|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace PhpOffice\PhpSpreadsheet\Reader; |
|
4
|
|
|
|
|
5
|
|
|
use DOMDocument; |
|
6
|
|
|
use DOMElement; |
|
7
|
|
|
use DOMNode; |
|
8
|
|
|
use DOMText; |
|
9
|
|
|
use PhpOffice\PhpSpreadsheet\Cell\Coordinate; |
|
10
|
|
|
use PhpOffice\PhpSpreadsheet\Spreadsheet; |
|
11
|
|
|
use PhpOffice\PhpSpreadsheet\Style\Border; |
|
12
|
|
|
use PhpOffice\PhpSpreadsheet\Style\Color; |
|
13
|
|
|
use PhpOffice\PhpSpreadsheet\Style\Fill; |
|
14
|
|
|
use PhpOffice\PhpSpreadsheet\Worksheet\Worksheet; |
|
15
|
|
|
|
|
16
|
|
|
/** PhpSpreadsheet root directory */ |
|
17
|
|
|
class Html extends BaseReader |
|
18
|
|
|
{ |
|
19
|
|
|
/** |
|
20
|
|
|
* Sample size to read to determine if it's HTML or not. |
|
21
|
|
|
*/ |
|
22
|
|
|
const TEST_SAMPLE_SIZE = 2048; |
|
23
|
|
|
|
|
24
|
|
|
/** |
|
25
|
|
|
* Input encoding. |
|
26
|
|
|
* |
|
27
|
|
|
* @var string |
|
28
|
|
|
*/ |
|
29
|
|
|
protected $inputEncoding = 'ANSI'; |
|
30
|
|
|
|
|
31
|
|
|
/** |
|
32
|
|
|
* Sheet index to read. |
|
33
|
|
|
* |
|
34
|
|
|
* @var int |
|
35
|
|
|
*/ |
|
36
|
|
|
protected $sheetIndex = 0; |
|
37
|
|
|
|
|
38
|
|
|
/** |
|
39
|
|
|
* Formats. |
|
40
|
|
|
* |
|
41
|
|
|
* @var array |
|
42
|
|
|
*/ |
|
43
|
|
|
protected $formats = [ |
|
44
|
|
|
'h1' => [ |
|
45
|
|
|
'font' => [ |
|
46
|
|
|
'bold' => true, |
|
47
|
|
|
'size' => 24, |
|
48
|
|
|
], |
|
49
|
|
|
], // Bold, 24pt |
|
50
|
|
|
'h2' => [ |
|
51
|
|
|
'font' => [ |
|
52
|
|
|
'bold' => true, |
|
53
|
|
|
'size' => 18, |
|
54
|
|
|
], |
|
55
|
|
|
], // Bold, 18pt |
|
56
|
|
|
'h3' => [ |
|
57
|
|
|
'font' => [ |
|
58
|
|
|
'bold' => true, |
|
59
|
|
|
'size' => 13.5, |
|
60
|
|
|
], |
|
61
|
|
|
], // Bold, 13.5pt |
|
62
|
|
|
'h4' => [ |
|
63
|
|
|
'font' => [ |
|
64
|
|
|
'bold' => true, |
|
65
|
|
|
'size' => 12, |
|
66
|
|
|
], |
|
67
|
|
|
], // Bold, 12pt |
|
68
|
|
|
'h5' => [ |
|
69
|
|
|
'font' => [ |
|
70
|
|
|
'bold' => true, |
|
71
|
|
|
'size' => 10, |
|
72
|
|
|
], |
|
73
|
|
|
], // Bold, 10pt |
|
74
|
|
|
'h6' => [ |
|
75
|
|
|
'font' => [ |
|
76
|
|
|
'bold' => true, |
|
77
|
|
|
'size' => 7.5, |
|
78
|
|
|
], |
|
79
|
|
|
], // Bold, 7.5pt |
|
80
|
|
|
'a' => [ |
|
81
|
|
|
'font' => [ |
|
82
|
|
|
'underline' => true, |
|
83
|
|
|
'color' => [ |
|
84
|
|
|
'argb' => Color::COLOR_BLUE, |
|
85
|
|
|
], |
|
86
|
|
|
], |
|
87
|
|
|
], // Blue underlined |
|
88
|
|
|
'hr' => [ |
|
89
|
|
|
'borders' => [ |
|
90
|
|
|
'bottom' => [ |
|
91
|
|
|
'borderStyle' => Border::BORDER_THIN, |
|
92
|
|
|
'color' => [ |
|
93
|
|
|
Color::COLOR_BLACK, |
|
94
|
|
|
], |
|
95
|
|
|
], |
|
96
|
|
|
], |
|
97
|
|
|
], // Bottom border |
|
98
|
|
|
]; |
|
99
|
|
|
|
|
100
|
|
|
protected $rowspan = []; |
|
101
|
|
|
|
|
102
|
|
|
/** |
|
103
|
|
|
* Create a new HTML Reader instance. |
|
104
|
|
|
*/ |
|
105
|
17 |
|
public function __construct() |
