Tue 08 Nov 2005 07:25:39 PM UTC, original submission:
Hi. First of all, I should say that I greatly appreciate the gnu.regexp package. I have done a lot of searching and I finally found a regular expression package for Jave that did pipelined stream searches, which is the feature that really wanted.
There are some features of other packages and some features that aren't in any package I know of, that I would like to see included in gnu.regexp:
1) unlimited capture groups and named capture groups
2) support for finding partial matches
3) access after a match to all substrings captured by a capture group rather than just the last substring
For example, if we're searching for a list of numbers
enclosed in parentheses and separated by commas (e.g. "(123,456,789)", the pattern might be "\((\d+)(?,(\d+))*\)". However, with ordinary regex capturing, this would only capture the first and last members of the list "123" and "789" as $1 and $2 respectively. I am proposing that captures be stored in some kind of tree structure so that there would be a way to retrieve all list elements, include "456" in my example.
Thank you. Eli <ecooper@mathstat.umass.edu>
|