在查询中添加可排序字段

Question

我尽我所知在PHP完成了搜索查询，但需要进行一些改进：

当我搜索说'什么是食物'并且我在数据库中有'什么是食物'时，所有结果都包含关键词'what'，'is'，'food'。期望的行为是显示包含确切短语“what is food”（第一个）
只突出显示查询中的最后一个单词，我想突出显示所有单词

期望的行为：正确的答案显示在顶部，无论其在数据库中的位置如何。

我目前的代码是这样的：

if (isset($_GET["mainSearch"]))
{
  $condition = '';
  $mainSearch = SQLite3::escapeString($_GET['mainSearch']);
  $keyword = $_GET['mainSearch'];
  $query = explode(" ", $keyword);
  $perpageview=7;

  if ($_GET["pageno"])
  {
      $page=$_GET["pageno"];
  }
  else
  {
      $page=1;
  }

  $frompage = $page*$perpageview-$perpageview;

  foreach ($query as $text)
  {
      $condition .= "question LIKE '%".SQLite3::escapeString($text)."%' OR answer LIKE '%".SQLite3::escapeString($text)."%' OR ";
  }
  foreach ($query as $text_2)
  {
      $condition_2 .= "bname LIKE '%".SQLite3::escapeString($text_2)."%' OR bankreq LIKE '%".SQLite3::escapeString($text_2)."%' OR ";
  }

  $condition = substr($condition, 0, -4);
  $condition_2 = substr($condition_2, 0, -4);


  $order = " ORDER BY quiz_id DESC ";
  $order_2 = " ORDER BY id DESC ";
  $sql_query = "SELECT * FROM questions WHERE " . $condition . ' '. $order.' LIMIT '.$frompage.','.$perpageview;
  $sql_query_count = "SELECT COUNT(*) as count FROM questions WHERE " . $condition .' '. $order;
  //$mainAnswer = "SELECT * FROM questions WHERE question LIKE '%$mainSearch%' or answer LIKE '%$mainSearch%'";
  $bank_query = "SELECT * FROM banks WHERE " . $condition_2 . ' LIMIT 1';
  $result = $db->query($sql_query);
  $resultCount = $db->querySingle($sql_query_count);
  $bankret = $db->query($bank_query);
  //$mainAnsRet = $db->query($mainAnswer);
  $pagecount = ceil($resultCount/$perpageview);

  if ($resultCount > 0)
  {
  if ($result && $bankret)
  {
      while ($row = $result->fetchArray(SQLITE3_ASSOC))
      {

          $wording = str_replace($text, "<span style='font-weight: bold; color: #1a0dab;'>".$text."</span>", $row['answer']);

           echo '<div class="quesbox_3">
            <div class="questitle">
                <h2>'.$row["question"].'</h2>
            </div>
            <div class="quesanswer">'.$wording.'</div>
        </div>';
      }
      while ($brow = $bankret->fetchArray(SQLITE3_ASSOC))
      {
            $bname = $brow['bname'];
            $bankbrief = $brow['bankbrief'];
            $bankreq = $brow['bankreq'];
            $bankaddress = $brow['bankaddress'];
            $banklogo = $brow['banklogo'];
            $founded = $brow['founded'];
            $owner = $brow['owner'];
            $available = $brow['available'];


           echo '<div class="modulecontent">
            <div class="modulename">
                <div class="mname">'.$bname.'</div>
                <div class="mlogo"><img src="'.$banklogo.'"></div>
            </div>';

            if (strlen($bankreq) > 300)
            {
                $bankcut = substr($bankreq, 0, 300);

                $bankreq = substr($bankcut, 0, strrpos($bankcut, ' ')).'... <a href="bankprofile.php?bname='.$bname.'">Read More</a>';
                echo '<div class="modulebrief">'.$bankreq.'</div>';
            }
            echo '<div class="modulelinks">
                <div class="mfound">Founded: <span>'.$founded.'</span></div>
                <div class="mowned">Ownd By: <span>'.$owner.'</span></div>
            </div>
        </div>';