|
106
|
|
|
{ |
|
107
|
17 |
|
$this->readFilter = new DefaultReadFilter(); |
|
108
|
17 |
|
} |
|
109
|
|
|
|
|
110
|
|
|
/** |
|
111
|
|
|
* Validate that the current file is an HTML file. |
|
112
|
|
|
* |
|
113
|
|
|
* @param string $pFilename |
|
114
|
|
|
* |
|
115
|
|
|
* @return bool |
|
116
|
|
|
*/ |
|
117
|
15 |
|
public function canRead($pFilename) |
|
118
|
|
|
{ |
|
119
|
|
|
// Check if file exists |
|
120
|
|
|
try { |
|
121
|
15 |
|
$this->openFile($pFilename); |
|
122
|
|
|
} catch (Exception $e) { |
|
123
|
|
|
return false; |
|
124
|
|
|
} |
|
125
|
|
|
|
|
126
|
15 |
|
$beginning = $this->readBeginning(); |
|
127
|
15 |
|
$startWithTag = self::startsWithTag($beginning); |
|
128
|
15 |
|
$containsTags = self::containsTags($beginning); |
|
129
|
15 |
|
$endsWithTag = self::endsWithTag($this->readEnding()); |
|
130
|
|
|
|
|
131
|
15 |
|
fclose($this->fileHandle); |
|
|
|
|
|
|
132
|
|
|
|
|
133
|
15 |
|
return $startWithTag && $containsTags && $endsWithTag; |
|
134
|
|
|
} |
|
135
|
|
|
|
|
136
|
15 |
|
private function readBeginning() |
|
137
|
|
|
{ |
|
138
|
15 |
|
fseek($this->fileHandle, 0); |
|
139
|
|
|
|
|
140
|
15 |
|
return fread($this->fileHandle, self::TEST_SAMPLE_SIZE); |
|
141
|
|
|
} |
|
142
|
|
|
|
|
143
|
15 |
|
private function readEnding() |
|
144
|
|
|
{ |
|
145
|
15 |
|
$meta = stream_get_meta_data($this->fileHandle); |
|
146
|
15 |
|
$filename = $meta['uri']; |
|
147
|
|
|
|
|
148
|
15 |
|
$size = filesize($filename); |
|
149
|
15 |
|
if ($size === 0) { |
|
150
|
1 |
|
return ''; |
|
151
|
|
|
} |
|
152
|
|
|
|
|
153
|
14 |
|
$blockSize = self::TEST_SAMPLE_SIZE; |
|
154
|
14 |
|
if ($size < $blockSize) { |
|
155
|
2 |
|
$blockSize = $size; |
|
156
|
|
|
} |
|
157
|
|
|
|
|
158
|
14 |
|
fseek($this->fileHandle, $size - $blockSize); |
|
159
|
|
|
|
|
160
|
14 |
|
return fread($this->fileHandle, $blockSize); |
|
161
|
|
|
} |
|
162
|
|
|
|
|
163
|
15 |
|
private static function startsWithTag($data) |
|
164
|
|
|
{ |
|
165
|
15 |
|
return '<' === substr(trim($data), 0, 1); |
|
166
|
|
|
} |
|
167
|
|
|
|
|
168
|
15 |
|
private static function endsWithTag($data) |
|
169
|
|
|
{ |
|
170
|
15 |
|
return '>' === substr(trim($data), -1, 1); |
|
171
|
|
|
} |
|
172
|
|
|
|
|
173
|
15 |
|
private static function containsTags($data) |
|
174
|
|
|
{ |
|
175
|
15 |
|
return strlen($data) !== strlen(strip_tags($data)); |
|
176
|
|
|
} |
|
177
|
|
|
|
|
178
|
|
|
/** |
|
179
|
|
|
* Loads Spreadsheet from file. |
|
180
|
|
|
* |
|
181
|
|
|
* @param string $pFilename |
|
182
|
|
|
* |
|
183
|
|
|
* @throws Exception |
|
184
|
|
|
* |
|
185
|
|
|
* @return Spreadsheet |
|
186
|
|
|
*/ |
|
187
|
8 |
|
public function load($pFilename) |
|
188
|
|
|
{ |
|
189
|
|
|
// Create new Spreadsheet |
|
190
|
8 |
|
$spreadsheet = new Spreadsheet(); |
|
191
|
|
|
|
|
192
|
|
|
// Load into this instance |
|
193
|
8 |
|
