Yes it does look like it should work. Please can you File->Export your project (containing DFDL xsds and test data) as an eclipse project interchange zip and mail it to me, and I'll take a look at the trace.

Regards

Steve Hanson
Architect, Data Format Description Language (DFDL)
Co-Chair, OGF DFDL Working Group
IBM SWG, Hursley, UK
smh@uk.ibm.com
tel:+44-1962-815848

From: Bradley Sexton <bradley.r.sexton@gmail.com>
To: Steve Hanson/UK/IBM@IBMGB
Cc: dfdl-wg@ogf.org
Date: 01/03/2012 18:06
Subject: Re: [DFDL-WG] DFDL Modeling Question

That was the plan, but I'm getting errors when I attempt to parse. When it runs into two separators in succession (e.g. "{\") I get "An unexpected non-postfix separator } occurs in a postfix position at offset . . .", causing it to fail on the first two characters of the file.

The schema is very basic:

<element name="RTF">
    <complexType>
        <sequence dfdl:separator="\ } {">
            <element name="field" type="string" minOccurs="0" maxOccurs="unbounded"/>
        </sequence>
    </complexType>
</element>

If I add separators for "{\", "}}", and other combinations the file parses fine, so I'm guessing there is a value or setting somewhere that I might have missed or set incorrectly in the format definition that could cause it to error out instead of seeing an empty record and skipping over it?

It seems so straightforward in theory that it's very frustrating to see it not working : )

Thanks,
Bradley

On Thu, Mar 1, 2012 at 11:48 AM, Steve Hanson <smh@uk.ibm.com> wrote:
Hi Bradley

I think this would work. Presumably the controlWord element would be minOccurs='0', maxOccurs='unbounded'? If so all occurrences are optional, and empty optional elements won't be added to the infoset. So you won't have unwanted empty elements in the infoset.

Regards

Steve Hanson
Architect, Data Format Description Language (DFDL)
Co-Chair, OGF DFDL Working Group
IBM SWG, Hursley, UK
smh@uk.ibm.com
tel:+44-1962-815848

From: Bradley Sexton <bradley.r.sexton@gmail.com>
To: Steve Hanson/UK/IBM@IBMGB
Cc: dfdl-wg@ogf.org, dfdl-wg-bounces@ogf.org
Date: 01/03/2012 14:48
Subject: Re: [DFDL-WG] DFDL Modeling Question
Sent by: dfdl-wg-bounces@ogf.org

After some internal discussion I believe we are going to put RTF on the shelf for the time being and look at some other formats. One question did come up that I was hoping someone here might be able to help with. I was asked if there was a way to flat model RTF such that it would work for any size file or depth or nested groups, similar to what Steve proposed earlier:

        dfdl:separator="\ }\ }}\ }}}\ {\ }{\ }}{\ }}}{\" dfdl:separatorPosition="prefix"

but suitable for any amount of "}" characters before the "\" or "{\". A possibility suggested to me was to use:

        dfdl:separator="\ { }"

to consider all instances of these symbols as separators, and in the cases such as "}}{\" consider the values in between each character as empty or null. If you have any thoughts on this method or alternatives to a general flat model they would be greatly appreciated.

Bradley

On Fri, Feb 24, 2012 at 10:31 AM, Bradley Sexton <bradley.r.sexton@gmail.com> wrote:
Steve,

The order of nested groups is somewhat fluid in RTF, and my concern is whether or not modeling everything completely flat would preserve the structure and formatting properly. If you were to modify the text format in a file such as inserting a comment a new group is created and any data entered within the comment or previously existing text that is highlighted by the comment would be moved in new groups to signify their link.

Feel free to put me down for the WG call, just let me know the time and call info.

Thanks,
Bradley Sexton

On Thu, Feb 23, 2012 at 4:31 PM, Steve Hanson <smh@uk.ibm.com> wrote:
Hi Bradley

Yes dfdl:lengthKind "pattern" is the ideal way to model this.

I'm struggling to find a way to model this that preserves the nested groups and separates the trailing data from the control word. However if you were prepared to lose the group structure and treat the trailing data as part of the control word, then you could model a completely flat structure with the various delimiters interpreted as a prefix separator.

dfdl:separator="\ }\ }}\ }}}\ {\ }{\ }}{\ }}}{\" dfdl:separatorPosition="prefix"

That would give you an infoset like:

<file>
<controlWord>rtf1</controlWord>
<controlWord>ansi</controlWord>
<controlWord>ansicpg1252</controlWord>
<controlWord>deff0</controlWord>
<controlWord>deflang1033</controlWord>
<controlWord>fonttbl</controlWord>
<controlWord>f0</controlWord>
   <controlWord>froman</controlWord>
    <controlWord>fprq2</controlWord>
    <controlWord>fcharset0 Times New Roman;</controlWord>
    <controlWord>f1</controlWord>
    <controlWord>fswiss</controlWord>
    <controlWord>fcharset0 Arial;</controlWord>
<controlWord>*</controlWord>
<controlWord>generator Msftedit 5.41.15.1515;</controlWord>
<controlWord>viewkind4</controlWord>
<controlWord>uc1</controlWord>
<controlWord>pard</controlWord>
<controlWord>f0</controlWord>
<controlWord>fs24 This is an example document of an RTF file.</controlWord>
<controlWord>f1</controlWord>
<controlWord>fs20</controlWord>
<controlWord>par</controlWord>
<controlWord>*</controlWord>
<controlWord>passwordhash 010000004c000000010000000480000050c3. . .</controlWord>
</file>

Not ideal. I'll carry on thinking about the problem.

