PHP从MySQL全文搜索中突出显示多项结果

Question

A lot of people have asked questions about how to highlight search terms in the results from a MySQL Full Text Search, but the solutions that I have found do not appear to be sufficient. The MySQL Full Text Search supports quoted multi-word search targets and ignores any punctuation. Thus, on my search form the user could enter the quoted term "quick brown" and MySQL will also return rows that include "quick, brown" (for example), but not "quick and brown". So when highlighting the search target in the returned text, it seems you need to do some regular expressions to make sure you identify all instances of the target. What I have so far is this, where $targ contains a possibly multiword search term such as "quick brown" (but without the quotes), and $blob is the large string of text that we are searching through. It works by replacing any spaces in the search target with a regular expression that matches any non-alphanumeric string.

$pattern = '/' . str_replace(" ", '[^A-Za-z0-9\"]', $targ) . '/i';
$replacement = '<span class="hilite">' . $targ . '</span>';
$blob = preg_replace($pattern, $replacement, $blob);

This mostly works, but with one unfortunate side-effect. It actually deletes the extra punctuation from the full string. So if $blob contains the string "quick, brown" this gets changed to

<span class="hilite">quick brown</span>

So it succeeded in adding the span tags around the term, but in the process, it removed the comma.

I think the solution probably involves using preg_replace with wildcards, but one difficulty is that $targ could have varying numbers of words in it.

Any suggestions? Thanks!

Answer 1

You can do this pretty easily with capture groups. Also when accepting input from the user for a pattern it's a good idea to escape it before using it.

<?php
$source = 'quick, brown" (for example), but not "quick and brown". So whe';
$test = "QUICK BROWN";
$temp = explode(" ", $test);
// use preg_quote to escape the literal strings
$temp = array_map(function ($val) { return preg_quote($val, "~");}, $temp);
// stitch the literals together with the variable pattern 
$pattern = '~('.implode("[^A-Za-z0-9]+", $temp).')~i';
// here the $1 means the result of the first capture group... 
// capture group is denoted in the pattern as the text in (). 
echo preg_replace( $pattern , '<span class="hilite">$1</span>', $source );

You can see this running here .