COBOL using XML Parser

pleonard · New User Joined: 20 Sep 2006 Posts: 19

I'm processing the XML through program COBOL z/OS (I guess we use 3.4 compiler version). On this XML I have something called "encoding=utf-8" which is a lot of problems to process. The parser is out with xml-code=70 (exception) and I already attempting to change the codepage option using CODEPAGE(1208) but I got xml-code=320.

Looks like in COBOL 4 there are WITH ENCODING to be used in XML Parser but I have COBOL 3.4. Are there a way to use xml parser to processing this encoding using cobol 3.4 ?

Regards
Paulo

Robert Sample · Posted: Sat Dec 13, 2008 6:13 pm

utf-8 is essentially ASCII, upgraded for Unicode. The XML encoding is determined by whatever generated the XML; if you don't change the encoding there you're wasting your time.

If you convert the file from ASCII to EBCDIC, you may be able to change the encoding from utf-i to ebcdic-cp-us and be able to read it in COBOL. However, that's not a viable long term solution IMHO since you need to be handling the code as specified by the generator.

pleonard · New User Joined: 20 Sep 2006 Posts: 19

How can I convert the message from ASCII to EBCDIC to manipulate it in COBOL ?

Robert Sample · Posted: Sat Dec 13, 2008 7:37 pm

Unix System Services command iconv will do the conversion, if the file is in Unix -- I'd probably OPUT it, convert it, and OGET it if this is a testing process underway.

If iconv isn't suitable, when you do the transfer do it as a text file.

pleonard · New User Joined: 20 Sep 2006 Posts: 19

It will be a problem. My program is called by a CICS transaction that passes a commonarea and XML DOCUMENT as parameters with encoding there.

Robert Sample · Posted: Sat Dec 13, 2008 9:29 pm

In that case, you can process it using UTF-8 or get the encoding changed on the source side.

pleonard · New User Joined: 20 Sep 2006 Posts: 19

but using XML PARSE I got XML-CODE=70. I tried to use CODEPAGE compile option with CODEPAGE(1208) NSYMBOL(NATIONAL) and is getting XML-CODE=320.

dick scherrer · Posted: Sun Dec 14, 2008 12:39 am

Hello,

You are changing "things" on the mainframe-side of the process. . . .

You need to have the file changed on the system that generates the file as Robert suggested.

pleonard · New User Joined: 20 Sep 2006 Posts: 19

In that case, you can process it using UTF-8 or get the encoding changed on the source side.

How can I do that ?

dick scherrer · Posted: Sun Dec 14, 2008 2:21 am

Hello,

Robert Sample · Posted: Sun Dec 14, 2008 3:20 am

I have not done a lot of work on XML so I can't say 100% for sure what will work for you, but I find in the COBOL Language Reference manual section 6.2.41.3.2:

pleonard · New User Joined: 20 Sep 2006 Posts: 19

After did the changes I got XML-CODE=317

317 The parser cannot determine the document encoding. The document might be damaged.

Robert Sample · Posted: Sun Dec 14, 2008 8:16 pm

Going back to your original post, the 70 is listed in the Programming Guide as meaning

pleonard · New User Joined: 20 Sep 2006 Posts: 19

Now I got XML-CODE=73

73 The actual document encoding was EBCDIC, but neither the CODEPAGE compiler option nor the document encoding declaration specified a supported EBCDIC code page. The parser uses EBCDIC code page 1140 (USA, Canada, . . . Euro Country Extended Code Page).

Robert Sample · Posted: Mon Dec 15, 2008 1:21 am

OK, recompile without the codepage option and change the encoding= string in the file from utf-8 to EBCDIC; basically the problem you keep getting is that the encoding= name must match the actual file encoding. Your file does not match; it appears to be an EBCDIC file with encoding=utf-8 specified. You can either change the file encoding by changing the file (and I do not know how to recode it from EBCDIC to utf-8 without additional research), or change what is expected by changing the encoding= string within the file.

pleonard · New User Joined: 20 Sep 2006 Posts: 19

I put EBCDIC in place of UTF-8 in the file, keep the MOVE FUNCTION DISPLAY OF with DBCS option (it is required when we use DISPLAY OF function) however, I got xml-code=317 again making move from ASCII to EBCDIC. Attempting to move from EBCDIC to ASCII I got xml-code=83.

Robert Sample · Posted: Mon Dec 15, 2008 6:43 pm

The 83 is promising -- it indicates the XML parser recognizes the file as being ASCII but the encoding didn't match -- try encoding='iso8859-1' for this file since that would indicate Latin-I ASCII.

Also, note that error codes 1 to 99 are error codes that XML handles and supposedly allows you to continue; have you checked the output field with the 70 or 83 error codes?

pleonard · New User Joined: 20 Sep 2006 Posts: 19

I changed the file to be 'iso8859-1' and resubmit and got xml-code=317. How can I interpose the xml-code 70 or 83 in my program ?

Robert Sample · Posted: Mon Dec 15, 2008 7:25 pm

You've hit the limits of my knowledge -- if there's anybody with XML expertise at your site, you need to talk over your problem with them. The 317 is not an error that can be continued; 70 or 83 allow you to set XML-CODE to zero and continue processing according to the manual.

The error codes you've posted are consistently encoding conflict errors -- where the file is EBCDIC and the encoding= option of XML does not reflect an EBCDIC encoding, or the file is ASCII and the encoding= option does not reflect an ASCII encoding. The difficulty is finding an encoding string (819 or iso8859-1 or whatever) that allows the ASCII file to be processed, or the EBCDIC file to be processed with encoding=1047 or whatever: EBCDIC would be better since the mainframe would allow you to view the output but getting it processed would be the first thing.

dick scherrer · Posted: Tue Dec 16, 2008 1:18 am

Hello,

To return to an earlier part of the topic - it may be helpful to try re-creating the file and using this new file for continued testing. . .

Robert Sample · Posted: Tue Dec 16, 2008 2:25 am

OK, I looked at an old program I wrote a year and a half ago -- the data does not include any encoding=. So make sure your file is EBCDIC, remove the encoding= from the first line, and run your XMLPARSE again. This should get you past the XML error -- I got XML-CODE=0 by doing so.

If you absolutely have to have the encoding, make sure the file is in EBCDIC, check the compile listing for the CODEPAGE used, and set the encoding to that CODEPAGE -- I thought we used 1047 but found out 1140 is selected by looking at a compile output.

pleonard · New User Joined: 20 Sep 2006 Posts: 19

I tried this one too. In my last test I removed the total header from XML file but for both (with header and without encoding=utf-8) I got XML-CODE=-1. I'm not sure which is the encoding made to this file because it is obtained from another program (that generates this XML to me). Maybe this program can be messed up this file I'm not sure. Indeed my program is attempting to process a XML message and their encoding are different between them.