diff --git a/html/index.php b/html/index.php index 6876ac0..d6f24af 100755 --- a/html/index.php +++ b/html/index.php @@ -38,12 +38,6 @@ else include 'form.html.php'; exit(); } - //check if doing a more general search - $general=false; - if (isset($_REQUEST['g'])) - { - $general=true; - } //fix phone users putting space at end if(strlen($query) > 1 && $query[strlen($query)-1]==" "){ @@ -220,7 +214,7 @@ else if ($word[0] == '+' && strlen($word) > 1){ $requiredword = substr($word,1); } - if ($word[0] == '-' && $word[0] == '+'){ + if ($word[0] == '-' || $word[0] == '+'){ $flags .= " $word"; } $i++; @@ -251,6 +245,27 @@ else } $wordcount++; } + } + + //Check if query contains a hyphenated word. MySQL is finicky about them. We will wrap quotes around hyphenated words that aren't part of a string which is already wraped in quotes. + if((strpos($queryNoQuotes,'-') !== false || strpos($queryNoQuotes,'+') !== false) && $urlDetected == false){ + $hyphenwords = explode(' ',$query); + $query = ''; + $quotes = 0; + $i = 0; + foreach ($hyphenwords as $word) { + if(strpos($queryNoQuotes,'"') !== false){ + $quotes++; + } + if(((strpos($queryNoQuotes,'-') !== false && $word[0] != '-') || (strpos($queryNoQuotes,'+') !== false && $word[0] != '+')) && $quotes%2 == 0){//if hyphen exists, not a flag, not wrapped in quotes already + $word = '"' . $word . '"'; + } + if($i > 0){ + $query .= ' '; + } + $query .= $word; + $i++; + } } if($filterHTTPS == true){ @@ -262,353 +277,178 @@ else $count = 0; - if(!$general){ - $queryWithQuotesAndFlags = '"'. $queryNoQuotes_SQLsafe.'"'.$flags.''; + $queryWithQuotesAndFlags = '"'. $queryNoQuotes_SQLsafe.'"'.$flags.''; + //perform full text search FOR InnoDB or MyISAM STORAGE ENGINE + if($exactMatch == false && $urlDetected==0 && strpos($query, ' ') == true){ + $outputFTS = mysqli_query($link, "SELECT id, url, title, description, body FROM windex WHERE enable = '1' $additions ORDER BY CASE WHEN MATCH(tags) AGAINST('$queryWithQuotesAndFlags' IN BOOLEAN MODE) THEN 30 WHEN MATCH(title) AGAINST('$queryWithQuotesAndFlags' IN BOOLEAN MODE) AND MATCH(title) AGAINST('$query' IN BOOLEAN MODE) THEN 20 WHEN MATCH(title) AGAINST('$queryWithQuotesAndFlags' IN BOOLEAN MODE) THEN 16 WHEN MATCH(body) AGAINST('$queryWithQuotesAndFlags' IN BOOLEAN MODE) OR MATCH(description) AGAINST('$queryWithQuotesAndFlags' IN BOOLEAN MODE) THEN 15 WHEN MATCH(title) AGAINST('$query' IN BOOLEAN MODE) THEN MATCH(title) AGAINST('$query' IN BOOLEAN MODE) WHEN MATCH(body) AGAINST('$query' IN BOOLEAN MODE) THEN 1 WHEN MATCH(url) AGAINST('$query' IN BOOLEAN MODE) THEN 0 END DESC, id DESC LIMIT $lim OFFSET $offset"); + }else{ + $outputFTS = mysqli_query($link, "SELECT id, url, title, description, body FROM windex WHERE MATCH(tags, body, description, title, url) AGAINST('$queryWithQuotesAndFlags' IN BOOLEAN MODE) AND enable = '1' $additions ORDER BY CASE WHEN MATCH(tags) AGAINST('$queryWithQuotesAndFlags' IN BOOLEAN MODE) THEN 30 WHEN MATCH(title) AGAINST('$queryWithQuotesAndFlags' IN BOOLEAN MODE) THEN 20 END DESC, id DESC LIMIT $lim OFFSET $offset"); + } + + /*if(!$outputFTS)//dont error out yet, will give another try below + { + $error = 'Error ' . mysqli_error($link); + include 'error.html.php'; + exit(); + }*/ + + if($urlDetected == 1) + { + $query = $queryOriginal; + } + + if(mysqli_num_rows($outputFTS) == 0 && $offset == 0 && $urlDetected == 0 && $exactMatch == false) + { + $starappend = 1; + $querystar = $query; + //innodb will get fussy over some things if put in like '''' or ****, uncomment below lines if using innoDB + $querystar = str_replace('*', "",$querystar); + $querystar = str_replace('"', "",$querystar); + $querystar = str_replace('"', "",$querystar); + $querystar = str_replace('\'', "",$querystar); + //----------------------------------------------- + + $querystar = $querystar . '*'; //perform full text search FOR InnoDB or MyISAM STORAGE ENGINE - $outputFTS = mysqli_query($link, "SELECT id, url, title, description, body FROM windex WHERE Match(tags, body, description, title, url) Against('$queryWithQuotesAndFlags' IN BOOLEAN MODE) AND enable = '1' $additions ORDER BY CASE WHEN MATCH(tags) AGAINST('$queryWithQuotesAndFlags' IN BOOLEAN MODE) THEN 30 WHEN MATCH(title) AGAINST('$queryWithQuotesAndFlags' IN BOOLEAN MODE) THEN 20 END DESC, id DESC LIMIT $lim OFFSET $offset"); + $outputFTSgeneral = mysqli_query($link, "SELECT id, url, title, description, body FROM windex WHERE Match(tags, body, description, title, url) Against('$querystar' IN BOOLEAN MODE) AND enable = '1' $additions ORDER BY CASE WHEN MATCH(tags) AGAINST('$queryWithQuotesAndFlags' IN BOOLEAN MODE) THEN 30 END DESC, id DESC LIMIT $lim OFFSET $offset"); - /*if(!$outputFTS)//dont error out yet, will give another try below + if(!$outputFTSgeneral) { $error = 'Error ' . mysqli_error($link); include 'error.html.php'; exit(); - }*/ + } + } + //this will get set if position of longest word of query is found within body + $pos = -1; + + //lets put contents of the full text search into the array + while($row = mysqli_fetch_array($outputFTS)) + { + //put the contents of the URL column within the DB into an array + $id[] = $row[0]; + $url[] = $row[1]; + $title[] = substr($row[2],0,150); + $description[] = substr($row[3],0,180); + $body = $row[4]; + $count++; + $lastID = $row[0]; - if($urlDetected == 1) + if($exactMatch == false) { - $query = $queryOriginal; - } - - //this will get set if position of longest word of query is found within body - $pos = -1; - - //lets put contents of the full text search into the array - while($row = mysqli_fetch_array($outputFTS)) - { - //put the contents of the URL column within the DB into an array - $id[] = $row[0]; - $url[] = $row[1]; - $title[] = substr($row[2],0,150); - $description[] = substr($row[3],0,180); - $body = $row[4]; - $count++; - $lastID = $row[0]; - - if($exactMatch == false) + //remove the '*' at the end of the longest word if present + if(strpos($longestWord,'*') == true) { - //remove the '*' at the end of the longest word if present - if(strpos($longestWord,'*') == true) - { - $longestWord = str_replace('*', "",$longestWord); - } - - //first find an exact - if(strlen($requiredword) > 0){ - $pos = stripos($body, $requiredword); - }else{ - $pos = stripos($body, $queryNoQuotes); - } - - //search within body for position of longest query word. If not found, try another word - if($pos == false){ - $pos = stripos($body, $longestWord); - if($pos == false && $wordcount > 1) - { - if($longestwordelementnum > 0) - { - if(strpos($words[0],'*') == true)//remove the '*' at the end of the query if present - $words[0] = str_replace('*', "",$words[0]); - $pos = stripos($body, $words[0]); - } - else if($longestwordelementnum == 0) - { - if(strpos($words[1],'*') == true)//remove the '*' at the end of the query if present - $words[1] = str_replace('*', "",$words[1]); - $pos = stripos($body, $words[1]); - } - } - } + $longestWord = str_replace('*', "",$longestWord); } - else - { + + //first find an exact + if(strlen($requiredword) > 0){ + $pos = stripos($body, $requiredword); + }else{ $pos = stripos($body, $queryNoQuotes); } - //still not found?, set position to 0 - if($pos == false){ - $pos = 0; - } - - //get all positions of all keywords in body - /* $lastPos = 0; - $positions = array(); - foreach($words as $word) - { - while (($lastPos = mb_strpos($body, $word, $lastPos))!