1
|
|
|
<?php |
2
|
|
|
/** |
3
|
|
|
* |
4
|
|
|
* (c) Ruben Dorado <[email protected]> |
5
|
|
|
* |
6
|
|
|
* For the full copyright and license information, please view the LICENSE |
7
|
|
|
* file that was distributed with this source code. |
8
|
|
|
*/ |
9
|
|
|
namespace SiteAnalyzer; |
10
|
|
|
|
11
|
|
|
use Exception; |
12
|
|
|
|
13
|
|
|
/** |
14
|
|
|
* class SiteAnalyzer |
15
|
|
|
* |
16
|
|
|
* @package SiteAnalyzer |
17
|
|
|
* @author Ruben Dorado <[email protected]> |
18
|
|
|
* @copyright 2018 Ruben Dorado |
19
|
|
|
* @license http://www.opensource.org/licenses/MIT The MIT License |
20
|
|
|
*/ |
21
|
|
|
class SiteAnalyzer |
22
|
|
|
{ |
23
|
|
|
|
24
|
|
|
/* |
25
|
|
|
* @param |
26
|
|
|
*/ |
27
|
|
|
public static function count($options = []) |
28
|
|
|
{ |
29
|
|
|
$config = SiteAnalyzer::loadConfig(array_key_exists('pdo', $options)); |
30
|
|
|
|
31
|
|
|
if (array_key_exists('pdo', $options)) { |
32
|
|
|
$pdo = $options['pdo']; |
33
|
|
|
} else { |
34
|
|
|
$pdo = Persistence::getPDO($config); |
35
|
|
|
} |
36
|
|
|
|
37
|
|
|
try { |
38
|
|
|
return Persistence::updateCount($pdo, $config, $options); |
39
|
|
|
} catch (Exception $e) { |
40
|
|
|
try { |
41
|
|
|
Persistence::crateDatabase($pdo, $config); |
42
|
|
|
return Persistence::updateCount($pdo, $config, $options); |
43
|
|
|
} catch (Exception $e) { |
44
|
|
|
throw new Exception("Site Analyzer could connect to the database.".$e->getMessage()); |
45
|
|
|
}; |
46
|
|
|
|
47
|
|
|
}; |
48
|
|
|
|
49
|
|
|
} |
50
|
|
|
|
51
|
|
|
/* |
52
|
|
|
* @param $format string, one of [php-array, xml, json, txt-csv] |
53
|
|
|
*/ |
54
|
|
|
public static function resetDatabase($options = []) |
55
|
|
|
{ |
56
|
|
|
$config = SiteAnalyzer::loadConfig(); |
57
|
|
|
$pdo = SiteAnalyzer::getPDO($config, $options); |
58
|
|
|
|
59
|
|
|
Persistence::deleteDatabase($pdo, $config); |
60
|
|
|
Persistence::crateDatabase($pdo, $config); |
61
|
|
|
} |
62
|
|
|
|
63
|
|
|
/* |
64
|
|
|
* @param $format string, one of [php-array, xml, json, txt-csv] |
65
|
|
|
*/ |
66
|
|
|
public static function loadConfig($pdoProvided = FALSE) |
67
|
|
|
{ |
68
|
|
|
try { |
69
|
|
|
$config = new Configuration(get_cfg_var("site_analyzer_config"), $pdoProvided); |
70
|
|
|
} catch (Exception $e) { |
71
|
|
|
try { |
72
|
|
|
$config = new Configuration("site-analyzer.ini", $pdoProvided); |
73
|
|
|
} catch (Exception $e) { |
74
|
|
|
throw new Exception("Configuration file not found. Provide a configuration file in the following directory: ".getcwd()." or configure it on the php.ini file by setting the variable 'site_analyzer_config = filename'"); |
75
|
|
|
} |
76
|
|
|
} |
77
|
|
|
return $config; |