return $this->loadIntoExisting($pFilename, $spreadsheet); |
|
194
|
|
|
} |
|
195
|
|
|
|
|
196
|
|
|
/** |
|
197
|
|
|
* Set input encoding. |
|
198
|
|
|
* |
|
199
|
|
|
* @param string $pValue Input encoding, eg: 'ANSI' |
|
200
|
|
|
*/ |
|
201
|
|
|
public function setInputEncoding($pValue) |
|
202
|
|
|
{ |
|
203
|
|
|
$this->inputEncoding = $pValue; |
|
204
|
|
|
|
|
205
|
|
|
return $this; |
|
206
|
|
|
} |
|
207
|
|
|
|
|
208
|
|
|
/** |
|
209
|
|
|
* Get input encoding. |
|
210
|
|
|
* |
|
211
|
|
|
* @return string |
|
212
|
|
|
*/ |
|
213
|
|
|
public function getInputEncoding() |
|
214
|
|
|
{ |
|
215
|
|
|
return $this->inputEncoding; |
|
216
|
|
|
} |
|
217
|
|
|
|
|
218
|
|
|
// Data Array used for testing only, should write to Spreadsheet object on completion of tests |
|
219
|
|
|
protected $dataArray = []; |
|
220
|
|
|
protected $tableLevel = 0; |
|
221
|
|
|
protected $nestedColumn = ['A']; |
|
222
|
|
|
|
|
223
|
8 |
|
protected function setTableStartColumn($column) |
|
224
|
|
|
{ |
|
225
|
8 |
|
if ($this->tableLevel == 0) { |
|
226
|
8 |
|
$column = 'A'; |
|
227
|
|
|
} |
|
228
|
8 |
|
++$this->tableLevel; |
|
229
|
8 |
|
$this->nestedColumn[$this->tableLevel] = $column; |
|
230
|
|
|
|
|
231
|
8 |
|
return $this->nestedColumn[$this->tableLevel]; |
|
232
|
|
|
} |
|
233
|
|
|
|
|
234
|
8 |
|
protected function getTableStartColumn() |
|
235
|
|
|
{ |
|
236
|
8 |
|
return $this->nestedColumn[$this->tableLevel]; |
|
237
|
|
|
} |
|
238
|
|
|
|
|
239
|
8 |
|
protected function releaseTableStartColumn() |
|
240
|
|
|
{ |
|
241
|
8 |
|
--$this->tableLevel; |
|
242
|
|
|
|
|
243
|
8 |
|
return array_pop($this->nestedColumn); |
|
244
|
|
|
} |
|
245
|
|
|
|
|
246
|
8 |
|
protected function flushCell(Worksheet $sheet, $column, $row, &$cellContent) |
|
247
|
|
|
{ |
|
248
|
8 |
|
if (is_string($cellContent)) { |
|
249
|
|
|
// Simple String content |
|
250
|
8 |
|
if (trim($cellContent) > '') { |
|
251
|
|
|
// Only actually write it if there's content in the string |
|
252
|
|
|
// Write to worksheet to be done here... |
|
253
|
|
|
// ... we return the cell so we can mess about with styles more easily |
|
254
|
8 |
|
$sheet->setCellValue($column . $row, $cellContent); |
|
255
|
8 |
|
$this->dataArray[$row][$column] = $cellContent; |
|
256
|
|
|
} |
|
257
|
|
|
} else { |
|
258
|
|
|
// We have a Rich Text run |
|
259
|
|
|
// TODO |
|
260
|
|
|
$this->dataArray[$row][$column] = 'RICH TEXT: ' . $cellContent; |
|
261
|
|
|
} |
|
262
|
8 |
|
$cellContent = (string) ''; |
|
263
|
8 |
|
} |
|
264
|
|
|
|
|
265
|
|
|
/** |
|
266
|
|
|
* @param DOMNode $element |
|
267
|
|
|
* @param Worksheet $sheet |
|
268
|
|
|
* @param int $row |
|
269
|
|
|
* @param string $column |
|
270
|
|
|
* @param string $cellContent |
|
271
|
|
|
*/ |
|
272
|
8 |
|
protected function processDomElement(DOMNode $element, Worksheet $sheet, &$row, &$column, &$cellContent) |
|