== false) { - $positions[$word][] = $lastPos; - $lastPos = $lastPos + strlen($word); - } - }*/ - - //figure out how much preceding text to use - if($pos < 32) - $starttext = 0; - else if($pos > 25) - $starttext = $pos - 25; - else if($pos > 20) - $starttext = $pos - 15; - //else $starttext = 0; - - //total length of the ballpark - $textlength = 180; - - //populate the ballpark - if($pos >= 0) - { - $ballparktext = substr($body,$starttext,$textlength); - } - else $ballpark = '0'; - - //find position of nearest Period - $foundPeriod = true; - $posPeriod = stripos($ballparktext, '. ') + $starttext +1; - - //find position of nearest Space - $foundSpace = true; - $posSpace = stripos($ballparktext, ' ') + $starttext; - - //if longest word in query is after a period+space within ballpark, reset $starttext to that point - if($pos-$starttext > $posPeriod) - { - $starttext = $posPeriod; - //populate the bodymatch - if($pos-$starttext >= 0) - { - $bodymatch[] = substr($body,$starttext,$textlength); - } - else $bodymatch[] = ''; - } - //else if($pos-starttext > $posSpace)//else if longest word in query is after a space within ballpark, reset $starttext to that point - else if($pos > $posSpace)//else if longest word in query is after a space within ballpark, reset $starttext to that point - { - $starttext = $posSpace; - //populate the bodymatch - if($pos-$starttext >= 0) - { - $bodymatch[] = substr($body,$starttext,$textlength); - } - else $bodymatch[] = ''; - } - else //else just set the bodymatch to the ballparktext - { - //populate the bodymatch - if($pos-$starttext >= 0) - { - $bodymatch[] = $ballparktext; - } - else $bodymatch[] = ''; - } - } - } - $outputFTSrows=$count; - //do a general search if exact results peter off - if($outputFTSrows < 8 && $exactMatch == false){ - $count = 0; - $general = true; - - //Check if query contains a hyphenated word. MySQL is finicky about them. We will wrap quotes around hyphenated words that aren't part of a string which is already wraped in quotes. - if((strpos($queryNoQuotes,'-') !== false || strpos($queryNoQuotes,'+') !== false) && $urlDetected == false){ - $hyphenwords = explode(' ',$query); - $query = ''; - $quotes = 0; - $i = 0; - foreach ($hyphenwords as $word) { - if(strpos($queryNoQuotes,'"') !== false){ - $quotes++; - } - if(((strpos($queryNoQuotes,'-') !== false && $word[0] != '-') || (strpos($queryNoQuotes,'+') !== false && $word[0] != '+')) && $quotes%2 == 0){//if hyphen exists, not a flag, not wrapped in quotes already - $word = '"' . $word . '"'; - } - if($i > 0){ - $query .= ' '; - } - $query .= $word; - $i++; - } - } - - //perform full text search FOR InnoDB or MyISAM STORAGE ENGINE - $outputFTSgeneral = mysqli_query($link, "SELECT id, url, title, description, body FROM windex WHERE Match(tags, body, description, title, url) Against('$query' IN BOOLEAN MODE) AND enable = '1' $additions ORDER BY CASE WHEN Match(title) AGAINST('$queryWithQuotesAndFlags' IN BOOLEAN MODE) AND Match(title) AGAINST('$query' IN BOOLEAN MODE) THEN 20 WHEN Match(title) AGAINST('$queryWithQuotesAndFlags' IN BOOLEAN MODE) THEN 16 WHEN Match(title) AGAINST('$query' IN BOOLEAN MODE) THEN Match(title) AGAINST('$query' IN BOOLEAN MODE) END DESC, id DESC LIMIT $lim OFFSET $offset"); - - - //if all else fails, try a full text search with * appended (better to get something than nothing I suppose) - if(mysqli_num_rows($outputFTSgeneral) == 0 && $offset == 0 && $urlDetected == 0) - { - $starappend = 1; - $querystar = $query; - //innodb will get fussy over some things if put in like '''' or ****, uncomment