78
|
|
|
} |
79
|
|
|
|
80
|
|
|
/* |
81
|
|
|
* @param |
82
|
|
|
*/ |
83
|
|
|
public static function getPDO($config, $options) |
84
|
|
|
{ |
85
|
|
|
if (array_key_exists("pdo",$options)) { |
86
|
|
|
return $options["pdo"]; |
87
|
|
|
} |
88
|
|
|
return Persistence::getPDO($config); |
89
|
|
|
} |
90
|
|
|
|
91
|
|
|
/* |
92
|
|
|
* @param $format string, one of [php-array, xml, json, txt-csv] |
93
|
|
|
*/ |
94
|
|
|
public static function getStats($options = []) |
95
|
|
|
{ |
96
|
|
|
$config = SiteAnalyzer::loadConfig(); |
97
|
|
|
$pdo = SiteAnalyzer::getPDO($config, $options); |
98
|
|
|
|
99
|
|
|
$data = Persistence::getCounts($pdo, $config); |
100
|
|
|
return $data; |
101
|
|
|
} |
102
|
|
|
|
103
|
|
|
/* |
104
|
|
|
* @param |
105
|
|
|
*/ |
106
|
|
|
public static function groupHitsByTime($options = []) |
107
|
|
|
{ |
108
|
|
|
$config = SiteAnalyzer::loadConfig(); |
109
|
|
|
$pdo = SiteAnalyzer::getPDO($config, $options); |
110
|
|
|
|
111
|
|
|
$data = OptionsDAO::getHitsWithOptions($pdo, $config); |
112
|
|
|
$resp = []; |
113
|
|
|
foreach ($data as $row) { |
114
|
|
|
$tmp = [$row[0]]; |
115
|
|
|
$tmp = array_merge($tmp, getdate($row[1])); |
116
|
|
|
$resp[] = $tmp; |
117
|
|
|
} |
118
|
|
|
return $resp; |
119
|
|
|
} |
120
|
|
|
|
121
|
|
|
/* |
122
|
|
|
* @param |
123
|
|
|
*/ |
124
|
|
|
public static function groupHitsByUser($options = []) |
125
|
|
|
{ |
126
|
|
|
$config = SiteAnalyzer::loadConfig(); |
127
|
|
|
$pdo = SiteAnalyzer::getPDO($config, $options); |
128
|
|
|
|
129
|
|
|
$data = OptionsDAO::getHitsWithOptions($pdo, $config); |
130
|
|
|
$count = []; |
131
|
|
|
foreach ($data as $row) { |
132
|
|
|
if (array_key_exists($row[2], $count)) { |
133
|
|
|
$count[$row[2]]++; |
134
|
|
|
} else { |
135
|
|
|
$count[$row[2]] = 1; |
136
|
|
|
} |
137
|
|
|
} |
138
|
|
|
|
139
|
|
|
$resp = []; |
140
|
|
|
foreach ($count as $user => $count) { |
141
|
|
|
$resp[] = [$user, $count]; |
142
|
|
|
} |
143
|
|
|
return $resp; |
144
|
|
|
} |
145
|
|
|
|
146
|
|
|
/* |
147
|
|
|
* @param $format string, one of [php-array, xml, json, txt-csv] |
148
|
|
|
*/ |
149
|
|
|
public static function transform($data, $format) |
150
|
|
|
{ |
151
|
|
|
if ($format=="html") { |
152
|
|
|
$resp = "<table style='border-collapse: collapse;border: 1px solid black;'>"; |
153
|
|
|
foreach ($data as $row) { |
154
|
|
|
$resp .= "<tr style='border: 1px solid black;'>"; |
155
|
|
|
foreach ($row as $cell) { |
156
|
|
|
$resp .= "<td style='border: 1px solid black;'>$cell</td>"; |
157
|
|
|
} |
158
|
|
|
$resp .= "</tr>"; |
159
|
|
|
} |
160
|
|
|
return $resp."</table>"; |
161
|
|
|
} |
162
|
|
|
return $data; |
163
|
|
|
} |
164
|
|
|
|
165
|
|
|
/* |
166
|
|
|
* @param |
167
|
|
|
*/ |
168
|
|
|
public static function getTransitionMatrix($options = []) |
169
|
|
|
{ |
170
|
|
|
$config = SiteAnalyzer::loadConfig(); |
171
|
|
|
$pdo = SiteAnalyzer::getPDO($config, $options); |
172
|
|
|
|
173
|
|
|
$targetCounts = Persistence::findByFrom($pdo, $config); |
174
|
|
|
$data = Matrix::submatrix($targetCounts, [1, 0, 2]); |
175
|
|
|
$data = Matrix::toSquareMatrix($data, 0, 1, 2); |
176
|
|
|
$labels = Matrix::arrayToMatrix($data["labels"]); |
177
|
|
|
$data = $data["data"]; |
178
|
|
|
$data = Matrix::toBinary($data); |
179
|
|
|
|
180
|
|
|
if (array_key_exists("level", $options)) { |
181
|
|
|
$level = $options["level"]; |
182
|
|
|
} |
183
|
|
|
else { |
184
|
|
|
$level = count($labels) - 1; |
185
|
|
|
} |
186
|
|
|
|
187
|
|
|
$result = $data; |
188
|
|
|
for ($i=1;$i<$level;$i++) { |
189
|
|
|
$tmp = Matrix::multiply($result, $data); |
190
|
|
|
$result = Matrix::sum($result, $tmp); |
191
|
|
|
} |
192
|
|
|
|
193
|
|
|
$result = Matrix::toBinary($result); |
194
|
|
|
return ["labels"=>$labels, "data"=>$result]; |
195
|
|
|
} |
196
|
|
|
|
197
|
|
|
/* |
198
|
|
|
* @param |
199
|
|
|
*/ |
200
|
|
|
public static function getPathCountMatrix($options = []) |
201
|
|
|
{ |
202
|
|
|
|
203
|
|
|
$config = SiteAnalyzer::loadConfig(); |
204
|
|
|
$pdo = SiteAnalyzer::getPDO($config, $options); |
205
|
|
|
|
206
|
|
|
$targetCounts = Persistence::findByFrom($pdo, $config); |
207
|
|
|
$data = Matrix::submatrix($targetCounts, [1, 0, 2]); |
208
|
|
|
$data = Matrix::toSquareMatrix($data, 0, 1, 2); |
209
|
|
|
$labels = Matrix::arrayToMatrix($data["labels"]); |
210
|
|
|
$data = $data["data"]; |
211
|
|
|
$data = Matrix::toBinary($data); |
212
|
|
|
|
213
|
|
|
if (array_key_exists("level", $options)) { |
214
|
|
|
$level = $options["level"]; |
215
|
|
|
} |
216
|
|
|
else { |
217
|
|
|
$level = count($labels) - 1; |
218
|
|
|
} |
219
|
|
|
|
220
|
|
|
$result = $data; |
221
|
|
|
for ($i=1;$i<$level;$i++) { |
222
|
|
|
$tmp = Matrix::multiply($result, $data); |
223
|
|
|
$result = Matrix::sum($result, $tmp); |
224
|
|
|
} |
225
|
|
|
|
226
|
|
|
return ["labels"=>$labels, "data"=>$result]; |
227
|
|
|
|
228
|
|
|
} |
229
|
|
|
|
230
|
|
|
/* |
231
|
|
|
* @param |
232
|
|
|
*/ |
233
|
|
|
public static function getTransitionCounts($options = []) |
234
|
|
|
{ |
235
|
|
|
$config = SiteAnalyzer::loadConfig(); |
236
|
|
|
$pdo = SiteAnalyzer::getPDO($config, $options); |
237
|
|
|
|
238
|
|
|
$targetCounts = Persistence::findByFrom($pdo, $config); |
239
|
|
|
|
240
|
|
|
$data = Matrix::submatrix($targetCounts, [1, 0, 2]); |
241
|
|
|
$data = Matrix::toSquareMatrix($data, 0, 1, 2); |
242
|
|
|
