How to write multiple lines with different layouts into a file?

[Martin Scheibe]

Hello,

(First and foremost, i like to thank Bruce Martin for his dedication, work and support. I think he helped a lot of people.)

Im trying to build a custom .bin file using the JRecord libary.
The file im trying to write has 3 kinds of lines (AbstractLines) each with a different layoutDetail.

I can get one layoutDetail from a copybook (in .CPY format) but the other two layouts i have to custom build.

So far i got code for the copybook layoutDetail, what seems to work as intended.

LayoutDetail layoutDetail =
JRecordInterface1.COBOL
                .newIOBuilder(copybook)
                .setFont("IBM01141")
                .setFileOrganization(IO_FIXED_LENGTH)
                .getLayout();

This is my attempt to build a custom layoutDetail:

LayoutDetail layoutDetail = JRecordInterface1.FIXED_WIDTH
                .newIOBuilder()
                .setFont("IBM01141")
                .defineFieldsByLength()
                .addFieldByLength("TEST1", Type.ftChar, 1, 0)
                .addFieldByLength("TEST2", Type.ftChar, 1000, 0)
                .endOfRecord()
                .getLayout();

This is the writer im using:

AbstractLineWriter writer = JRecordInterface1
                .COBOL
                .newIOBuilder(copybook)
                .setFont("IBM01141")
                .setFileOrganization(IO_FIXED_LENGTH)
                .newWriter(fileOut);

This is how im creating a new AbstractLine and feed it with the respective layoutDetail:

AbstractLine line =
LineIOProvider.getInstance().getLineProvider(layoutDetail).getLine(layoutDetail);

The problem i run into is, that i cant seem to write these lines into the same file, without one layout, overwriting the other.
For example the first line has a length of for example 1000, and this length gets applied to the other lines im trying to write, even when theyre shorter or longer.

-Am i on the right approach or am i doing it the wrong way?
-How do I achieve, that a line is dynamically sometimes shorter, sometimes longer, depending if a field is filled or not, is this possible?

Also on another note:
-How do i read a copybook in XML format, like this:

<?xml version="1.0" encoding="UTF-8" standalone="no"?>
<copybook dialect="Mainframe" filename="XXXX.CPY">
    <item display-length="189" level="30" name="X" position="1" storage-length="177">
        <item display-length="2" level="35" name="XX" numeric="true" picture="9(002)" position="1" storage-length="2"/>
        <item display-length="15" level="35" name="XXX" numeric="true" picture="9(015)" position="3" storage-length="15"/>
        <item display-length="10" level="35" name="XXXX" picture="X(010)" position="18" 
    etc...
    etc...
    etc...   
    </item>
</copybook>

Thanks in advance for any answers.

[Bruce Martin]

You are using IBM01141 - EBCDIC encoding

so I presume you are either

• Going to send the file to mainframe (most likely)
• Final Application to run on the mainframe (very unlikely)

---

Assuming you are sending to the mainframe, The two easiest options are

• Write in EBCDIC FB as you are doing -easy safest transfer, will handle binary files
• Write in ASCII/UTF-8 and use IFileStructureConstants.IO_STANDARD_TEXT_FILE
  and do the UTF-8 ===> EBCDIC translation when transfering the file.
  This only works for text files, binary files get corrupted. There can be Translation issues.

For small files I would just use the larger record-length and IO_FIXED_LENGTH
as you are doing an accept the wasted space.

---

Fixed Length IO_FIXED_LENGTH

The File-Structure of IO_FIXED_LENGTH means every record is the same length
with no End-of-line makings (i.e. no \n or \r characters). i.e.

   <====  Record-1  =================================>
   <====  Record-2  =================================>
                     ...
   <====  Record-1  =================================>

This is normal in Cobol with a single copybook used in the file.

Windows/Linux Style IO_STANDARD_TEXT_FILE

If you use IFileStructureConstants.IO_STANDARD_TEXT_FILE with IBM01141, you get
a standard windows/linux style file but it probably will not be understood
on the mainframe.

VB Files (IO_VB)

You could use IFileStructureConstants.IO_VB, you might have problems
transfering to the mainframe (my experience is from a long time ago, so things
might of changed). For FTP use the RDW option

Format of VB File

    <RDW><====  Record-1  =================================>
    <RDW><====  Record-2  ========>
    <RDW><====  Record-3  ==================>
      ...

Where RDW is the record-descriptor word (4 bytes, it includes a 2 byte legth).

---

## Xml copybook

It looks like the xml produced by cb2xml. To process it

     LayoutDetail bldr =
           JRecordInterface1.CB2XML
                .newIOBuilder(copybook)
                .setFont("IBM01141")
                .setFileOrganization(IO_FIXED_LENGTH);
     LayoutDetail layoutDetail = bldr .getLayout();

It basically works like JRecordInterface1.COBOL

[Martin Scheibe]

Hello Bruce,
Thanks a lot for your reply.

I did as you mentioned and just accepted the wasted space. I just filled the shorter lines with spaces, so they are the same length as the longest line.

JRecordInterface1.CB2XML worked with the xml copybooks aswell. Thanks again.

You are correct, i want to send a the files i create to a mainframe, kind of. I need the files to be the exact same as the ones the mainframe can work with. But I didn't actually do this yet.

[Bruce Martin]

There should not be any problem working with a FB file on the mainframe. If the program is expecting a VB file, you can use Sort or IEBCOPY etc utilites to do the conversion from FB to VB

• See https://www.mainframestechhelp.com/utilities/sort/outfil-converting-vb-to-fb.htm

This should work.

---

If the file is Only text, you could write as a normal ascii text file and let FTP do the conversion (and write a VB file

---

I did do a some google searches and could not find any way to send a VB file to the mainframe; you could ask some experienced mainframe people

---

Finally running Java/JRecord on the mainframe would allow the creation of VB files using java-zos services