|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* vipnytt/RobotsTxtParser |
|
4
|
|
|
* |
|
5
|
|
|
* @link https://github.com/VIPnytt/RobotsTxtParser |
|
6
|
|
|
* @license https://github.com/VIPnytt/RobotsTxtParser/blob/master/LICENSE The MIT License (MIT) |
|
7
|
|
|
*/ |
|
8
|
|
|
|
|
9
|
|
|
namespace vipnytt\RobotsTxtParser; |
|
10
|
|
|
|
|
11
|
|
|
/** |
|
12
|
|
|
* Class Import |
|
13
|
|
|
* |
|
14
|
|
|
* @package vipnytt\RobotsTxtParser |
|
15
|
|
|
*/ |
|
16
|
|
|
class Import extends TxtClient |
|
17
|
|
|
{ |
|
18
|
|
|
/** |
|
19
|
|
|
* Root level export template |
|
20
|
|
|
*/ |
|
21
|
|
|
const TEMPLATE_ROOT = [ |
|
22
|
|
|
self::DIRECTIVE_CLEAN_PARAM => [], |
|
23
|
|
|
self::DIRECTIVE_HOST => null, |
|
24
|
|
|
self::DIRECTIVE_SITEMAP => [], |
|
25
|
|
|
self::DIRECTIVE_USER_AGENT => [] |
|
26
|
|
|
]; |
|
27
|
|
|
|
|
28
|
|
|
/** |
|
29
|
|
|
* User-agent level export template |
|
30
|
|
|
*/ |
|
31
|
|
|
const TEMPLATE_SUB = [ |
|
32
|
|
|
self::DIRECTIVE_ALLOW => [], |
|
33
|
|
|
self::DIRECTIVE_CACHE_DELAY => null, |
|
34
|
|
|
self::DIRECTIVE_COMMENT => [], |
|
35
|
|
|
self::DIRECTIVE_CRAWL_DELAY => null, |
|
36
|
|
|
self::DIRECTIVE_DISALLOW => [], |
|
37
|
|
|
self::DIRECTIVE_NO_INDEX => [], |
|
38
|
|
|
self::DIRECTIVE_REQUEST_RATE => [], |
|
39
|
|
|
self::DIRECTIVE_ROBOT_VERSION => null, |
|
40
|
|
|
self::DIRECTIVE_VISIT_TIME => [], |
|
41
|
|
|
]; |
|
42
|
|
|
|
|
43
|
|
|
/** |
|
44
|
|
|
* Array |
|
45
|
|
|
* @var array |
|
46
|
|
|
*/ |
|
47
|
|
|
private $array; |
|
48
|
|
|
|
|
49
|
|
|
/** |
|
50
|
|
|
* Import constructor. |
|
51
|
|
|
* |
|
52
|
|
|
* @param array $export |
|
53
|
|
|
* @param string $baseUri |
|
54
|
|
|
*/ |
|
55
|
|
|
public function __construct(array $export, $baseUri = 'https://example.com') |
|
56
|
|
|
{ |
|
57
|
|
|
$this->array = $this->arrayMergeRecursiveEx(self::TEMPLATE_ROOT, $export); |
|
58
|
|
|
parent::__construct($baseUri, null, implode(PHP_EOL, array_merge( |
|
59
|
|
|
$this->buildHost($this->array[self::DIRECTIVE_HOST]), |
|
60
|
|
|
$this->buildCleanParam($this->array[self::DIRECTIVE_CLEAN_PARAM]), |
|
61
|
|
|
$this->buildArray($this->array[self::DIRECTIVE_SITEMAP], self::DIRECTIVE_SITEMAP), |
|
62
|
|
|
$this->buildUserAgent($this->array[self::DIRECTIVE_USER_AGENT]) |
|
63
|
|
|
))); |
|
64
|
|
|
} |
|
65
|
|
|
|
|
66
|
|
|
/** |
|
67
|
|
|
* array_merge_recursive_ex |
|
68
|
|
|
* @link http://stackoverflow.com/a/25712428/4396537 |
|
69
|
|
|
* |
|
70
|
|
|
* @param array $array1 |
|
71
|
|
|
* @param array $array2 |
|
72
|
|
|
* @return array |
|
73
