|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace PhpOffice\PhpSpreadsheet\Reader; |
|
4
|
|
|
|
|
5
|
|
|
use DOMAttr; |
|
6
|
|
|
use DOMDocument; |
|
7
|
|
|
use DOMElement; |
|
8
|
|
|
use DOMNode; |
|
9
|
|
|
use DOMText; |
|
10
|
|
|
use PhpOffice\PhpSpreadsheet\Calculation\Calculation; |
|
11
|
|
|
use PhpOffice\PhpSpreadsheet\Cell\Coordinate; |
|
12
|
|
|
use PhpOffice\PhpSpreadsheet\Cell\DataType; |
|
13
|
|
|
use PhpOffice\PhpSpreadsheet\Comment; |
|
14
|
|
|
use PhpOffice\PhpSpreadsheet\Document\Properties; |
|
15
|
|
|
use PhpOffice\PhpSpreadsheet\Exception as SpreadsheetException; |
|
16
|
|
|
use PhpOffice\PhpSpreadsheet\Helper\Dimension as CssDimension; |
|
17
|
|
|
use PhpOffice\PhpSpreadsheet\Helper\Html as HelperHtml; |
|
18
|
|
|
use PhpOffice\PhpSpreadsheet\Reader\Security\XmlScanner; |
|
19
|
|
|
use PhpOffice\PhpSpreadsheet\Spreadsheet; |
|
20
|
|
|
use PhpOffice\PhpSpreadsheet\Style\Border; |
|
21
|
|
|
use PhpOffice\PhpSpreadsheet\Style\Color; |
|
22
|
|
|
use PhpOffice\PhpSpreadsheet\Style\Fill; |
|
23
|
|
|
use PhpOffice\PhpSpreadsheet\Style\Font; |
|
24
|
|
|
use PhpOffice\PhpSpreadsheet\Style\Style; |
|
25
|
|
|
use PhpOffice\PhpSpreadsheet\Worksheet\Drawing; |
|
26
|
|
|
use PhpOffice\PhpSpreadsheet\Worksheet\Worksheet; |
|
27
|
|
|
use Throwable; |
|
28
|
|
|
|
|
29
|
|
|
class Html extends BaseReader |
|
30
|
|
|
{ |
|
31
|
|
|
/** |
|
32
|
|
|
* Sample size to read to determine if it's HTML or not. |
|
33
|
|
|
*/ |
|
34
|
|
|
const TEST_SAMPLE_SIZE = 2048; |
|
35
|
|
|
|
|
36
|
|
|
private const STARTS_WITH_BOM = '/^(?:\xfe\xff|\xff\xfe|\xEF\xBB\xBF)/'; |
|
37
|
|
|
|
|
38
|
|
|
private const DECLARES_CHARSET = '/\\bcharset=/i'; |
|
39
|
|
|
|
|
40
|
|
|
/** |
|
41
|
|
|
* Input encoding. |
|
42
|
|
|
*/ |
|
43
|
|
|
protected string $inputEncoding = 'ANSI'; |
|
44
|
|
|
|
|
45
|
|
|
/** |
|
46
|
|
|
* Sheet index to read. |
|
47
|
|
|
*/ |
|
48
|
|
|
protected int $sheetIndex = 0; |
|
49
|
|
|
|
|
50
|
|
|
/** |
|
51
|
|
|
* Formats. |
|
52
|
|
|
*/ |
|
53
|
|
|
protected array $formats = [ |
|
54
|
|
|
'h1' => [ |
|
55
|
|
|
'font' => [ |
|
56
|
|
|
'bold' => true, |
|
57
|
|
|
'size' => 24, |
|
58
|
|
|
], |
|
59
|
|
|
], // Bold, 24pt |
|
60
|
|
|
'h2' => [ |
|
61
|
|
|
'font' => [ |
|
62
|
|
|
'bold' => true, |
|
63
|
|
|
'size' => 18, |
|
64
|
|
|
], |
|
65
|
|
|
], // Bold, 18pt |
|
66
|
|
|
'h3' => [ |
|
67
|
|
|
'font' => [ |
|
68
|
|
|
'bold' => true, |
|
69
|
|
|
'size' => 13.5, |
|
70
|
|
|
], |
|
71
|
|
|
], // Bold, 13.5pt |
|
72
|
|
|
'h4' => [ |
|
73
|
|
|
'font' => [ |
|
74
|
|
|
'bold' => true, |
|
75
|
|
|
'size' => 12, |
|
76
|
|
|
], |
|
77
|
|
|
], // Bold, 12pt |
|
78
|
|
|
'h5' => [ |
|
79
|
|
|
'font' => [ |
|
80
|
|
|
'bold' => true, |
|
81
|
|
|
'size' => 10, |
|
82
|
|
|
], |
|
83
|
|
|
], // Bold, 10pt |
|
84
|
|
|
'h6' => [ |
|
85
|
|
|
'font' => [ |
|
86
|
|
|
'bold' => true, |
|
87
|
|
|
'size' => 7.5, |
|
88
|
|
|
], |
|
89
|
|
|
], // Bold, 7.5pt |
|
90
|
|
|
'a' => [ |
|
91
|
|
|
'font' => [ |
|
92
|
|
|
'underline' => true, |
|
93
|
|
|
'color' => [ |
|
94
|
|
|
'argb' => Color::COLOR_BLUE, |
|
95
|
|
|
], |
|
96
|
|
|
], |
|
97
|
|
|
], // Blue underlined |
|
98
|
|
|
'hr' => [ |
|
99
|
|
|
'borders' => [ |
|
100
|
|
|
'bottom' => [ |
|
101
|
|
|
'borderStyle' => Border::BORDER_THIN, |
|
102
|
|
|
'color' => [ |
|
103
|
|
|
Color::COLOR_BLACK, |
|
104
|
|
|
], |
|
105
|
|
|
], |
|
106
|
|
|
], |
|
107
|
|
|
], // Bottom border |
|
108
|
|
|
'strong' => [ |
|
109
|
|
|
'font' => [ |
|
110
|
|
|
'bold' => true, |
|
111
|
|
|
], |
|
112
|
|
|
], // Bold |
|
113
|
|
|
'b' => [ |
|
114
|
|
|
'font' => [ |
|
115
|
|
|
'bold' => true, |
|
116
|
|
|
], |
|
117
|
|
|
], // Bold |
|
118
|
|
|
'i' => [ |
|
119
|
|
|
'font' => [ |
|
120
|
|
|
'italic' => true, |
|
121
|
|
|
], |
|
122
|
|
|
], // Italic |
|
123
|
|
|
'em' => [ |
|
124
|
|
|
'font' => [ |
|
125
|
|
|
'italic' => true, |
|
126
|
|
|
], |
|
127
|
|
|
], // Italic |
|
128
|
|
|
]; |
|
129
|
|
|
|
|
130
|
|
|
protected array $rowspan = []; |
|
131
|
|
|
|
|
132
|
|
|
/** |
|
133
|
|
|
* Create a new HTML Reader instance. |
|
134
|
|
|
*/ |
|
135
|
516 |
|
public function __construct() |
|
136
|
|
|
{ |
|
137
|
516 |
|
parent::__construct(); |
|
138
|
516 |
|
$this->securityScanner = XmlScanner::getInstance($this); |
|
139
|
|
|
} |
|
140
|
|
|
|
|
141
|
|
|
/** |
|
142
|
|
|
* Validate that the current file is an HTML file. |
|
143
|
|
|
*/ |
|
144
|
495 |
|
public function canRead(string $filename): bool |
|
145
|
|
|
{ |
|
146
|
|
|
// Check if file exists |
|
147
|
|
|
try { |
|
148
|
495 |
|
$this->openFile($filename); |
|
149
|
1 |
|
} catch (Exception) { |
|
150
|
1 |
|
return false; |
|
151
|
|
|
} |
|
152
|
|
|
|
|
153
|
494 |
|
$beginning = preg_replace(self::STARTS_WITH_BOM, '', $this->readBeginning()) ?? ''; |
|
154
|
|
|
|
|
155
|
494 |
|
$startWithTag = self::startsWithTag($beginning); |
|
156
|
494 |
|
$containsTags = self::containsTags($beginning); |
|
157
|
494 |
|
$endsWithTag = self::endsWithTag($this->readEnding()); |
|
158
|
|
|
|
|
159
|
494 |
|
fclose($this->fileHandle); |
|
160
|
|
|
|
|
161
|
494 |
|
return $startWithTag && $containsTags && $endsWithTag; |
|
162
|
|
|
} |
|
163
|
|
|
|
|
164
|
494 |
|
private function readBeginning(): string |
|
165
|
|
|
{ |
|
166
|
494 |
|
fseek($this->fileHandle, 0); |
|
167
|
|
|
|
|
168
|
494 |
|
return (string) fread($this->fileHandle, self::TEST_SAMPLE_SIZE); |
|
169
|
|
|
} |
|
170
|
|
|
|
|
171
|
494 |
|
private function readEnding(): string |
|
172
|
|
|
{ |
|
173
|
494 |
|
$meta = stream_get_meta_data($this->fileHandle); |
|
174
|
|
|
// Phpstan incorrectly flags following line for Php8.2-, corrected in 8.3 |
|
175
|
494 |
|
$filename = $meta['uri']; //@phpstan-ignore-line |
|
176
|
|
|
|
|
177
|
494 |
|
clearstatcache(true, $filename); |
|
178
|
494 |
|
$size = (int) filesize($filename); |
|
179
|
494 |
|
if ($size === 0) { |
|
180
|
1 |
|
return ''; |
|
181
|
|
|
} |
|
182
|
|
|
|
|
183
|
493 |
|
$blockSize = self::TEST_SAMPLE_SIZE; |
|
184
|
493 |
|
if ($size < $blockSize) { |
|
185
|
42 |
|
$blockSize = $size; |
|
186
|
|
|
} |
|
187
|
|
|
|
|
188
|
493 |
|
fseek($this->fileHandle, $size - $blockSize); |
|
189
|
|
|
|
|
190
|
493 |
|
return (string) fread($this->fileHandle, $blockSize); |
|
191
|
|
|
} |
|
192
|
|
|
|
|
193
|
494 |
|
private static function startsWithTag(string $data): bool |
|
194
|
|
|
{ |
|
195
|
494 |
|
return str_starts_with(trim($data), '<'); |
|
196
|
|
|
} |
|
197
|
|