273
|
|
|
{ |
|
274
|
8 |
|
foreach ($element->childNodes as $child) { |
|
275
|
8 |
|
if ($child instanceof DOMText) { |
|
276
|
8 |
|
$domText = preg_replace('/\s+/u', ' ', trim($child->nodeValue)); |
|
277
|
8 |
|
if (is_string($cellContent)) { |
|
278
|
|
|
// simply append the text if the cell content is a plain text string |
|
279
|
8 |
|
$cellContent .= $domText; |
|
280
|
|
|
} |
|
281
|
|
|
// but if we have a rich text run instead, we need to append it correctly |
|
282
|
|
|
// TODO |
|
283
|
8 |
|
} elseif ($child instanceof DOMElement) { |
|
284
|
8 |
|
$attributeArray = []; |
|
285
|
8 |
|
foreach ($child->attributes as $attribute) { |
|
286
|
8 |
|
$attributeArray[$attribute->name] = $attribute->value; |
|
287
|
|
|
} |
|
288
|
|
|
|
|
289
|
8 |
|
switch ($child->nodeName) { |
|
290
|
8 |
|
case 'meta': |
|
291
|
8 |
|
foreach ($attributeArray as $attributeName => $attributeValue) { |
|
292
|
|
|
switch ($attributeName) { |
|
293
|
8 |
|
case 'content': |
|
294
|
|
|
// TODO |
|
295
|
|
|
// Extract character set, so we can convert to UTF-8 if required |
|
296
|
8 |
|
break; |
|
297
|
|
|
} |
|
298
|
|
|
} |
|
299
|
8 |
|
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
|
300
|
|
|
|
|
301
|
8 |
|
break; |
|
302
|
8 |
|
case 'title': |
|
303
|
8 |
|
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
|
304
|
8 |
|
$sheet->setTitle($cellContent, true, false); |
|
305
|
8 |
|
$cellContent = ''; |
|
306
|
|
|
|
|
307
|
8 |
|
break; |
|
308
|
8 |
|
case 'span': |
|
309
|
8 |
|
case 'div': |
|
310
|
8 |
|
case 'font': |
|
311
|
8 |
|
case 'i': |
|
312
|
8 |
|
case 'em': |
|
313
|
8 |
|
case 'strong': |
|
314
|
8 |
|
case 'b': |
|
315
|
5 |
|
if (isset($attributeArray['class']) && $attributeArray['class'] === 'comment') { |
|
316
|
5 |
|
$sheet->getComment($column . $row) |
|
317
|
5 |
|
->getText() |
|
318
|
5 |
|
->createTextRun($child->textContent); |
|
319
|
|
|
|
|
320
|
5 |
|
break; |
|
321
|
|
|
} |
|
322
|
|
|
|
|
323
|
|
|
if ($cellContent > '') { |
|
324
|
|
|
$cellContent .= ' '; |
|
325
|
|
|
} |
|
326
|
|
|
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
|
327
|
|
|
if ($cellContent > '') { |
|
328
|
|
|
$cellContent .= ' '; |
|
329
|
|
|
} |
|
330
|
|
|
|
|
331
|
|
|
break; |
|
332
|
8 |
|
case 'hr': |
|
333
|
|
|
$this->flushCell($sheet, $column, $row, $cellContent); |
|
334
|
|
|
++$row; |
|
335
|
|
|
if (isset($this->formats[$child->nodeName])) { |
|
336
|
|
|
$sheet->getStyle($column . $row)->applyFromArray($this->formats[$child->nodeName]); |
|
337
|
|
|
} else { |
|
338
|
|
|
$cellContent = '----------'; |
|
339
|
|
|
$this->flushCell($sheet, $column, $row, $cellContent); |
|
340
|
|
|
} |
|
341
|
|
|
++$row; |
|
342
|
|
|
// Add a break after a horizontal rule, simply by allowing the code to dropthru |
|
343
|
|
|
// no break |
|
344
|
8 |
|
case 'br': |
|
345
|
|
|
if ($this->tableLevel > 0) { |
|
346
|
|
