<?php

namespace PHPStan\PhpDocParser\Lexer;

use function implode;
use function preg_match_all;

use const PREG_SET_ORDER;

/**
 * Implementation based on Nette Tokenizer (New BSD License; https://github.com/nette/tokenizer)
 */
class Lexer
{
    public const TOKEN_REFERENCE = 0;
    public const TOKEN_UNION = 1;
    public const TOKEN_INTERSECTION = 2;
    public const TOKEN_NULLABLE = 3;
    public const TOKEN_OPEN_PARENTHESES = 4;
    public const TOKEN_CLOSE_PARENTHESES = 5;
    public const TOKEN_OPEN_ANGLE_BRACKET = 6;
    public const TOKEN_CLOSE_ANGLE_BRACKET = 7;
    public const TOKEN_OPEN_SQUARE_BRACKET = 8;
    public const TOKEN_CLOSE_SQUARE_BRACKET = 9;
    public const TOKEN_COMMA = 10;
    public const TOKEN_VARIADIC = 11;
    public const TOKEN_DOUBLE_COLON = 12;
    public const TOKEN_DOUBLE_ARROW = 13;
    public const TOKEN_EQUAL = 14;
    public const TOKEN_OPEN_PHPDOC = 15;
    public const TOKEN_CLOSE_PHPDOC = 16;
    public const TOKEN_PHPDOC_TAG = 17;
    public const TOKEN_FLOAT = 18;
    public const TOKEN_INTEGER = 19;
    public const TOKEN_SINGLE_QUOTED_STRING = 20;
    public const TOKEN_DOUBLE_QUOTED_STRING = 21;
    public const TOKEN_IDENTIFIER = 22;
    public const TOKEN_THIS_VARIABLE = 23;
    public const TOKEN_VARIABLE = 24;
    public const TOKEN_HORIZONTAL_WS = 25;
    public const TOKEN_PHPDOC_EOL = 26;
    public const TOKEN_OTHER = 27;
    public const TOKEN_END = 28;
    public const TOKEN_COLON = 29;
    public const TOKEN_WILDCARD = 30;
    public const TOKEN_OPEN_CURLY_BRACKET = 31;
    public const TOKEN_CLOSE_CURLY_BRACKET = 32;
    public const TOKEN_NEGATED = 33;
    public const TOKEN_ARROW = 34;

    public const TOKEN_LABELS = [
        self::TOKEN_REFERENCE => '\'&\'',
        self::TOKEN_UNION => '\'|\'',
        self::TOKEN_INTERSECTION => '\'&\'',
        self::TOKEN_NULLABLE => '\'?\'',
        self::TOKEN_NEGATED => '\'!\'',
        self::TOKEN_OPEN_PARENTHESES => '\'(\'',
        self::TOKEN_CLOSE_PARENTHESES => '\')\'',
        self::TOKEN_OPEN_ANGLE_BRACKET => '\'<\'',
        self::TOKEN_CLOSE_ANGLE_BRACKET => '\'>\'',
        self::TOKEN_OPEN_SQUARE_BRACKET => '\'[\'',
        self::TOKEN_CLOSE_SQUARE_BRACKET => '\']\'',
        self::TOKEN_OPEN_CURLY_BRACKET => '\'{\'',
        self::TOKEN_CLOSE_CURLY_BRACKET => '\'}\'',
        self::TOKEN_COMMA => '\',\'',
        self::TOKEN_COLON => '\':\'',
        self::TOKEN_VARIADIC => '\'...\'',
        self::TOKEN_DOUBLE_COLON => '\'::\'',
        self::TOKEN_DOUBLE_ARROW => '\'=>\'',
        self::TOKEN_ARROW => '\'->\'',
        self::TOKEN_EQUAL => '\'=\'',
        self::TOKEN_OPEN_PHPDOC => '\'/**\'',
        self::TOKEN_CLOSE_PHPDOC => '\'*/\'',
        self::TOKEN_PHPDOC_TAG => 'TOKEN_PHPDOC_TAG',
        self::TOKEN_PHPDOC_EOL => 'TOKEN_PHPDOC_EOL',
        self::TOKEN_FLOAT => 'TOKEN_FLOAT',
        self::TOKEN_INTEGER => 'TOKEN_INTEGER',
        self::TOKEN_SINGLE_QUOTED_STRING => 'TOKEN_SINGLE_QUOTED_STRING',
        self::TOKEN_DOUBLE_QUOTED_STRING => 'TOKEN_DOUBLE_QUOTED_STRING',
        self::TOKEN_IDENTIFIER => 'type',
        self::TOKEN_THIS_VARIABLE => '\'$this\'',
        self::TOKEN_VARIABLE => 'variable',
        self::TOKEN_HORIZONTAL_WS => 'TOKEN_HORIZONTAL_WS',
        self::TOKEN_OTHER => 'TOKEN_OTHER',
        self::TOKEN_END => 'TOKEN_END',
        self::TOKEN_WILDCARD => '*',
    ];

    public const VALUE_OFFSET = 0;
    public const TYPE_OFFSET = 1;