|
|
|
*/ |
|
74
|
|
|
private function arrayMergeRecursiveEx(array $array1, array &$array2) |
|
75
|
|
|
{ |
|
76
|
|
|
foreach ($array2 as $key => &$value) { |
|
77
|
|
|
if (is_array($value) && |
|
78
|
|
|
isset($array1[$key]) && |
|
79
|
|
|
is_array($array1[$key]) |
|
80
|
|
|
) { |
|
81
|
|
|
$array1[$key] = $this->arrayMergeRecursiveEx($array1[$key], $value); |
|
82
|
|
|
} elseif (!is_int($key)) { |
|
83
|
|
|
$array1[$key] = $value; |
|
84
|
|
|
} elseif (!in_array($value, $array1)) { |
|
85
|
|
|
$array1[] = $value; |
|
86
|
|
|
} |
|
87
|
|
|
} |
|
88
|
|
|
return $array1; |
|
89
|
|
|
} |
|
90
|
|
|
|
|
91
|
|
|
/** |
|
92
|
|
|
* Host |
|
93
|
|
|
* |
|
94
|
|
|
* @param string[]|string|null $array |
|
95
|
|
|
* @return string[] |
|
96
|
|
|
*/ |
|
97
|
|
|
private function buildHost($array) |
|
98
|
|
|
{ |
|
99
|
|
|
if (!is_array($array)) { |
|
100
|
|
|
$array = [$array]; |
|
101
|
|
|
} |
|
102
|
|
|
return preg_filter('/^/', self::DIRECTIVE_HOST . ':', $array); |
|
103
|
|
|
} |
|
104
|
|
|
|
|
105
|
|
|
/** |
|
106
|
|
|
* Clean-param |
|
107
|
|
|
* |
|
108
|
|
|
* @param string[][] $array |
|
109
|
|
|
* @return string[] |
|
110
|
|
|
*/ |
|
111
|
|
|
private function buildCleanParam($array) |
|
112
|
|
|
{ |
|
113
|
|
|
$result = []; |
|
114
|
|
|
foreach ($array as $param => $paths) { |
|
115
|
|
|
foreach ($paths as $path) { |
|
116
|
|
|
$result[] = self::DIRECTIVE_CLEAN_PARAM . ':' . $param . ' ' . $path; |
|
117
|
|
|
} |
|
118
|
|
|
} |
|
119
|
|
|
return $result; |
|
120
|
|
|
} |
|
121
|
|
|
|
|
122
|
|
|
/** |
|
123
|
|
|
* Comment | Sitemap |
|
124
|
|
|
* |
|
125
|
|
|
* @param string[] $array |
|
126
|
|
|
* @param string $directive |
|
127
|
|
|
* @return string[] |
|
128
|
|
|
*/ |
|
129
|
|
|
private function buildArray($array, $directive) |
|
130
|
|
|
{ |
|
131
|
|
|
return preg_filter('/^/', $directive . ':', $array); |
|
132
|
|
|
} |
|
133
|
|
|
|
|
134
|
|
|
/** |
|
135
|
|
|
* User-agent |
|
136
|
|
|
* |
|
137
|
|
|
* @param array $array |
|
138
|
|
|
* @return string[] |
|
139
|
|
|
*/ |
|
140
|
|
|
private function buildUserAgent($array) |
|
141
|
|
|
{ |
|
142
|
|
|
$result = []; |
|
143
|
|
|
foreach ($array as $userAgent => $rules) { |
|
144
|
|
|
$rules = $this->arrayMergeRecursiveEx(self::TEMPLATE_SUB, $rules); |
|
145
|
|
|
$result = array_merge( |
|
146
|
|
|
$result, |
|
147
|
|
|
[self::DIRECTIVE_USER_AGENT . ':' . $userAgent], |
|
148
|
|
|
$this->buildString($rules[self::DIRECTIVE_ROBOT_VERSION], self::DIRECTIVE_ROBOT_VERSION), |
|
149
|
|
|
$this->buildVisitTime($rules[self::DIRECTIVE_VISIT_TIME]), |