               // <div class="mavailable">Available for Export Loan: <span>'.$available.'</span></div>
      }
      ?>
      <div class="page_num">
      <?php
      for ($i=1; $i <= $pagecount; $i++) {
         echo '<a href="searchresult.php?mainSearch='.$mainSearch.'&pageno='.$i.'">'.$i.'</a>';
      }
      ?>
      </div>
      <?php
  }
  }
  else
  {
      $session_n = $_SESSION['log_id'];
      $sesdate = date('d/M/Y');
      echo "<div class='searchNone'><p>No results found</p></div>
      <div class='sendSearchQ'>
      <p>Please send us your question.</p>
      <form action='sendquestion.php' method='post' encytype='multipart/form-data'>
      <div class='searchQinputs'>
          <input type='text' name='searchQuestion' id='searchQuestion'placeholder='Whats your question'><br>
          <input type='submit' name='sendQuestion' id='sendQuestion' value='Send'>
          <input type='text' name='user' id='user' value='$session_n' style='display: none'>
          <input type='text' name='qDate' id='qDate' value='$sesdate' style='display: none'>
          <input type='text' name='status' id='status' value='0' style='display: none'>
          </div>
      </form>
      </div>";
  }
}

Answer 1

在查询中添加可排序字段

首先我们需要简化您的问题

您可以将问题视为简单的关键字匹配，其中最重要的结果必须与输入的所有关键字匹配。

搜索：some search text应该返回包含任何单词['some', 'search', 'text']的所有结果，其中的结果与完全一致 "some search text"。

这意味着您需要创建一个聚合字段，允许根据匹配对结果进行排序。我知道这样做的唯一方法是在没有重构数据和/或代码的情况下使用MySql Case语句。

您的查询已简化

SELECT * FROM questions WHERE question LIKE '%[word1]%' OR answer LIKE '%[word1]%' OR question LIKE '%[word2]%' OR answer LIKE '%[word2]%' OR question LIKE '%[word3]%' OR answer LIKE '%[word3]%' ORDER BY quiz_id DESC

案例[完全匹配]和排序

我们需要构建的是一个看起来有点像这样的查询：

SELECT *, (CASE WHEN question LIKE '%[full-search-query]%' OR answer LIKE '%[full-search-query]%' THEN 1 ELSE 0 END) as fullmatch FROM questions WHERE question LIKE '%[word1]%' OR answer LIKE '%[word1]%' OR question LIKE '%[word2]%' OR answer LIKE '%[word2]%' OR question LIKE '%[word3]%' OR answer LIKE '%[word3]%' ORDER BY fullmatch DESC, quiz_id DESC

调整代码

// your initial storage of the full search, before you split it on spaces $keyword = $_GET['mainSearch']; . . . // build our sorting field $sortFullMatch = "(CASE WHEN question LIKE '%".SQLite3::escapeString($keyword)."%' OR answer LIKE '%".SQLite3::escapeString($keyword)."%' THEN 1 ELSE 0 END) as fullmatch"; . . . // adjust the query and sort $order = " ORDER BY fullmatch DESC, quiz_id DESC "; $sql_query = "SELECT *,". $sortFullMatch ." FROM questions WHERE ".$condition.' '.$order.' LIMIT '.$frompage.','.$perpageview;

这是做什么的？

我们在SELECT语句fulltext中添加了一个新字段。当问题或答案完全包含完整搜索时，此字段将为1，否则为0.然后只需对此字段进行排序。

高亮

至于您的突出显示问题，您只能替换$text中mainSearch中每个字的循环中设置的$wording = str_replace($text, "".$text."", $row['answer']);。因此，它只是集合中的最后一个单词。相反，你需要在这里做一个类似的循环。

您的代码

foreach($query as $text) { $wording = str_replace($text, "".$text."", $row['answer']); }

调整

public class SomeClient { public static void main(String args[]) { TransactionXmlWriter txw = new TransactionXmlWriter(); TransactionType tranType = getNextTransaction(); try { txw.openXmlOutput("someFileName.xml"); while(tranType != null) { txw.processObject(tranType); tranType = getNextTransaction(); } txw.closeXmlOutput(); } catch(JAXBException e) { } catch(FileNotFoundException e) { } catch(XMLStreamExceptoin e) { } } }

Answer 2

您可以使用Google的NLP apis来检测只是连词和单词的单词并且具有显着性的单词，这样您发送到数据库的查询就会忽略像＆＃34; what＆＃这样的单词34;或＆＃34;是＆＃34;这样您就可以为用户提供更好的体验。还有其他工具，如NLTK等，它们也可以用来使查询变得更聪明。

