From Ze'ev Atlas and contains a port of the PCRE2 (Perl-Compatible Regular Expressions) product to z/OS, a later release of PCRE.
NOTE: This Git Repository was created and managed by ZIGI (the z/OS ISPF Git Interface. It is strongly suggested that before installing thie repository that you install ZIGI from https://github.com/wizardofzos/zigi
(This note pertains to PCRE release 10.35)
Please see the documents:
EBCDIC_Horror.txt member $EBCDIC
REXXAPI.txt member REXXAPI
If you are using code-page 1047 (common in the U.S.), a quick
install is being presented here, in the form of 6 XMIT-format
files. These need to be RECEIVE'd under TSO, and they will be
almost ready to go. You still need to read the documentation on
the PC (and the member PCRE2DOC).
It is very important to read below, which will explain what to do
if you need other code pages.
Port is of PCRE Version 10.35, known as PCRE2.
Current release is marked with ISPF statistics as 10.35.
email: zatlas1@yahoo.com
web site: http://www.zaconsultants.net
PCRE info: http://www.pcre.org (please go there)
=======================================================================
Short description:
Regular Expressions is a technology that allows text search and manipulation in ways that go above and beyond what is known even to extreme Rexx programmers. The lack of Regular Expressions in COBOL contrasts with their availability in Perl and Java and is always frustrating to me. Thus, I decided to port a Regular Expressions library and make it available on native z/OS. I chose the pcre2 which is considered to be the best publicly available such library in that it is compatible with the standard bearer, Perl. pcre2 is used in PHP and other projects.
This is the fifteenth version and is compatible with pcre2 10.35. The package is distributed as Open Source, as is with no warranty and under the BSD license that is pretty open and non-limiting.
=======================================================================
Clarification for other EBCDIC code pages
As per Sam's request, I'll try to clarify the code page issue a bit farther. There are two, related but not totally connected issues.
-
IBM set their C compiler to work with IBM1047, however all the machines that I have access to seem to be set to IBM1140 which is the old '037' code page. I had to do some acrobatics and replace the caret sign with the IBM logical not (which is the caret in IBM1047) for the thing to work.
-
Other languages' EBCDIC code pages issue.
2.1. Originally, I thought to set some switch in the config.h header file, in order to allow some logic for various languages and code pages. This was proven to be unnecessary, but I did not remove the switches from the header in case I will need them sometime in the future. You should ignore those switches.
2.2. OTH, since the only machines that I have access to are in the USA, I could not compile or test other languages. Therefore, the supplied LOADLIB is for USA English. If you want to work with other languages like Russian, Hebrew or Turkish, you must compile your own version of the LOADLIB.
The good news is that if you run the COMPCRE2 member in the JCLLIB and all is compiled and bound with RC=4 or less, then you are good. The first step builds the correct character table for your machine and the second step compiles it.
The bad news is that if your caret is in the other place, you will have to transform it all over the place, especially in the SRCE and TESTLIB libraries.
Note that there is one line in the source code, in the PCRZOSCS header file that is marked with ''. The caret sign in that line must stay as caret in the source code.
Ze'ev Atlas