|
// If we're inside a table, replace with a \n |
|
347
|
|
|
$cellContent .= "\n"; |
|
348
|
|
|
} else { |
|
349
|
|
|
// Otherwise flush our existing content and move the row cursor on |
|
350
|
|
|
$this->flushCell($sheet, $column, $row, $cellContent); |
|
351
|
|
|
++$row; |
|
352
|
|
|
} |
|
353
|
|
|
|
|
354
|
|
|
break; |
|
355
|
8 |
|
case 'a': |
|
356
|
5 |
|
foreach ($attributeArray as $attributeName => $attributeValue) { |
|
357
|
|
|
switch ($attributeName) { |
|
358
|
5 |
|
case 'href': |
|
359
|
|
|
$sheet->getCell($column . $row)->getHyperlink()->setUrl($attributeValue); |
|
360
|
|
|
if (isset($this->formats[$child->nodeName])) { |
|
361
|
|
|
$sheet->getStyle($column . $row)->applyFromArray($this->formats[$child->nodeName]); |
|
362
|
|
|
} |
|
363
|
|
|
|
|
364
|
|
|
break; |
|
365
|
5 |
|
case 'class': |
|
366
|
5 |
|
if ($attributeValue === 'comment-indicator') { |
|
367
|
5 |
|
break; // Ignore - it's just a red square. |
|
368
|
|
|
} |
|
369
|
|
|
} |
|
370
|
|
|
} |
|
371
|
5 |
|
$cellContent .= ' '; |
|
372
|
5 |
|
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
|
373
|
|
|
|
|
374
|
5 |
|
break; |
|
375
|
8 |
|
case 'h1': |
|
376
|
8 |
|
case 'h2': |
|
377
|
8 |
|
case 'h3': |
|
378
|
8 |
|
case 'h4': |
|
379
|
8 |
|
case 'h5': |
|
380
|
8 |
|
case 'h6': |
|
381
|
8 |
|
case 'ol': |
|
382
|
8 |
|
case 'ul': |
|
383
|
8 |
|
case 'p': |
|
384
|
|
|
if ($this->tableLevel > 0) { |
|
385
|
|
|
// If we're inside a table, replace with a \n |
|
386
|
|
|
$cellContent .= "\n"; |
|
387
|
|
|
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
|
388
|
|
|
} else { |
|
389
|
|
|
if ($cellContent > '') { |
|
390
|
|
|
$this->flushCell($sheet, $column, $row, $cellContent); |
|
391
|
|
|
++$row; |
|
392
|
|
|
} |
|
393
|
|
|
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
|
394
|
|
|
$this->flushCell($sheet, $column, $row, $cellContent); |
|
395
|
|
|
|
|
396
|
|
|
if (isset($this->formats[$child->nodeName])) { |
|
397
|
|
|
$sheet->getStyle($column . $row)->applyFromArray($this->formats[$child->nodeName]); |
|
398
|
|
|
} |
|
399
|
|
|
|
|
400
|
|
|
++$row; |
|
401
|
|
|
$column = 'A'; |
|
402
|
|
|
} |
|
403
|
|
|
|
|
404
|
|
|
break; |
|
405
|
8 |
|
case 'li': |
|
406
|
|
|
if ($this->tableLevel > 0) { |
|
407
|
|
|
// If we're inside a table, replace with a \n |
|
408
|
|
|
$cellContent .= "\n"; |
|
409
|
|
|
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
|
410
|
|
|
} else { |
|
411
|
|
|
if ($cellContent > '') { |
|
412
|
|
|
$this->flushCell($sheet, $column, $row, $cellContent); |
|
413
|
|
|
} |
|
414
|
|
|
++$row; |
|
415
|
|
|
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
|
416
|
|
|
$this->flushCell($sheet, $column, $row, $cellContent); |
|
417
|
|
|
$column = 'A'; |
|
418
|
|
|
} |
|
419
|
|
|
|
|
420
|
|
|
break; |
|
421
|
8 |
|
case 'table': |
|
422
|
8 |
|
$this->flushCell($sheet, $column, $row, $cellContent); |
