141414',
226        'gray80' => 'cccccc',
227        'gray81' => 'cfcfcf',
228        'gray82' => 'd1d1d1',
229        'gray83' => 'd4d4d4',
230        'gray84' => 'd6d6d6',
231        'gray85' => 'd9d9d9',
232        'gray86' => 'dbdbdb',
233        'gray87' => 'dedede',
234        'gray88' => 'e0e0e0',
235        'gray89' => 'e3e3e3',
236        'gray9' => '171717',
237        'gray90' => 'e5e5e5',
238        'gray91' => 'e8e8e8',
239        'gray92' => 'ebebeb',
240        'gray93' => 'ededed',
241        'gray94' => 'f0f0f0',
242        'gray95' => 'f2f2f2',
243        'gray97' => 'f7f7f7',
244        'gray98' => 'fafafa',
245        'gray99' => 'fcfcfc',
246        'green' => '00ff00',
247        'green1' => '00ff00',
248        'green2' => '00ee00',
249        'green3' => '00cd00',
250        'green4' => '008b00',
251        'greenyellow' => 'adff2f',
252        'honeydew1' => 'f0fff0',
253        'honeydew2' => 'e0eee0',
254        'honeydew3' => 'c1cdc1',
255        'honeydew4' => '838b83',
256        'hotpink' => 'ff69b4',
257        'hotpink1' => 'ff6eb4',
258        'hotpink2' => 'ee6aa7',
259        'hotpink3' => 'cd6090',
260        'hotpink4' => '8b3a62',
261        'indianred' => 'cd5c5c',
262        'indianred1' => 'ff6a6a',
263        'indianred2' => 'ee6363',
264        'indianred3' => 'cd5555',
265        'indianred4' => '8b3a3a',
266        'ivory1' => 'fffff0',
267        'ivory2' => 'eeeee0',
268        'ivory3' => 'cdcdc1',
269        'ivory4' => '8b8b83',
270        'khaki' => 'f0e68c',
271        'khaki1' => 'fff68f',
272        'khaki2' => 'eee685',
273        'khaki3' => 'cdc673',
274        'khaki4' => '8b864e',
275        'lavender' => 'e6e6fa',
276        'lavenderblush1' => 'fff0f5',
277        'lavenderblush2' => 'eee0e5',
278        'lavenderblush3' => 'cdc1c5',
279        'lavenderblush4' => '8b8386',
280        'lawngreen' => '7cfc00',
281        'lemonchiffon1' => 'fffacd',
282        'lemonchiffon2' => 'eee9bf',
283        'lemonchiffon3' => 'cdc9a5',
284        'lemonchiffon4' => '8b8970',
285        'light' => 'eedd82',
286        'lightblue' => 'add8e6',
287        'lightblue1' => 'bfefff',
288        'lightblue2' => 'b2dfee',
289        'lightblue3' => '9ac0cd',
290        'lightblue4' => '68838b',
291        'lightcoral' => 'f08080',
292        'lightcyan1' => 'e0ffff',
293        'lightcyan2' => 'd1eeee',
294        'lightcyan3' => 'b4cdcd',
295        'lightcyan4' => '7a8b8b',
296        'lightgoldenrod1' => 'ffec8b',
297        'lightgoldenrod2' => 'eedc82',
298        'lightgoldenrod3' => 'cdbe70',
299        'lightgoldenrod4' => '8b814c',
300        'lightgoldenrodyellow' => 'fafad2',
301        'lightgray' => 'd3d3d3',
302        'lightpink' => 'ffb6c1',
303        'lightpink1' => 'ffaeb9',
304        'lightpink2' => 'eea2ad',
305        'lightpink3' => 'cd8c95',
306        'lightpink4' => '8b5f65',
307        'lightsalmon1' => 'ffa07a',
308        'lightsalmon2' => 'ee9572',
309        'lightsalmon3' => 'cd8162',
310        'lightsalmon4' => '8b5742',
311        'lightseagreen' => '20b2aa',
312        'lightskyblue' => '87cefa',
313        'lightskyblue1' => 'b0e2ff',
314        'lightskyblue2' => 'a4d3ee',
315        'lightskyblue3' => '8db6cd',
316        'lightskyblue4' => '607b8b',
317        'lightslateblue' => '8470ff',
318        'lightslategray' => '778899',
319        'lightsteelblue' => 'b0c4de',
320        'lightsteelblue1' => 'cae1ff',
321        'lightsteelblue2' => 'bcd2ee',
322        'lightsteelblue3' => 'a2b5cd',
323        'lightsteelblue4' => '6e7b8b',
324        'lightyellow1' => 'ffffe0',
325        'lightyellow2' => 'eeeed1',
326        'lightyellow3' => 'cdcdb4',
327        'lightyellow4' => '8b8b7a',
328        'lime' => '00ff00',
329        'limegreen' => '32cd32',
330        'linen' => 'faf0e6',
331        'magenta' => 'ff00ff',
332        'magenta2' => 'ee00ee',
333        'magenta3' => 'cd00cd',
334        'magenta4' => '8b008b',
335        'maroon' => 'b03060',
336        'maroon1' => 'ff34b3',
