|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Box\Spout\Reader\XLSX\Helper\SharedStringsCaching; |
|
4
|
|
|
|
|
5
|
|
|
/** |
|
6
|
|
|
* Class CachingStrategyFactory |
|
7
|
|
|
* |
|
8
|
|
|
* @package Box\Spout\Reader\XLSX\Helper\SharedStringsCaching |
|
9
|
|
|
*/ |
|
10
|
|
|
class CachingStrategyFactory |
|
11
|
|
|
{ |
|
12
|
|
|
/** |
|
13
|
|
|
* The memory amount needed to store a string was obtained empirically from this data: |
|
14
|
|
|
* |
|
15
|
|
|
* ------------------------------------ |
|
16
|
|
|
* | Number of chars⁺ | Memory needed | |
|
17
|
|
|
* ------------------------------------ |
|
18
|
|
|
* | 3,000 | 1 MB | |
|
19
|
|
|
* | 15,000 | 2 MB | |
|
20
|
|
|
* | 30,000 | 5 MB | |
|
21
|
|
|
* | 75,000 | 11 MB | |
|
22
|
|
|
* | 150,000 | 21 MB | |
|
23
|
|
|
* | 300,000 | 43 MB | |
|
24
|
|
|
* | 750,000 | 105 MB | |
|
25
|
|
|
* | 1,500,000 | 210 MB | |
|
26
|
|
|
* | 2,250,000 | 315 MB | |
|
27
|
|
|
* | 3,000,000 | 420 MB | |
|
28
|
|
|
* | 4,500,000 | 630 MB | |
|
29
|
|
|
* ------------------------------------ |
|
30
|
|
|
* |
|
31
|
|
|
* ⁺ All characters were 1 byte long |
|
32
|
|
|
* |
|
33
|
|
|
* This gives a linear graph where each 1-byte character requires about 150 bytes to be stored. |
|
34
|
|
|
* Given that some characters can take up to 4 bytes, we need 600 bytes per character to be safe. |
|
35
|
|
|
* Also, there is on average about 20 characters per cell (this is entirely empirical data...). |
|
36
|
|
|
* |
|
37
|
|
|
* This means that in order to store one shared string in memory, the memory amount needed is: |
|
38
|
|
|
* => 20 * 600 ≈ 12KB |
|
39
|
|
|
*/ |
|
40
|
|
|
const AMOUNT_MEMORY_NEEDED_PER_STRING_IN_KB = 12; |
|
41
|
|
|
|
|
42
|
|
|
/** |
|
43
|
|
|
* To avoid running out of memory when extracting a huge number of shared strings, they can be saved to temporary files |
|
44
|
|
|
* instead of in memory. Then, when accessing a string, the corresponding file contents will be loaded in memory |
|
45
|
|
|
* and the string will be quickly retrieved. |
|
46
|
|
|
* The performance bottleneck is not when creating these temporary files, but rather when loading their content. |
|
47
|
|
|
* Because the contents of the last loaded file stays in memory until another file needs to be loaded, it works |
|
48
|
|
|
* best when the indexes of the shared strings are sorted in the sheet data. |
|
49
|
|
|
* 10,000 was chosen because it creates small files that are fast to be loaded in memory. |
|
50
|
|
|
*/ |
|
51
|
|
|
const MAX_NUM_STRINGS_PER_TEMP_FILE = 10000; |
|
52
|
|
|
|
|
53
|
|
|
/** @var CachingStrategyFactory|null Singleton instance */ |
|
54
|
|
|
protected static $instance = null; |
|
55
|
|
|
|
|
56
|
|
|
/** |
|
57
|
|
|
* Private constructor for singleton |
|
58
|
|
|
*/ |
|
59
|
3 |
|
private function __construct() |
|
60
|
|
|
{ |
|
61
|
3 |
|
} |
|
62
|
|
|
|
|
63
|
|
|
/** |
|
64
|
|
|
* Returns the singleton instance of the factory |
|
65
|
|
|
* |
|
66
|
|
|
* @return CachingStrategyFactory |
|
67
|
|
|
*/ |
|
68
|
96 |
|
public static function getInstance() |
|
69
|
|
|
{ |
|
70
|
96 |
|
if (self::$instance === null) { |
|
71
|
3 |
|
self::$instance = new CachingStrategyFactory(); |
|
72
|
3 |
|
} |
|
73
|
|
|
|
|
74
|
96 |
|
return self::$instance; |
|
75
|
|
|
} |
|
76
|
|
|
|
|
77
|
|
|
/** |
|
78
|
|
|
* Returns the best caching strategy, given the number of unique shared strings |
|
79
|
|
|
* and the amount of memory available. |