|
|
|
198
|
494 |
|
private static function endsWithTag(string $data): bool |
|
199
|
|
|
{ |
|
200
|
494 |
|
return str_ends_with(trim($data), '>'); |
|
201
|
|
|
} |
|
202
|
|
|
|
|
203
|
494 |
|
private static function containsTags(string $data): bool |
|
204
|
|
|
{ |
|
205
|
494 |
|
return strlen($data) !== strlen(strip_tags($data)); |
|
206
|
|
|
} |
|
207
|
|
|
|
|
208
|
|
|
/** |
|
209
|
|
|
* Loads Spreadsheet from file. |
|
210
|
|
|
*/ |
|
211
|
476 |
|
public function loadSpreadsheetFromFile(string $filename): Spreadsheet |
|
212
|
|
|
{ |
|
213
|
|
|
// Create new Spreadsheet |
|
214
|
476 |
|
$spreadsheet = new Spreadsheet(); |
|
215
|
476 |
|
$spreadsheet->setValueBinder($this->valueBinder); |
|
216
|
|
|
|
|
217
|
|
|
// Load into this instance |
|
218
|
476 |
|
return $this->loadIntoExisting($filename, $spreadsheet); |
|
219
|
|
|
} |
|
220
|
|
|
|
|
221
|
|
|
// Data Array used for testing only, should write to Spreadsheet object on completion of tests |
|
222
|
|
|
|
|
223
|
|
|
protected array $dataArray = []; |
|
224
|
|
|
|
|
225
|
|
|
protected int $tableLevel = 0; |
|
226
|
|
|
|
|
227
|
|
|
protected array $nestedColumn = ['A']; |
|
228
|
|
|
|
|
229
|
491 |
|
protected function setTableStartColumn(string $column): string |
|
230
|
|
|
{ |
|
231
|
491 |
|
if ($this->tableLevel == 0) { |
|
232
|
491 |
|
$column = 'A'; |
|
233
|
|
|
} |
|
234
|
491 |
|
++$this->tableLevel; |
|
235
|
491 |
|
$this->nestedColumn[$this->tableLevel] = $column; |
|
236
|
|
|
|
|
237
|
491 |
|
return $this->nestedColumn[$this->tableLevel]; |
|
238
|
|
|
} |
|
239
|
|
|
|
|
240
|
487 |
|
protected function getTableStartColumn(): string |
|
241
|
|
|
{ |
|
242
|
487 |
|
return $this->nestedColumn[$this->tableLevel]; |
|
243
|
|
|
} |
|
244
|
|
|
|
|
245
|
490 |
|
protected function releaseTableStartColumn(): string |
|
246
|
|
|
{ |
|
247
|
490 |
|
--$this->tableLevel; |
|
248
|
|
|
|
|
249
|
490 |
|
return array_pop($this->nestedColumn); |
|
250
|
|
|
} |
|
251
|
|
|
|
|
252
|
|
|
/** |
|
253
|
|
|
* Flush cell. |
|
254
|
|
|
*/ |
|
255
|
492 |
|
protected function flushCell(Worksheet $sheet, string $column, int|string $row, mixed &$cellContent, array $attributeArray): void |
|
256
|
|
|
{ |
|
257
|
492 |
|
if (is_string($cellContent)) { |
|
258
|
|
|
// Simple String content |
|
259
|
492 |
|
if (trim($cellContent) > '') { |
|
260
|
|
|
// Only actually write it if there's content in the string |
|
261
|
|
|
// Write to worksheet to be done here... |
|
262
|
|
|
// ... we return the cell, so we can mess about with styles more easily |
|
263
|
|
|
|
|
264
|
|
|
// Set cell value explicitly if there is data-type attribute |
|
265
|
480 |
|
if (isset($attributeArray['data-type'])) { |
|
266
|
5 |
|
$datatype = $attributeArray['data-type']; |
|
267
|
5 |
|
if (in_array($datatype, [DataType::TYPE_STRING, DataType::TYPE_STRING2, DataType::TYPE_INLINE])) { |
|
268
|
|
|
//Prevent to Excel treat string with beginning equal sign or convert big numbers to scientific number |
|
269
|
5 |
|
if (str_starts_with($cellContent, '=')) { |
|
270
|
1 |
|
$sheet->getCell($column . $row) |
|
271
|
1 |
|
->getStyle() |
|
272
|
1 |
|
->setQuotePrefix(true); |
|
273
|
|
|
} |
|
274
|
|
|
} |
|
275
|
5 |
|
if ($datatype === DataType::TYPE_BOOL) { |
|
276
|
5 |
|
$cellContent = self::convertBoolean($cellContent); |
|
277
|
5 |
|
if (!is_bool($cellContent)) { |
|
278
|
1 |
|
$attributeArray['data-type'] = DataType::TYPE_STRING; |
|
279
|
|
|
} |
|
280
|
|
|
} |
|
281
|
|
|
|
|
282
|
|
|
//catching the Exception and ignoring the invalid data types |
|
283
|
|
|
try { |
|
284
|
5 |
|
$sheet->setCellValueExplicit($column . $row, $cellContent, $attributeArray['data-type']); |
|
285
|
1 |
|
} catch (SpreadsheetException) { |
|
286
|
1 |
|
$sheet->setCellValue($column . $row, $cellContent); |
|
287
|
|
|
} |
|
288
|
|
|
} else { |
|
289
|
479 |
|
$sheet->setCellValue($column . $row, $cellContent); |
|
290
|
|
|
} |
|
291
|
480 |
|
$this->dataArray[$row][$column] = $cellContent; |
|
292
|
|
|
} |
|
293
|
|
|
} else { |
|
294
|
|
|
// We have a Rich Text run |
|
295
|
|
|
// TODO |
|
296
|
|
|
$this->dataArray[$row][$column] = 'RICH TEXT: ' . $cellContent; |
|
297
|
|
|
} |
|
298
|
492 |
|
$cellContent = (string) ''; |
|
299
|
|
|
} |
|
300
|
|
|
|
|
301
|
|
|
/** @var array<int, array<int, string>> */ |
|
302
|
|
|
private static array $falseTrueArray = []; |
|
303
|
|
|
|
|
304
|
5 |
|
private static function convertBoolean(?string $cellContent): bool|string |
|
305
|
|
|
{ |
|
306
|
5 |
|
if ($cellContent === '1') { |
|
307
|
1 |
|
return true; |
|
308
|
|
|
} |
|
309
|
4 |
|
if ($cellContent === '0' || $cellContent === '' || $cellContent === null) { |
|
310
|
|
|
return false; |
|
311
|
|
|
} |
|
312
|
4 |
|
if (empty(self::$falseTrueArray)) { |
|
313
|
1 |
|
$calc = Calculation::getInstance(); |
|
314
|
1 |
|
self::$falseTrueArray = $calc->getFalseTrueArray(); |
|
315
|
|
|
} |
|
316
|
4 |
|
if (in_array(mb_strtoupper($cellContent), self::$falseTrueArray[1], true)) { |
|
317
|
4 |
|
return true; |
|
318
|
|
|
} |
|
319
|
4 |
|
if (in_array(mb_strtoupper($cellContent), self::$falseTrueArray[0], true)) { |
|
320
|
4 |
|
return false; |
|
321
|
|
|
} |
|
322
|
|
|
|
|
323
|
1 |
|
return $cellContent; |
|
324
|
|
|
} |
|
325
|
|
|
|
|
326
|
492 |
|
private function processDomElementBody(Worksheet $sheet, int &$row, string &$column, string &$cellContent, DOMElement $child): void |
|
327
|
|
|
{ |
|
328
|
492 |
|
$attributeArray = []; |
|
329
|
|
|
/** @var DOMAttr $attribute */ |
|
330
|
492 |
|
foreach ($child->attributes as $attribute) { |
|
331
|
478 |
|
$attributeArray[$attribute->name] = $attribute->value; |
|
332
|
|
|
} |
|
333
|
|
|
|
|
334
|
492 |
|
if ($child->nodeName === 'body') { |
|
335
|
492 |
|
$row = 1; |
|
336
|
492 |
|
$column = 'A'; |
|
337
|
492 |
|
$cellContent = ''; |
|
338
|
492 |
|
$this->tableLevel = 0; |
|
339
|
492 |
|
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
|
340
|
|
|
} else { |
|
341
|
492 |
|
$this->processDomElementTitle($sheet, $row, $column, $cellContent, $child, $attributeArray); |
|
342
|
|
|
} |
|
343
|
|
|
} |
|
344
|
|
|
|
|
345
|
492 |
|
private function processDomElementTitle(Worksheet $sheet, int &$row, string &$column, string &$cellContent, DOMElement $child, array &$attributeArray): void |
|
346
|
|
|
{ |
|
347
|
492 |
|
if ($child->nodeName === 'title') { |
|
348
|
459 |
|
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
|
349
|
|
|
|
|
350
|
|
|
try { |
|
351
|
