smart eregi_replace possible?

[jkeppens]

For a template engine, I'd like to subsitute the following :

Hi <var:user_name />

to

Hi Jeroen

when the variable user_name = "Jeroen".

I've come up with following pattern :

(<var:)([a-zA-Z0-9_])[ ](/>)

And I get "user_name" perfectly in my \2 for eregi_replace :

eregi_replace("(<var:)([a-zA-Z0-9_])[ ](/>)", "\2", $val);

Now, it isn't really what I want... I would like to do this :

eregi_replace("(<tag:)([a-zA-Z0-9_])[ ](/>)", ${"\2"}, $val);

but it isn't possible. I'd rather not parse the string after replacing, so anyone have an idea how to accomplish this?

so no eval(eregi_replace("(<tag:)([a-zA-Z0-9_])[ ](/>)", "\$\2", $val));

nx

[MrPixel]

There is a considerably simpler way to do this:
Template:

Hi, <NAME>!

PHP:

//$template = the code/text shown above
$name = 'MrPixel';
print str_replace('<NAME>', $name, $template);    //output: Hi, MrPixel!

Really easy, really fast. You can use arrays, with str_replace() as well. Once I discovered this function, I forswore other templating systems - it is easy, fast, flexible (with arrays), and facilitates the separation of logic and presentation.

[jkeppens]

Yes, but then you have to check for all possible variables... In that system you have a list of variables you do a replace for, what I'd like to do is get the variables in the template itself that need to be replaced and replace those.

[MrPixel]

Getting that list of tags dynamically is simple enough (simpler with non-XML style tags.):

$template = 'The hypotenuse of a right triangle with sides {A} and {B} is {C}';
//Let's use the following tag format: only caps or an underscore inside curly braces - it'll be harder to get mismatches, this way.
preg_match_all('/\{[A-Z_]\}/', $template, $tags);
$values = array(3, 4);
$values[] = hypot($values[0], $values[1]);
print str_replace($tags[0], $values, $template); //Output: They hypotenuse of a right triangle with sides 3 and 4 is 5.

But no matter what you do, you will always have to make sure that the values belong where they are inserted. Doing this in a truly robust manner (ie., without preknowledge of the tags, what goes into them, or the context) is an AI task; it relies on the meaning and context of the output.

I don't have the time for that, so I choose to rely on order ... with slightly more fragile scripts as a result.

[Bloodwine]

Originally posted by jkeppens

eregi_replace("(<var:)([a-zA-Z0-9_])[ ](/>)", ${"\2"}, $val);

preg_replace('/(<var:)([a-z0-9_])[ ](\/>)/ie', ${\2}, $val);

Using the 'i' modifier with preg_replace makes the matching case-insensitive

using the 'e' modifier with preg_replace causes the replacement value to be evaluated