1: <?php
 2: 
 3: /**
 4:  * This file is part of the Nette Framework (https://nette.org)
 5:  * Copyright (c) 2004 David Grudl (https://davidgrudl.com)
 6:  */
 7: 
 8: namespace Nette\Tokenizer;
 9: 
10: 
11: /**
12:  * Simple lexical analyser.
13:  */
14: class Tokenizer
15: {
16:     /** @var string */
17:     private $re;
18: 
19:     /** @var array */
20:     private $types;
21: 
22: 
23:     /**
24:      * @param  array of [(int|string) token type => (string) pattern]
25:      * @param  string  regular expression flags
26:      */
27:     public function __construct(array $patterns, $flags = '')
28:     {
29:         $this->re = '~(' . implode(')|(', $patterns) . ')~A' . $flags;
30:         $this->types = array_keys($patterns);
31:     }
32: 
33: 
34:     /**
35:      * Tokenizes string.
36:      * @param  string
37:      * @return Stream
38:      * @throws Exception
39:      */
40:     public function tokenize($input)
41:     {
42:         preg_match_all($this->re, $input, $tokens, PREG_SET_ORDER);
43:         $len = 0;
44:         $count = count($this->types);
45:         foreach ($tokens as &$token) {
46:             $type = null;
47:             for ($i = 1; $i <= $count; $i++) {
48:                 if (!isset($token[$i])) {
49:                     break;
50:                 } elseif ($token[$i] != null) {
51:                     $type = $this->types[$i - 1];
52:                     break;
53:                 }
54:             }
55:             $token = new Token($token[0], $type, $len);
56:             $len += strlen($token->value);
57:         }
58:         if ($len !== strlen($input)) {
59:             list($line, $col) = $this->getCoordinates($input, $len);
60:             $token = str_replace("\n", '\n', substr($input, $len, 10));
61:             throw new Exception("Unexpected '$token' on line $line, column $col.");
62:         }
63:         return new Stream($tokens);
64:     }
65: 
66: 
67:     /**
68:      * Returns position of token in input string.
69:      * @param  string
70:      * @param  int
71:      * @return array of [line, column]
72:      */
73:     public static function getCoordinates($text, $offset)
74:     {
75:         $text = substr($text, 0, $offset);
76:         return [substr_count($text, "\n") + 1, $offset - strrpos("\n" . $text, "\n") + 1];
77:     }
78: }
79: