1 | <?php |
||
13 | class RobotsTxtParser implements RobotsTxtInterface |
||
14 | { |
||
15 | use DirectiveParserCommons; |
||
16 | use UrlParser; |
||
17 | |||
18 | /** |
||
19 | * Directive white list |
||
20 | */ |
||
21 | const TOP_LEVEL_DIRECTIVES = [ |
||
22 | self::DIRECTIVE_CLEAN_PARAM => 'cleanParam', |
||
23 | self::DIRECTIVE_HOST => 'host', |
||
24 | self::DIRECTIVE_SITEMAP => 'sitemap', |
||
25 | self::DIRECTIVE_USER_AGENT => 'userAgent', |
||
26 | ]; |
||
27 | |||
28 | /** |
||
29 | * Root directive handler |
||
30 | * @var RootDirectiveHandler |
||
31 | */ |
||
32 | protected $handler; |
||
33 | |||
34 | /** |
||
35 | * Current user-agent(s) |
||
36 | * @var array |
||
37 | */ |
||
38 | private $userAgents; |
||
39 | |||
40 | /** |
||
41 | * Previous directive |
||
42 | * @var string |
||
43 | */ |
||
44 | private $previousDirective; |
||
45 | |||
46 | /** |
||
47 | * Core constructor. |
||
48 | * |
||
49 | * @param string $baseUri |
||
50 | * @param string $content |
||
51 | */ |
||
52 | public function __construct($baseUri, $content) |
||
58 | |||
59 | /** |
||
60 | * Client robots.txt |
||
61 | * |
||
62 | * @param string $txt |
||
63 | * @return void |
||
64 | */ |
||
65 | private function parseTxt($txt) |
||
78 | |||
79 | /** |
||
80 | * Add line |
||
81 | * |
||
82 | * @param string $line |
||
83 | * @return bool |
||
84 | */ |
||
85 | public function add($line) |
||
103 | |||
104 | /** |
||
105 | * Render |
||
106 | * |
||
107 | * @param string $lineSeparator |
||
108 | * @return string |
||
109 | */ |
||
110 | public function render($lineSeparator = "\n") |
||
119 | |||
120 | /** |
||
121 | * Export rules |
||
122 | * |
||
123 | * @return array |
||
124 | */ |
||
125 | public function export() |
||
134 | } |
||
135 |