    /** @var string|null */
    private $regexp;

    public function tokenize(string $s): array
    {
        if ($this->regexp === null) {
            $this->regexp = $this->generateRegexp();
        }

        preg_match_all($this->regexp, $s, $matches, PREG_SET_ORDER);

        $tokens = [];
        foreach ($matches as $match) {
            $tokens[] = [$match[0], (int) $match['MARK']];
        }

        $tokens[] = ['', self::TOKEN_END];

        return $tokens;
    }

    private function generateRegexp(): string
    {
        $patterns = [
            self::TOKEN_HORIZONTAL_WS => '[\\x09\\x20]++',

            self::TOKEN_IDENTIFIER => '(?:[\\\\]?+[a-z_\\x80-\\xFF][0-9a-z_\\x80-\\xFF-]*+)++',
            self::TOKEN_THIS_VARIABLE => '\\$this(?![0-9a-z_\\x80-\\xFF])',
            self::TOKEN_VARIABLE => '\\$[a-z_\\x80-\\xFF][0-9a-z_\\x80-\\xFF]*+',

            // '&' followed by TOKEN_VARIADIC, TOKEN_VARIABLE, TOKEN_EQUAL, TOKEN_EQUAL or TOKEN_CLOSE_PARENTHESES
            self::TOKEN_REFERENCE => '&(?=\\s*+(?:[.,=)]|(?:\\$(?!this(?![0-9a-z_\\x80-\\xFF])))))',
            self::TOKEN_UNION => '\\|',
            self::TOKEN_INTERSECTION => '&',
            self::TOKEN_NULLABLE => '\\?',
            self::TOKEN_NEGATED => '!',

            self::TOKEN_OPEN_PARENTHESES => '\\(',
            self::TOKEN_CLOSE_PARENTHESES => '\\)',
            self::TOKEN_OPEN_ANGLE_BRACKET => '<',
            self::TOKEN_CLOSE_ANGLE_BRACKET => '>',
            self::TOKEN_OPEN_SQUARE_BRACKET => '\\[',
            self::TOKEN_CLOSE_SQUARE_BRACKET => '\\]',
            self::TOKEN_OPEN_CURLY_BRACKET => '\\{',
            self::TOKEN_CLOSE_CURLY_BRACKET => '\\}',

            self::TOKEN_COMMA => ',',
            self::TOKEN_VARIADIC => '\\.\\.\\.',
            self::TOKEN_DOUBLE_COLON => '::',
            self::TOKEN_DOUBLE_ARROW => '=>',
            self::TOKEN_ARROW => '->',
            self::TOKEN_EQUAL => '=',
            self::TOKEN_COLON => ':',

            self::TOKEN_OPEN_PHPDOC => '/\\*\\*(?=\\s)\\x20?+',
            self::TOKEN_CLOSE_PHPDOC => '\\*/',
            self::TOKEN_PHPDOC_TAG => '@(?:[a-z][a-z0-9-\\\\]+:)?[a-z][a-z0-9-\\\\]*+',
            self::TOKEN_PHPDOC_EOL => '\\r?+\\n[\\x09\\x20]*+(?:\\*(?!/)\\x20?+)?',

            self::TOKEN_FLOAT => '(?:-?[0-9]++\\.[0-9]*+(?:e-?[0-9]++)?)|(?:-?[0-9]*+\\.[0-9]++(?:e-?[0-9]++)?)|(?:-?[0-9]++e-?[0-9]++)',
            self::TOKEN_INTEGER => '-?(?:(?:0b[0-1]++)|(?:0o[0-7]++)|(?:0x[0-9a-f]++)|(?:[0-9]++))',
            self::TOKEN_SINGLE_QUOTED_STRING => '\'(?:\\\\[^\\r\\n]|[^\'\\r\\n\\\\])*+\'',
            self::TOKEN_DOUBLE_QUOTED_STRING => '"(?:\\\\[^\\r\\n]|[^"\\r\\n\\\\])*+"',

            self::TOKEN_WILDCARD => '\\*',

            // anything but TOKEN_CLOSE_PHPDOC or TOKEN_HORIZONTAL_WS or TOKEN_EOL
            self::TOKEN_OTHER => '(?:(?!\\*/)[^\\s])++',
        ];

        foreach ($patterns as $type => &$pattern) {
            $pattern = '(?:' . $pattern . ')(*MARK:' . $type . ')';
        }

        return '~' . implode('|', $patterns) . '~Asi';
    }
}
__halt_compiler();----SIGNATURE:----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----ATTACHMENT:----ODMwOTUxOTM4NzM1NzYxNiA4OTg5Njc5MDM0MTkzOTUyIDQ4NDQ5Mzg4MzkzMDgyMzM=