JRecord and Mainframe Assembler DSECTs

[Greg Senia]

Bruce,
Not sure if this is an option but is it possible to use something like JRecord to convert PDS that are stored as Assembler DSECTs vs Cobol Copybooks?

[Bruce Martin]

There is nothing builtin for handling DSECT's. Options

• If only one or two; manually coding The file definitions; this should be easy enough
• Use a scripting language (rexx / Python to convert the DSEC to an Xml copybook / description.
• JRecord does support writing your own parser. So in theory a DSECT parser is possible

The Scripting option might be the easiest.

[Bruce Martin]

You can define records in java like:

         AbstractLineReader reader = JRecordInterface1.FIXED_WIDTH.newIOBuilder()
                                 .defineFieldsByLength()
                                     .addFieldByLength("Sku"  , Type.ftChar,   8, 017                                         .addFieldByLength("Store", Type.ftNumRightJustified, 3, 0)
                                     .addFieldByLength("Date" , Type.ftNumRightstified, 6, 0)
                                     .addFieldByLength("Dept" , Type.ftNumRightJustified, 3, 0)
                                     .addFieldByLength("Qty"  , Type.ftNumRightJustified, 2, 0)
                                     .addFieldByLength("Price", Type.ftNumRightJustified, 6, 2)
                                 .endOfRecord()
                                 .newReader(this.getClass().getResource("DTAR020_tst1.bin.txt").getFile());
         AbstractLine saleRecord;

         while ((saleRecord = reader.read()) != null) {
             System.out.println(
                              saleRecord.getFieldValue("Sku").asString()
                     + "\t" + saleRecord.getFieldValue("Store").asString()
                     + "\t" + saleRecord.getFieldValue("Date").asString()
                     + "\t" + saleRecord.getFieldValue("Dept").asString()
                     + "\t" + saleRecord.getFieldValue("Qty").asString()
                     + "\t" + saleRecord.getFieldValue("Price").asString());

         }
         reader.close();

[Bruce Martin]

You can define a Record in xml. You could try converting your DSECT to Xml using Rexx / Python etc:

<RECORD RECORDNAME="DTAR020" COPYBOOK="DTAR020" DELIMITER="&lt;Tab&gt;" 
         FONTNAME="CP037" FILESTRUCTURE="Default" STYLE="0" RECORDTYPE="RecordLayout"
         LIST="Y" QUOTE="" RecSep="default">
         <FIELDS>
                 <FIELD NAME="KEYCODE-NO" POSITION="1"  LENGTH="8" TYPE="Char" />
                 <FIELD NAME="STORE-NO"   POSITION="9"  LENGTH="2" TYPE="Mainframe Packed Decimal (comp-3)" />
                 <FIELD NAME="DATE"       POSITION="11" LENGTH="4" TYPE="Mainframe Packed Decimal (comp-3)" />
                 <FIELD NAME="DEPT-NO"    POSITION="15" LENGTH="2" TYPE="Mainframe Packed Decimal (comp-3)" />
                 <FIELD NAME="QTY-SOLD"   POSITION="17" LENGTH="5" TYPE="Mainframe Packed Decimal (comp-3)" />
                 <FIELD NAME="SALE-PRICE" POSITION="22" LENGTH="6" DECIMAL="2" TYPE="Mainframe Packed Decimal (comp-3)" />
         </FIELDS>
</RECORD>

[Greg Senia]

This was very insightful. I think what we have to do now is attempt to convert these Variable Block DSECTs into cobol copyrecords. An example of one:

         MACRO                                                          00000100
         INFO                                                           00000200
INFO     DSECT                                                          00000300
*********************************************************************** 00000400
*                                                                     * 00000500
* FUNCTIONAL AREA         :                                           * 00000600
* MODULES AFFECTING MACRO :                                           * 00000700
* TRANSACTION ENGINE      :  YES                                      * 00000800
* TE DISPLAY NAME         :  INFO                                     * 00000900
* DISBURSEMENT CODES      :  GC                                       * 00001000
* TRANSACTION CODES       :  GC                                       * 00001100
*                                                                     * 00001200
*********************************************************************** 00001300
RCLG     DS    XL4      # RECORD LENGTH                                 00001400
PLNO     DS    PL5      # POLICY NUMBER                                 00001500
SCDT     DS    XL28     # SCHEDULE DATE                                 00001600
EFFECTCG DS    PL4      # EFFECTIVE DATE OF LAST TRANSACTION CHANGE     00001700
TRANSDT  DS    PL4      # CYCLE DATE - TRANSACTION CHANGE DATE          00001800
FIRST    DS    CL66     # 1ST                                         00001900
SECND    DS    CL66     # SECOND                                   00002000
THIRD    DS    CL68     # 3RD                                           00002100
OWNRD    DS    CL31     # OWNER                                   00002200
         MEND                                                           00002300

Bruce let me know your thoughts. I guess my question would be what would be used to determine record type like in cobol?

[Greg Senia]

Also Bruce does this give the entire length of the line including the RDW or does it ignore the RDW piece?
Integer copyRecLength = copyRecord.getData().length;

byte[] net.sf.JRecord.Details.AbstractLine.getData().length

get The dat in the line as an Array of Bytes

Returns:
Returns the record.

[Bruce Martin]

With JRecord you do not specify a Record-Length Field. When definiing the RecordLayout you set the FileStructure / FileOrganisation IO_VB.

int copyRecLength = copyRecord.getData().length;

will give the length with-out the record-length field.