459 |
|
$sheet->setTitle($cellContent, true, true); |
|
352
|
456 |
|
$sheet->getParent()?->getProperties()?->setTitle($cellContent); |
|
353
|
3 |
|
} catch (SpreadsheetException) { |
|
354
|
|
|
// leave default title if too long or illegal chars |
|
355
|
|
|
} |
|
356
|
459 |
|
$cellContent = ''; |
|
357
|
|
|
} else { |
|
358
|
492 |
|
$this->processDomElementSpanEtc($sheet, $row, $column, $cellContent, $child, $attributeArray); |
|
359
|
|
|
} |
|
360
|
|
|
} |
|
361
|
|
|
|
|
362
|
|
|
private const SPAN_ETC = ['span', 'div', 'font', 'i', 'em', 'strong', 'b']; |
|
363
|
|
|
|
|
364
|
492 |
|
private function processDomElementSpanEtc(Worksheet $sheet, int &$row, string &$column, string &$cellContent, DOMElement $child, array &$attributeArray): void |
|
365
|
|
|
{ |
|
366
|
492 |
|
if (in_array((string) $child->nodeName, self::SPAN_ETC, true)) { |
|
367
|
444 |
|
if (isset($attributeArray['class']) && $attributeArray['class'] === 'comment') { |
|
368
|
9 |
|
$sheet->getComment($column . $row) |
|
369
|
9 |
|
->getText() |
|
370
|
9 |
|
->createTextRun($child->textContent); |
|
371
|
9 |
|
if (isset($attributeArray['dir']) && $attributeArray['dir'] === 'rtl') { |
|
372
|
1 |
|
$sheet->getComment($column . $row)->setTextboxDirection(Comment::TEXTBOX_DIRECTION_RTL); |
|
373
|
|
|
} |
|
374
|
9 |
|
if (isset($attributeArray['style'])) { |
|
375
|
2 |
|
$alignStyle = $attributeArray['style']; |
|
376
|
2 |
|
if (preg_match('/\\btext-align:\\s*(left|right|center|justify)\\b/', $alignStyle, $matches) === 1) { |
|
377
|
2 |
|
$sheet->getComment($column . $row)->setAlignment($matches[1]); |
|
378
|
|
|
} |
|
379
|
|
|
} |
|
380
|
|
|
} else { |
|
381
|
444 |
|
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
|
382
|
|
|
} |
|
383
|
|
|
|
|
384
|
444 |
|
if (isset($this->formats[$child->nodeName])) { |
|
385
|
2 |
|
$sheet->getStyle($column . $row)->applyFromArray($this->formats[$child->nodeName]); |
|
386
|
|
|
} |
|
387
|
|
|
} else { |
|
388
|
492 |
|
$this->processDomElementHr($sheet, $row, $column, $cellContent, $child, $attributeArray); |
|
389
|
|
|
} |
|
390
|
|
|
} |
|
391
|
|
|
|
|
392
|
492 |
|
private function processDomElementHr(Worksheet $sheet, int &$row, string &$column, string &$cellContent, DOMElement $child, array &$attributeArray): void |
|
393
|
|
|
{ |
|
394
|
492 |
|
if ($child->nodeName === 'hr') { |
|
395
|
1 |
|
$this->flushCell($sheet, $column, $row, $cellContent, $attributeArray); |
|
396
|
1 |
|
++$row; |
|
397
|
1 |
|
if (isset($this->formats[$child->nodeName])) { |
|
398
|
1 |
|
$sheet->getStyle($column . $row)->applyFromArray($this->formats[$child->nodeName]); |
|
399
|
|
|
} |
|
400
|
1 |
|
++$row; |
|
401
|
|
|
} |
|
402
|
|
|
// fall through to br |
|
403
|
492 |
|
$this->processDomElementBr($sheet, $row, $column, $cellContent, $child, $attributeArray); |
|
404
|
|
|
} |
|
405
|
|
|
|
|
406
|
492 |
|
private function processDomElementBr(Worksheet $sheet, int &$row, string &$column, string &$cellContent, DOMElement $child, array &$attributeArray): void |
|
407
|
|
|
{ |
|
408
|
492 |
|
if ($child->nodeName === 'br' || $child->nodeName === 'hr') { |
|
409
|
4 |
|
if ($this->tableLevel > 0) { |
|
410
|
|
|
// If we're inside a table, replace with a newline and set the cell to wrap |
|
411
|
4 |
|
$cellContent .= "\n"; |
|
412
|
4 |
|
$sheet->getStyle($column . $row)->getAlignment()->setWrapText(true); |
|
413
|
|
|
} else { |
|
414
|
|
|
// Otherwise flush our existing content and move the row cursor on |
|
415
|
1 |
|
$this->flushCell($sheet, $column, $row, $cellContent, $attributeArray); |
|
416
|
1 |
|
++$row; |
|
417
|
|
|
} |
|
418
|
|
|
} else { |
|
419
|
492 |
|
$this->processDomElementA($sheet, $row, $column, $cellContent, $child, $attributeArray); |
|
420
|
|
|
} |
|
421
|
|
|
} |
|
422
|
|
|
|
|
423
|
492 |
|
private function processDomElementA(Worksheet $sheet, int &$row, string &$column, string &$cellContent, DOMElement $child, array &$attributeArray): void |
|
424
|
|
|
{ |
|
425
|
492 |
|
if ($child->nodeName === 'a') { |
|
426
|
12 |
|
foreach ($attributeArray as $attributeName => $attributeValue) { |
|
427
|
|
|
switch ($attributeName) { |
|
428
|
12 |
|
case 'href': |
|
429
|
3 |
|
$sheet->getCell($column . $row)->getHyperlink()->setUrl($attributeValue); |
|
430
|
3 |
|
if (isset($this->formats[$child->nodeName])) { |
|
431
|
3 |
|
$sheet->getStyle($column . $row)->applyFromArray($this->formats[$child->nodeName]); |
|
432
|
|
|
} |
|
433
|
|
|
|
|
434
|
3 |
|
break; |
|
435
|
10 |
|
case 'class': |
|
436
|
9 |
|
if ($attributeValue === 'comment-indicator') { |
|
437
|
9 |
|
break; // Ignore - it's just a red square. |
|
438
|
|
|
} |
|
439
|
|
|
} |
|
440
|
|
|
} |
|
441
|
|
|
// no idea why this should be needed |
|
442
|
|
|
//$cellContent .= ' '; |
|
443
|
12 |
|
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
|
444
|
|
|
} else { |
|
445
|
492 |
|
$this->processDomElementH1Etc($sheet, $row, $column, $cellContent, $child, $attributeArray); |
|
446
|
|
|
} |
|
447
|
|
|
} |
|
448
|
|
|
|
|
449
|
|
|
private const H1_ETC = ['h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'ol', 'ul', 'p']; |
|
450
|
|
|
|
|
451
|
492 |
|
private function processDomElementH1Etc(Worksheet $sheet, int &$row, string &$column, string &$cellContent, DOMElement $child, array &$attributeArray): void |
|
452
|
|
|
{ |
|
453
|
492 |
|
if (in_array((string) $child->nodeName, self::H1_ETC, true)) { |
|
454
|
2 |
|
if ($this->tableLevel > 0) { |
|
455
|
|
|
// If we're inside a table, replace with a newline |
|
456
|
1 |
|
$cellContent .= $cellContent ? "\n" : ''; |
|
457
|
1 |
|
$sheet->getStyle($column . $row)->getAlignment()->setWrapText(true); |
|
458
|
1 |
|
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
|
459
|
|
|
} else { |
|
460
|
2 |
|
if ($cellContent > '') { |
|
461
|
1 |
|
$this->flushCell($sheet, $column, $row, $cellContent, $attributeArray); |
|
462
|
1 |
|
++$row; |
|
463
|
|
|
} |
|
464
|
2 |
|
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
|
465
|
2 |
|
$this->flushCell($sheet, $column, $row, $cellContent, $attributeArray); |
|
466
|
|
|
|
|
467
|
2 |
|
if (isset($this->formats[$child->nodeName])) { |
|
468
|
1 |
|
$sheet->getStyle($column . $row)->applyFromArray($this->formats[$child->nodeName]); |
|
469
|
|
|
} |
|
470
|
|
|
|
|
471
|
2 |
|
++$row; |
|
472
|
2 |
|
$column = 'A'; |
|
473
|
|
|
} |
|
474
|
|
|
} else { |
|
475
|
492 |
|
$this->processDomElementLi($sheet, $row, $column, $cellContent, $child, $attributeArray); |
