1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
declare(strict_types=1); |
4
|
|
|
|
5
|
|
|
/** |
6
|
|
|
* Parser class. |
7
|
|
|
* |
8
|
|
|
* @package YetiForcePDF\Html |
9
|
|
|
* |
10
|
|
|
* @copyright YetiForce Sp. z o.o |
11
|
|
|
* @license MIT |
12
|
|
|
* @author Rafal Pospiech <[email protected]> |
13
|
|
|
*/ |
14
|
|
|
|
15
|
|
|
namespace YetiForcePDF\Html; |
16
|
|
|
|
17
|
|
|
use YetiForcePDF\Base; |
18
|
|
|
use YetiForcePDF\Layout\PageGroupBox; |
19
|
|
|
|
20
|
|
|
/** |
21
|
|
|
* Class Parser. |
22
|
|
|
*/ |
23
|
|
|
class Parser extends Base |
24
|
|
|
{ |
25
|
|
|
/** |
26
|
|
|
* @var \DOMDocument |
27
|
|
|
*/ |
28
|
|
|
protected $domDocument; |
29
|
|
|
|
30
|
|
|
/** |
31
|
|
|
* @var string |
32
|
|
|
*/ |
33
|
|
|
protected $html = ''; |
34
|
|
|
|
35
|
|
|
/** |
36
|
|
|
* @var array page groups with html content divided |
37
|
|
|
*/ |
38
|
|
|
protected $htmlPageGroups = []; |
39
|
|
|
|
40
|
|
|
/** |
41
|
|
|
* @var array |
42
|
|
|
*/ |
43
|
|
|
protected $pageGroups = []; |
44
|
|
|
|
45
|
|
|
/** |
46
|
|
|
* Cleanup html. |
47
|
|
|
* |
48
|
|
|
* @param string $html |
49
|
|
|
* @param string $fromEncoding |
50
|
|
|
* |
51
|
|
|
* @return string |
52
|
|
|
*/ |
53
|
|
|
protected function cleanUpHtml(string $html) |
54
|
|
|
{ |
55
|
|
|
$html = preg_replace('/\r\n/', "\r", $html); |
56
|
|
|
return preg_replace('/\n/', "\r", $html); |
57
|
|
|
} |
58
|
|
|
|
59
|
|
|
/** |
60
|
|
|
* Load html string. |
61
|
|
|
* |
62
|
|
|
* @param string $html |
63
|
|
|
* @param string $fromEncoding |
64
|
|
|
* |
65
|
|
|
* @return Parser |
66
|
|
|
*/ |
67
|
|
|
public function loadHtml(string $html, string $fromEncoding): self |
68
|
|
|
{ |
69
|
|
|
$this->html = htmlspecialchars_decode($html, ENT_HTML5); |
70
|
|
|
$this->html = $this->cleanUpHtml($html); |
71
|
|
|
|
72
|
|
|
// 0x80 - start of unicode range |
73
|
|
|
// 0x10FFFF - end of unicode range |
74
|
|
|
// 0 - do not ommit any unicode char |
75
|
|
|
// ~0 - negated 0 - convert negation of nothing (so convert all) |
76
|
|
|
$this->html = mb_encode_numericentity($this->html, [0x80, 0x10FFFF, 0, ~0], $fromEncoding); |
77
|
|
|
|
78
|
|
|
return $this; |
79
|
|
|
} |
80
|
|
|
|
81
|
|
|
/** |
82
|
|
|
* Get html. |
83
|
|
|
* |
84
|
|
|
* @return string |
85
|
|
|
*/ |
86
|
|
|
public function getHtml(): string |
87
|
|
|
{ |
88
|
|
|
return $this->html; |
89
|
|
|
} |
90
|
|
|
|
91
|
|
|
/** |
92
|
|
|
* Remove comment blocks. |
93
|
|
|
* |
94
|
|
|
* @param string $html |
95
|
|
|
* |
96
|
|
|
* @return string |
97
|
|
|
*/ |
98
|
|
|
public function removeComments(string $html) |
99
|
|
|
{ |
100
|
|
|
return preg_replace('/<!--((?!-->))[\w\W]+-->/uUi', '', $html); |
101
|
|
|
} |
102
|
|
|
|
103
|
|
|
/** |
104
|
|
|
* Divide html into page groups. |
105
|
|
|
* |
106
|
|
|
* @param string $html |
107
|
|
|
* |
108
|
|
|
* @return array |
109
|
|
|
*/ |
110
|
|
|
public function getHtmlPageGroups(string $html) |
111
|
|
|
{ |
112
|
|
|
$pageGroups = []; |
113
|
|
|
$matches = []; |
114
|
|
|
preg_match_all('/\<div\s+data-page-group\s?/ui', $html, $matches, PREG_OFFSET_CAPTURE); |
115
|
|
|
$matches = $matches[0]; |
116
|
|
|
$groupsCount = \count($matches); |
117
|
|
|
for ($i = 0; $i < $groupsCount; ++$i) { |
118
|
