Converting a Google search query to a PostgreSQL "tsquery"
Honest, I think regular expressions are the way to go with something like this. Just the same, this was a fun exercise. The code below is very prototypal - in fact, you'll see that I didn't even implement the lexer itself - I just faked the output. I'd like to continue it but I just don't have more spare time today.
Also, there definitely a lot more work to be done here in terms of supporting other types of search operators and the like.
Basically, the idea is that a certain type of query is lexed then parsed into a common format (in this case, a QueryExpression instance) which is then rendered back out as another type of query.
<?phpini_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();