|
423
|
8 |
|
$column = $this->setTableStartColumn($column); |
|
424
|
8 |
|
if ($this->tableLevel > 1) { |
|
425
|
|
|
--$row; |
|
426
|
|
|
} |
|
427
|
8 |
|
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
|
428
|
8 |
|
$column = $this->releaseTableStartColumn(); |
|
429
|
8 |
|
if ($this->tableLevel > 1) { |
|
430
|
|
|
++$column; |
|
431
|
|
|
} else { |
|
432
|
8 |
|
++$row; |
|
433
|
|
|
} |
|
434
|
|
|
|
|
435
|
8 |
|
break; |
|
436
|
8 |
|
case 'thead': |
|
437
|
8 |
|
case 'tbody': |
|
438
|
7 |
|
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
|
439
|
|
|
|
|
440
|
7 |
|
break; |
|
441
|
8 |
|
case 'tr': |
|
442
|
8 |
|
$column = $this->getTableStartColumn(); |
|
443
|
8 |
|
$cellContent = ''; |
|
444
|
8 |
|
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
|
445
|
8 |
|
++$row; |
|
446
|
|
|
|
|
447
|
8 |
|
break; |
|
448
|
8 |
|
case 'th': |
|
449
|
8 |
|
case 'td': |
|
450
|
8 |
|
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
|
451
|
|
|
|
|
452
|
|
|
// apply inline style |
|
453
|
8 |
|
$this->applyInlineStyle($sheet, $row, $column, $attributeArray); |
|
454
|
|
|
|
|
455
|
8 |
|
while (isset($this->rowspan[$column . $row])) { |
|
456
|
|
|
++$column; |
|
457
|
|
|
} |
|
458
|
|
|
|
|
459
|
8 |
|
$this->flushCell($sheet, $column, $row, $cellContent); |
|
460
|
|
|
|
|
461
|
8 |
|
if (isset($attributeArray['rowspan'], $attributeArray['colspan'])) { |
|
462
|
|
|
//create merging rowspan and colspan |
|
463
|
|
|
$columnTo = $column; |
|
464
|
|
|
for ($i = 0; $i < $attributeArray['colspan'] - 1; ++$i) { |
|
465
|
|
|
++$columnTo; |
|
466
|
|
|
} |
|
467
|
|
|
$range = $column . $row . ':' . $columnTo . ($row + $attributeArray['rowspan'] - 1); |
|
468
|
|
|
foreach (Coordinate::extractAllCellReferencesInRange($range) as $value) { |
|
469
|
|
|
$this->rowspan[$value] = true; |
|
470
|
|
|
} |
|
471
|
|
|
$sheet->mergeCells($range); |
|
472
|
|
|
$column = $columnTo; |
|
473
|
8 |
|
} elseif (isset($attributeArray['rowspan'])) { |
|
474
|
|
|
//create merging rowspan |
|
475
|
|
|
$range = $column . $row . ':' . $column . ($row + $attributeArray['rowspan'] - 1); |
|
476
|
|
|
foreach (Coordinate::extractAllCellReferencesInRange($range) as $value) { |
|
477
|
|
|
$this->rowspan[$value] = true; |
|
478
|
|
|
} |
|
479
|
|
|
$sheet->mergeCells($range); |
|
480
|
8 |
|
} elseif (isset($attributeArray['colspan'])) { |
|
481
|
|
|
//create merging colspan |
|
482
|
1 |
|
$columnTo = $column; |
|
483
|
1 |
|
for ($i = 0; $i < $attributeArray['colspan'] - 1; ++$i) { |
|
484
|
1 |
|
++$columnTo; |
|
485
|
|
|
} |
|
486
|
1 |
|
$sheet->mergeCells($column . $row . ':' . $columnTo . $row); |
|
487
|
1 |
|
$column = $columnTo; |
|
488
|
8 |
|
} elseif (isset($attributeArray['bgcolor'])) { |
|
489
|
|
|
$sheet->getStyle($column . $row)->applyFromArray( |
|
490
|
|
|
[ |
|
491
|
|
|
'fill' => [ |
|
492
|
