Find a word usind DXL script in the microsoft word document attached as OLE

mcnairk - Tue Apr 26 10:47:50 EDT 2011
Elam,

I have the exact same need as you, and previous posted here in the following thread.

http://www.ibm.com/developerworks/forums/thread.jspa?threadID=313869&tstart=0

If the above link doesn't work, email me and I will send you a summary. I got some really useful replies, but didn't follow up due to lack of resources to work on it.

I am not sure what the output of this search tool would be: would it output the Word paragraph that the string it was found in (ideally with the string highlighted), or just the module and object ID (and maybe number of hits).

Let me know if you make any progress on this.

Ken.

Elamkumaran - Tue Apr 26 11:45:25 EDT 2011
Hai,
Thanks for your reply. I couldn't able to find any search tool which you mentioned in your reply. I am new to DXL. Please let me know step by step procedure to find a string in Word OLE object. Thanks You

Elamkumaran - Tue Apr 26 11:45:25 EDT 2011
Hai,
Thanks for your reply. I couldn't able to find any search tool which you mentioned in your reply. I am new to DXL. Please let me know step by step procedure to find a string in Word OLE object. Thanks You

SystemAdmin - Wed Apr 27 12:10:09 EDT 2011
As an alternative, consider a typical practice in document management systems.

In document management systems, lots of content is OCRed (imported with optical character recognition). For each record, the image itself (or proprietary document) is stored plus some search meta-data is stored.

If you add a text "Search Meta-Data" attribute to your DOORS module schema, you can store the "plain text" content of your OLE objects. This way search is fast and the user does not necessarily need Word to search. The user will need Word to open the OLE object to see inside.

For OLE Word.Document you can store the "plain text" of the Word document. I'm talking about content that you would get if (from Word) you did from the menu

File \ Save As

With Save As Type = Plain Text (*.txt)

For other OLE objects (e.g., pictures) you can store some useful search keywords so users can find a given picture. An example would be "ABC Company Organizational Chart".

The "Search Meta-Data" attribute gets populated at insert or update time so search is fast.

SystemAdmin - Wed Apr 27 12:10:09 EDT 2011
As an alternative, consider a typical practice in document management systems.

In document management systems, lots of content is OCRed (imported with optical character recognition). For each record, the image itself (or proprietary document) is stored plus some search meta-data is stored.

If you add a text "Search Meta-Data" attribute to your DOORS module schema, you can store the "plain text" content of your OLE objects. This way search is fast and the user does not necessarily need Word to search. The user will need Word to open the OLE object to see inside.

For OLE Word.Document you can store the "plain text" of the Word document. I'm talking about content that you would get if (from Word) you did from the menu

File \ Save As

With Save As Type = Plain Text (*.txt)

For other OLE objects (e.g., pictures) you can store some useful search keywords so users can find a given picture. An example would be "ABC Company Organizational Chart".

The "Search Meta-Data" attribute gets populated at insert or update time so search is fast.

llandale - Wed Apr 27 13:24:23 EDT 2011
I was thinking along these lines as well, exporting the Word to a Text then searching the text. Storing that info in an attribute is even a better idea.

I cannot do it but surely someone here can export the embedded OLE to a temp text file, open the temp file and store in an attribute.

• Louie

SystemAdmin - Wed Apr 27 14:36:06 EDT 2011
There are too many unknowns to suggest an optimum way to achieve this. Are these OLE Word.Document items of the linked or unlinked variety?

I'd make the policy:
If you insert an OLE Word.Document, you insert the plain text at that time.
If you change the Word doc, you fix the Search Metadata as well.

I suppose some exist already in DOORS. Just open them. Select All. Copy to clipboard. Save somewhere like a text editor (or notepad). Reselect and copy to clipboard. Paste into "Search Metadata" attribute.

For automation you could paste clipboard into the "Search Metadata" attribute and then run a DXL script to remove OLE Rich Text and undo Rich Text formatting.

You'll have to decide if the automation is worth it.

If the word documents exist externally, there are loads of options to start Winword via batch and specify a macro to run to save a copy as a text file. Search the web for "word command line run macro" and you'll get lots of help.