|
150
|
|
|
$this->buildArray($rules[self::DIRECTIVE_NO_INDEX], self::DIRECTIVE_NO_INDEX), |
|
151
|
|
|
$this->buildArray($rules[self::DIRECTIVE_DISALLOW], self::DIRECTIVE_DISALLOW), |
|
152
|
|
|
$this->buildArray($rules[self::DIRECTIVE_ALLOW], self::DIRECTIVE_ALLOW), |
|
153
|
|
|
$this->buildString($rules[self::DIRECTIVE_CRAWL_DELAY], self::DIRECTIVE_CRAWL_DELAY), |
|
154
|
|
|
$this->buildString($rules[self::DIRECTIVE_CACHE_DELAY], self::DIRECTIVE_CACHE_DELAY), |
|
155
|
|
|
$this->buildRequestRate($rules[self::DIRECTIVE_REQUEST_RATE]), |
|
156
|
|
|
$this->buildArray($rules[self::DIRECTIVE_COMMENT], self::DIRECTIVE_COMMENT) |
|
157
|
|
|
); |
|
158
|
|
|
} |
|
159
|
|
|
return $result; |
|
160
|
|
|
} |
|
161
|
|
|
|
|
162
|
|
|
/** |
|
163
|
|
|
* Cache-delay | Comment | Crawl-delay | Robot-version |
|
164
|
|
|
* |
|
165
|
|
|
* @param float|int|string|null $value |
|
166
|
|
|
* @param string $directive |
|
167
|
|
|
* @return string[] |
|
168
|
|
|
*/ |
|
169
|
|
|
private function buildString($value, $directive) |
|
170
|
|
|
{ |
|
171
|
|
|
return [$directive . ':' . $value]; |
|
172
|
|
|
} |
|
173
|
|
|
|
|
174
|
|
|
/** |
|
175
|
|
|
* Visit-time |
|
176
|
|
|
* |
|
177
|
|
|
* @param int[]|string[] $array |
|
178
|
|
|
* @return string[] |
|
179
|
|
|
*/ |
|
180
|
|
|
private function buildVisitTime($array) |
|
181
|
|
|
{ |
|
182
|
|
|
$result = []; |
|
183
|
|
|
foreach ($array as $pair) { |
|
184
|
|
|
$result[] = self::DIRECTIVE_VISIT_TIME . ':' . $pair['from'] . '-' . $pair['to']; |
|
185
|
|
|
} |
|
186
|
|
|
return $result; |
|
187
|
|
|
} |
|
188
|
|
|
|
|
189
|
|
|
/** |
|
190
|
|
|
* Request-rate |
|
191
|
|
|
* |
|
192
|
|
|
* @param array $array |
|
193
|
|
|
* @return string[] |
|
194
|
|
|
*/ |
|
195
|
|
|
private function buildRequestRate($array) |
|
196
|
|
|
{ |
|
197
|
|
|
$result = []; |
|
198
|
|
|
foreach ($array as $pair) { |
|
199
|
|
|
$string = self::DIRECTIVE_REQUEST_RATE . ':1/' . $pair['rate'] . 's'; |
|
200
|
|
|
if (isset($pair['from']) && |
|
201
|
|
|
isset($pair['to']) |
|
202
|
|
|
) { |
|
203
|
|
|
$string .= ' ' . $pair['from'] . '-' . $pair['to']; |
|
204
|
|
|
} |
|
205
|
|
|
$result[] = $string; |
|
206
|
|
|
} |
|
207
|
|
|
return $result; |
|
208
|
|
|
} |
|
209
|
|
|
|
|
210
|
|
|
/** |
|
211
|
|
|
* Get difference |
|
212
|
|
|
* |
|
213
|
|
|
* @return array |
|
214
|
|
|
*/ |
|
215
|
|
|
public function getIgnoredImportData() |
|
216
|
|
|
{ |
|
217
|
|
|
$pair = [ |
|
218
|
|
|
'source' => $this->array, |
|
219
|
|
|
'parsed' => $this->export(), |
|
220
|
|
|
]; |
|
