01-lexer.php

<?php

namespace vierbergenlars\Norch\QueryParser;

class Lexer
{

    /**
     * This class should not be instanciated
     */
    private function __construct()
    {
        ;
    }

    public static function tokenize($string)
    {
        $len = strlen($string);
        $tokens = array();

        $current_token = new Token(Token::T_NONE, 0);
        $i = 0;
        while($i < $len) {
            $c = $string[$i];
            switch($c) {
                case '\\': // Escape character
                    $current_token->addData($string[++$i]);
                    break;
                case ' ':
                    self::push($tokens, $current_token, $i);
                    break;
                case ':':
                    if($current_token->getData() == null)
                        throw new ParseException('Expected T_FIELD_NAME, got nothing', $string, $i);
                    if(!$current_token->isTypeNoneOr(Token::T_FIELD_NAME))
                        throw new ParseException('Expected T_FIELD_NAME, got ' . Token::getName($current_token->getType()), $string, $i);
                    $current_token->setType(Token::T_FIELD_NAME);
                    self::push($tokens, $current_token, $i);
                    $current_token->setType(Token::T_FIELD_VALUE);
                    break;
                case '^':
                    if($current_token->getData() == null)
                        throw new ParseException('Expected T_FIELD_NAME, got nothing', $string, $i);
                    if(!$current_token->isTypeNoneOr(Token::T_FIELD_NAME))
                        throw new ParseException('Expected T_FIELD_NAME, got ' . Token::getName($current_token->getType()), $string, $i);
                    $current_token->setType(Token::T_FIELD_NAME);
                    $field_token = $current_token;
                    self::push($tokens, $current_token, $i);
                    $current_token->setType(Token::T_FIELD_WEIGHT);
                    self::readInt($current_token, $string, $i);
                    self::push($tokens, $current_token, $i);
                    if($i + 1 < $len && $string[$i + 1] == ':') // Peek one ahead. Duplicate T_FIELD_NAME token if a T_FIELD_VALUE follows.
                        $current_token = $field_token;
                    break;
                case '@':
                    if($current_token->getData() != null)
                        throw new ParseException('Expected nothing, got ' . Token::getName($current_token->getType()), $string, $i);
                    $current_token->setType(Token::T_FIELD_SEARCH);
                    break;
                case '"':
                    if($current_token->getData() == null) {
                        $current_token->setTypeIfNone(Token::T_STRING);
                        self::readEncString($current_token, $string, $i);
                        if($i + 1 < $len && $string[$i + 1] != ' ') // Peek one ahead. Should be empty
                            throw new ParseException('Unexpected T_STRING', $string, $i + 1);
                    } else {
                        throw new ParseException('Unexpected T_STRING', $string, $i);
                    }
                    break;
                default:
                    $current_token->addData($c);
            }
            $i++;
        }
        self::push($tokens, $current_token, $i);
        return $tokens;
    }

    static private function push(&$tokens, &$current_token, $i)
    {
        if($current_token->getData() === null)
            return;
        $current_token->setTypeIfNone(Token::T_STRING);
        $tokens[] = $current_token;
        $current_token = new Token(Token::T_NONE, $i);
    }

    static private function readEncString(Token $current_token, $string, &$i)
    {
        while(++$i < strlen($string)) {
            if($string[$i] == '\\') {
                $current_token->addData($string[++$i]);
            } else if($string[$i] != '"') {
                $current_token->addData($string[$i]);
            } else {
                break;
            }
        }
    }

    static private function readInt(Token $current_token, $string, &$i)
    {
        while(++$i < strlen($string)) {
            if(in_array($string[$i], array('0', '1', '2', '3', '4', '5', '6', '7',
                        '8', '9', '-'), true)) {
                $current_token->addData($string[$i]);
            } else {
                $i--;
                break;
            }
        }
    }

}

02-compiler.php

<?php

namespace vierbergenlars\Norch\QueryParser;

use vierbergenlars\Norch\SearchQuery\QueryBuilder;
use vierbergenlars\Norch\QueryParser\Token;

class Compiler extends QueryBuilder
{

    public function updateQuery($queryExpr)
    {
        $tokens = Lexer::tokenize($queryExpr);
        $searchQuery = '';
        while(false !== ($token = current($tokens))) {
            switch($token->getType()) {
                case Token::T_STRING:
                    $searchQuery.= ' ' . $token->getData();
                    break;
                case Token::T_FIELD_NAME:
                    $nextToken = next($tokens);
                    if($nextToken === false)
                        throw new ParseException('Unexpected end of token stream');
                    switch($nextToken->getType()) {
                        case Token::T_FIELD_VALUE:
                            $this->addFilter($token->getData(), $nextToken->getData());
                            break;
                        case Token::T_FIELD_WEIGHT:
                            $this->addWeight($token->getData(), $nextToken->getData());
                            break;
                        default:
                            throw new ParseException('Unexpected ' . Token::getName($nextToken->getType()), $queryExpr, $token->getStartPosition());
                    }
                    break;
                case Token::T_FIELD_SEARCH:
                    $this->addSearchField($token->getData());
                    break;
                default:
                    throw new ParseException('Unexpected ' . Token::getName($token->getType()) . ' (This is a lexer bug, please report it)', $queryExpr, $token->getStartPosition());
            }
            next($tokens);
        }
        $this->setSearchQuery(substr($searchQuery, 1));
        return $this;
    }

}

token.php

<?php

namespace vierbergenlars\Norch\QueryParser;

class Token
{

    const T_NONE = 0;

    const T_FIELD_NAME = 1;

    const T_STRING = 2;

    const T_FIELD_WEIGHT = 3;

    const T_FIELD_VALUE = 4;

    const T_FIELD_SEARCH = 5;

    protected $type;

    protected $data = null;

    protected $startPos;

    function __construct($type, $startPos)
    {
        $this->type = $type;
        $this->startPos = $startPos;
    }

    function addData($data)
    {
        $this->data.=$data;
    }

    function setType($type)
    {
        $this->type = $type;
    }

    function setTypeIfNone($type)
    {
        if($this->type == self::T_NONE)
            $this->type = $type;
    }

    function isTypeNoneOr($type)
    {
        return ($this->type == self::T_NONE || $this->type == $type);
    }

    function getType()
    {
        return $this->type;
    }

    function getData()
    {
        return $this->data;
    }

    function getStartPosition()
    {
        return $this->startPos;
    }

    static function getName($token)
    {
        $refl = new \ReflectionClass(__CLASS__);
        $constants = $refl->getConstants();
        $token_name = array_search($token, $constants);
        if($token_name)
            return $token_name;
        return 'UNKNOWN_TOKEN';
    }

}