|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace kalanis\kw_forums\Cutting; |
|
4
|
|
|
|
|
5
|
|
|
|
|
6
|
|
|
/** |
|
7
|
|
|
* Class CuttingContent |
|
8
|
|
|
* @package kalanis\kw_forums\Cutting |
|
9
|
|
|
* Cutting messages |
|
10
|
|
|
*/ |
|
11
|
|
|
class Content |
|
12
|
|
|
{ |
|
13
|
|
|
// it's necessary to have both arrays sorted this way - it's for lookup by positions |
|
14
|
|
|
/** @var string[] */ |
|
15
|
|
|
protected static array $OPENING_TAGS = ['<b ', '<b>', '<i ', '<i>', '<u ', '<u>', '<center>', '<span>', '<span ', '<font color', '<font face', '<font size', '<font ', '<font', '<table ', '<table', '<tr>', '<td>', '<a ', '<a>']; |
|
16
|
|
|
/** @var string[] */ |
|
17
|
|
|
protected static array $CLOSING_TAGS = ['</b>', '</b>', '</i>', '</i>', '</u>', '</u>', '</center>', '</span>', '</span>', '</font>', '</font>', '</font>', '</font>', '</font>', '</table>', '</table>', '</tr>', '</td>', '</a>', '</a>']; |
|
18
|
|
|
|
|
19
|
|
|
protected int $wordLengthNeedle = 20; |
|
20
|
|
|
protected int $maxLength = 0; |
|
21
|
|
|
protected string $content = ''; |
|
22
|
|
|
/** |
|
23
|
|
|
* Contains positions of tags in |
|
24
|
|
|
* @var int[] |
|
25
|
|
|
*/ |
|
26
|
|
|
protected array $tagPositionsStack = []; |
|
27
|
|
|
|
|
28
|
6 |
|
public function __construct(int $maxLength) |
|
29
|
|
|
{ |
|
30
|
6 |
|
$this->maxLength = $maxLength; |
|
31
|
6 |
|
} |
|
32
|
|
|
|
|
33
|
6 |
|
public function setContent(string $content): self |
|
34
|
|
|
{ |
|
35
|
6 |
|
$this->content = $content; |
|
36
|
6 |
|
$this->tagPositionsStack = []; |
|
37
|
6 |
|
return $this; |
|
38
|
|
|
} |
|
39
|
|
|
|
|
40
|
6 |
|
public function getContent(): string |
|
41
|
|
|
{ |
|
42
|
6 |
|
return $this->content; |
|
43
|
|
|
} |
|
44
|
|
|
|
|
45
|
6 |
|
public function process(): self |
|
46
|
|
|
{ |
|
47
|
6 |
|
if (mb_strlen($this->content) <= $this->maxLength) { |
|
48
|
1 |
|
return $this; |
|
49
|
|
|
} |
|
50
|
|
|
|
|
51
|
|
|
// lookup for available end, the values has been found empirically |
|
52
|
5 |
|
$content = mb_substr($this->content, 0, $this->maxLength + $this->wordLengthNeedle); |
|
53
|
5 |
|
$len = mb_strrpos($content, ' '); |
|
54
|
5 |
|
$len = false !== $len ? intval($len) : null; |
|
55
|
5 |
|
$content = mb_substr($content, 0, $len) . ' '; |
|
56
|
|
|
|
|
57
|
5 |
|
$this->content = $this->closeQuotationMarks($this->contentCut($content)) . $this->fillCutTags() . ' ...'; |
|
58
|
5 |
|
return $this; |
|
59
|
|
|
} |
|
60
|
|
|
|
|
61
|
|
|
/** |
|
62
|
|
|
* @param string $content content to cut |
|
63
|
|
|
* @return string updated string |
|
64
|
|
|
*/ |
|
65
|
5 |
|
protected function contentCut(string $content): string |
|
66
|
|
|
{ |
|
67
|
5 |
|
$this->tagPositionsStack = []; |
|
68
|
5 |
|
for ($i = 0; $i <= mb_strlen($content); $i++) { |
|
69
|
|
|
// start tag sequence |
|
70
|
5 |
|
if ('<' == mb_substr($content, $i, 1)) { |
|
71
|
|
|
//print_r(['found open', $i]); |
|
72
|
4 |
|
if ('/' == mb_substr($content, $i + 1, 1)) { |
|
73
|
|
|
// closing tag |
|
74
|
1 |
|
$tagPosition = $this->tagPosition(mb_strtolower(mb_substr($content, $i, 7)), static::$CLOSING_TAGS); |
|
75
|
1 |
|
if (!is_null($tagPosition)) { |
|
76
|
|
|
// is correct one in array of closing tags? |
|
77
|
1 |
|