|
476
|
|
|
} |
|
477
|
|
|
} |
|
478
|
|
|
|
|
479
|
492 |
|
private function processDomElementLi(Worksheet $sheet, int &$row, string &$column, string &$cellContent, DOMElement $child, array &$attributeArray): void |
|
480
|
|
|
{ |
|
481
|
492 |
|
if ($child->nodeName === 'li') { |
|
482
|
2 |
|
if ($this->tableLevel > 0) { |
|
483
|
|
|
// If we're inside a table, replace with a newline |
|
484
|
1 |
|
$cellContent .= $cellContent ? "\n" : ''; |
|
485
|
1 |
|
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
|
486
|
|
|
} else { |
|
487
|
2 |
|
if ($cellContent > '') { |
|
488
|
1 |
|
$this->flushCell($sheet, $column, $row, $cellContent, $attributeArray); |
|
489
|
|
|
} |
|
490
|
2 |
|
++$row; |
|
491
|
2 |
|
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
|
492
|
2 |
|
$this->flushCell($sheet, $column, $row, $cellContent, $attributeArray); |
|
493
|
2 |
|
$column = 'A'; |
|
494
|
|
|
} |
|
495
|
|
|
} else { |
|
496
|
492 |
|
$this->processDomElementImg($sheet, $row, $column, $cellContent, $child, $attributeArray); |
|
497
|
|
|
} |
|
498
|
|
|
} |
|
499
|
|
|
|
|
500
|
492 |
|
private function processDomElementImg(Worksheet $sheet, int &$row, string &$column, string &$cellContent, DOMElement $child, array &$attributeArray): void |
|
501
|
|
|
{ |
|
502
|
492 |
|
if ($child->nodeName === 'img') { |
|
503
|
9 |
|
$this->insertImage($sheet, $column, $row, $attributeArray); |
|
504
|
|
|
} else { |
|
505
|
492 |
|
$this->processDomElementTable($sheet, $row, $column, $cellContent, $child, $attributeArray); |
|
506
|
|
|
} |
|
507
|
|
|
} |
|
508
|
|
|
|
|
509
|
|
|
private string $currentColumn = 'A'; |
|
510
|
|
|
|
|
511
|
492 |
|
private function processDomElementTable(Worksheet $sheet, int &$row, string &$column, string &$cellContent, DOMElement $child, array &$attributeArray): void |
|
512
|
|
|
{ |
|
513
|
492 |
|
if ($child->nodeName === 'table') { |
|
514
|
491 |
|
if (isset($attributeArray['class'])) { |
|
515
|
444 |
|
$classes = explode(' ', $attributeArray['class']); |
|
516
|
444 |
|
$sheet->setShowGridlines(in_array('gridlines', $classes, true)); |
|
517
|
444 |
|
$sheet->setPrintGridlines(in_array('gridlinesp', $classes, true)); |
|
518
|
|
|
} |
|
519
|
491 |
|
$this->currentColumn = 'A'; |
|
520
|
491 |
|
$this->flushCell($sheet, $column, $row, $cellContent, $attributeArray); |
|
521
|
491 |
|
$column = $this->setTableStartColumn($column); |
|
522
|
491 |
|
if ($this->tableLevel > 1 && $row > 1) { |
|
523
|
2 |
|
--$row; |
|
524
|
|
|
} |
|
525
|
491 |
|
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
|
526
|
490 |
|
$column = $this->releaseTableStartColumn(); |
|
527
|
490 |
|
if ($this->tableLevel > 1) { |
|
528
|
2 |
|
++$column; |
|
529
|
|
|
} else { |
|
530
|
490 |
|
++$row; |
|
531
|
|
|
} |
|
532
|
|
|
} else { |
|
533
|
492 |
|
$this->processDomElementTr($sheet, $row, $column, $cellContent, $child, $attributeArray); |
|
534
|
|
|
} |
|
535
|
|
|
} |
|
536
|
|
|
|
|
537
|
492 |
|
private function processDomElementTr(Worksheet $sheet, int &$row, string &$column, string &$cellContent, DOMElement $child, array &$attributeArray): void |
|
538
|
|
|
{ |
|
539
|
492 |
|
if ($child->nodeName === 'col') { |
|
540
|
442 |
|
$this->applyInlineStyle($sheet, -1, $this->currentColumn, $attributeArray); |
|
541
|
442 |
|
++$this->currentColumn; |
|
542
|
492 |
|
} elseif ($child->nodeName === 'tr') { |
|
543
|
487 |
|
$column = $this->getTableStartColumn(); |
|
544
|
487 |
|
$cellContent = ''; |
|
545
|
487 |
|
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
|
546
|
|
|
|
|
547
|
486 |
|
if (isset($attributeArray['height'])) { |
|
548
|
1 |
|
$sheet->getRowDimension($row)->setRowHeight($attributeArray['height']); |
|
549
|
|
|
} |
|
550
|
|
|
|
|
551
|
486 |
|
++$row; |
|
552
|
|
|
} else { |
|
553
|
492 |
|
$this->processDomElementThTdOther($sheet, $row, $column, $cellContent, $child, $attributeArray); |
|
554
|
|
|
} |
|
555
|
|
|
} |
|
556
|
|
|
|
|
557
|
492 |
|
private function processDomElementThTdOther(Worksheet $sheet, int &$row, string &$column, string &$cellContent, DOMElement $child, array &$attributeArray): void |
|
558
|
|
|
{ |
|
559
|
492 |
|
if ($child->nodeName !== 'td' && $child->nodeName !== 'th') { |
|
560
|
492 |
|
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
|
561
|
|
|
} else { |
|
562
|
487 |
|
$this->processDomElementThTd($sheet, $row, $column, $cellContent, $child, $attributeArray); |
|
563
|
|
|
} |
|
564
|
|
|
} |
|
565
|
|
|
|
|
566
|
486 |
|
private function processDomElementBgcolor(Worksheet $sheet, int $row, string $column, array $attributeArray): void |
|
567
|
|
|
{ |
|
568
|
486 |
|
if (isset($attributeArray['bgcolor'])) { |
|
569
|
1 |
|
$sheet->getStyle("$column$row")->applyFromArray( |
|
570
|
1 |
|
[ |
|
571
|
1 |
|
'fill' => [ |
|
572
|
1 |
|
'fillType' => Fill::FILL_SOLID, |
|
573
|
1 |
|
'color' => ['rgb' => $this->getStyleColor($attributeArray['bgcolor'])], |
|
574
|
1 |
|
], |
|
575
|
1 |
|
] |
|
576
|
1 |
|
); |
|
577
|
|
|
} |
|
578
|
|
|
} |
|
579
|
|
|
|
|
580
|
486 |
|
private function processDomElementWidth(Worksheet $sheet, string $column, array $attributeArray): void |
|
581
|
|
|
{ |
|
582
|
486 |
|
if (isset($attributeArray['width'])) { |
|
583
|
1 |
|
$sheet->getColumnDimension($column)->setWidth((new CssDimension($attributeArray['width']))->width()); |
|
584
|
|
|
} |
|
585
|
|
|
} |
|
586
|
|
|
|
|
587
|
486 |
|
private function processDomElementHeight(Worksheet $sheet, int $row, array $attributeArray): void |
|
588
|
|
|
{ |
|
589
|
486 |
|
if (isset($attributeArray['height'])) { |
|
590
|
1 |
|
$sheet->getRowDimension($row)->setRowHeight((new CssDimension($attributeArray['height']))->height()); |
|
591
|
|
|
} |
|
592
|
|
|
} |
|
593
|
|
|
|
|
594
|
486 |
|
private function processDomElementAlign(Worksheet $sheet, int $row, string $column, array $attributeArray): void |
|
595
|
|
|
{ |
|
596
|
486 |
|
if (isset($attributeArray['align'])) { |
|
597
|
1 |
|
$sheet->getStyle($column . $row)->getAlignment()->setHorizontal($attributeArray['align']); |
|
598
|
|
|
} |
|
599
|
|
|
} |
|
600
|
|
|
|
|
601
|
486 |
|
private function processDomElementVAlign(Worksheet $sheet, int $row, string $column, array $attributeArray): void |
|
602
|
|
|
{ |
|
603
|
486 |
|
if (isset($attributeArray['valign'])) { |
|
604
|
1 |
|
$sheet->getStyle($column . $row)->getAlignment()->setVertical($attributeArray['valign']); |
|
605
|
|
|
} |
|
606
|
|
|
} |
|
607
|
|
|
|
|
608
|
486 |
|
