php - 将 Google 搜索查询转换为 PostgreSQL "tsquery"

Question

如何将 Google 搜索查询转换为可以提供给 PostgreSQL 的 to_tsquery() 的内容？

如果没有现成的库，我应该如何使用 PHP 等语言解析 Google 搜索查询？

例如，我想采用以下类似 Google 的搜索查询:

("used cars" OR "new cars") -ford -mistubishi

并将其转换为对 to_tsquery() 友好的字符串:

('used cars' | 'new cars') & !ford & !mistubishi

我可以用正则表达式捏造这个，但这是我能做的最好的。是否有一些强大的词法分析方法可以解决这个问题？我也希望能够支持扩展搜索运算符(如 Google 的 site: 和 intitle:)，它们将应用于不同的数据库字段，因此需要与 tsquery 字符串分开。

更新:我意识到使用特殊运算符，这变成了 Google 到 SQL WHERE 子句的转换，而不是 Google 到 tsquery 的转换。但是 WHERE 子句可能包含一个或多个 tsqueries。

例如，Google 风格的查询:

((color:blue OR "4x4") OR style:coupe) -color:red used

应该产生这样的 SQL WHERE 子句:

WHERE to_tsvector(description) MATCH to_tsquery('used')
  AND color <> 'red'
  AND ( (color = 'blue' OR to_tsvector(description) MATCH to_tsquery('4x4') )
    OR style = 'coupe'
  );

我不确定以上是否可以用正则表达式实现？

Answer 1

老实说，我认为正则表达式是解决此类问题的方法。同样，这是一个有趣的练习。下面的代码非常原型(prototype)——事实上，你会看到我什至没有实现词法分析器本身——我只是伪造了输出。我想继续，但我今天没有更多的空闲时间。

此外，在支持其他类型的搜索运算符等方面肯定还有很多工作要做。

基本上，这个想法是对某种类型的查询进行词法分析，然后将其解析为一种通用格式(在本例中为 QueryExpression 实例)，然后将其呈现为另一种类型的查询。

<?php

ini_set( "display_errors", "on" );
error_reporting( E_ALL );

interface ILexer
{
    public function execute( $str );
    public function getTokens();
}

interface IParser
{
    public function __construct( iLexer $lexer );
    public function parse( $input );
    public function addToken( $token );
}

class GoogleQueryLexer implements ILexer
{
    private $tokenStack = array();

    public function execute( $str )
    {
        $chars = str_split( $str );
        foreach ( $chars as $char )
        {
            //  add to self::$tokenStack per your rules
        }

        //'("used cars" OR "new cars") -ford -mistubishi'
        $this->tokenStack = array(
                '('
            ,   'used cars'
            ,   'or new cars'
            ,   ')'
            ,   '-ford'
            ,   '-mitsubishi'
        );
    }

    public function getTokens()
    {
        return $this->tokenStack;
    }
}

class GoogleQueryParser implements IParser
{
    protected $lexer;

    public function __construct( iLexer $lexer )
    {
        $this->lexer = $lexer;
    }

    public function addToken( $token )
    {
        $this->tokenStack[] = $token;
    }

    public function parse( $input )
    {
        $this->lexer->execute( $input );
        $tokens = $this->lexer->getTokens();

        $expression = new QueryExpression();

        foreach ( $tokens as $token )
        {
            $expression = $this->processToken( $token, $expression );
        }

        return $expression;
    }

    protected function processToken( $token, QueryExpression $expression )
    {
        switch ( $token )
        {
            case '(':
                return $expression->initiateSubExpression();
                break;
            case ')':
                return $expression->getParentExpression();
                break;
            default:
                $modifier   = $token[0];
                $phrase     = substr( $token, 1 );
                switch ( $modifier )
                {
                    case '-':
                        $expression->addExclusionPhrase( $phrase );
                        break;
                    case '+':
                        $expression->addPhrase( $phrase );
                        break;
                    default:
                        $operator   = trim( substr( $token, 0, strpos( $token, ' ' ) ) );
                        $phrase     = trim( substr( $token, strpos( $token, ' ' ) ) );
                        switch ( strtolower( $operator ) )
                        {
                            case 'and':
                                $expression->addAndPhrase( $phrase );
                                break;
                            case 'or':
                                $expression->addOrPhrase( $phrase );
                                break;
                            default:
                                $expression->addPhrase( $token );
                        }
                }
        }
        return $expression;
    }
}

class QueryExpression
{
    protected $phrases = array();
    protected $subExpressions = array();
    protected $parent;

    public function __construct( $parent=null )
    {
        $this->parent = $parent;
    }

    public function initiateSubExpression()
    {
        $expression = new self( $this );
        $this->subExpressions[] = $expression;
        return $expression;
    }

    public function getPhrases()
    {
        return $this->phrases;
    }

    public function getSubExpressions()
    {
        return $this->subExpressions;
    }

    public function getParentExpression()
    {
        return $this->parent;
    }

    protected function addQueryPhrase( QueryPhrase $phrase )
    {
        $this->phrases[] = $phrase;
    }

    public function addPhrase( $input )
    {
        $this->addQueryPhrase( new QueryPhrase( $input ) );
    }

    public function addOrPhrase( $input )
    {
        $this->addQueryPhrase( new QueryPhrase( $input, QueryPhrase::MODE_OR ) );
    }

    public function addAndPhrase( $input )
    {
        $this->addQueryPhrase( new QueryPhrase( $input, QueryPhrase::MODE_AND ) );
    }

    public function addExclusionPhrase( $input )
    {
        $this->addQueryPhrase( new QueryPhrase( $input, QueryPhrase::MODE_EXCLUDE ) );
    }
}

class QueryPhrase
{
    const MODE_DEFAULT = 1;
    const MODE_OR = 2;
    const MODE_AND = 3;
    const MODE_EXCLUDE = 4;

    protected $phrase;
    protected $mode;

    public function __construct( $input, $mode=self::MODE_DEFAULT )
    {
        $this->phrase = $input;
        $this->mode = $mode;
    }

    public function getMode()
    {
        return $this->mode;
    }

    public function __toString()
    {
        return $this->phrase;
    }
}

class TsqueryBuilder
{
    protected $expression;
    protected $query;

    public function __construct( QueryExpression $expression )
    {
        $this->query = trim( $this->processExpression( $expression ), ' &|' );
    }

    public function getResult()
    {
        return $this->query;
    }

    protected function processExpression( QueryExpression $expression )
    {
        $query = '';
        $phrases = $expression->getPhrases();
        $subExpressions = $expression->getSubExpressions();

        foreach ( $phrases as $phrase )
        {
            $format = "'%s' ";
            switch ( $phrase->getMode() )
            {
                case QueryPhrase::MODE_AND :
                    $format = "& '%s' ";
                    break;
                case QueryPhrase::MODE_OR :
                    $format = "| '%s' ";
                    break;
                case QueryPhrase::MODE_EXCLUDE :
                    $format = "& !'%s' ";
                    break;
            }
            $query .= sprintf( $format, str_replace( "'", "\\'", $phrase ) );
        }

        foreach ( $subExpressions as $subExpression )
        {
            $query .= "& (" . $this->processExpression( $subExpression ) . ")";
        }
        return $query;
    }
}

$parser = new GoogleQueryParser( new GoogleQueryLexer() );

$queryBuilder = new TsqueryBuilder( $parser->parse( '("used cars" OR "new cars") -ford -mistubishi' ) );

echo $queryBuilder->getResult();