$lastPosition = array_pop($this->tagPositionsStack); |
|
78
|
1 |
|
if (!is_null($lastPosition)) { |
|
79
|
1 |
|
if (static::$CLOSING_TAGS[$tagPosition] == static::$CLOSING_TAGS[$lastPosition]) { // because there are multiple openings |
|
80
|
1 |
|
$i = $i + mb_strlen(static::$CLOSING_TAGS[$tagPosition]) - 2; // move to next available position in string |
|
81
|
|
|
} else { |
|
82
|
1 |
|
$this->tagPositionsStack[] = $lastPosition; |
|
83
|
|
|
} |
|
84
|
|
|
} |
|
85
|
|
|
} |
|
86
|
|
|
//print_r(['is closing', $tagPosition, is_null($tagPosition) ? 'x' : static::$CLOSING_TAGS[$tagPosition]]); |
|
87
|
|
|
} else { |
|
88
|
|
|
// opening tag |
|
89
|
|
|
// cut by longest available tag in array |
|
90
|
4 |
|
$tagPosition = $this->tagPosition(mb_substr($content, $i, 11), static::$OPENING_TAGS); |
|
91
|
4 |
|
if (!is_null($tagPosition)) { // is as pair in array and I know that |
|
92
|
4 |
|
$this->tagPositionsStack[] = $tagPosition; |
|
93
|
4 |
|
$i = $i + mb_strlen(static::$OPENING_TAGS[$tagPosition]) - 1; // move to next available position in string |
|
94
|
|
|
} |
|
95
|
|
|
//print_r(['is opening', $tagPosition, is_null($tagPosition) ? 'x' : static::$OPENING_TAGS[$tagPosition]]); |
|
96
|
|
|
} |
|
97
|
|
|
} |
|
98
|
|
|
} |
|
99
|
|
|
|
|
100
|
5 |
|
return $content; |
|
101
|
|
|
} |
|
102
|
|
|
|
|
103
|
|
|
/** |
|
104
|
|
|
* Returns tag position |
|
105
|
|
|
* @param string $lookupTag |
|
106
|
|
|
* @param string[] $definedTags |
|
107
|
|
|
* @return int|null |
|
108
|
|
|
* int as position in defined tags, null as not found |
|
109
|
|
|
*/ |
|
110
|
4 |
|
protected function tagPosition(string $lookupTag, array $definedTags): ?int |
|
111
|
|
|
{ |
|
112
|
4 |
|
foreach ($definedTags as $position => $definedTag) { |
|
113
|
4 |
|
$cutTag = mb_substr($lookupTag, 0, mb_strlen($definedTag)); |
|
114
|
4 |
|
if ($cutTag == $definedTag) { |
|
115
|
4 |
|
return $position; |
|
116
|
|
|
} |
|
117
|
|
|
} |
|
118
|
2 |
|
return null; |
|
119
|
|
|
} |
|
120
|
|
|
|
|
121
|
5 |
|
protected function closeQuotationMarks(string $content): string |
|
122
|
|
|
{ |
|
123
|
|
|
// check for closing tags with spaces and quotation marks inside |
|
124
|
5 |
|
$tagStarts = mb_strrpos($content, '<'); |
|
125
|
5 |
|
$tagEnds = mb_strrpos($content, '>'); |
|
126
|
5 |
|
if ((false !== $tagStarts) && ((false === $tagEnds) || (false !== $tagEnds) && ($tagStarts > $tagEnds))) { // unclosed tag |
|
127
|
3 |
|
$foundOdd = mb_substr_count(mb_substr($content, $tagStarts), "'"); |
|
128
|
3 |
|
if ($foundOdd % 2) { |
|
129
|
1 |
|
$content .= "'"; |
|
130
|
|
|
} |
|
131
|
3 |
|
$foundOdd = mb_substr_count(mb_substr($content, $tagStarts), '"'); |
|
132
|
3 |
|
if ($foundOdd % 2) { |
|
133
|
1 |
|
$content .= '"'; |
|
134
|
|
|
} |
|
135
|
3 |
|
$content .= '>'; |
|
136
|
|
|
} |
|
137
|
5 |
|
return $content; |
|
138
|
|
|
} |
|
139
|
|
|
|
|
140
|
5 |
|
protected function fillCutTags(): string |
|
141
|
|
|
{ |
|
142
|
|
|
//var_dump(['rev', $this->tagPositionsStack]); |
|
143
|
5 |
|
$content = ''; |
|
144
|
5 |
|
foreach (array_reverse($this->tagPositionsStack) as $positions) { |
|
145
|
4 |
|
$content .= static::$CLOSING_TAGS[$positions]; |
|
146
|
|
|
} |
|
147
|
|
|
|
|
148
|
5 |
|
return $content; |
|
149
|
|
|
} |
|
150
|
|
|
} |
|
151
|
|
|
|
alex-kalanis /
kw_forums