private function processDomElementDataFormat(Worksheet $sheet, int $row, string $column, array $attributeArray): void |
|
609
|
|
|
{ |
|
610
|
486 |
|
if (isset($attributeArray['data-format'])) { |
|
611
|
1 |
|
$sheet->getStyle($column . $row)->getNumberFormat()->setFormatCode($attributeArray['data-format']); |
|
612
|
|
|
} |
|
613
|
|
|
} |
|
614
|
|
|
|
|
615
|
487 |
|
private function processDomElementThTd(Worksheet $sheet, int &$row, string &$column, string &$cellContent, DOMElement $child, array &$attributeArray): void |
|
616
|
|
|
{ |
|
617
|
487 |
|
while (isset($this->rowspan[$column . $row])) { |
|
618
|
3 |
|
++$column; |
|
619
|
|
|
} |
|
620
|
487 |
|
$this->processDomElement($child, $sheet, $row, $column, $cellContent); |
|
621
|
|
|
|
|
622
|
|
|
// apply inline style |
|
623
|
486 |
|
$this->applyInlineStyle($sheet, $row, $column, $attributeArray); |
|
624
|
|
|
|
|
625
|
486 |
|
$this->flushCell($sheet, $column, $row, $cellContent, $attributeArray); |
|
626
|
|
|
|
|
627
|
486 |
|
$this->processDomElementBgcolor($sheet, $row, $column, $attributeArray); |
|
628
|
486 |
|
$this->processDomElementWidth($sheet, $column, $attributeArray); |
|
629
|
486 |
|
$this->processDomElementHeight($sheet, $row, $attributeArray); |
|
630
|
486 |
|
$this->processDomElementAlign($sheet, $row, $column, $attributeArray); |
|
631
|
486 |
|
$this->processDomElementVAlign($sheet, $row, $column, $attributeArray); |
|
632
|
486 |
|
$this->processDomElementDataFormat($sheet, $row, $column, $attributeArray); |
|
633
|
|
|
|
|
634
|
486 |
|
if (isset($attributeArray['rowspan'], $attributeArray['colspan'])) { |
|
635
|
|
|
//create merging rowspan and colspan |
|
636
|
2 |
|
$columnTo = $column; |
|
637
|
2 |
|
for ($i = 0; $i < (int) $attributeArray['colspan'] - 1; ++$i) { |
|
638
|
2 |
|
++$columnTo; |
|
639
|
|
|
} |
|
640
|
2 |
|
$range = $column . $row . ':' . $columnTo . ($row + (int) $attributeArray['rowspan'] - 1); |
|
641
|
2 |
|
foreach (Coordinate::extractAllCellReferencesInRange($range) as $value) { |
|
642
|
2 |
|
$this->rowspan[$value] = true; |
|
643
|
|
|
} |
|
644
|
2 |
|
$sheet->mergeCells($range); |
|
645
|
2 |
|
$column = $columnTo; |
|
646
|
486 |
|
} elseif (isset($attributeArray['rowspan'])) { |
|
647
|
|
|
//create merging rowspan |
|
648
|
3 |
|
$range = $column . $row . ':' . $column . ($row + (int) $attributeArray['rowspan'] - 1); |
|
649
|
3 |
|
foreach (Coordinate::extractAllCellReferencesInRange($range) as $value) { |
|
650
|
3 |
|
$this->rowspan[$value] = true; |
|
651
|
|
|
} |
|
652
|
3 |
|
$sheet->mergeCells($range); |
|
653
|
486 |
|
} elseif (isset($attributeArray['colspan'])) { |
|
654
|
|
|
//create merging colspan |
|
655
|
3 |
|
$columnTo = $column; |
|
656
|
3 |
|
for ($i = 0; $i < (int) $attributeArray['colspan'] - 1; ++$i) { |
|
657
|
3 |
|
++$columnTo; |
|
658
|
|
|
} |
|
659
|
3 |
|
$sheet->mergeCells($column . $row . ':' . $columnTo . $row); |
|
660
|
3 |
|
$column = $columnTo; |
|
661
|
|
|
} |
|
662
|
|
|
|
|
663
|
486 |
|
++$column; |
|
664
|
|
|
} |
|
665
|
|
|
|
|
666
|
492 |
|
protected function processDomElement(DOMNode $element, Worksheet $sheet, int &$row, string &$column, string &$cellContent): void |
|
667
|
|
|
{ |
|
668
|
492 |
|
foreach ($element->childNodes as $child) { |
|
669
|
492 |
|
if ($child instanceof DOMText) { |
|
670
|
489 |
|
$domText = (string) preg_replace('/\s+/', ' ', trim($child->nodeValue ?? '')); |
|
671
|
489 |
|
if ($domText === "\u{a0}") { |
|
672
|
12 |
|
$domText = ''; |
|
673
|
|
|
} |
|
674
|
489 |
|
if (is_string($cellContent)) { |
|
675
|
|
|
// simply append the text if the cell content is a plain text string |
|
676
|
489 |
|
$cellContent .= $domText; |
|
677
|
|
|
} |
|
678
|
|
|
// but if we have a rich text run instead, we need to append it correctly |
|
679
|
|
|
// TODO |
|
680
|
492 |
|
} elseif ($child instanceof DOMElement) { |
|
681
|
492 |
|
$this->processDomElementBody($sheet, $row, $column, $cellContent, $child); |
|
682
|
|
|
} |
|
683
|
|
|
} |
|
684
|
|
|
} |
|
685
|
|
|
|
|
686
|
|
|
/** |
|
687
|
|
|
* Loads PhpSpreadsheet from file into PhpSpreadsheet instance. |
|
688
|
|
|
*/ |
|
689
|
476 |
|
public function loadIntoExisting(string $filename, Spreadsheet $spreadsheet): Spreadsheet |
|
690
|
|
|
{ |
|
691
|
|
|
// Validate |
|
692
|
476 |
|
if (!$this->canRead($filename)) { |
|
693
|
1 |
|
throw new Exception($filename . ' is an Invalid HTML file.'); |
|
694
|
|
|
} |
|
695
|
|
|
|
|
696
|
|
|
// Create a new DOM object |
|
697
|
475 |
|
$dom = new DOMDocument(); |
|
698
|
|
|
|
|
699
|
|
|
// Reload the HTML file into the DOM object |
|
700
|
|
|
try { |
|
701
|
475 |
|
$convert = $this->getSecurityScannerOrThrow()->scanFile($filename); |
|
702
|
474 |
|
$convert = self::replaceNonAsciiIfNeeded($convert); |
|
703
|
474 |
|
$loaded = ($convert === null) ? false : $dom->loadHTML($convert); |
|
704
|
2 |
|
} catch (Throwable $e) { |
|
705
|
2 |
|
$loaded = false; |
|
706
|
|
|
} |
|
707
|
475 |
|
if ($loaded === false) { |
|
708
|
2 |
|
throw new Exception('Failed to load ' . $filename . ' as a DOM Document', 0, $e ?? null); |
|
709
|
|
|
} |
|
710
|
473 |
|
self::loadProperties($dom, $spreadsheet); |
|
711
|
|
|
|
|
712
|
473 |
|
return $this->loadDocument($dom, $spreadsheet); |
|
713
|
|
|
} |
|
714
|
|
|
|
|
715
|
492 |
|
private static function loadProperties(DOMDocument $dom, Spreadsheet $spreadsheet): void |
|
716
|
|
|
{ |
|
717
|
492 |
|
$properties = $spreadsheet->getProperties(); |
|
718
|
492 |
|
foreach ($dom->getElementsByTagName('meta') as $meta) { |
|
719
|
452 |
|
$metaContent = (string) $meta->getAttribute('content'); |
|
720
|
452 |
|
if ($metaContent !== '') { |
|
721
|
444 |
|
$metaName = (string) $meta->getAttribute('name'); |
|
722
|
|
|
switch ($metaName) { |
|
723
|
444 |
|
case 'author': |
|
724
|
442 |
|
$properties->setCreator($metaContent); |
|
725
|
|
|
|
|
726
|
442 |
|
break; |
|
727
|
444 |
|
case 'category': |
|
728
|
1 |
|
$properties->setCategory($metaContent); |
|
729
|
|
|
|
|
730
|
1 |
|
break; |
|
731
|
444 |
|
case 'company': |
|
732
|
1 |
|
$properties->setCompany($metaContent); |
|
733
|
|
|
|
|
734
|
1 |
|
break; |
|
735
|
444 |
|
case 'created': |
|
736
|
442 |
|
$properties->setCreated($metaContent); |
|
737
|
|
|
|
|
738
|
442 |
|
break; |
|
739
|
444 |
|
case 'description': |
|
740
|
1 |
|
$properties->setDescription($metaContent); |
|
741
|
|
|
|
|
742
|
1 |
|
break; |
|
743
|
444 |
|
case 'keywords': |
|
744
|
1 |
|
$properties->setKeywords($metaContent); |
|
745
|
|
|
|
|
746
|
