RegEx to match expression even if it is not 100% the same - php

Question

How would you write a regular expression pattern that matches a string even if it is 90% accurate?
For example:
$search_string = "Two years in, the <a href='site.com'>company</a> has expanded to 35 cities, five of which are outside the U.S. "
$subject = "Two years in,the company has expanded to 35 cities, five of which are outside the U.S."
The end result is that the $search_string matches the $subject and returns true even though they are not 100% the same.

Answer 1

You can have some optional parts on the regex pattern. For example:
$search_string = "A tiny little bear";
$regex = "A ([a-zA-Z]+)? little bear";
The ? character there says that the group before it goes optional, and the [a-zA-Z]+ indicates there will be one or more letters inside it.
Thus, using preg_match you can get a validation not 100% restrictive.

Answer 2

in case any one comes around looking for the right way to do it
$search_string = "Two years in, the <a href='site.com'>company</a> has expanded to 35 cities, five of which are outside the U.S. ";
$subject = "Two years in,the company has expanded to 35 cities, five of which are outside the U.S.";
similar_text ($search_string,$subject,$sim);
echo 'text is: ' .round($sim). '% similar';
result:
text is:85% similar
you can use the result to determine what value is a match in your particular circumstances like so:
similar_text($search_string,$subject,$sim);
if($sim >=85){
echo 'MATCH';
}

Answer 3

Just for grins, I tried this out using Perl.
All the warnings about using regex to parse html apply:
(Should not use on html).
This will split the Search string on either html or entities or whitespace.
After that, the parts are joined with .*? using the modifiers (?is).
This is not a true partial matching substring regex because
it requires all the parts to exist.
This does overcome the distance or content between them however.
Possibly, with a little algorithm work, it could be tweaked in such
a way that parts are optional, in the form of clustering.
use strict;
use warnings;
my $search_string = "Two years in, the <a href='site.com'>company</a> has expanded to 35 cities, five of which are outside the U.S. ";
my $subject = "Two years in,the company has expanded to 35 cities, five of which are outside the U.S.";
## Trim leading/trailing whitespace from $search_string
$search_string =~ s/^\s+|\s+$//g;
## Split the $search_string on html tags or entities or whitespaces ..
my #SearchParts = split m~
\s+|
(?i)[&%](?:[a-z]+|(?:\#(?:[0-9]+|x[0-9a
-f]+)));|<(?:script(?:\s+(?:"[\S\s]*?"|'
[\S\s]*?'|[^>]*?)+)?\s*>[\S\s]*?</script
\s*|(?:/?[\w:]+\s*/?)|(?:[\w:]+\s+(?:(?:
(?:"[\S\s]*?")|(?:'[\S\s]*?'))|(?:[^>]*?
))+\s*/?)|\?[\S\s]*?\?|(?:!(?:(?:DOCTYPE
[\S\s]*?)|(?:\[CDATA\[[\S\s]*?\]\])|(?:-
-[\S\s]*?--)|(?:ATTLIST[\S\s]*?)|(?:ENTI
TY[\S\s]*?)|(?:ELEMENT[\S\s]*?))))>
~x, $search_string;
## Escape the metacharacters from SearchParts
#SearchParts = grep { $_ = quotemeta } #SearchParts;
## Join the SearchParts into a regex
my $rx = '(?si)(?:' . ( join '.*?', #SearchParts ) . ')';
## Try to match SearchParts in the $subject
if ( $subject =~ /$rx/ )
{
print "Match in subject:\n'$&' \n";
}
Output:
Match in subject:
'Two years in,the company has expanded to 35 cities, five of which are outside the U.S.'
edit:
As a side note, each element of #SearchParts could be further split//
once again (on each character), joining with .*?.
This would get into the realm of a true partial match.
Not quite there though as each character is required to match.
The order is maintained, but each one would have to be optional.
Usually, without capture groups, there is no way to tell the percentage
of actual letter's matched.
If you were to use Perl however, it's fairly easy to count in
regex Code construct (?{{..}}) where a counter can be incremented.
I guess, at that point it becomes non-portable. Better to use C++.