below lines if using innoDB - $querystar = str_replace('*', "",$querystar); - $querystar = str_replace('"', "",$querystar); - $querystar = str_replace('"', "",$querystar); - $querystar = str_replace('\'', "",$querystar); - //----------------------------------------------- - - $querystar = $querystar . '*'; - - //perform full text search FOR InnoDB or MyISAM STORAGE ENGINE - $outputFTSgeneral = mysqli_query($link, "SELECT id, url, title, description, body FROM windex WHERE Match(tags, body, description, title, url) Against('$querystar' IN BOOLEAN MODE) AND enable = '1' $additions ORDER BY CASE WHEN MATCH(tags) AGAINST('$queryWithQuotesAndFlags' IN BOOLEAN MODE) THEN 30 END DESC, id DESC LIMIT $lim OFFSET $offset"); - - if(!$outputFTSgeneral) - { - $error = 'Error ' . mysqli_error($link); - include 'error.html.php'; - exit(); - } - } - - //this will get set if position of longest word of query is found within body - $pos = -1; - - //lets put contents of the full text search into the array - while($row = mysqli_fetch_array($outputFTSgeneral)) - { - $count++; - //check for duplicates if appending general search matches on the same page where exact matches were found - $duplicate = false; - if($outputFTSrows < 8 && $outputFTSrows > 0){ - foreach($id as $idtocheck){ - if($idtocheck==$row[0]){ - $duplicate=true; - break; - } - - } - } - if($duplicate==false){ - //put the contents of the URL column within the DB into an array - $id[] = $row[0]; - $url[] = $row[1]; - $title[] = substr($row[2],0,150); - $description[] = substr($row[3],0,180); - $body = $row[4]; - $lastID = $row[0]; - if($exactMatch == false) + //search within body for position of longest query word. If not found, try another word + if($pos == false){ + $pos = stripos($body, $longestWord); + if($pos == false && $wordcount > 1) { - //remove the '*' at the end of the longest word if present - if(strpos($longestWord,'*') == true) + if($longestwordelementnum > 0) { - $longestWord = str_replace('*', "",$longestWord); + if(strpos($words[0],'*') == true)//remove the '*' at the end of the query if present + $words[0] = str_replace('*', "",$words[0]); + $pos = stripos($body, $words[0]); } - - //first find an exact - if(strlen($requiredword) > 0){ - $pos = stripos($body, $requiredword); - }else{ - $pos = stripos($body, $queryNoQuotes); - } - - //search within body for position of longest query word. If not found, try another word - if($pos == false){ - $pos = stripos($body, $longestWord); - if($pos == false && $wordcount > 1) - { - if($longestwordelementnum > 0) - { - if(strpos($words[0],'*') == true)//remove the '*' at the end of the query if present - $words[0] = str_replace('*', "",$words[0]); - $pos = stripos($body, $words[0]); - } - else if($longestwordelementnum == 0) - { - if(strpos($words[1],'*') == true)//remove the '*' at the end of the query if present - $words[1] = str_replace('*', "",$words[1]); - $pos = stripos($body, $words[1]); - } - } - } - } - else - { - $pos = stripos($body, $queryNoQuotes); - } - //still not found?, set position to 0 - if($pos == false){ - $pos = 0; - } - - //get all positions of all keywords in body - /* $lastPos = 0; - $positions = array(); - foreach($words as $word) - { - while (($lastPos = mb_strpos($body, $word, $lastPos))!