#4 CobolIoProvider.getLineReader with InputStream instead of String filename?

[Nick Briggs]

I would find it quite helpful to have a CobolIoProvider.getLineReader() which took an InputStream instead of a file name, especially for the copybook, but also for the input data... it's much more flexible.

[Bruce Martin]

I will have look at doing this

[Bruce Martin]

I am little concerned at having similar methods with a lot of parameters. I was thinking a Builder type approach might be better:

IOBuilder builder = CobolIoProvider.newCobolIOBuilder(...);
AbstractLineReader reader = builder.newLineReader(stream);
AbstreactLineWriter writer = builder.newLineWriter(stream);

     /* or */

AbstractLineReader reader = builder.newLineReader(filename);
AbstreactLineWriter writer = builder.newLineWriter(filename);

     /* you could also do */

AbstractLineReader reader = CobolIoProvider
                                 .newCobolIOBuilder(...)
                                 .newLineWriter(stream /* or filename */)

It should lead to simpler clearer code !!!

---

Also at the moment, You can do:

ByteArrayInputStream bs = new ByteArrayInputStream(copybook.getBytes);
LayoutDetail schema = new CobolCopybookLoader()
                   .loadCopyBook(   bs, "RedefTest", 
                            CopybookLoader.SPLIT_01_LEVEL, 0, "",
                        Cb2xmlConstants.USE_STANDARD_COLUMNS, 
                            Convert.FMT_INTEL, 0, new TextLog())
                ..asLayoutDetail();

AbstractLineReader reader = LineIOProvider.getInstance().getLineReader(schema);

reader.open(inputStream, schema);

I realize this is more long winded than the CobolIoProvider. CobolIoProvider was always intended handle just the simple cases, I would prefer not to complicate it to much

[Nick Briggs]

Thanks! I think the Builder approach would work well.

In general when designing library interfaces which are processing some stream of bytes(characters) I try to stay far away from String filenames, File objects, or anything else which implies a particular set of operations that need to be applied before you can read(write) the data. I figure that the caller knows their source/sink better than my library, so they can arrange to apply the operations to get it into an Input(Output)Stream before calling the library.

[Bruce Martin]

I agree with what you said, if I was writing JRecord now I would do things quite differently. Basically with JRecord

• I did not intend to write JRecord as a standalone library, The cope originally was part of the RecordEditor. JRecord was a Response to various requests
• JRecord was originally written for Java 1.3/1.4 so no Generics, Enums etc.

---

Please give me your thoughts on Builder approach. I am thinking of making it the standard way of accessing JRecord.

I do have thoughts of doing a complete rewrite of JRecord with all the, so please pass through any suggestions

[Bruce Martin]

In version 0.80.8 I have created a new IOBuilder classes which can be used like

ICobolIOBuilder iob = CobolIoProvider.getInstance()
                           .newIOBuilder(copybookStream, copybookname)
                                .setSplitCopybook(CopybookLoader.SPLIT_REDEFINE); 
AbstractLineReader reader = iob.newReader(salesstream);

The IOBuilder will accept either filenames or streams as input for Cobol-Copyybooks, Input-Files and Output-Files.

[Bruce Martin]

This has been implemented now

[Anonymous]

Hi, I amusing jrecord 0.82 but could not find newIOBuilder. Iwant to process input streams
Please help

[Bruce Martin]

For a Cobol file you would start with JRecordInterface1

i.e.

  JRecordInterface1.COBOL.newIOBuilder(...)

or using the fully qualified class name

  net.sf.JRecord.JRecordInterface1
         .COBOL.newIOBuilder(...)

---

For a Fixed-Width file (no Cobol Copybook):

  JRecordInterface1.FIXED_WIDTH.newIOBuilder(...)

For a Csv File

  JRecordInterface1.CSV.newIOBuilder(...)

etc

[Anonymous]

Hi Thanks for ur quick response. I tried above option but getting JRecordInterface1 cannt be resolved. Iam ajava developer and want to validate cobol datafile againstcobol copybook fileand both are in inputstreams . Iam using jrecord0.82. please letme know ifabove thing can be achieved using this version.

[Bruce Martin]

The latest JRecord is 0.81.4:

• https://sourceforge.net/projects/jrecord/files/latest/download?source=files

Have a look in the Source/JRecord_IO_Builder_Examples this sub projects contains examples of using JRecord. In particular, src\net\sf\JRecord\zExamples\iob\cobol contains Cobol examples.

I will come back to this later

[Anonymous]

Hi Bruce, Is there a way to validate data against the copybook schema. Now am able to parse the data but it does not seem to validate the datatypes and patterns. Please suggest

[Bruce Martin]

There is not anything builtin to validate the data. the Best option is to the CobolSchemaDetails class. You can use it like:

• CobolSchemaDetails.newCobolSchemaReader("copybook-file-name")
• CobolSchemaDetails.newCobolSchemaReader(Reader reader, "copybook-Name")

This will return:

• JRecord Layout (schema)
• Cobol Item Tree (copybook and recordItems) which has the Cobol Tree + JRecord Field definitions added in to it.

The IItem has a lot of Cobol definitions (e.g. Cobol Picture, Display-Length, decimal etc). It can be cast back to Item for extra details (like Cobol 88 level details). If you do need more details from Item let me know I can update IItem if neen be. Alternalively take a copy of net.sf.JRecord.schema. There are not a huge number of classes and it is based on the cb2xml Jaxb example.

So you would need to write your own validation.

---

Note: you can use

JRecordInterface1.SCHEMA.newIOBuilder(cobolSchemaDetails.schema);

to create an IOBuilder from a layout or schema.