|
|
$start = $matches[$i][1]; |
119
|
|
|
if (isset($matches[$i + 1])) { |
120
|
|
|
$stop = $matches[$i + 1][1]; |
121
|
|
|
$len = $stop - $start; |
122
|
|
|
$pageGroups[] = substr($html, $start, $len); |
123
|
|
|
} else { |
124
|
|
|
$pageGroups[] = substr($html, $start); |
125
|
|
|
} |
126
|
|
|
} |
127
|
|
|
if (empty($pageGroups)) { |
128
|
|
|
return [$html]; |
129
|
|
|
} |
130
|
|
|
return $pageGroups; |
131
|
|
|
} |
132
|
|
|
|
133
|
|
|
/** |
134
|
|
|
* Set page group options. |
135
|
|
|
* |
136
|
|
|
* @param PageGroupBox $root |
137
|
|
|
* @param \DOMDocument $domDocument |
138
|
|
|
* |
139
|
|
|
* @return $this |
140
|
|
|
*/ |
141
|
|
|
public function setGroupOptions(PageGroupBox $root, \DOMDocument $domDocument) |
142
|
|
|
{ |
143
|
|
|
$childDomElement = $domDocument->documentElement->firstChild; |
144
|
|
|
if (!$childDomElement instanceof \DOMElement) { |
145
|
|
|
return $this; |
146
|
|
|
} |
147
|
|
|
if ($childDomElement->hasAttribute('data-format')) { |
148
|
|
|
$root->format = $childDomElement->getAttribute('data-format'); |
149
|
|
|
if (!$root->format) { |
150
|
|
|
$root->format = 'A4'; |
151
|
|
|
} |
152
|
|
|
} |
153
|
|
|
if ($childDomElement->hasAttribute('data-orientation')) { |
154
|
|
|
$root->orientation = $childDomElement->getAttribute('data-orientation'); |
155
|
|
|
if (!$root->orientation) { |
156
|
|
|
$root->orientation = 'P'; |
157
|
|
|
} |
158
|
|
|
} |
159
|
|
|
if ($childDomElement->hasAttribute('data-margin-left')) { |
160
|
|
|
$root->marginLeft = (float) $childDomElement->getAttribute('data-margin-left'); |
161
|
|
|
if (!$root->marginLeft) { |
162
|
|
|
$root->marginLeft = 30; |
163
|
|
|
} |
164
|
|
|
} |
165
|
|
|
if ($childDomElement->hasAttribute('data-margin-right')) { |
166
|
|
|
$root->marginRight = (float) $childDomElement->getAttribute('data-margin-right'); |
167
|
|
|
if (!$root->marginRight) { |
168
|
|
|
$root->marginRight = 30; |
169
|
|
|
} |
170
|
|
|
} |
171
|
|
|
if ($childDomElement->hasAttribute('data-margin-top')) { |
172
|
|
|
$root->marginTop = (float) $childDomElement->getAttribute('data-margin-top'); |
173
|
|
|
if (!$root->marginTop) { |
174
|
|
|
$root->marginTop = 40; |
175
|
|
|
} |
176
|
|
|
} |
177
|
|
|
if ($childDomElement->hasAttribute('data-margin-bottom')) { |
178
|
|
|
$root->marginBottom = (float) $childDomElement->getAttribute('data-margin-bottom'); |
179
|
|
|
if (!$root->marginBottom) { |
180
|
|
|
$root->marginBottom = 40; |
181
|
|
|
} |
182
|
|
|
} |
183
|
|
|
if ($childDomElement->hasAttribute('data-header-top')) { |
184
|
|
|
$root->headerTop = (float) $childDomElement->getAttribute('data-header-top'); |
185
|
|
|
if (!$root->headerTop) { |
186
|
|
|
$root->headerTop = 10; |
187
|
|
|
} |
188
|
|
|
} |
189
|
|
|
if ($childDomElement->hasAttribute('data-footer-bottom')) { |
190
|
|
|
$root->footerBottom = (float) $childDomElement->getAttribute('data-footer-bottom'); |
191
|
|
|
if (!$root->footerBottom) { |
192
|
|
|
$root->footerBottom = 10; |
193
|
|
|
} |
194
|
|
|
} |
195
|
|
|
return $this; |
196
|
|
|
} |
197
|
|
|
|
198
|
|
|
/** |
199
|
|
|
* Convert loaded html to pdf objects. |
200
|
|
|
*/ |
201
|
|
|
public function parse() |
202
|
|
|
{ |
203
|
|
|
if ('' === $this->html) { |
204
|
|
|
return null; |
205
|
