Question

Here’s the problem: user is presented with a text field, unto which may he or she type the filter. A filter, to filter unfiltered data. User, experiencing an Oracle Forms brainwashing, excpets no special characters other than %, which I guess more or less stands for “.*” regex in Java.

If the User Person is well behaved, given person will type stuff like “CTHULH%”, in which case I may construct a pattern:

Pattern.compile(inputText.replaceAll("%", ".*"));

But if the User Person hails from Innsmouth, insurmountably will he type “.+\[a-#$%^&*(” destroying my scheme with a few simple keystrokes. This will not work:

Pattern.compile(Pattern.quote(inputText).replaceAll("%", ".*"));

as it will put \Q at the beginning and \E at the end of the String, rendereing my % -> .* switch moot.

The question is: do I have to look up every special character in the Pattern code and escape it myself by adding “\\” in front, or can this be done automagically? Or am I so deep into the problem, I’m omitting some obvious way of resolution?

Answer 1

What about Pattern.compile(Pattern.quote(inputText).replaceAll("%", "\\E.*\\Q"));?

This should result in the following pattern:

input:   ".+\[a-#$%^&*(" 
quote:   \Q".+\[a-#$%^&*("\E 
replace: \Q".+\[a-#$\E.*\Q^&*("\E

In case the % character was the first or last character you would get a \Q\E (if you only have the input % the expression would end up being \Q\E.*\Q\E) but this should still be a valid expression.

Update:

I forgot the difference between replace(...) and replaceAll(...): the replacement parameter in the former is a literal while the replacement in the latter is an expression itself. Thus – as you already stated in your comment – you need to call Pattern.compile(Pattern.quote(inputText).replaceAll("%", "\\\\E.*\\\\Q")); (quote the backslashes in the string and in the expression).

From the documentation on String#replaceAll(...):

Note that backslashes in the replacement string may cause the results to be different than if it were being treated as a literal replacement string.