[SOLVED] preg_replace help

[SharkBait]

Ok I've been trying to figure out the use of regex and hit a bit of a snag.

 

i am trying to massage some data and if the string after the comma is larger than 2 characters I want to drop the comma

 

Example:

<?php
$string = 'My Item, N-female';

$pattern = '\(\w+),(\s+)(\w{3})/i';
$replacement = '$1 $3';

$temp = preg_replace($pattern, $replacement, $string);
?>

But I think the issue I am having is with the hyphen and the above pattern will not drop the comma when but when the string is

 

<?php
$string = 'My Item, Fun';
?>

Then it does drop the comma.

 

I am trying to make it so that it drop the comma if the characters after the comma are 2 and less. But if the 2nd character is a hyphen it doesn't drop the comma.

 

What do I need to do?

 

[DarkWater]

/,(\s+.{2}?)/

$1

 

<?php
$string = 'My Item, N-female';

$pattern = '/,(\s+.{2}?)/s';
$replacement = '$1';
echo preg_replace($pattern, $replacement, $string);
?>

[SharkBait]

Ok, though if

 

<?php
$string = 'My Item, OT';
?>

 

I don't want to drop the comma. Though if it has a hyphen then i want to drop the comma.

[DarkWater]

<?php
$string = 'My Item, N-female';

$pattern = '/,(\s+.{3}?)/s';
$replacement = '$1';
echo preg_replace($pattern, $replacement, $string);
?>

[effigy]

<pre>
<?php
$string = 'My Item, N-female';
echo preg_replace('/,(?=\s+.{3})/s', '', $string);
?>
</pre>

[DarkWater]

Ah, forgot about using a positive lookahead.  Good call.

[SharkBait]

Ok I'm slowly getting the idea... perhaps this will help more in explaining what I am trying to do

 

<?php
$string = '9/22/2008,Sales,Sales History Ranging From  9/15/2008  To  9/19/2008,Page #  0001,Date,Customer,City & State,ZIP Code   Qty, UM,Part Number:,TR-5amp-Nf,"5GHz AP/PtP/CPE, N-female",9/16/2008,"Company. LLC","Nevada, MO",64772,1,EA,Brand Total,799';
?>

 

I want to drop the comma in the "5GHz AP/PtP/CPE, N-Female" but keep the comma in the "Nevada, MO" so when I explode the string it will hopefully seperate the Nevada and MO only.

[DarkWater]

Is it that you don't want things in " " being exploded, like a CSV?

[SharkBait]

I do and I don't. That's the trick.

 

There is a field that is City, State and I need to leave the comma in, but all the other fields  i would like to remove the comma from.

 

I can do this manually of course, but the end result is that a persona can upload a file, have it corrected for these issues, then write it to a new file which is then loaded into a MySQL table. Unfortunately 1 field needs to be split, the others do not.

 

its why i was trying to match against 2 characters after the comma because in the data files the state is always 2 characters, anything more and it's not a state (not the right field).

[DarkWater]

Did you try my revised regex or effigy's?

[SharkBait]

Yes I've tried both.

[DarkWater]

<?php

$string = '9/22/2008,Sales,Sales History Ranging From  9/15/2008  To  9/19/2008,Page #  0001,Date,Customer,City & State,ZIP Code   Qty, UM,Part Number:,TR-5amp-Nf,"5GHz AP/PtP/CPE, N-female",9/16/2008,"Company. LLC","Nevada, MO",64772,1,EA,Brand Total,799';

$pattern = '/,(?=\s+.{3})/';
$replacement = '';
echo preg_replace($pattern, $replacement, $string);
?>

 

That seems to work for me.

[SharkBait]

weird so it removes the comma from in front of N-Female but leaves it in for the Nevada, MO ??

[DarkWater]

Ah, I see the problem.  We all wrote out regexes assuming that your sample string was the whole string.  You'd need:

 

<?php

$string = '9/22/2008,Sales,Sales History Ranging From  9/15/2008  To  9/19/2008,Page #  0001,Date,Customer,City & State,ZIP Code   Qty, UM,Part Number:,TR-5amp-Nf,"5GHz AP/PtP/CPE, N-female",9/16/2008,"Company. LLC","Nevada, MO",64772,1,EA,Brand Total,799';

$pattern = '/,(?=\s+[\w\d-]{3})/';
$replacement = '';
echo preg_replace($pattern, $replacement, $string);
?>

 

That actually works as you want (I just tested this one.  I tested the other one too, but I forgot to look at the Nevada part of the output. =P)

[SharkBait]

Ah cool that works nicely. So let me see if I can explain it in english as to what the regex does so i can understand what it does.

 

 

?=\s+

if after a comma and whitespace

[\w\d-]

can be word, numbers or hyphens

{3} 3 or above characters
[code]

right?

[/code]

[DarkWater]

Actually:

 

/ - start regex
, - match a single comma
(?= - begin positive lookahead
\s+ = match one or more space characters after the comma
[\w\d-]{3} - match 3 characters in A-Za-z_0-9-
) - end positive lookahead
/ - end regex

 

Then, since a positive lookahead doesn't actually match anything, you can use '' as the replacement.

[SharkBait]

Thanks for the help! Much appreciated