|
|
} |
206
|
|
|
$this->html = $this->removeComments($this->html); |
207
|
|
|
$this->htmlPageGroups = $this->getHtmlPageGroups($this->html); |
208
|
|
|
foreach ($this->htmlPageGroups as $groupIndex => $htmlPageGroup) { |
209
|
|
|
$domDocument = new \DOMDocument(); |
210
|
|
|
$domDocument->encoding = 'UTF-8'; |
211
|
|
|
$domDocument->strictErrorChecking = false; |
212
|
|
|
$domDocument->substituteEntities = false; |
213
|
|
|
$domDocument->recover = false; |
214
|
|
|
$domDocument->loadHTML('<div id="yetiforcepdf">' . $htmlPageGroup . '</div>', LIBXML_HTML_NOIMPLIED | LIBXML_NOWARNING | LIBXML_NOBLANKS | LIBXML_NOERROR); |
215
|
|
|
$pageGroup = (new PageGroupBox()) |
216
|
|
|
->setDocument($this->document) |
217
|
|
|
->setRoot(true) |
218
|
|
|
->init(); |
219
|
|
|
$pageGroup->format = $this->document->getDefaultFormat(); |
220
|
|
|
$margins = $this->document->getDefaultMargins(); |
221
|
|
|
$pageGroup->marginLeft = $margins['left']; |
222
|
|
|
$pageGroup->marginTop = $margins['top']; |
223
|
|
|
$pageGroup->marginRight = $margins['right']; |
224
|
|
|
$pageGroup->marginBottom = $margins['bottom']; |
225
|
|
|
$pageGroup->orientation = $this->document->getDefaultOrientation(); |
226
|
|
|
$this->setGroupOptions($pageGroup, $domDocument); |
227
|
|
|
$page = $this->document->addPage($pageGroup->format, $pageGroup->orientation); |
228
|
|
|
$page->setPageNumber(1); |
229
|
|
|
$page->setGroup($groupIndex); |
230
|
|
|
$page->setMargins($pageGroup->marginLeft, $pageGroup->marginTop, $pageGroup->marginRight, $pageGroup->marginBottom); |
231
|
|
|
$rootElement = (new \YetiForcePDF\Html\Element()) |
232
|
|
|
->setDocument($this->document) |
233
|
|
|
->setDOMElement($domDocument->documentElement); |
234
|
|
|
// root element must be defined before initialisation |
235
|
|
|
$rootElement->init(); |
236
|
|
|
$pageGroup->setElement($rootElement); |
237
|
|
|
$pageGroup->setStyle($rootElement->parseStyle()); |
238
|
|
|
|
239
|
|
|
$pageGroup->buildTree(); |
240
|
|
|
$pageGroup->fixTables(); |
241
|
|
|
$pageGroup->getStyle()->fixDomTree(); |
242
|
|
|
$pageGroup->layout(); |
243
|
|
|
$page->setBox($pageGroup); |
244
|
|
|
|
245
|
|
|
foreach ($this->document->getPages($groupIndex) as $page) { |
246
|
|
|
$page->getBox()->breakPageAfter(); |
|
|
|
|
247
|
|
|
} |
248
|
|
|
foreach ($this->document->getPages($groupIndex) as $page) { |
249
|
|
|
$page->getBox()->spanAllRows(); |
|
|
|
|
250
|
|
|
} |
251
|
|
|
foreach ($this->document->getPages($groupIndex) as $page) { |
252
|
|
|
$page->breakOverflow(); |
253
|
|
|
} |
254
|
|
|
$page->getBox()->getStyle()->fixDomTree(); |
255
|
|
|
$this->document->fixPageNumbers(); |
256
|
|
|
foreach ($this->document->getPages($groupIndex) as $page) { |
257
|
|
|
$this->document->setCurrentPage($page); |
258
|
|
|
$children = []; |
259
|
|
|
$page->setUpAbsoluteBoxes(); |
260
|
|
|
$page->getBox()->replacePageNumbers(); |
|
|
|
|
261
|
|
|
$page->getBox()->getAllChildren($children); |
262
|
|
|
foreach ($children as $box) { |
263
|
|
|
if (!$box instanceof \YetiForcePDF\Layout\LineBox && $box->isRenderable()) { |
264
|
|
|
$page->getContentStream()->addRawContent($box->getInstructions()); |
265
|
|
|
} |
266
|
|
|
} |
267
|
|
|
} |
268
|
|
|
} |
269
|
|
|
} |
270
|
|
|
} |
271
|
|
|
|