337        'maroon2' => 'ee30a7',
338        'maroon3' => 'cd2990',
339        'maroon4' => '8b1c62',
340        'medium' => '66cdaa',
341        'mediumaquamarine' => '66cdaa',
342        'mediumblue' => '0000cd',
343        'mediumorchid' => 'ba55d3',
344        'mediumorchid1' => 'e066ff',
345        'mediumorchid2' => 'd15fee',
346        'mediumorchid3' => 'b452cd',
347        'mediumorchid4' => '7a378b',
348        'mediumpurple' => '9370db',
349        'mediumpurple1' => 'ab82ff',
350        'mediumpurple2' => '9f79ee',
351        'mediumpurple3' => '8968cd',
352        'mediumpurple4' => '5d478b',
353        'mediumseagreen' => '3cb371',
354        'mediumslateblue' => '7b68ee',
355        'mediumspringgreen' => '00fa9a',
356        'mediumturquoise' => '48d1cc',
357        'mediumvioletred' => 'c71585',
358        'midnightblue' => '191970',
359        'mintcream' => 'f5fffa',
360        'mistyrose1' => 'ffe4e1',
361        'mistyrose2' => 'eed5d2',
362        'mistyrose3' => 'cdb7b5',
363        'mistyrose4' => '8b7d7b',
364        'moccasin' => 'ffe4b5',
365        'navajowhite1' => 'ffdead',
366        'navajowhite2' => 'eecfa1',
367        'navajowhite3' => 'cdb38b',
368        'navajowhite4' => '8b795e',
369        'navy' => '000080',
370        'navyblue' => '000080',
371        'oldlace' => 'fdf5e6',
372        'olive' => '808000',
373        'olivedrab' => '6b8e23',
374        'olivedrab1' => 'c0ff3e',
375        'olivedrab2' => 'b3ee3a',
376        'olivedrab4' => '698b22',
377        'orange' => 'ffa500',
378        'orange1' => 'ffa500',
379        'orange2' => 'ee9a00',
380        'orange3' => 'cd8500',
381        'orange4' => '8b5a00',
382        'orangered1' => 'ff4500',
383        'orangered2' => 'ee4000',
384        'orangered3' => 'cd3700',
385        'orangered4' => '8b2500',
386        'orchid' => 'da70d6',
387        'orchid1' => 'ff83fa',
388        'orchid2' => 'ee7ae9',
389        'orchid3' => 'cd69c9',
390        'orchid4' => '8b4789',
391        'pale' => 'db7093',
392        'palegoldenrod' => 'eee8aa',
393        'palegreen' => '98fb98',
394        'palegreen1' => '9aff9a',
395        'palegreen2' => '90ee90',
396        'palegreen3' => '7ccd7c',
397        'palegreen4' => '548b54',
398        'paleturquoise' => 'afeeee',
399        'paleturquoise1' => 'bbffff',
400        'paleturquoise2' => 'aeeeee',
401        'paleturquoise3' => '96cdcd',
402        'paleturquoise4' => '668b8b',
403        'palevioletred' => 'db7093',
404        'palevioletred1' => 'ff82ab',
405        'palevioletred2' => 'ee799f',
406        'palevioletred3' => 'cd6889',
407        'palevioletred4' => '8b475d',
408        'papayawhip' => 'ffefd5',
409        'peachpuff1' => 'ffdab9',
410        'peachpuff2' => 'eecbad',
411        'peachpuff3' => 'cdaf95',
412        'peachpuff4' => '8b7765',
413        'pink' => 'ffc0cb',
414        'pink1' => 'ffb5c5',
415        'pink2' => 'eea9b8',
416        'pink3' => 'cd919e',
417        'pink4' => '8b636c',
418        'plum' => 'dda0dd',
419        'plum1' => 'ffbbff',
420        'plum2' => 'eeaeee',
421        'plum3' => 'cd96cd',
422        'plum4' => '8b668b',
423        'powderblue' => 'b0e0e6',
424        'purple' => 'a020f0',
425        'rebeccapurple' => '663399',
426        'purple1' => '9b30ff',
427        'purple2' => '912cee',
428        'purple3' => '7d26cd',
429        'purple4' => '551a8b',
430        'red' => 'ff0000',
431        'red1' => 'ff0000',
432        'red2' => 'ee0000',
433        'red3' => 'cd0000',
434        'red4' => '8b0000',
435        'rosybrown' => 'bc8f8f',
436        'rosybrown1' => 'ffc1c1',
437        'rosybrown2' => 'eeb4b4',
438        'rosybrown3' => 'cd9b9b',
439        'rosybrown4' => '8b6969',
440        'royalblue' => '4169e1',
441        'royalblue1' => '4876ff',
442        'royalblue2' => '436eee',
443        'royalblue3' => '3a5fcd',
444        'royalblue4' => '27408b',
445        'saddlebrown' => '8b4513',
446        'salmon' => 'fa8072',
