regex to remove variable prefix with or without a delimiter - php

Question

I am trying to process historic military service numbers which have a very variable format. The key thing is to remove any prefix, but also to keep any suffix. Prefixes most commonly have a delimiter of a space, slash or dash, but sometimes they do not. In these cases the prefix is always one or more uppercase letters. In all other cases both prefixes and suffixes can contain letters or numbers and whilst typically uppercase, can be lower!
Currently my php code is
$cleanServiceNumber = preg_replace("/^.*[\/\s-]/","",$serviceNumber)
and typical values and desired results are
AB/12345 => 12345
CD-23456 => 23456
EF 34567 => 34567
5/45678 => 45678
GH/56789/A =>56789/A
GH/56789B => 56789B
XY67890 => 67890 <<< fails to do any replace and returns XY67890
I'm afraid my basic regex skills are failing me in terms of sorting the last example!

Answer 1

This regex replaces the combination of 0 to n digits and n non-digits at the beginning of the string: /^\d*\D+/
Demo
$serviceNumbers = array(
'AB/12345',
'CD-23456',
'EF 34567',
'5/45678',
'GH/56789/A',
'GH/56789B',
'XY67890');
foreach ($serviceNumbers as $serviceNumber) {
$cleanServiceNumber = preg_replace("/^\d*\D+/","",$serviceNumber);
echo $cleanServiceNumber . "\n";
}
Output:
12345
23456
34567
45678
56789/A
56789B
67890

Answer 2

You can add an alternation of [A-Z]+, but you should also make the other alternation more efficient by searching for non-delimiter characters followed by a delimiter:
$cleanServiceNumber = preg_replace("/^(?:[^\/ -]+[\/ -]|[A-Z]+)/","",$serviceNumber);
Demo on regex101
PHP demo on 3v4l.org

Answer 3

Here is another try for a regex which looks like:
/^([A-Za-z]+(\d+\W|\W)?|\d+\W)/
It has 2 parts which detects the type of prefixes you have:
[A-Za-z]+(\d+\W|\W)? => Any alphabets ending with non word character or alphabets having numbers and then ending with non word character. However, this ending game is optional with a ? at the end.
\d+\W => Any digits followed by a non word character.
Snippet:
<?php
$tests = [
'AB/12345',
'CD-23456',
'EF 34567',
'5/45678',
'GH/56789/A',
'GH/56789B',
'XY67890',
'XY67890/90/A'
];
foreach($tests as $test){
echo $test," => ",preg_replace("/^([A-Za-z]+(\d+\W|\W)?|\d+\W)/","",$test),PHP_EOL;
}
Demo: https://3v4l.org/9hJLJ

Answer 4

The pattern you tried ^.*[\/\s-] first matches until the end of the string because the dot is greedy. Then it will backtrack until it can match either a /, - or a whitespace char.
This will not work for GH/56789/A as it will backtrack until the last / and it will not work for XY67890 as it does not match any of the characters in the character class.
You could match from the start of the string either 1 or more chars a-zA-Z or 1 or more digits 0-9 and at the end match an optional /, - or a horizontal whitespace character.
^(?:[A-Za-z]+|\d+)[/\h-]?
Regex demo | Php demo
For example
$serviceNumbers = [
"AB/12345",
"CD-23456",
"EF 34567",
"5/45678",
"GH/56789/A",
"GH/56789B",
"XY67890"
];
foreach ($serviceNumbers as $serviceNumber) {
echo preg_replace("~^(?:[A-Za-z]+|\d+)[/\h-]?~","",$serviceNumber) . PHP_EOL;
}
Output
12345
23456
34567
45678
56789/A
56789B
67890