1 |
|
break; |
|
747
|
444 |
|
case 'lastModifiedBy': |
|
748
|
442 |
|
$properties->setLastModifiedBy($metaContent); |
|
749
|
|
|
|
|
750
|
442 |
|
break; |
|
751
|
444 |
|
case 'manager': |
|
752
|
1 |
|
$properties->setManager($metaContent); |
|
753
|
|
|
|
|
754
|
1 |
|
break; |
|
755
|
444 |
|
case 'modified': |
|
756
|
442 |
|
$properties->setModified($metaContent); |
|
757
|
|
|
|
|
758
|
442 |
|
break; |
|
759
|
444 |
|
case 'subject': |
|
760
|
1 |
|
$properties->setSubject($metaContent); |
|
761
|
|
|
|
|
762
|
1 |
|
break; |
|
763
|
444 |
|
case 'title': |
|
764
|
440 |
|
$properties->setTitle($metaContent); |
|
765
|
|
|
|
|
766
|
440 |
|
break; |
|
767
|
444 |
|
case 'viewport': |
|
768
|
1 |
|
$properties->setViewport($metaContent); |
|
769
|
|
|
|
|
770
|
1 |
|
break; |
|
771
|
|
|
default: |
|
772
|
444 |
|
if (preg_match('/^custom[.](bool|date|float|int|string)[.](.+)$/', $metaName, $matches) === 1) { |
|
773
|
1 |
|
match ($matches[1]) { |
|
774
|
1 |
|
'bool' => $properties->setCustomProperty($matches[2], (bool) $metaContent, Properties::PROPERTY_TYPE_BOOLEAN), |
|
775
|
1 |
|
'float' => $properties->setCustomProperty($matches[2], (float) $metaContent, Properties::PROPERTY_TYPE_FLOAT), |
|
776
|
1 |
|
'int' => $properties->setCustomProperty($matches[2], (int) $metaContent, Properties::PROPERTY_TYPE_INTEGER), |
|
777
|
1 |
|
'date' => $properties->setCustomProperty($matches[2], $metaContent, Properties::PROPERTY_TYPE_DATE), |
|
778
|
|
|
// string |
|
779
|
1 |
|
default => $properties->setCustomProperty($matches[2], $metaContent, Properties::PROPERTY_TYPE_STRING), |
|
780
|
1 |
|
}; |
|
781
|
|
|
} |
|
782
|
|
|
} |
|
783
|
|
|
} |
|
784
|
|
|
} |
|
785
|
492 |
|
if (!empty($dom->baseURI)) { |
|
786
|
1 |
|
$properties->setHyperlinkBase($dom->baseURI); |
|
787
|
|
|
} |
|
788
|
|
|
} |
|
789
|
|
|
|
|
790
|
11 |
|
private static function replaceNonAscii(array $matches): string |
|
791
|
|
|
{ |
|
792
|
11 |
|
return '&#' . mb_ord($matches[0], 'UTF-8') . ';'; |
|
793
|
|
|
} |
|
794
|
|
|
|
|
795
|
493 |
|
private static function replaceNonAsciiIfNeeded(string $convert): ?string |
|
796
|
|
|
{ |
|
797
|
493 |
|
if (preg_match(self::STARTS_WITH_BOM, $convert) !== 1 && preg_match(self::DECLARES_CHARSET, $convert) !== 1) { |
|
798
|
40 |
|
$lowend = "\u{80}"; |
|
799
|
40 |
|
$highend = "\u{10ffff}"; |
|
800
|
40 |
|
$regexp = "/[$lowend-$highend]/u"; |
|
801
|
|
|
/** @var callable $callback */ |
|
802
|
40 |
|
$callback = [self::class, 'replaceNonAscii']; |
|
803
|
40 |
|
$convert = preg_replace_callback($regexp, $callback, $convert); |
|
804
|
|
|
} |
|
805
|
|
|
|
|
806
|
493 |
|
return $convert; |
|
807
|
|
|
} |
|
808
|
|
|
|
|
809
|
|
|
/** |
|
810
|
|
|
* Spreadsheet from content. |
|
811
|
|
|
*/ |
|
812
|
19 |
|
public function loadFromString(string $content, ?Spreadsheet $spreadsheet = null): Spreadsheet |
|
813
|
|
|
{ |
|
814
|
|
|
// Create a new DOM object |
|
815
|
19 |
|
$dom = new DOMDocument(); |
|
816
|
|
|
|
|
817
|
|
|
// Reload the HTML file into the DOM object |
|
818
|
|
|
try { |
|
819
|
19 |
|
$convert = $this->getSecurityScannerOrThrow()->scan($content); |
|
820
|
19 |
|
$convert = self::replaceNonAsciiIfNeeded($convert); |
|
821
|
19 |
|
$loaded = ($convert === null) ? false : $dom->loadHTML($convert); |
|
822
|
|
|
} catch (Throwable $e) { |
|
823
|
|
|
$loaded = false; |
|
824
|
|
|
} |
|
825
|
19 |
|
if ($loaded === false) { |
|
826
|
|
|
throw new Exception('Failed to load content as a DOM Document', 0, $e ?? null); |
|
827
|
|
|
} |
|
828
|
19 |
|
$spreadsheet = $spreadsheet ?? new Spreadsheet(); |
|
829
|
19 |
|
$spreadsheet->setValueBinder($this->valueBinder); |
|
830
|
19 |
|
self::loadProperties($dom, $spreadsheet); |
|
831
|
|
|
|
|
832
|
19 |
|
return $this->loadDocument($dom, $spreadsheet); |
|
833
|
|
|
} |
|
834
|
|
|
|
|
835
|
|
|
/** |
|
836
|
|
|
* Loads PhpSpreadsheet from DOMDocument into PhpSpreadsheet instance. |
|
837
|
|
|
*/ |
|
838
|
492 |
|
private function loadDocument(DOMDocument $document, Spreadsheet $spreadsheet): Spreadsheet |
|
839
|
|
|
{ |
|
840
|
492 |
|
while ($spreadsheet->getSheetCount() <= $this->sheetIndex) { |
|
841
|
2 |
|
$spreadsheet->createSheet(); |
|
842
|
|
|
} |
|
843
|
492 |
|
$spreadsheet->setActiveSheetIndex($this->sheetIndex); |
|
844
|
|
|
|
|
845
|
|
|
// Discard white space |
|
846
|
492 |
|
$document->preserveWhiteSpace = false; |
|
847
|
|
|
|
|
848
|
492 |
|
$row = 0; |
|
849
|
492 |
|
$column = 'A'; |
|
850
|
492 |
|
$content = ''; |
|
851
|
492 |
|
$this->rowspan = []; |
|
852
|
492 |
|
$this->processDomElement($document, $spreadsheet->getActiveSheet(), $row, $column, $content); |
|
853
|
|
|
|
|
854
|
|
|
// Return |
|
855
|
491 |
|
return $spreadsheet; |
|
856
|
|
|
} |
|
857
|
|
|
|
|
858
|
|
|
/** |
|
859
|
|
|
* Get sheet index. |
|
860
|
|
|
*/ |
|
861
|
1 |
|
public function getSheetIndex(): int |
|
862
|
|
|
{ |
|
863
|
1 |
|
return $this->sheetIndex; |
|
864
|
|
|
} |
|
865
|
|
|
|
|
866
|
|
|
/** |
|
867
|
|
|
* Set sheet index. |
|
868
|
|
|
* |
|
869
|
|
|
* @param int $sheetIndex Sheet index |
|
870
|
|
|
* |
|
871
|
|
|
* @return $this |
|
872
|
|
|
*/ |
|
873
|
2 |
|
public function setSheetIndex(int $sheetIndex): static |
|
874
|
|
|
{ |
|
875
|
2 |
|
$this->sheetIndex = $sheetIndex; |
|
876
|
|
|
|
|
877
|
2 |
|
return $this; |
|
878
|
|
|
} |
|
879
|
|
|
|
|
880
|
|
|
/** |
|
881
|
|
|
* Apply inline css inline style. |
|
882
|
|
|
* |
|
883
|
|
|
* NOTES : |
|
884
|
|
|
* Currently only intended for td & th element, |
|
885
|
|
|
* and only takes 'background-color' and 'color'; property with HEX color |
|
886
|
|
|
* |
|
887
|
|
|
* TODO : |
|
888
|
|
|
* - Implement to other propertie, such as border |
|
889
|
|
|
*/ |
|
890
|
486 |
|
private function applyInlineStyle(Worksheet &$sheet, int $row, string $column, array $attributeArray): void |
|
891
|
|
|
{ |
|
892
|
486 |
|
if (!isset($attributeArray['style'])) { |
|
893
|
480 |
|
return; |
|
894
|
|
|
} |
|
895
|
|
|
|
|
896
|
16 |
|
if ($row <= 0 || $column === '') { |
|
897
|
1 |
|
$cellStyle = new Style(); |
|
898
|
16 |
|
} elseif (isset($attributeArray['rowspan'], $attributeArray['colspan'])) { |
|
899
|
1 |
|
$columnTo = $column; |
|
900
|
1 |
|
for ($i = 0; $i < (int) $attributeArray['colspan'] - 1; ++$i) { |
|
901
|
1 |
|
++$columnTo; |
|
902
|
|
|
} |
|
903
|
1 |
|
$range = $column . $row . ':' . $columnTo . ($row + (int) $attributeArray['rowspan'] - 1); |
|
904
|
1 |
|
$cellStyle = $sheet->getStyle($range); |