David_G_Bond - Wed Apr 27 15:12:50 EDT 2011

If you really do only have OLEs that are word documents, the following function should work. I hacked it down from a more complex version.
 

bool searchOleInObject(Object o, string sKeyword)
{
    OleAutoObj objDoc = null, objApp = null, objSelection = null, objFind = null
    bool bIsStatic
    bool bFoundKeyword = false
    bool bIsActivated = false
    string sErr, sName
    
    if (containsStaticOle(o."Object Text")) return(bFoundKeyword)
    if (containsUnregisteredOle(o."Object Text")) return(bFoundKeyword)
    if (oleCount(o."Object Text") < 1) return(bFoundKeyword)
    
    bIsActivated = oleActivate(o)
    if (!bIsActivated) return(bFoundKeyword)
    objDoc = oleGetAutoObject(o)
    if (null(objDoc))
    {
        oleDeactivate(o)
        return(bFoundKeyword)
    }
    oleGet(objDoc, "Application", objApp)
    if (null(objApp))
    {
        if (!null(objDoc)) oleCloseAutoObject(objDoc)
        oleDeactivate(o)
        return(bFoundKeyword)
    }
    oleGet(objApp, "Name", sName)
    if (!matches("Word", sName))
    {
        // not a word application, so close out and return
        oleDeactivate(o)
        return(bFoundKeyword)
    }
    oleGet(objApp, "Selection", objSelection)
    oleGet(objSelection, "Find", objFind)
    oleMethod(objFind, "ClearFormatting")
    olePut(objFind, "Text", sKeyword)
    olePut(objFind, "MatchWholeWord", true)
    olePut(objFind, "Format", false)
    oleMethod(objFind, "Execute")
    oleGet(objFind, "Found", bFoundKeyword)
    oleDeactivate(o)
    
    if (!null(objDoc)) oleCloseAutoObject(objDoc)
    if (!null(objApp)) oleCloseAutoObject(objApp)
    if (!null(objSelection)) oleCloseAutoObject(objSelection)
    if (!null(objFind)) oleCloseAutoObject(objFind)
    return(bFoundKeyword)
}

• David Bond

Elamkumaran - Sun May 01 11:54:06 EDT 2011
Thank you very much
David_G_Bond. it is working... :)

Elamkumaran - Wed May 04 08:26:28 EDT 2011
New req.

David_G_Bond - Wed May 04 13:05:28 EDT 2011
If it were me I would modify the function to add as a parameter a skip list that has the words to search for and return a skip list of words found. The function can be modified to loop through the input list of words and add words to the return skip list as they are found.

• David Bond

Elamkumaran - Wed May 04 13:49:33 EDT 2011
Thanks Bond. Also can you pls tell me the method of writing a word into text type attribute column. example

ID || Object Text || Text type attribute

ID_1 Text.... Text1,
Text2

Elamkumaran - Wed May 04 13:53:40 EDT 2011
Thanks Bond. Also can you pls tell me the method of writing a word into text type attribute column. example

ID || Object Text || Text type attribute

David_G_Bond - Thu May 05 14:28:17 EDT 2011
I don't know your situation, so I don't know what approach is best for you. You indicate that this is a very large module with many OLEs. It may be best to use a scheduled task to run a script nightly to update the attribute in question. That way there is no slow-down for the user. You could use a layout dxl, but that might lead to slow response for that view. You might also consider a dxl attribute. Perhaps others can suggest what approach is best.

• David Bond

David_G_Bond - Wed Apr 27 15:12:50 EDT 2011

If you really do only have OLEs that are word documents, the following function should work. I hacked it down from a more complex version.
 

bool searchOleInObject(Object o, string sKeyword)
{
    OleAutoObj objDoc = null, objApp = null, objSelection = null, objFind = null
    bool bIsStatic
    bool bFoundKeyword = false
    bool bIsActivated = false
    string sErr, sName
    
    if (containsStaticOle(o."Object Text")) return(bFoundKeyword)
    if (containsUnregisteredOle(o."Object Text")) return(bFoundKeyword)
    if (oleCount(o."Object Text") < 1) return(bFoundKeyword)
    
    bIsActivated = oleActivate(o)
    if (!bIsActivated) return(bFoundKeyword)
    objDoc = oleGetAutoObject(o)
    if (null(objDoc))
    {
        oleDeactivate(o)
        return(bFoundKeyword)
    }
    oleGet(objDoc, "Application", objApp)
    if (null(objApp))
    {
        if (!null(objDoc)) oleCloseAutoObject(objDoc)
        oleDeactivate(o)
        return(bFoundKeyword)
    }
    oleGet(objApp, "Name", sName)
    if (!matches("Word", sName))
    {
        // not a word application, so close out and return
        oleDeactivate(o)
        return(bFoundKeyword)
    }
    oleGet(objApp, "Selection", objSelection)
    oleGet(objSelection, "Find", objFind)
    oleMethod(objFind, "ClearFormatting")
    olePut(objFind, "Text", sKeyword)
    olePut(objFind, "MatchWholeWord", true)
    olePut(objFind, "Format", false)
    oleMethod(objFind, "Execute")
    oleGet(objFind, "Found", bFoundKeyword)
    oleDeactivate(o)
    
    if (!null(objDoc)) oleCloseAutoObject(objDoc)
    if (!null(objApp)) oleCloseAutoObject(objApp)
    if (!null(objSelection)) oleCloseAutoObject(objSelection)
    if (!null(objFind)) oleCloseAutoObject(objFind)
    return(bFoundKeyword)
}

• David Bond