regex operators View as Markdown

Hi David,
2 weeks ago Hartmut Holzgraefe had similar idea :
switch ($data, "preg_match") {
case '/foo.?bar/i' : / computation here */
break;
}

The second argument is a callback.

Andrey

David Sklar wrote:

I was thinking about adding one or two regex-related features to the engine:

1. "preg_case": this would behave just like case but instead of doing an
   equality comparison, would match against a regular expression, e.g.

switch($data) {
preg_case '/^\d{5}(-\d{4})?$/':
print "US Postal Code";
break;
preg_case '/^[a-z]\d[a-z] ?\d[a-z]\d$/i';
print "Canadian Postal Code";
break;
default:
print "something else!";
}

Where should any captured subpatterns go?

2. A regex match operator that returns an array containing subpatterns. If
   there is no match against the regex, then an empty array (or just false?)
   would be returned.

if ($data =~ '/^(\d{5})(-\d{4})?$/') {
print "The whole postal code is $data[0].";
print "The first five digits is $data[1].";
if ($data[2]) { print "The ZIP+4 is $data[2].";}
}

Some issues with adding these features:

• It creates an engine dependency on the PCRE library.
• There would have to be some new opcodes and parser tokens
• Ideally the code that implements these operators could share as much as
  possible with what's already been done in the PCRE extension -- is that
  possible?

Comments?

Thanks,
David

David Sklar wrote:

I was thinking about adding one or two regex-related features to the engine:

1. "preg_case": this would behave just like case but instead of doing an
   equality comparison, would match against a regular expression, e.g.

switch($data) {
preg_case '/^\d{5}(-\d{4})?$/':
print "US Postal Code";
break;
preg_case '/^[a-z]\d[a-z] ?\d[a-z]\d$/i';
print "Canadian Postal Code";
break;
default:
print "something else!";
}

i've started to play with a more general way to handle this:

switch(mixed data [, callback compare_function])
{
...

"compare_function" would default to the functionality of the
"==" operator to show the current behavior

"compare_function" will be called with two parameters:
the switch expression and the case value and should
return true if both match, false otherwise

as a special case "compare_function" should also accept
"==="

this way we do not have to add new keywords and are not
limited to preg expresions here,
possible uses not possible now would include:

switch($foobar, "===") {
case false: ...
case 0: ...
case '': ...

switch($foobar, "preg_match") {
case '/^\d{5}$/': ...
case '/(foo)?bar/': ...

switch($foobar, "fnmatch") {
case '.gif': ...
case '.jpg': ...

switch($foober, "my_custom_function") {
...

switch($foobar, array($this, "compare_function") {
...

--
Hartmut Holzgraefe <hartmut@php.net

• Thus wrote David Sklar (sklar@sklar.com):

I was thinking about adding one or two regex-related features to the engine:

1. "preg_case": this would behave just like case but instead of doing an
   equality comparison, would match against a regular expression, e.g.

switch($data) {
preg_case '/^\d{5}(-\d{4})?$/':
print "US Postal Code";
break;
preg_case '/^[a-z]\d[a-z] ?\d[a-z]\d$/i';
print "Canadian Postal Code";
break;
default:
print "something else!";
}
[...]

Some issues with adding these features:

• It creates an engine dependency on the PCRE library.
• There would have to be some new opcodes and parser tokens
• Ideally the code that implements these operators could share as much as
  possible with what's already been done in the PCRE extension -- is that
  possible?

I'm thinking a more viable solution would be to do something like:

mixed preg_match ( mixed pattern, string subject [, array matches [, int flags]])

if pattern is an array, preg_match will cycle through each one till
a match is made; returns the index of the array passed (that
matched) or false if not found.

This should also be BC if you just pass one string.

Curt

"I used to think I was indecisive, but now I'm not so sure."