|
905
|
16 |
|
} elseif (isset($attributeArray['rowspan'])) { |
|
906
|
1 |
|
$range = $column . $row . ':' . $column . ($row + (int) $attributeArray['rowspan'] - 1); |
|
907
|
1 |
|
$cellStyle = $sheet->getStyle($range); |
|
908
|
16 |
|
} elseif (isset($attributeArray['colspan'])) { |
|
909
|
1 |
|
$columnTo = $column; |
|
910
|
1 |
|
for ($i = 0; $i < (int) $attributeArray['colspan'] - 1; ++$i) { |
|
911
|
1 |
|
++$columnTo; |
|
912
|
|
|
} |
|
913
|
1 |
|
$range = $column . $row . ':' . $columnTo . $row; |
|
914
|
1 |
|
$cellStyle = $sheet->getStyle($range); |
|
915
|
|
|
} else { |
|
916
|
16 |
|
$cellStyle = $sheet->getStyle($column . $row); |
|
917
|
|
|
} |
|
918
|
|
|
|
|
919
|
|
|
// add color styles (background & text) from dom element,currently support : td & th, using ONLY inline css style with RGB color |
|
920
|
16 |
|
$styles = explode(';', $attributeArray['style']); |
|
921
|
16 |
|
foreach ($styles as $st) { |
|
922
|
16 |
|
$value = explode(':', $st); |
|
923
|
16 |
|
$styleName = isset($value[0]) ? trim($value[0]) : null; |
|
924
|
16 |
|
$styleValue = isset($value[1]) ? trim($value[1]) : null; |
|
925
|
16 |
|
$styleValueString = (string) $styleValue; |
|
926
|
|
|
|
|
927
|
16 |
|
if (!$styleName) { |
|
928
|
12 |
|
continue; |
|
929
|
|
|
} |
|
930
|
|
|
|
|
931
|
|
|
switch ($styleName) { |
|
932
|
16 |
|
case 'background': |
|
933
|
16 |
|
case 'background-color': |
|
934
|
3 |
|
$styleColor = $this->getStyleColor($styleValueString); |
|
935
|
|
|
|
|
936
|
3 |
|
if (!$styleColor) { |
|
937
|
1 |
|
continue 2; |
|
938
|
|
|
} |
|
939
|
|
|
|
|
940
|
3 |
|
$cellStyle->applyFromArray(['fill' => ['fillType' => Fill::FILL_SOLID, 'color' => ['rgb' => $styleColor]]]); |
|
941
|
|
|
|
|
942
|
3 |
|
break; |
|
943
|
16 |
|
case 'color': |
|
944
|
3 |
|
$styleColor = $this->getStyleColor($styleValueString); |
|
945
|
|
|
|
|
946
|
3 |
|
if (!$styleColor) { |
|
947
|
1 |
|
continue 2; |
|
948
|
|
|
} |
|
949
|
|
|
|
|
950
|
3 |
|
$cellStyle->applyFromArray(['font' => ['color' => ['rgb' => $styleColor]]]); |
|
951
|
|
|
|
|
952
|
3 |
|
break; |
|
953
|
|
|
|
|
954
|
13 |
|
case 'border': |
|
955
|
3 |
|
$this->setBorderStyle($cellStyle, $styleValueString, 'allBorders'); |
|
956
|
|
|
|
|
957
|
3 |
|
break; |
|
958
|
|
|
|
|
959
|
11 |
|
case 'border-top': |
|
960
|
1 |
|
$this->setBorderStyle($cellStyle, $styleValueString, 'top'); |
|
961
|
|
|
|
|
962
|
1 |
|
break; |
|
963
|
|
|
|
|
964
|
11 |
|
case 'border-bottom': |
|
965
|
1 |
|
$this->setBorderStyle($cellStyle, $styleValueString, 'bottom'); |
|
966
|
|
|
|
|
967
|
1 |
|
break; |
|
968
|
|
|
|
|
969
|
11 |
|
case 'border-left': |
|
970
|
1 |
|
$this->setBorderStyle($cellStyle, $styleValueString, 'left'); |
|
971
|
|
|
|
|
972
|
1 |
|
break; |
|
973
|
|
|
|
|
974
|
11 |
|
case 'border-right': |
|
975
|
1 |
|
$this->setBorderStyle($cellStyle, $styleValueString, 'right'); |
|
976
|
|
|
|
|
977
|
1 |
|
break; |
|
978
|
|
|
|
|
979
|
10 |
|
case 'font-size': |
|
980
|
1 |
|
$cellStyle->getFont()->setSize( |
|
981
|
1 |
|
(float) $styleValue |
|
982
|
1 |
|
); |
|
983
|
|
|
|
|
984
|
1 |
|
break; |
|
985
|
|
|
|
|
986
|
10 |
|
case 'font-weight': |
|
987
|
1 |
|
if ($styleValue === 'bold' || $styleValue >= 500) { |
|
988
|
1 |
|
$cellStyle->getFont()->setBold(true); |
|
989
|
|
|
} |
|
990
|
|
|
|
|
991
|
1 |
|
break; |
|
992
|
|
|
|
|
993
|
10 |
|
case 'font-style': |
|
994
|
1 |
|
if ($styleValue === 'italic') { |
|
995
|
1 |
|
$cellStyle->getFont()->setItalic(true); |
|
996
|
|
|
} |
|
997
|
|
|
|
|
998
|
1 |
|
break; |
|
999
|
|
|
|
|
1000
|
10 |
|
case 'font-family': |
|
1001
|
1 |
|
$cellStyle->getFont()->setName(str_replace('\'', '', $styleValueString)); |
|
1002
|
|
|
|
|
1003
|
1 |
|
break; |
|
1004
|
|
|
|
|
1005
|
10 |
|
case 'text-decoration': |
|
1006
|
|
|
switch ($styleValue) { |
|
1007
|
1 |
|
case 'underline': |
|
1008
|
1 |
|
$cellStyle->getFont()->setUnderline(Font::UNDERLINE_SINGLE); |
|
1009
|
|
|
|
|
1010
|
1 |
|
break; |
|
1011
|
1 |
|
case 'line-through': |
|
1012
|
1 |
|
$cellStyle->getFont()->setStrikethrough(true); |
|
1013
|
|
|
|
|
1014
|
1 |
|
break; |
|
1015
|
|
|
} |
|
1016
|
|
|
|
|
1017
|
1 |
|
break; |
|
1018
|
|
|
|
|
1019
|
9 |
|
case 'text-align': |
|
1020
|
1 |
|
$cellStyle->getAlignment()->setHorizontal($styleValueString); |
|
1021
|
|
|
|
|
1022
|
1 |
|
break; |
|
1023
|
|
|
|
|
1024
|
9 |
|
case 'vertical-align': |
|
1025
|
2 |
|
$cellStyle->getAlignment()->setVertical($styleValueString); |
|
1026
|
|
|
|
|
1027
|
2 |
|
break; |
|
1028
|
|
|
|
|
1029
|
9 |
|
case 'width': |
|
1030
|
2 |
|
if ($column !== '') { |
|
1031
|
2 |
|
$sheet->getColumnDimension($column)->setWidth( |
|
1032
|
2 |
|
(new CssDimension($styleValue ?? ''))->width() |
|
1033
|
2 |
|
); |
|
1034
|
|
|
} |
|
1035
|
|
|
|
|
1036
|
2 |
|
break; |
|
1037
|
|
|
|
|
1038
|
7 |
|
case 'height': |
|
1039
|
1 |
|
if ($row > 0) { |
|
1040
|
1 |
|
$sheet->getRowDimension($row)->setRowHeight( |
|
1041
|
1 |
|
(new CssDimension($styleValue ?? ''))->height() |
|
1042
|
1 |
|
); |
|
1043
|
|
|
} |
|
1044
|
|
|
|
|
1045
|
1 |
|
break; |
|
1046
|
|
|
|
|
1047
|
6 |
|
case 'word-wrap': |
|
1048
|
1 |
|
$cellStyle->getAlignment()->setWrapText( |
|
1049
|
1 |
|
$styleValue === 'break-word' |
|
1050
|
1 |
|
); |
|
1051
|
|
|
|
|
1052
|
1 |
|
break; |
|
1053
|
|
|
|
|
1054
|
6 |
|
case 'text-indent': |
|
1055
|
2 |
|
$cellStyle->getAlignment()->setIndent( |
|
1056
|
2 |
|
(int) str_replace(['px'], '', $styleValueString) |
|
1057
|
2 |
|
); |
|
1058
|
|
|
|
|
1059
|
2 |
|
break; |
|
1060
|
|
|
} |
|
1061
|
|
|
} |
|
1062
|
|
|
} |
|
1063
|
|
|
|
|
1064
|
|
|
/** |
|
1065
|
|
|
* Check if has #, so we can get clean hex. |
|
1066
|
|
|
*/ |
|
1067
|
7 |
|
public function getStyleColor(?string $value): string |
|
1068
|
|
|
{ |
|
1069
|
7 |
|
$value = (string) $value; |
|
1070
|
7 |
|
if (str_starts_with($value, '#')) { |
|
1071
|
5 |
|
return substr($value, 1); |
|
1072
|
|
|
} |
|
1073
|
|
|
|
|
1074
|
4 |
|
return HelperHtml::colourNameLookup($value); |
|
1075
|
|
|
} |
|
1076
|
|
|
|
|
1077
|
9 |
|
private function insertImage(Worksheet $sheet, string $column, int $row, array $attributes): void |
|
1078
|
|
|
{ |
|
1079
|
9 |
|
if (!isset($attributes['src'])) { |
|
1080
|
1 |
|
return; |
|
1081
|
|
|
} |
|
1082
|
8 |
|
$styleArray = self::getStyleArray($attributes); |
|
1083
|
|
|
|
|
1084
|
8 |
|
$src = $attributes['src']; |
|
1085
|
8 |
|
if (substr($src, 0, 5) !== 'data:') { |
|
1086
|
4 |
|
$src = urldecode($src); |
|
1087
|
|
|