221
|
|
|
foreach ($pair as &$array) { |
|
222
|
|
|
$array = $this->arrayFilterRecursive($array); |
|
223
|
|
|
array_multisort($array); |
|
224
|
|
|
$this->kSortRecursive($array); |
|
225
|
|
|
} |
|
226
|
|
|
return $this->arrayDiffAssocRecursive($pair['source'], $pair['parsed']); |
|
227
|
|
|
} |
|
228
|
|
|
|
|
229
|
|
|
/** |
|
230
|
|
|
* array_filter_recursive |
|
231
|
|
|
* @link http://php.net/manual/en/function.array-filter.php#87581 |
|
232
|
|
|
* |
|
233
|
|
|
* @param array $array |
|
234
|
|
|
* @return array |
|
235
|
|
|
*/ |
|
236
|
|
|
private function arrayFilterRecursive(array &$array) |
|
237
|
|
|
{ |
|
238
|
|
|
foreach ($array as $key => &$item) { |
|
239
|
|
|
is_array($item) && $array[$key] = $this->arrayFilterRecursive($item); |
|
240
|
|
|
if ($array[$key] === null || |
|
241
|
|
|
$array[$key] === [] |
|
242
|
|
|
) { |
|
243
|
|
|
unset($array[$key]); |
|
244
|
|
|
} |
|
245
|
|
|
} |
|
246
|
|
|
return $array; |
|
247
|
|
|
} |
|
248
|
|
|
|
|
249
|
|
|
/** |
|
250
|
|
|
* ksort_recursive |
|
251
|
|
|
* @link http://stackoverflow.com/a/2543447/4396537 |
|
252
|
|
|
* |
|
253
|
|
|
* @param array $array |
|
254
|
|
|
* @return bool |
|
255
|
|
|
*/ |
|
256
|
|
|
private function kSortRecursive(array &$array) |
|
257
|
|
|
{ |
|
258
|
|
|
foreach ($array as &$current) { |
|
259
|
|
|
if (is_array($current) && |
|
260
|
|
|
$this->kSortRecursive($current) === false |
|
261
|
|
|
) { |
|
262
|
|
|
return false; |
|
263
|
|
|
} |
|
264
|
|
|
} |
|
265
|
|
|
return ksort($array); |
|
266
|
|
|
} |
|
267
|
|
|
|
|
268
|
|
|
/** |
|
269
|
|
|
* array_diff_assoc_recursive |
|
270
|
|
|
* @link http://php.net/manual/en/function.array-diff-assoc.php#111675 |
|
271
|
|
|
* |
|
272
|
|
|
* @param array $array1 |
|
273
|
|
|
* @param array $array2 |
|
274
|
|
|
* @return array |
|
275
|
|
|
*/ |
|
276
|
|
|
private function arrayDiffAssocRecursive(array &$array1, array &$array2) |
|
277
|
|
|
{ |
|
278
|
|
|
$difference = []; |
|
279
|
|
|
foreach ($array1 as $key => &$value) { |
|
280
|
|
|
if (is_array($value)) { |
|
281
|
|
|
if (!isset($array2[$key]) || !is_array($array2[$key])) { |
|
282
|
|
|
$difference[$key] = $value; |
|
283
|
|
|
} elseif (!empty($newDiff = $this->arrayDiffAssocRecursive($value, $array2[$key]))) { |
|
284
|
|
|
$difference[$key] = $newDiff; |
|
285
|
|
|
} |
|
286
|
|
|
} elseif (!array_key_exists($key, $array2) || $array2[$key] !== $value) { |
|
287
|
|
|
$difference[$key] = $value; |
|
288
|
|
|
} |
|
289
|
|
|
} |
|
290
|
|
|
return $difference; |
|
291
|
|
|
} |
|
292
|
|
|
} |
|
293
|
|
|
|