447        'salmon1' => 'ff8c69',
448        'salmon2' => 'ee8262',
449        'salmon3' => 'cd7054',
450        'salmon4' => '8b4c39',
451        'sandybrown' => 'f4a460',
452        'seagreen1' => '54ff9f',
453        'seagreen2' => '4eee94',
454        'seagreen3' => '43cd80',
455        'seagreen4' => '2e8b57',
456        'seashell1' => 'fff5ee',
457        'seashell2' => 'eee5de',
458        'seashell3' => 'cdc5bf',
459        'seashell4' => '8b8682',
460        'sienna' => 'a0522d',
461        'sienna1' => 'ff8247',
462        'sienna2' => 'ee7942',
463        'sienna3' => 'cd6839',
464        'sienna4' => '8b4726',
465        'silver' => 'c0c0c0',
466        'skyblue' => '87ceeb',
467        'skyblue1' => '87ceff',
468        'skyblue2' => '7ec0ee',
469        'skyblue3' => '6ca6cd',
470        'skyblue4' => '4a708b',
471        'slateblue' => '6a5acd',
472        'slateblue1' => '836fff',
473        'slateblue2' => '7a67ee',
474        'slateblue3' => '6959cd',
475        'slateblue4' => '473c8b',
476        'slategray' => '708090',
477        'slategray1' => 'c6e2ff',
478        'slategray2' => 'b9d3ee',
479        'slategray3' => '9fb6cd',
480        'slategray4' => '6c7b8b',
481        'snow1' => 'fffafa',
482        'snow2' => 'eee9e9',
483        'snow3' => 'cdc9c9',
484        'snow4' => '8b8989',
485        'springgreen1' => '00ff7f',
486        'springgreen2' => '00ee76',
487        'springgreen3' => '00cd66',
488        'springgreen4' => '008b45',
489        'steelblue' => '4682b4',
490        'steelblue1' => '63b8ff',
491        'steelblue2' => '5cacee',
492        'steelblue3' => '4f94cd',
493        'steelblue4' => '36648b',
494        'tan' => 'd2b48c',
495        'tan1' => 'ffa54f',
496        'tan2' => 'ee9a49',
497        'tan3' => 'cd853f',
498        'tan4' => '8b5a2b',
499        'teal' => '008080',
500        'thistle' => 'd8bfd8',
501        'thistle1' => 'ffe1ff',
502        'thistle2' => 'eed2ee',
503        'thistle3' => 'cdb5cd',
504        'thistle4' => '8b7b8b',
505        'tomato1' => 'ff6347',
506        'tomato2' => 'ee5c42',
507        'tomato3' => 'cd4f39',
508        'tomato4' => '8b3626',
509        'turquoise' => '40e0d0',
510        'turquoise1' => '00f5ff',
511        'turquoise2' => '00e5ee',
512        'turquoise3' => '00c5cd',
513        'turquoise4' => '00868b',
514        'violet' => 'ee82ee',
515        'violetred' => 'd02090',
516        'violetred1' => 'ff3e96',
517        'violetred2' => 'ee3a8c',
518        'violetred3' => 'cd3278',
519        'violetred4' => '8b2252',
520        'wheat' => 'f5deb3',
521        'wheat1' => 'ffe7ba',
522        'wheat2' => 'eed8ae',
523        'wheat3' => 'cdba96',
524        'wheat4' => '8b7e66',
525        'white' => 'ffffff',
526        'whitesmoke' => 'f5f5f5',
527        'yellow' => 'ffff00',
528        'yellow1' => 'ffff00',
529        'yellow2' => 'eeee00',
530        'yellow3' => 'cdcd00',
531        'yellow4' => '8b8b00',
532        'yellowgreen' => '9acd32',
533    ];
534
535    protected $face;
536
537    protected $size;
538
539    protected $color;
540
541    protected $bold = false;
542
543    protected $italic = false;
544
545    protected $underline = false;
546
547    protected $superscript = false;
548
549    protected $subscript = false;
550
551    protected $strikethrough = false;
552
553    protected $startTagCallbacks = [
554        'font' => 'startFontTag',
555        'b' => 'startBoldTag',
556        'strong' => 'startBoldTag',
557        'i' => 'startItalicTag',
558        'em' => 'startItalicTag',
559        'u' => 'startUnderlineTag',
560        'ins' => 'startUnderlineTag',
561        'del' => 'startStrikethruTag',
562        'sup' => 'startSuperscriptTag',
563        'sub' => 'startSubscriptTag',
564    ];
565
566    protected $endTagCallbacks = [
567        'font' => 'endFontTag',
568        'b' => 'endBoldTag',
569        'strong' => 'endBoldTag',
570        'i' => 'endItalicTag',
571        'em' => 'endItalicTag',
572        'u' => 'endUnderlineTag',