} |
|
1088
|
8 |
|
$width = isset($attributes['width']) ? (float) $attributes['width'] : ($styleArray['width'] ?? null); |
|
1089
|
8 |
|
$height = isset($attributes['height']) ? (float) $attributes['height'] : ($styleArray['height'] ?? null); |
|
1090
|
8 |
|
$name = $attributes['alt'] ?? null; |
|
1091
|
|
|
|
|
1092
|
8 |
|
$drawing = new Drawing(); |
|
1093
|
8 |
|
$drawing->setPath($src, false); |
|
1094
|
7 |
|
if ($drawing->getPath() === '') { |
|
1095
|
|
|
return; |
|
1096
|
|
|
} |
|
1097
|
7 |
|
$drawing->setWorksheet($sheet); |
|
1098
|
7 |
|
$drawing->setCoordinates($column . $row); |
|
1099
|
7 |
|
$drawing->setOffsetX(0); |
|
1100
|
7 |
|
$drawing->setOffsetY(10); |
|
1101
|
7 |
|
$drawing->setResizeProportional(true); |
|
1102
|
|
|
|
|
1103
|
7 |
|
if ($name) { |
|
1104
|
6 |
|
$drawing->setName($name); |
|
1105
|
|
|
} |
|
1106
|
|
|
|
|
1107
|
7 |
|
if ($width) { |
|
1108
|
5 |
|
if ($height) { |
|
1109
|
2 |
|
$drawing->setWidthAndHeight((int) $width, (int) $height); |
|
1110
|
|
|
} else { |
|
1111
|
3 |
|
$drawing->setWidth((int) $width); |
|
1112
|
|
|
} |
|
1113
|
2 |
|
} elseif ($height) { |
|
1114
|
1 |
|
$drawing->setHeight((int) $height); |
|
1115
|
|
|
} |
|
1116
|
|
|
|
|
1117
|
7 |
|
$sheet->getColumnDimension($column)->setWidth( |
|
1118
|
7 |
|
$drawing->getWidth() / 6 |
|
1119
|
7 |
|
); |
|
1120
|
|
|
|
|
1121
|
7 |
|
$sheet->getRowDimension($row)->setRowHeight( |
|
1122
|
7 |
|
$drawing->getHeight() * 0.9 |
|
1123
|
7 |
|
); |
|
1124
|
|
|
|
|
1125
|
7 |
|
if (isset($styleArray['opacity'])) { |
|
1126
|
|
|
$opacity = $styleArray['opacity']; |
|
1127
|
|
|
if (is_numeric($opacity)) { |
|
1128
|
|
|
$drawing->setOpacity((int) ($opacity * 100000)); |
|
1129
|
|
|
} |
|
1130
|
|
|
} |
|
1131
|
|
|
} |
|
1132
|
|
|
|
|
1133
|
8 |
|
private static function getStyleArray(array $attributes): array |
|
1134
|
|
|
{ |
|
1135
|
8 |
|
$styleArray = []; |
|
1136
|
8 |
|
if (isset($attributes['style'])) { |
|
1137
|
4 |
|
$styles = explode(';', $attributes['style']); |
|
1138
|
4 |
|
foreach ($styles as $style) { |
|
1139
|
4 |
|
$value = explode(':', $style); |
|
1140
|
4 |
|
if (count($value) === 2) { |
|
1141
|
4 |
|
$arrayKey = trim($value[0]); |
|
1142
|
4 |
|
$arrayValue = trim($value[1]); |
|
1143
|
4 |
|
if ($arrayKey === 'width') { |
|
1144
|
4 |
|
if (substr($arrayValue, -2) === 'px') { |
|
1145
|
4 |
|
$arrayValue = (string) (((float) substr($arrayValue, 0, -2))); |
|
1146
|
|
|
} else { |
|
1147
|
|
|
$arrayValue = (new CssDimension($arrayValue))->width(); |
|
1148
|
|
|
} |
|
1149
|
4 |
|
} elseif ($arrayKey === 'height') { |
|
1150
|
2 |
|
if (substr($arrayValue, -2) === 'px') { |
|
1151
|
2 |
|
$arrayValue = substr($arrayValue, 0, -2); |
|
1152
|
|
|
} else { |
|
1153
|
|
|
$arrayValue = (new CssDimension($arrayValue))->height(); |
|
1154
|
|
|
} |
|
1155
|
|
|
} |
|
1156
|
4 |
|
$styleArray[$arrayKey] = $arrayValue; |
|
1157
|
|
|
} |
|
1158
|
|
|
} |
|
1159
|
|
|
} |
|
1160
|
|
|
|
|
1161
|
8 |
|
return $styleArray; |
|
1162
|
|
|
} |
|
1163
|
|
|
|
|
1164
|
|
|
private const BORDER_MAPPINGS = [ |
|
1165
|
|
|
'dash-dot' => Border::BORDER_DASHDOT, |
|
1166
|
|
|
'dash-dot-dot' => Border::BORDER_DASHDOTDOT, |
|
1167
|
|
|
'dashed' => Border::BORDER_DASHED, |
|
1168
|
|
|
'dotted' => Border::BORDER_DOTTED, |
|
1169
|
|
|
'double' => Border::BORDER_DOUBLE, |
|
1170
|
|
|
'hair' => Border::BORDER_HAIR, |
|
1171
|
|
|
'medium' => Border::BORDER_MEDIUM, |
|
1172
|
|
|
'medium-dashed' => Border::BORDER_MEDIUMDASHED, |
|
1173
|
|
|
'medium-dash-dot' => Border::BORDER_MEDIUMDASHDOT, |
|
1174
|
|
|
'medium-dash-dot-dot' => Border::BORDER_MEDIUMDASHDOTDOT, |
|
1175
|
|
|
'none' => Border::BORDER_NONE, |
|
1176
|
|
|
'slant-dash-dot' => Border::BORDER_SLANTDASHDOT, |
|
1177
|
|
|
'solid' => Border::BORDER_THIN, |
|
1178
|
|
|
'thick' => Border::BORDER_THICK, |
|
1179
|
|
|
]; |
|
1180
|
|
|
|
|
1181
|
15 |
|
public static function getBorderMappings(): array |
|
1182
|
|
|
{ |
|
1183
|
15 |
|
return self::BORDER_MAPPINGS; |
|
1184
|
|
|
} |
|
1185
|
|
|
|
|
1186
|
|
|
/** |
|
1187
|
|
|
* Map html border style to PhpSpreadsheet border style. |
|
1188
|
|
|
*/ |
|
1189
|
3 |
|
public function getBorderStyle(string $style): ?string |
|
1190
|
|
|
{ |
|
1191
|
3 |
|
return self::BORDER_MAPPINGS[$style] ?? null; |
|
1192
|
|
|
} |
|
1193
|
|
|
|
|
1194
|
3 |
|
private function setBorderStyle(Style $cellStyle, string $styleValue, string $type): void |
|
1195
|
|
|
{ |
|
1196
|
3 |
|
if (trim($styleValue) === Border::BORDER_NONE) { |
|
1197
|
1 |
|
$borderStyle = Border::BORDER_NONE; |
|
1198
|
1 |
|
$color = null; |
|
1199
|
|
|
} else { |
|
1200
|
3 |
|
$borderArray = explode(' ', $styleValue); |
|
1201
|
3 |
|
$borderCount = count($borderArray); |
|
1202
|
3 |
|
if ($borderCount >= 3) { |
|
1203
|
3 |
|
$borderStyle = $borderArray[1]; |
|
1204
|
3 |
|
$color = $borderArray[2]; |
|
1205
|
|
|
} else { |
|
1206
|
1 |
|
$borderStyle = $borderArray[0]; |
|
1207
|
1 |
|
$color = $borderArray[1] ?? null; |
|
1208
|
|
|
} |
|
1209
|
|
|
} |
|
1210
|
|
|
|
|
1211
|
3 |
|
$cellStyle->applyFromArray([ |
|
1212
|
3 |
|
'borders' => [ |
|
1213
|
3 |
|
$type => [ |
|
1214
|
3 |
|
'borderStyle' => $this->getBorderStyle($borderStyle), |
|
1215
|
3 |
|
'color' => ['rgb' => $this->getStyleColor($color)], |
|
1216
|
3 |
|
], |
|
1217
|
3 |
|
], |
|
1218
|
3 |
|
]); |
|
1219
|
|
|
} |
|
1220
|
|
|
|
|
1221
|
|
|
/** |
|
1222
|
|
|
* Return worksheet info (Name, Last Column Letter, Last Column Index, Total Rows, Total Columns). |
|
1223
|
|
|
*/ |
|
1224
|
1 |
|
public function listWorksheetInfo(string $filename): array |
|
1225
|
|
|
{ |
|
1226
|
1 |
|
$info = []; |
|
1227
|
1 |
|
$spreadsheet = new Spreadsheet(); |
|
1228
|
1 |
|
$this->loadIntoExisting($filename, $spreadsheet); |
|
1229
|
1 |
|
foreach ($spreadsheet->getAllSheets() as $sheet) { |
|
1230
|
1 |
|
$newEntry = ['worksheetName' => $sheet->getTitle()]; |
|
1231
|
1 |
|
$newEntry['lastColumnLetter'] = $sheet->getHighestDataColumn(); |
|
1232
|
1 |
|
$newEntry['lastColumnIndex'] = Coordinate::columnIndexFromString($sheet->getHighestDataColumn()) - 1; |
|
1233
|
1 |
|
$newEntry['totalRows'] = $sheet->getHighestDataRow(); |
|
1234
|
1 |
|
$newEntry['totalColumns'] = $newEntry['lastColumnIndex'] + 1; |
|
1235
|
1 |
|
$info[] = $newEntry; |
|
1236
|
|
|
} |
|
1237
|
1 |
|
$spreadsheet->disconnectWorksheets(); |
|
1238
|
|
|
|
|
1239
|
1 |
|
return $info; |
|
1240
|
|
|
} |
|
1241
|
|
|
} |
|
1242
|
|
|
|
PHPOffice /
PhpSpreadsheet
