Speaking just for myself (and not cfengine today), I think this is a good idea.
As long as you're at it, you probably want a tolower() and toupper() function.
You can get "ignore case" on regular expression matching with PCRE '(?i)',
but converting to lower or to upper on output is harder (and if you want
convergence, you DON'T want what Perl does... :-)
Also, in your model, regreplace needs a 4th argument - "first" or "all", to
give you the equivalent of /g substitution. While "first" can be a bad
thing, there is already precendent for that in Cfengine in the replacement of
lines in files...
-Dan
> On Tue, Mar 02, 2010 at 08:33:15AM -0500, nwat...@symcor.com wrote:
> >I don't think there is a function at present for string search a replace
> >actions out side of file editing. Perhaps look at it from a higher level.
> > Why is string manipulation needed?
>
> There are two reasons, one specific, and one general.
>
> The specific reason is that I am trying to build a somewhat more robust
> template library (which if I ever get to work, I will post to the list).
>
> To do this, I am trying to convert a file from the form of
> "/some/path/to/file" into something like
> "/path/to/repo/_some_path_to_file".
>
> This is trivial with two regexes. Using quasi-cfengine notation, and
> a hypothetical regreplace(string,find,replace) function, something like
> this would work:
>
> vars:
> "orig" = "/some/path/to/file";
> "new" = regreplace(
> regreplace("${orig}", "/", "_" ),
> "^",
> "/path/to/repo"
> );
>
>
>
> The general reason is that string search and replace is a very useful
> tool. It doesn't really matter the what the specific problem actually
> is, since this is a more general issue. A good tool is useful in lots
> of different situations, including those in which it was not originally
> intended.
>
> There have been several occasions where a simple string replacement would
> make certain promises simpler, but I have attempted to work around them.
> But having to shell out and make two calls to outside binaries--one
> of them being echo!--just to do a simple string manipulation seems
> counter-productive, inefficient, and much more error prone (especially
> due to the hazards of shell quoting).
>
> Here's another example, albeit slightly contrived. You have a system
> where you want to start hashing user home directories in the form of
> "jbecker -> /home/j/jbecker" (for example). Based on some
> list of users, there is a set of promises to ensure that the accounts
> exist, including calling 'useradd'. Having a short set of promises to
> convert the string "jbecker" into "/home/j/jbecker" is fairly easy using
> a search and replace function. I can envision a situation where
> cfengine could actually handle the migration of existing "unhashed" user
> home directories to "hashed" ones, but only if it supports better string
> handling routines.
>
> Of course, one can always just hand off a username to a Perl script as
> well.
>
> I really want is a general solution (which is why I did not simply ask
> how to munge a path). Once I--and everyone else--has that, the
> specifics should be simple. :)
>
>
>
> --
> Jesse Becker
> NHGRI Linux support (Digicon Contractor)
> _______________________________________________
> Help-cfengine mailing list
> Help-cfengine@cfengine.org
> https://cfengine.org/mailman/listinfo/help-cfengine
_______________________________________________
Help-cfengine mailing list
Help-cfengine@cfengine.org
https://cfengine.org/mailman/listinfo/help-cfengine