|
80
|
|
|
* |
|
81
|
|
|
* @param int|null $sharedStringsUniqueCount Number of unique shared strings (NULL if unknown) |
|
82
|
|
|
* @param string|void $tempFolder Temporary folder where the temporary files to store shared strings will be stored |
|
83
|
|
|
* @return CachingStrategyInterface The best caching strategy |
|
84
|
|
|
*/ |
|
85
|
117 |
|
public function getBestCachingStrategy($sharedStringsUniqueCount, $tempFolder = null) |
|
86
|
|
|
{ |
|
87
|
117 |
|
if ($this->isInMemoryStrategyUsageSafe($sharedStringsUniqueCount)) { |
|
88
|
93 |
|
return new InMemoryStrategy($sharedStringsUniqueCount); |
|
89
|
|
|
} else { |
|
90
|
24 |
|
return new FileBasedStrategy($tempFolder, self::MAX_NUM_STRINGS_PER_TEMP_FILE); |
|
91
|
|
|
} |
|
92
|
|
|
} |
|
93
|
|
|
|
|
94
|
|
|
/** |
|
95
|
|
|
* Returns whether it is safe to use in-memory caching, given the number of unique shared strings |
|
96
|
|
|
* and the amount of memory available. |
|
97
|
|
|
* |
|
98
|
|
|
* @param int|null $sharedStringsUniqueCount Number of unique shared strings (NULL if unknown) |
|
99
|
|
|
* @return bool |
|
100
|
|
|
*/ |
|
101
|
117 |
|
protected function isInMemoryStrategyUsageSafe($sharedStringsUniqueCount) |
|
102
|
|
|
{ |
|
103
|
|
|
// if the number of shared strings in unknown, do not use "in memory" strategy |
|
104
|
117 |
|
if ($sharedStringsUniqueCount === null) { |
|
105
|
9 |
|
return false; |
|
106
|
|
|
} |
|
107
|
|
|
|
|
108
|
108 |
|
$memoryAvailable = $this->getMemoryLimitInKB(); |
|
109
|
|
|
|
|
110
|
108 |
|
if ($memoryAvailable === -1) { |
|
111
|
|
|
// if cannot get memory limit or if memory limit set as unlimited, don't trust and play safe |
|
112
|
12 |
|
return ($sharedStringsUniqueCount < self::MAX_NUM_STRINGS_PER_TEMP_FILE); |
|
113
|
|
|
} else { |
|
114
|
96 |
|
$memoryNeeded = $sharedStringsUniqueCount * self::AMOUNT_MEMORY_NEEDED_PER_STRING_IN_KB; |
|
115
|
96 |
|
return ($memoryAvailable > $memoryNeeded); |
|
116
|
|
|
} |
|
117
|
|
|
} |
|
118
|
|
|
|
|
119
|
|
|
/** |
|
120
|
|
|
* Returns the PHP "memory_limit" in Kilobytes |
|
121
|
|
|
* |
|
122
|
|
|
* @return float |
|
123
|
|
|
*/ |
|
124
|
123 |
|
protected function getMemoryLimitInKB() |
|
125
|
|
|
{ |
|
126
|
123 |
|
$memoryLimitFormatted = $this->getMemoryLimitFromIni(); |
|
127
|
123 |
|
$memoryLimitFormatted = strtolower(trim($memoryLimitFormatted)); |
|
128
|
|
|
|
|
129
|
|
|
// No memory limit |
|
130
|
123 |
|
if ($memoryLimitFormatted === '-1') { |
|
131
|
6 |
|
return -1; |
|
132
|
|
|
} |
|
133
|
|
|
|
|
134
|
117 |
|
if (preg_match('/(\d+)([bkmgt])b?/', $memoryLimitFormatted, $matches)) { |
|
135
|
114 |
|
$amount = intval($matches[1]); |
|
136
|
114 |
|
$unit = $matches[2]; |
|
137
|
|
|
|
|
138
|
|
|
switch ($unit) { |
|
139
|
114 |
|
case 'b': return ($amount / 1024); |
|
140
|
111 |
|
case 'k': return $amount; |
|
141
|
105 |
|
case 'm': return ($amount * 1024); |
|
142
|
99 |
|
case 'g': return ($amount * 1024 * 1024); |
|
143
|
6 |
|
case 't': return ($amount * 1024 * 1024 * 1024); |
|
144
|
|
|
} |
|
145
|
|
|
} |
|
146
|
|
|
|
|
147
|
3 |
|
return -1; |
|
148
|
|
|
} |
|
149
|
|
|
|
|
150
|
|
|
/** |
|
151
|
|
|
* Returns the formatted "memory_limit" value |
|
152
|
|
|
* |
|
153
|
|
|
* @return string |
|
154
|
|
|
*/ |
|
155
|
90 |
|
protected function getMemoryLimitFromIni() |
|
156
|
|
|
{ |
|
157
|
90 |
|
return ini_get('memory_limit'); |
|
158
|
|
|
} |
|
159
|
|
|
} |
|
160
|
|
|
|
box /
spout
