|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Wulkanowy\TimetableParser; |
|
4
|
|
|
|
|
5
|
|
|
use DOMWrap\Document; |
|
6
|
|
|
use DOMWrap\Element; |
|
7
|
|
|
use DOMWrap\NodeList; |
|
8
|
|
|
|
|
9
|
|
|
class Table |
|
10
|
|
|
{ |
|
11
|
|
|
private $doc; |
|
12
|
|
|
|
|
13
|
|
|
public function __construct(string $html) |
|
14
|
|
|
{ |
|
15
|
|
|
$this->doc = new Document(); |
|
16
|
|
|
$this->doc->html($html); |
|
17
|
|
|
} |
|
18
|
|
|
|
|
19
|
|
|
public function getTable(): array |
|
20
|
|
|
{ |
|
21
|
|
|
$table = $this->doc->find('.tabela')->first(); |
|
22
|
|
|
$days = $this->getDays($table->find('tr th')); |
|
23
|
|
|
|
|
24
|
|
|
$this->setLessonHoursToDays($table, $days); |
|
25
|
|
|
|
|
26
|
|
|
$title = explode(' ', $this->doc->find('title')->text()); |
|
27
|
|
|
$generated = preg_split('/\s+/', trim($this->doc->find('td[align=right]')->end()->text())); |
|
28
|
|
|
|
|
29
|
|
|
return [ |
|
30
|
|
|
'name' => $this->doc->find('.tytulnapis')->text(), |
|
31
|
|
|
'generated' => trim($generated[1]), |
|
32
|
|
|
'description' => trim($this->doc->find('td[align=left]')->first()->text()), |
|
33
|
|
|
'typeName' => $title[2], |
|
34
|
|
|
'days' => $days, |
|
35
|
|
|
]; |
|
36
|
|
|
} |
|
37
|
|
|
|
|
38
|
|
|
private function getDays(NodeList $headerCells): array |
|
39
|
|
|
{ |
|
40
|
|
|
$headerCells->shift(); |
|
41
|
|
|
$headerCells->shift(); |
|
42
|
|
|
$days = []; |
|
43
|
|
|
|
|
44
|
|
|
/** @var \DOMNode $cell */ |
|
45
|
|
|
foreach ($headerCells as $cell) { |
|
46
|
|
|
$days[] = ['name' => $cell->textContent]; |
|
47
|
|
|
} |
|
48
|
|
|
|
|
49
|
|
|
return $days; |
|
50
|
|
|
} |
|
51
|
|
|
|
|
52
|
|
|
private function setLessonHoursToDays(Element $table, array &$days): void |
|
53
|
|
|
{ |
|
54
|
|
|
$rows = $table->find('tr'); |
|
55
|
|
|
$rows->shift(); |
|
56
|
|
|
|
|
57
|
|
|
foreach ($rows as $row) { |
|
58
|
|
|
$rowCells = $row->find('td'); |
|
59
|
|
|
|
|
60
|
|
|
// fill hours in day |
|
61
|
|
|
/** @var NodeList $rowCells */ |
|
62
|
|
|
for ($i = 2; $i < $rowCells->count(); $i++) { |
|
63
|
|
|
$days[$i - 2]['hours'][] = $this->getHourWithLessons($rowCells, $i); |
|
64
|
|
|
} |
|
65
|
|
|
} |
|
66
|
|
|
} |
|
67
|
|
|
|
|
68
|
|
|
private function getHourWithLessons(NodeList $rowCells, int $index): array |
|
69
|
|
|
{ |
|
70
|
|
|
$hours = explode('-', $rowCells->get(1)->textContent); |
|
71
|
|
|
|
|
72
|
|
|
return [ |
|
73
|
|
|
'number' => $rowCells->get(0)->textContent, |
|
74
|
|
|
'start' => trim($hours[0]), |
|
75
|
|
|
'end' => trim($hours[1]), |
|
76
|
|
|
'lessons' => $this->getExtractedLessons($rowCells->get($index)), |
|
77
|
|
|
]; |
|
78
|
|
|
} |
|
79
|
|
|
|
|
80
|
|
|
private function getExtractedLessons(Element $current): array |
|
81
|
|
|
{ |
|
82
|
|
|
$lessons = []; |
|
83
|
|
|
|
|
84
|
|
|
$chunks = explode('<br>', $current->getOuterHtml()); |
|
85
|
|
|
$spans = $current->find('span[style]'); |
|
86
|
|
|
$subject = $current->findXPath('./*[@class="p"]'); |
|
87
|
|
|
|
|
88
|
|
|
if ($spans->count() > 0 && $subject->count() === 0) { |
|
89
|
|
|
foreach ($spans as $group) { |
|
90
|
|
|
$lessons[] = array_merge($this->getLesson($group, true)); |
|
91
|
|
|
} |
|
92
|
|
|
} elseif (\count($chunks) > 0 && $subject->count() > 0) { |
|
93
|
|
|
foreach ($chunks as $item) { |
|
94
|
|
|
$this->setLessonFromChunk($lessons, $item); |
|
95
|
|
|
} |
|
96
|
|
|
} |
|
97
|
|
|
|
|
98
|
|
|
$this->updateLessonWithMultipleClasses($current, $lessons); |
|
99
|
|
|
$this->setFallbackLesson($current, $lessons); |
|
100
|
|
|
|
|
101
|
|
|
return $lessons; |
|
102
|
|
|
} |
|
103
|
|
|
|
|
104
|
|
|
private function setLessonFromChunk(array &$lessons, string $chunk): void |
|
105
|
|
|
{ |
|
106
|
|
|
$doc = new Document(); |
|
107
|
|
|
$doc->html($chunk); |
