Re: Question about CL-PPCRE

>>>>> On Tue, 23 Jun 2026 20:47:32 +0200, Marco Antoniotti (as marco dot antoniotti at unimib dot it) said:
> 
> Hi
> 
> I am sure I missed something in the CL-PPCRE documentation, but here it is.
> 
> I am fooling around with a C tokenizer and I am running into the following
> problem.
> 
> The string esre contains the "escape sequences" for C (and Perl, etc etc)
> 
> CL Prompt > esre
> "(\\\\(['\\\"\\?\\\\abfnrtv]|[0-7]{1,3}|x[a-fA-F0-9]+))"
> 
> If I use CL-PPCRE:REPLACE-ALL I get the following (I tried to boil it down):
> 
> CL Prompt > (regex-replace-all "\\$<ES>" "/$<ES>/" esre)
> "/(\\(['\\\"\\?\\abfnrtv]|[0-7]{1,3}|x[a-fA-F0-9]+))/"
> T
> 
> Apart from the bounding slashes, I would have expected the result of
> CL-PPCRE:REPLACE-ALL not to return a shorter string.
> 
> What exactly (backslash escaping for sure) am I missing?

It looks like an undocumented feature of the replacement string: \\ is
converted to a single backslash by CL-PPCRE::BUILD-REPLACEMENT-TEMPLATE.  That
means it will always return a shorter string when \\ is present in the
replacement string and if you want \\ in the output then you'll need to use
\\\\ (i.e. 8 backslashes in the Lisp syntax of that string!).

Also, to make it more confusing, if \ is followed by anything other than \ & `
' digits or {digits} then it is also treated as a single backslash.  i.e. \a
is \a.

-- 
Martin Simmons
LispWorks Ltd
http://www.lispworks.com/

_______________________________________________
Lisp Hug - the mailing list for LispWorks users
[email protected]
http://www.lispworks.com/support/lisp-hug.html