|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* This file is part of the beebot package. |
|
4
|
|
|
* For the full copyright and license information, please view the LICENSE |
|
5
|
|
|
* file that was distributed with this source code. |
|
6
|
|
|
* |
|
7
|
|
|
* @copyright Bee4 2016 |
|
8
|
|
|
* @author Stephane HULARD <[email protected]> |
|
9
|
|
|
*/ |
|
10
|
|
|
|
|
11
|
|
|
namespace Bee4\RobotsTxt\Test\Units; |
|
12
|
|
|
|
|
13
|
|
|
use mageekguy\atoum; |
|
14
|
|
|
use Bee4\RobotsTxt\Parser as SUT; |
|
15
|
|
|
use Bee4\RobotsTxt as LUT; |
|
16
|
|
|
|
|
17
|
|
|
/** |
|
18
|
|
|
* Parser unit test |
|
19
|
|
|
*/ |
|
20
|
|
|
class Parser extends atoum |
|
21
|
|
|
{ |
|
22
|
|
|
public function testEmptyStringContent() |
|
23
|
|
|
{ |
|
24
|
|
|
$this |
|
25
|
|
|
->given($rules = SUT::parse("")) |
|
26
|
|
|
->when($rule = $rules->get(LUT\Rules::DEFAULT_UA)) |
|
27
|
|
|
->then |
|
28
|
|
|
->object($rules) |
|
29
|
|
|
->isInstanceOf('Bee4\RobotsTxt\Rules') |
|
30
|
|
|
->object($rule) |
|
31
|
|
|
->isInstanceOf('Bee4\RobotsTxt\Rule'); |
|
32
|
|
|
} |
|
33
|
|
|
|
|
34
|
|
|
public function testEmptyContentInstance() |
|
35
|
|
|
{ |
|
36
|
|
|
$this |
|
37
|
|
|
->given( |
|
38
|
|
|
$content = new LUT\Content(""), |
|
39
|
|
|
$rules = SUT::parse($content) |
|
40
|
|
|
) |
|
41
|
|
|
->when($rule = $rules->get(LUT\Rules::DEFAULT_UA)) |
|
42
|
|
|
->then |
|
43
|
|
|
->object($rules) |
|
44
|
|
|
->isInstanceOf('Bee4\RobotsTxt\Rules') |
|
45
|
|
|
->object($rule) |
|
46
|
|
|
->isInstanceOf('Bee4\RobotsTxt\Rule'); |
|
47
|
|
|
} |
|
48
|
|
|
|
|
49
|
|
|
public function testInvalidContentException() |
|
50
|
|
|
{ |
|
51
|
|
|
$this->exception(function () { |
|
52
|
|
|
SUT::parse(new \stdClass); |
|
53
|
|
|
}) |
|
54
|
|
|
->isInstanceOf('Bee4\RobotsTxt\Exception\InvalidContentException'); |
|
55
|
|
|
} |
|
56
|
|
|
|
|
57
|
|
|
public function testDuplicateRuleException() |
|
58
|
|
|
{ |
|
59
|
|
|
$robot = <<<ROBOTS |
|
60
|
|
|
User-Agent: * |
|
61
|
|
|
Allow: / |
|
62
|
|
|
|
|
63
|
|
|
User-Agent: * |
|
64
|
|
|
Disallow: /toto |
|
65
|
|
|
ROBOTS; |
|
66
|
|
|
|
|
67
|
|
|
$this |
|
68
|
|
|
->given($content = new LUT\Content($robot)) |
|
69
|
|
|
->exception(function () use ($content) { |
|
70
|
|
|
SUT::parse($content); |
|
71
|
|
|
}) |
|
72
|
|
|
->isInstanceOf('Bee4\RobotsTxt\Exception\DuplicateRuleException'); |
|
73
|
|
|
} |
|
74
|
|
|
|
|
75
|
|
|
public function testParse() |
|
76
|
|
|
{ |
|
77
|
|
|
$robot = <<<ROBOTS |
|
78
|
|
|
User-Agent: * |
|
79
|
|
|
Allow: / |
|
80
|
|
|
|
|
81
|
|
|
User-Agent: Google |
|
82
|
|
|
Disallow: /toto |
|
83
|
|
|
ROBOTS; |
|
84
|
|
|
|
|
85
|
|
|
$this |
|
86
|
|
|
->given($rules = SUT::parse($robot)) |
|
87
|
|
|
->then |
|
88
|
|
|
->sizeOf($rules) |
|
89
|
|
|
->isEqualTo(2) |
|
90
|
|
|
->boolean($rules->match('*', '/page')) |
|
91
|
|
|
->isTrue() |
|
92
|
|
|
->boolean($rules->match('Google', '/page')) |
|
93
|
|
|
->isTrue() |
|
94
|
|
|
->boolean($rules->match('Google', '/toto/page')) |
|
95
|
|
|
->isFalse(); |
|
96
|
|
|
} |
|
97
|
|
|
} |
|
98
|
|
|
|
bee4 /
robots.txt