== false) { - $positions[$word][] = $lastPos; - $lastPos = $lastPos + strlen($word); + else if($longestwordelementnum == 0) + { + if(strpos($words[1],'*') == true)//remove the '*' at the end of the query if present + $words[1] = str_replace('*', "",$words[1]); + $pos = stripos($body, $words[1]); } - }*/ - - //figure out how much preceding text to use - if($pos < 32) - $starttext = 0; - else if($pos > 25) - $starttext = $pos - 25; - else if($pos > 20) - $starttext = $pos - 15; - //else $starttext = 0; - - //total length of the ballpark - $textlength = 180; - - //populate the ballpark - if($pos >= 0) - { - $ballparktext = substr($body,$starttext,$textlength); - } - else $ballpark = '0'; - - //find position of nearest Period - $foundPeriod = true; - $posPeriod = stripos($ballparktext, '. ') + $starttext +1; - - //find position of nearest Space - $foundSpace = true; - $posSpace = stripos($ballparktext, ' ') + $starttext; - - //if longest word in query is after a period+space within ballpark, reset $starttext to that point - if($pos-$starttext > $posPeriod) - { - $starttext = $posPeriod; - //populate the bodymatch - if($pos-$starttext >= 0) - { - $bodymatch[] = substr($body,$starttext,$textlength); - } - else $bodymatch[] = ''; - } - //else if($pos-starttext > $posSpace)//else if longest word in query is after a space within ballpark, reset $starttext to that point - else if($pos > $posSpace)//else if longest word in query is after a space within ballpark, reset $starttext to that point - { - $starttext = $posSpace; - //populate the bodymatch - if($pos-$starttext >= 0) - { - $bodymatch[] = substr($body,$starttext,$textlength); - } - else $bodymatch[] = ''; - } - else //else just set the bodymatch to the ballparktext - { - //populate the bodymatch - if($pos-$starttext >= 0) - { - $bodymatch[] = $ballparktext; - } - else $bodymatch[] = ''; } } - } + else + { + $pos = stripos($body, $queryNoQuotes); + } + //still not found?, set position to 0 + if($pos == false){ + $pos = 0; + } + + //get all positions of all keywords in body + /* $lastPos = 0; + $positions = array(); + foreach($words as $word) + { + while (($lastPos = mb_strpos($body, $word, $lastPos))!== false) { + $positions[$word][] = $lastPos; + $lastPos = $lastPos + strlen($word); + } + }*/ + + //figure out how much preceding text to use + if($pos < 32) + $starttext = 0; + else if($pos > 25) + $starttext = $pos - 25; + else if($pos > 20) + $starttext = $pos - 15; + //else $starttext = 0; + + //total length of the ballpark + $textlength = 180; + + //populate the ballpark + if($pos >= 0) + { + $ballparktext = substr($body,$starttext,$textlength); + } + else $ballpark = '0'; + + //find position of nearest Period + $foundPeriod = true; + $posPeriod = stripos($ballparktext, '. ') + $starttext +1; + + //find position of nearest Space + $foundSpace = true; + $posSpace = stripos($ballparktext, ' ') + $starttext; + + //if longest word in query is after a period+space within ballpark, reset $starttext to that point + if($pos-$starttext > $posPeriod) + { + $starttext = $posPeriod; + //populate the bodymatch + if($pos-$starttext >= 0) + { + $bodymatch[] = substr($body,$starttext,$textlength); + } + else $bodymatch[] = ''; + } + //else if($pos-starttext > $posSpace)//else if longest word in query is after a space within ballpark, reset $starttext to that point + else if($pos > $posSpace)//else if longest word in query is after a space within ballpark, reset $starttext to that point + { + $starttext = $posSpace; + //populate the bodymatch + if($pos-$starttext >= 0) + { + $bodymatch[] = substr($body,$starttext,$textlength); + } + else $bodymatch[] = ''; + } + else //else just set the bodymatch to the ballparktext + { + //populate the bodymatch + if($pos-$starttext >= 0) + { + $bodymatch[] = $ballparktext; + } + else $bodymatch[] = ''; + } + } $query = $_GET['q']; diff --git a/html/results.html.php b/html/results.html.php index 011c9b8..3787a20 100755 --- a/html/results.html.php +++ b/html/results.html.php @@ -2,15 +2,15 @@
-