|
|
'fillType' => Fill::FILL_SOLID, |
|
493
|
|
|
'color' => ['rgb' => $attributeArray['bgcolor']], |
|
494
|
|
|
], |
|
495
|
|
|
] |
|
496
|
|
|
); |
|
497
|
|
|
} |
|
498
|
8 |
|
++$column; |
|
499
|
|
|
|
|
500
|
8 |
|
break; |
|
501
|
8 |
|
case 'body': |
|
502
|
8 |
|
$row = 1; |
|
503
|
8 |
|
$column = 'A'; |
|
504
|
8 |
|
$cellContent = ''; |
|
505
|
8 |
|
$this->tableLevel = 0; |
|
506
|
8 |
|
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
|
507
|
|
|
|
|
508
|
8 |
|
break; |
|
509
|
|
|
default: |
|
510
|
8 |
|
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
|
511
|
|
|
} |
|
512
|
|
|
} |
|
513
|
|
|
} |
|
514
|
8 |
|
} |
|
515
|
|
|
|
|
516
|
|
|
/** |
|
517
|
|
|
* Loads PhpSpreadsheet from file into PhpSpreadsheet instance. |
|
518
|
|
|
* |
|
519
|
|
|
* @param string $pFilename |
|
520
|
|
|
* @param Spreadsheet $spreadsheet |
|
521
|
|
|
* |
|
522
|
|
|
* @throws Exception |
|
523
|
|
|
* |
|
524
|
|
|
* @return Spreadsheet |
|
525
|
|
|
*/ |
|
526
|
8 |
|
public function loadIntoExisting($pFilename, Spreadsheet $spreadsheet) |
|
527
|
|
|
{ |
|
528
|
|
|
// Validate |
|
529
|
8 |
|
if (!$this->canRead($pFilename)) { |
|
530
|
|
|
throw new Exception($pFilename . ' is an Invalid HTML file.'); |
|
531
|
|
|
} |
|
532
|
|
|
|
|
533
|
|
|
// Create new sheet |
|
534
|
8 |
|
while ($spreadsheet->getSheetCount() <= $this->sheetIndex) { |
|
535
|
|
|
$spreadsheet->createSheet(); |
|
536
|
|
|
} |
|
537
|
8 |
|
$spreadsheet->setActiveSheetIndex($this->sheetIndex); |
|
538
|
|
|
|
|
539
|
|
|
// Create a new DOM object |
|
540
|
8 |
|
$dom = new DOMDocument(); |
|
|
|
|
|
|
541
|
|
|
// Reload the HTML file into the DOM object |
|
542
|
8 |
|
$loaded = $dom->loadHTML(mb_convert_encoding($this->securityScanFile($pFilename), 'HTML-ENTITIES', 'UTF-8')); |
|
543
|
8 |
|
if ($loaded === false) { |
|
544
|
|
|
throw new Exception('Failed to load ' . $pFilename . ' as a DOM Document'); |
|
545
|
|
|
} |
|
546
|
|
|
|
|
547
|
|
|
// Discard white space |
|
548
|
8 |
|
$dom->preserveWhiteSpace = false; |
|
549
|
|
|
|
|
550
|
8 |
|
$row = 0; |
|
551
|
8 |
|
$column = 'A'; |
|
552
|
8 |
|
$content = ''; |
|
553
|
8 |
|
$this->processDomElement($dom, $spreadsheet->getActiveSheet(), $row, $column, $content); |
|
554
|
|
|
|
|
555
|
|
|
// Return |
|
556
|
8 |
|
return $spreadsheet; |
|
557
|
|
|
} |
|
558
|
|
|
|
|
559
|
|
|
/** |
|
560
|
|
|
* Get sheet index. |
|
561
|
|
|
* |
|
562
|
|
|
* @return int |
|
563
|
|
|
*/ |
|
564
|
|
|
public function getSheetIndex() |
|
565
|
|
|
{ |
|
566
|
|
|
return $this->sheetIndex; |
|
567
|
|
|
} |
|
568
|
|
|
|
|
569
|
|
|
/** |
|
570
|
|
|
* Set sheet index. |
|
571
|
|
|
* |
|
572
|
|
|
* @param int $pValue Sheet index |
|
573
|
|
|
* |
|
574
|
|
|
* @return HTML |
|
575
|
|
|
*/ |
|
576
|
|
|
public function setSheetIndex($pValue) |
|
577
|
|
|
{ |
|
578
|
|
|
$this->sheetIndex = $pValue; |
|