|
108
|
|
|
|
|
109
|
|
|
$span = $doc->find('span[style]'); |
|
110
|
|
|
$cell = $doc->find('.l'); |
|
111
|
|
|
$body = $doc->find('body'); |
|
112
|
|
|
|
|
113
|
|
|
if ($span->count() > 0) { |
|
114
|
|
|
$lessons[] = array_merge($this->getLesson($span->first(), true)); |
|
115
|
|
|
} elseif ($cell->count() > 0) { |
|
116
|
|
|
$lessons[] = $this->getLesson($cell->first()); |
|
117
|
|
|
} elseif ($body->count() > 0) { |
|
118
|
|
|
$lessons[] = $this->getLesson($body->first()); |
|
119
|
|
|
} |
|
120
|
|
|
} |
|
121
|
|
|
|
|
122
|
|
|
private function setFallbackLesson(Element $current, array &$lessons): void |
|
123
|
|
|
{ |
|
124
|
|
|
if (\count($lessons) === 0 && trim($current->text(), "\xC2\xA0\n") !== '') { |
|
125
|
|
|
$lessons[] = $this->getLesson($current); |
|
126
|
|
|
} |
|
127
|
|
|
} |
|
128
|
|
|
|
|
129
|
|
|
private function updateLessonWithMultipleClasses(Element $current, array &$lessons): void |
|
130
|
|
|
{ |
|
131
|
|
|
if ($current->findXPath('./*[@class="o"]')->count() < 2) { |
|
132
|
|
|
return; |
|
133
|
|
|
} |
|
134
|
|
|
|
|
135
|
|
|
$lastIndex = \count($lessons) - 1; |
|
136
|
|
|
|
|
137
|
|
|
unset($lessons[$lastIndex]['className'], $lessons[$lastIndex]['alt']); |
|
138
|
|
|
|
|
139
|
|
|
/** @var Element $item */ |
|
140
|
|
|
foreach (explode(',', $current->getOuterHtml()) as $item) { |
|
141
|
|
|
$doc = new Document(); |
|
142
|
|
|
$doc->html(str_replace('<td class="l">', '', $item)); |
|
143
|
|
|
$el = $doc->find('body')->first(); |
|
144
|
|
|
$lessons[$lastIndex]['className'][] = [ |
|
145
|
|
|
'name' => trim($el->find('a.o')->text()), |
|
146
|
|
|
'value' => $this->getUrlValue($el->find('a.o')->first(), 'o'), |
|
147
|
|
|
'alt' => trim($el->findXPath('./text()')->text()), |
|
148
|
|
|
]; |
|
149
|
|
|
} |
|
150
|
|
|
} |
|
151
|
|
|
|
|
152
|
|
|
private function getLesson(Element $cell, bool $diversion = false): array |
|
153
|
|
|
{ |
|
154
|
|
|
$subject = $cell->findXPath('./*[@class="p"]'); |
|
155
|
|
|
|
|
156
|
|
|
$lesson = [ |
|
157
|
|
|
'teacher' => $this->getLessonPartValues($cell->findXPath('./*[@class="n"]'), 'n'), |
|
158
|
|
|
'room' => $this->getLessonPartValues($cell->findXPath('./*[@class="s"]'), 's'), |
|
159
|
|
|
'className' => $this->getLessonPartValues($cell->findXPath('./*[@class="o"]'), 'o'), |
|
160
|
|
|
'subject' => $subject->text(), |
|
161
|
|
|
'diversion' => $diversion, |
|
162
|
|
|
'alt' => trim($cell->findXPath('./text()')->text()), |
|
163
|
|
|
]; |
|
164
|
|
|
|
|
165
|
|
|
$subjects = $cell->findXPath('./*[@class="p"]'); |
|
166
|
|
|
if ($subjects->count() > 1) { |
|
167
|
|
|
$textBetweenSubject = $cell->findXPath('./text()[(preceding::*[@class="p"])]'); |
|
168
|
|
|
if (trim($textBetweenSubject->text()) !== '') { |
|
169
|
|
|
$lesson['subject'] = $subject->first()->text().trim($textBetweenSubject->text()).' '.$subject->end()->text(); |
|
170
|
|
|
} else { |
|
171
|
|
|
unset($lesson['subject']); |
|
172
|
|
|
foreach ($subjects as $subject) { |
|
173
|
|
|
$lesson['subject'][] = $subject->text(); |
|
174
|
|
|
} |
|
175
|
|
|
} |
|
176
|
|
|
} |
|
177
|
|
|
|
|
178
|
|
|
return $lesson; |
|
179
|
|
|
} |
|
180
|
|
|
|
|
181
|
|
|
private function getLessonPartValues(NodeList $part, string $prefix): array |
|
182
|
|
|
{ |
|
183
|
|
|
return [ |
|
184
|
|
|
'name' => $part->text(), |
|
185
|
|
|
'value' => $this->getUrlValue($part->first(), $prefix), |
|
186
|
|
|
]; |
|
187
|
|
|
} |
|
188
|
|
|
|
|
189
|
|
|
private function getUrlValue(?Element $el, string $prefix): string |
|
190
|
|
|
{ |
|
191
|
|
|
if (null === $el) { |
|
192
|
|
|
return ''; |
|
193
|
|
|
} |
|
194
|
|
|
|
|
195
|
|
|
return str_replace([$prefix, '.html'], '', $el->attr('href')); |
|
196
|
|
|
} |
|
197
|
|
|
} |
|
198
|
|
|
|
wulkanowy /
timetable-parser-php