Answer 3

出于性能原因，您应该使用REGEXP而不是多个LIKE说明。订购可以使用简单的ORDER BY指令完成，如下所示：

$keyword=$_GET['mainSearch'];
$pattern=join("|", array_filter(explode(" ", $keyword)));

$sql_query="SELECT * FROM questions
     WHERE (question REGEXP '$pattern') OR (answer REGEXP '$pattern')
     ORDER BY CONCAT(question, answer) LIKE '%$keyword%' DESC;";

使用preg_replace

可以轻松完成突出显示

 preg_replace("/($pattern)/",
       "<span style='font-weight: bold; color: #1a0dab;'>$1</span>",
       $row['answer']);

Answer 4

首先在数据库中运行以下查询。

ALTER TABLE  questions  ADD FULLTEXT(question, answer);

然后使用以下查询进行搜索

if (isset($_GET["mainSearch"]))
{
  $condition = '';
  $mainSearch = SQLite3::escapeString($_GET['mainSearch']);
  $keyword = $_GET['mainSearch'];
  $query = explode(" ", $keyword);


  $puresearch=implode("*+*", $query);

  $myquery = "SELECT *,MATCH(question,answer) AGAINST('*".$keyword."*' IN BOOLEAN MODE) as relavance FROM questions WHERE MATCH(question,answer) AGAINST('*".$keyword."*' IN BOOLEAN MODE) ORDER BY ralavance DESC";

  $result = $db->query($myquery);

}

这是全文搜索，根据相关性为您提供结果，可以按降序排列，以便在第一点显示搜索更可靠的搜索结果。

Answer 5

鉴于此test表

create table test (
  tid integer PRIMARY KEY,
  ttext text NOT NULL
  );
  INSERT INTO test (
  tid,
  ttext
  ) VALUES (1,'what are you doing?'),(2,'this is your house?'),(3,'food chain'),(4,'what is food');

您的结束查询应如下所示

select *,
(case
when ttext like '%food%' then 1
else 0
end+
case
when ttext like '%chain%' then 1
else 0
end) as tsum
from test where ttext like '%food%' or ttext like '%chain%' 
order by tsum desc;

id  text           sum
 3  food chain      2
 4  what is food    1    

select *,
(case
when ttext like '%what%' then 1
else 0
end+
case
when ttext like '%is%' then 1
else 0
end+
case
when ttext like '%food%' then 1
else 0
end) as tsum
from test where ttext like '%what%' or ttext like '%is%' or ttext like '%food%' 
order by tsum desc;

id  text                sum
 4  what is food         3
 1  what are you doing?  1
 2  this is your house?  1
 3  food chain           1

select *,
(case
when ttext like '%this%' then 1
else 0
end+
case
when ttext like '%is%' then 1
else 0
end) as tsum
from test where ttext like '%this%' or ttext like '%is%' 
order by tsum desc;

id  text                sum
 2  this is your house?  2
 4  what is food         1

select *,
(case
when ttext like '%what%' then 1
else 0
end+
case
when ttext like '%is%' then 1
else 0
end+
case
when ttext like '%house%' then 1
else 0
end) as tsum
from test where ttext like '%what%' or ttext like '%is%' or ttext like '%house%' 
order by tsum desc;

id  text                sum
 2  this is your house?  2
 4  what is food         2
 1  what are you doing?  1

您可以在DB Fiddle

上测试上述所有查询

要构建此类查询并假设您使用的是PDO，那么您的PHP代码应该如下所示。

...
// Sample code for 1st example query
$keywrd1 = '%food%';
$keywrd2 = '%chain%';
// Prepare the command
$sth = $dbh->prepare('select *,
(case
when ttext like :keyword1 then 1
else 0
end+
case
when ttext like :keyword2 then 1
else 0
end) as tsum
from test where ttext like :keyword1 or ttext like :keyword2 
order by tsum desc');
// Bind the parameters
$sth->bindParam(':keyword1', $keywrd1, PDO::PARAM_STR);
$sth->bindParam(':keyword2', $keywrd2, PDO::PARAM_STR);
$sth->execute();
...

搜索查询不够准确

5 个答案:

在查询中添加可排序字段

首先我们需要简化您的问题

您的查询已简化

案例[完全匹配]和排序

调整代码

这是做什么的？

高亮

您的代码

调整