$labels = Matrix::arrayToMatrix($data["labels"]); |
243
|
|
|
|
244
|
|
|
return ["data"=>$data["data"], "labels"=>$labels]; |
245
|
|
|
} |
246
|
|
|
|
247
|
|
|
/* |
248
|
|
|
* @param |
249
|
|
|
*/ |
250
|
|
|
public static function performABTest($tests, $options = []) |
251
|
|
|
{ |
252
|
|
|
$config = SiteAnalyzer::loadConfig(); |
253
|
|
|
$pdo = SiteAnalyzer::getPDO($config, $options); |
254
|
|
|
|
255
|
|
|
$testCounts = Persistence::getCountsFromTest($pdo, $config, $tests); |
256
|
|
|
$targetCounts = Persistence::getFromByTest($pdo, $config, $tests); |
257
|
|
|
$result = Statistics::ABtest($testCounts, $targetCounts); |
258
|
|
|
|
259
|
|
|
return $result; |
260
|
|
|
} |
261
|
|
|
|
262
|
|
|
/* |
263
|
|
|
* @param |
264
|
|
|
*/ |
265
|
|
|
public static function findVisitTimeProfiles($nprofiles, $options = []) |
266
|
|
|
{ |
267
|
|
|
$config = SiteAnalyzer::loadConfig(); |
268
|
|
|
$pdo = SiteAnalyzer::getPDO($config, $options); |
269
|
|
|
$table = OptionsDAO::getHitsWithOptions($pdo, $config); |
270
|
|
|
|
271
|
|
|
|
272
|
|
|
$data = []; |
273
|
|
|
foreach ($table as $row) { |
274
|
|
|
$tmp = getdate($row[1]); |
275
|
|
|
$data[] = [$tmp['weekday'], $tmp['hours']]; |
276
|
|
|
} |
277
|
|
|
|
278
|
|
|
|
279
|
|
|
$cdata = new CategoricalDataset($data); |
280
|
|
|
$cdata->setEncodedFeatures([0, 1]); |
281
|
|
|
$tdata = $cdata->encode(); |
282
|
|
|
$kmResult = ML::kmeans($tdata, $nprofiles); |
283
|
|
|
$resp = []; |
284
|
|
|
$resp["clusters"] = $kmResult["clusters"]; |
285
|
|
|
$resp["centroids"] = $kmResult["centroids"]; |
286
|
|
|
$resp["labels"] = $cdata->getLabelsAsArray(); |
287
|
|
|
return $resp; |
288
|
|
|
} |
289
|
|
|
|
290
|
|
|
/* |
291
|
|
|
* @param |
292
|
|
|
*/ |
293
|
|
|
public static function findUserProfiles($nprofiles, $options = []) |
294
|
|
|
{ |
295
|
|
|
$config = SiteAnalyzer::loadConfig(); |
296
|
|
|
$pdo = SiteAnalyzer::getPDO($config, $options); |
297
|
|
|
$table = OptionsDAO::getHitsWithOptions($pdo, $config); |
298
|
|
|
|
299
|
|
|
$data = []; |
300
|
|
|
foreach ($table as $row) { |
301
|
|
|
$tmp = getdate($row[1]); |
302
|
|
|
$data[] = [$row[2], $tmp['weekday'], $tmp['hours']]; |
303
|
|
|
} |
304
|
|
|
|
305
|
|
|
|
306
|
|
|
$cdata = new CategoricalDataset($data); |
307
|
|
|
$cdata->setEncodedFeatures([0, 1, 2]); |
308
|
|
|
$tdata = $cdata->encode(); |
309
|
|
|
$kmResult = ML::kmeans($tdata, $nprofiles); |
310
|
|
|
$resp = []; |
311
|
|
|
$resp["clusters"] = $kmResult["clusters"]; |
312
|
|
|
$resp["centroids"] = $kmResult["centroids"]; |
313
|
|
|
$resp["labels"] = $cdata->getLabelsAsArray(); |
314
|
|
|
return $resp; |
315
|
|
|
} |
316
|
|
|
|
317
|
|
|
} |
318
|
|
|
|
319
|
|
|
|