573        'ins' => 'endUnderlineTag',
574        'del' => 'endStrikethruTag',
575        'sup' => 'endSuperscriptTag',
576        'sub' => 'endSubscriptTag',
577        'br' => 'breakTag',
578        'p' => 'breakTag',
579        'h1' => 'breakTag',
580        'h2' => 'breakTag',
581        'h3' => 'breakTag',
582        'h4' => 'breakTag',
583        'h5' => 'breakTag',
584        'h6' => 'breakTag',
585    ];
586
587    protected $stack = [];
588
589    protected $stringData = '';
590
591    /**
592     * @var RichText
593     */
594    protected $richTextObject;
595
596    protected function initialise()
597    {
598        $this->face = $this->size = $this->color = null;
599        $this->bold = $this->italic = $this->underline = $this->superscript = $this->subscript = $this->strikethrough = false;
600
601        $this->stack = [];
602
603        $this->stringData = '';
604    }
605
606    /**
607     * Parse HTML formatting and return the resulting RichText.
608     *
609     * @param string $html
610     *
611     * @return RichText
612     */
613    public function toRichTextObject($html)
614    {
615        $this->initialise();
616
617        //    Create a new DOM object
618        $dom = new DOMDocument();
619        //    Load the HTML file into the DOM object
620        //  Note the use of error suppression, because typically this will be an html fragment, so not fully valid markup
621        $prefix = '<?xml encoding="UTF-8">';
622        @$dom->loadHTML($prefix . $html, LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD);
623        //    Discard excess white space
624        $dom->preserveWhiteSpace = false;
625
626        $this->richTextObject = new RichText();
627        $this->parseElements($dom);
628
629        // Clean any further spurious whitespace
630        $this->cleanWhitespace();
631
632        return $this->richTextObject;
633    }
634
635    protected function cleanWhitespace()
636    {
637        foreach ($this->richTextObject->getRichTextElements() as $key => $element) {
638            $text = $element->getText();
639            // Trim any leading spaces on the first run
640            if ($key == 0) {
641                $text = ltrim($text);
642            }
643            // Trim any spaces immediately after a line break
644            $text = preg_replace('/\n */mu', "\n", $text);
645            $element->setText($text);
646        }
647    }
648
649    protected function buildTextRun()
650    {
651        $text = $this->stringData;
652        if (trim($text) === '') {
653            return;
654        }
655
656        $richtextRun = $this->richTextObject->createTextRun($this->stringData);
657        if ($this->face) {
658            $richtextRun->getFont()->setName($this->face);
659        }
660        if ($this->size) {
661            $richtextRun->getFont()->setSize($this->size);
662        }
663        if ($this->color) {
664            $richtextRun->getFont()->setColor(new Color('ff' . $this->color));
665        }
666        if ($this->bold) {
667            $richtextRun->getFont()->setBold(true);
668        }
669        if ($this->italic) {
670            $richtextRun->getFont()->setItalic(true);
671        }
672        if ($this->underline) {
673            $richtextRun->getFont()->setUnderline(Font::UNDERLINE_SINGLE);
674        }
675        if ($this->superscript) {
676            $richtextRun->getFont()->setSuperscript(true);
677        }
678        if ($this->subscript) {
679            $richtextRun->getFont()->setSubscript(true);
680        }
681        if ($this->strikethrough) {
682            $richtextRun->getFont()->setStrikethrough(true);
683        }
684        $this->stringData = '';
685    }
686
687    protected function rgbToColour($rgb)
688    {
689        preg_match_all('/\d+/', $rgb, $values);
690        foreach ($values[0] as &$value) {
691            $value = str_pad(dechex($value), 2, '0', STR_PAD_LEFT);
692        }
693
694        return implode($values[0]);
695    }
696