If you like I'll add you to the invite list for the DFDL WG call next Tuesday and we can discuss further?

Regards

Steve Hanson
Architect, Data Format Description Language (DFDL)
Co-Chair, OGF DFDL Working Group
IBM SWG, Hursley, UK
smh@uk.ibm.com
tel:+44-1962-815848

From: Bradley Sexton <bradley.r.sexton@gmail.com>
To: dfdl-wg@ogf.org
Date: 23/02/2012 19:07
Subject: [DFDL-WG] DFDL Modeling Question
Sent by: dfdl-wg-bounces@ogf.org

Hello,

I've been looking at modeling Rich Text Format (RTF) files using the IBM Message Broker DFDL implementation, and ran into an issue. For some background, here's a small example of an RTF file:

{\rtf1\ansi\ansicpg1252\deff0\deflang1033{\fonttbl{\f0\froman\fprq2\fcharset0 Times New Roman;}{\f1\fswiss\fcharset0 Arial;}}{\*\generator Msftedit 5.41.15.1515;}\viewkind4\uc1\pard\f0\fs24 This is an example document of an RTF file.\f1\fs20\par{\*\passwordhash 010000004c000000010000000480000050c300001400000010000000f89c360d0c9d360d000000008bc29e2f78a2144122ed68a1701e2ea50bbbbeaf7333c40dfe048ccf55f709b8cc7e8b49}}

'\' and '\*\' mark the beginning of control words, and the curly braces mark the beginning and end of control groups that contain control words and data. My issue is that control words and data do not have suitable terminators for parsing. The end of control words is signified by a space when trailing data is present, but typically they are ended by '\' signalling the beginning of a new word or a curly brace signalling the end of the current of beginning of a new control group. Similarly data is typically ended by the '}' of the parent control group.

With the exception of a small header the value and placement of control words, groups, and data varies by file.

My issue with modeling this is that I was going to use dfdl:lengthKind="pattern" in lieu of suitable delimiters, but this feature is not implemented by IBM. I'm looking for an alternative way to model the data, and was hoping someone on the mailing list might have suggestions. My goal is to model control words and groups in as general a manner as possible given IBMs implementation restrictions, since RTF has over 1800 defined control words and gives you the ability to create your own.

Ideal output for the above sample would be something along these lines:

<file>
<controlWord>rtf1</controlWord>
<controlWord>ansi</controlWord>
<controlWord>ansicpg1252</controlWord>
<controlWord>deff0</controlWord>
<controlWord>deflang1033</controlWord>
<controlGroup>
       <name>fonttbl</name>
       <controlGroup>
           <name>f0</name>
           <controlWord>froman</controlWord>
           <controlWord>fprq2</controlWord>
           <controlWord>fcharset0</controlWord>
           <data>Times New Roman;</data>
       </controlGroup>
       <controlGroup>
           <name>f1</name>
           <controlWord>fswiss</controlWord>
           <controlWord>fcharset0</controlWord>
           <data>Arial;</data>
       </controlGroup>
   </controlGroup>
<controlGroup>
       <name>generator</name>
       <data>Msftedit 5.41.15.1515;</data>
   </controlGroup>
<controlWord>viewkind4</controlWord>
<controlWord>uc1</controlWord>
<controlWord>pard</controlWord>
<controlWord>f0</controlWord>
<controlWord>fs24</controlWord>
<text>This is an example document of an RTF file.</text>
<controlWord>f1</controlWord>
<controlWord>fs20</controlWord>
<controlWord>par</controlWord>
<controlGroup>
       <name>passwordhash</name>
       <data>010000004c000000010000000480000050c3. . .</data>
   </controlGroup>
</file>

IBM Unsupported Features:
http://publib.boulder.ibm.com/infocenter/wmbhelp/v8r0m0/index.jsp?topic=%2Fcom.ibm.dfdl.editor.messagebroker.doc%2Fdf00150_.html

I know that's a lot of info out of left field, but I wanted to try and explain it as thoroughly as possible to avoid any confusion. Thanks in advance for any advice you might have and let me know if I've been unclear in any areas.

Bradley Sexton-- dfdl-wg mailing list dfdl-wg@ogf.org https://www.ogf.org/mailman/listinfo/dfdl-wg

Unless stated otherwise above:
IBM United Kingdom Limited - Registered in England and Wales with number 741598.
Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6 3AU

-- dfdl-wg mailing list dfdl-wg@ogf.org https://www.ogf.org/mailman/listinfo/dfdl-wg

Unless stated otherwise above:
IBM United Kingdom Limited - Registered in England and Wales with number 741598.
Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6 3AU