579
|
|
|
|
|
580
|
|
|
return $this; |
|
581
|
|
|
} |
|
582
|
|
|
|
|
583
|
|
|
/** |
|
584
|
|
|
* Scan theXML for use of <!ENTITY to prevent XXE/XEE attacks. |
|
585
|
|
|
* |
|
586
|
|
|
* @param string $xml |
|
587
|
|
|
* |
|
588
|
|
|
* @throws Exception |
|
589
|
|
|
*/ |
|
590
|
8 |
|
public function securityScan($xml) |
|
591
|
|
|
{ |
|
592
|
8 |
|
$pattern = '/\\0?' . implode('\\0?', str_split('<!ENTITY')) . '\\0?/'; |
|
593
|
8 |
|
if (preg_match($pattern, $xml)) { |
|
594
|
|
|
throw new Exception('Detected use of ENTITY in XML, spreadsheet file load() aborted to prevent XXE/XEE attacks'); |
|
595
|
|
|
} |
|
596
|
|
|
|
|
597
|
8 |
|
return $xml; |
|
598
|
|
|
} |
|
599
|
|
|
|
|
600
|
|
|
/** |
|
601
|
|
|
* Apply inline css inline style. |
|
602
|
|
|
* |
|
603
|
|
|
* NOTES : |
|
604
|
|
|
* Currently only intended for td & th element, |
|
605
|
|
|
* and only takes 'background-color' and 'color'; property with HEX color |
|
606
|
|
|
* |
|
607
|
|
|
* TODO : |
|
608
|
|
|
* - Implement to other propertie, such as border |
|
609
|
|
|
* |
|
610
|
|
|
* @param Worksheet $sheet |
|
611
|
|
|
* @param array $attributeArray |
|
612
|
|
|
* @param int $row |
|
613
|
|
|
* @param string $column |
|
614
|
|
|
*/ |
|
615
|
8 |
|
private function applyInlineStyle(&$sheet, $row, $column, $attributeArray) |
|
616
|
|
|
{ |
|
617
|
8 |
|
if (!isset($attributeArray['style'])) { |
|
618
|
8 |
|
return; |
|
619
|
|
|
} |
|
620
|
|
|
|
|
621
|
1 |
|
$supported_styles = ['background-color', 'color']; |
|
622
|
|
|
|
|
623
|
|
|
// add color styles (background & text) from dom element,currently support : td & th, using ONLY inline css style with RGB color |
|
624
|
1 |
|
$styles = explode(';', $attributeArray['style']); |
|
625
|
1 |
|
foreach ($styles as $st) { |
|
626
|
1 |
|
$value = explode(':', $st); |
|
627
|
|
|
|
|
628
|
1 |
|
if (empty(trim($value[0])) || !in_array(trim($value[0]), $supported_styles)) { |
|
629
|
|
|
continue; |
|
630
|
|
|
} |
|
631
|
|
|
|
|
632
|
|
|
//check if has #, so we can get clean hex |
|
633
|
1 |
|
if (substr(trim($value[1]), 0, 1) == '#') { |
|
634
|
1 |
|
$style_color = substr(trim($value[1]), 1); |
|
635
|
|
|
} |
|
636
|
|
|
|
|
637
|
1 |
|
if (empty($style_color)) { |
|
638
|
|
|
continue; |
|
639
|
|
|
} |
|
640
|
|
|
|
|
641
|
1 |
|
switch (trim($value[0])) { |
|
642
|
1 |
|
case 'background-color': |
|
643
|
1 |
|
$sheet->getStyle($column . $row)->applyFromArray(['fill' => ['fillType' => Fill::FILL_SOLID, 'color' => ['rgb' => "{$style_color}"]]]); |
|
|
|
|
|
|
644
|
|
|
|
|
645
|
1 |
|
break; |
|
646
|
1 |
|
case 'color': |
|
647
|
1 |
|
$sheet->getStyle($column . $row)->applyFromArray(['font' => ['color' => ['rgb' => "$style_color}"]]]); |
|
648
|
|
|
|
|
649
|
1 |
|
break; |
|
650
|
|
|
} |
|
651
|
|
|
} |
|
652
|
1 |
|
} |
|
653
|
|
|
} |
|
654
|
|
|
|
PHPOffice /
PhpSpreadsheet
Loading history...