DasBlog has a ContentFilter feature that allows you to replace a regular expression with a match. For example, $g(expression) is rewritten as <a href="http://www.google.com/search=expression">expression</a>.

My problem is when "expression" contains spaces. For the querystring part, the spaces should be replaced by spaces; but not in the visible text of the anchor tag. In other words:
    $g(regex advice)
should map to
    <a href="http://www.google.com/search=regex+advice">regex advice</a>

The current rule looks like
    \$g\((?<expr>[\w\s\d]+)\)
    <a href="http://www.google.com/search?q=${expr}">${expr}</a>
which will leave spaces in the querystring.

Any ideas?

/George V. Reilly, Seattle, Wash.

Writing this as two or more rules is a possibility, but I haven't figured out how to express it like that.

What platform are you on? PHP provides preg_replace(), and .NET's RegularExpressions namespace has a MatchEvaluator class that allows you to programmatically handle the replacement. You want to do two entirely unrelated things to the match: HTML-encode it (just spaces?!?), and enclose it in a URL as in your example. No one expression is going to do that for you. If what's inside the parentheses is the match, then that's what the expression will return; you can't also find all the spaces (of which there could be any number, at any location) and replace them with '+' in a single step. I'd be shocked to learn that PHP didn't provide an HTML-encode function that you could invoke within your replacement function. You could also encode spaces by means of a second regex, from within the above replacement function, but that would be unconscionably cute if the platform already provided a way to do it.

---

"Some day, and that day may never come, I will call upon you to do a service for me." — Don Vito Corleone

DasBlog is written in ASP.NET: http://www.dasblog.info/. So yes, there's an HTML encoder available, which is a good idea if the expression contains any punctuation characters. As it stands, it only matches [\w\d\s]+, so only the \s needs escaping.

I was hoping (though with little expectation of success) that a pure regex solution existed. The regex and the replacement both live in an XML config file. The C# source can be modified, of course, to do something with MatchEvaluators.

The workaround for now is to write out the Google search URL explicitly.