697    protected function colourNameLookup($rgb)
698    {
699        return self::$colourMap[$rgb];
700    }
701
702    protected function startFontTag($tag)
703    {
704        foreach ($tag->attributes as $attribute) {
705            $attributeName = strtolower($attribute->name);
706            $attributeValue = $attribute->value;
707
708            if ($attributeName == 'color') {
709                if (preg_match('/rgb\s*\(/', $attributeValue)) {
710                    $this->$attributeName = $this->rgbToColour($attributeValue);
711                } elseif (strpos(trim($attributeValue), '#') === 0) {
712                    $this->$attributeName = ltrim($attributeValue, '#');
713                } else {
714                    $this->$attributeName = $this->colourNameLookup($attributeValue);
715                }
716            } else {
717                $this->$attributeName = $attributeValue;
718            }
719        }
720    }
721
722    protected function endFontTag()
723    {
724        $this->face = $this->size = $this->color = null;
725    }
726
727    protected function startBoldTag()
728    {
729        $this->bold = true;
730    }
731
732    protected function endBoldTag()
733    {
734        $this->bold = false;
735    }
736
737    protected function startItalicTag()
738    {
739        $this->italic = true;
740    }
741
742    protected function endItalicTag()
743    {
744        $this->italic = false;
745    }
746
747    protected function startUnderlineTag()
748    {
749        $this->underline = true;
750    }
751
752    protected function endUnderlineTag()
753    {
754        $this->underline = false;
755    }
756
757    protected function startSubscriptTag()
758    {
759        $this->subscript = true;
760    }
761
762    protected function endSubscriptTag()
763    {
764        $this->subscript = false;
765    }
766
767    protected function startSuperscriptTag()
768    {
769        $this->superscript = true;
770    }
771
772    protected function endSuperscriptTag()
773    {
774        $this->superscript = false;
775    }
776
777    protected function startStrikethruTag()
778    {
779        $this->strikethrough = true;
780    }
781
782    protected function endStrikethruTag()
783    {
784        $this->strikethrough = false;
785    }
786
787    protected function breakTag()
788    {
789        $this->stringData .= "\n";
790    }
791
792    protected function parseTextNode(DOMText $textNode)
793    {
794        $domText = preg_replace(
795            '/\s+/u',
796            ' ',
797            str_replace(["\r", "\n"], ' ', $textNode->nodeValue)
798        );
799        $this->stringData .= $domText;
800        $this->buildTextRun();
801    }
802
803    /**
804     * @param DOMElement $element
805     * @param string $callbackTag
806     * @param array $callbacks
807     */
808    protected function handleCallback(DOMElement $element, $callbackTag, array $callbacks)
809    {
810        if (isset($callbacks[$callbackTag])) {
811            $elementHandler = $callbacks[$callbackTag];
812            if (method_exists($this, $elementHandler)) {
813                call_user_func([$this, $elementHandler], $element);
814            }
815        }
816    }
817
818    protected function parseElementNode(DOMElement $element)
819    {
820        $callbackTag = strtolower($element->nodeName);
821        $this->stack[] = $callbackTag;
822
823        $this->handleCallback($element, $callbackTag, $this->startTagCallbacks);
824
825        $this->parseElements($element);
826        array_pop($this->stack);
827
828        $this->handleCallback($element, $callbackTag, $this->endTagCallbacks);
829    }
830
831    protected function parseElements(DOMNode $element)
832    {
833        foreach ($element->childNodes as $child) {
834            if ($child instanceof DOMText) {
835                $this->parseTextNode($child);
836            } elseif ($child instanceof DOMElement) {
837                $this->parseElementNode($child);
838            }
839        }
840    }
841}
842