[users] Re: API to retrieve document metadata

[Ariel Constenla-Haile]

Hi Alona,

On Mon, Oct 24, 2011 at 04:41:50PM -0400, Alona Rossen wrote:
> metadata listed under File->Properties->Statistics in OpenOffice (number
> of pages, paragraph count, word count, etc).

I see I didn't include those in the example.
See the version I attach now.

Regards
-- 
Ariel Constenla-Haile
La Plata, Argentina

#define C2U( constAsciiStr ) \
        ( ::rtl::OUString( RTL_CONSTASCII_USTRINGPARAM( constAsciiStr ) ) )

#define U2C( ouString ) \
        ( ::rtl::OUStringToOString( ouString, RTL_TEXTENCODING_UTF8 ).getStr() )

// 
-------------------------------------------------------------------------------------

#include <vector>
#include <iostream>
#include <cstdio>
#include <osl/file.hxx>
#include <cppuhelper/bootstrap.hxx>
#include <com/sun/star/beans/PropertyValue.hpp>
#include <com/sun/star/document/XDocumentPropertiesSupplier.hpp>
#include <com/sun/star/frame/XComponentLoader.hpp>
#include <com/sun/star/frame/FrameSearchFlag.hpp>
#include <com/sun/star/uno/Reference.h>
#include <com/sun/star/uno/XComponentContext.hpp>

using namespace ::com::sun::star;


int SAL_CALL main ( const int argc, const char* argv[] )
{
    int ret = 0;

    if ( argc != 2 )
    {
        std::cerr << "Usage: " << std::endl <<
                     "       " << argv[0] << " DOC_TO_LOAD" << std::endl <<
                     "where  " << std::endl <<
                     "       DOC_TO_LOAD is the system path of the document to 
load" << std::endl;
        return 1;
    }

    rtl::OUString path = rtl::OUString::createFromAscii( argv[1] );
    rtl::OUString path_url;
    if ( osl_File_E_None != osl_getFileURLFromSystemPath( path.pData, 
&path_url.pData ) )
    {
        std::cerr << "Can not convert " << argv[1] << " to URL." << std::endl;
        return 1;
    }

    try
    {
        uno::Reference< uno::XComponentContext > rxContext ( 
::cppu::bootstrap() );

        uno::Reference< frame::XComponentLoader > rxComponentLoader(
            rxContext->getServiceManager()->createInstanceWithContext(
                C2U( "com.sun.star.frame.Desktop" ),
                rxContext ),
            uno::UNO_QUERY );

        uno::Reference< lang::XComponent > xComponent =
            rxComponentLoader->loadComponentFromURL(
                path_url, 
                C2U( "_default" ), 
                frame::FrameSearchFlag::ALL,
                uno::Sequence< beans::PropertyValue >() );

        uno::Reference< document::XDocumentPropertiesSupplier> xDocPropsSup( 
xComponent, uno::UNO_QUERY );
        if ( !xDocPropsSup.is() )
        {
            std::cerr << "No Document Props. Supplier!" << std::endl;
            return 1;
        }

        uno::Reference< document::XDocumentProperties > xDocProps = 
xDocPropsSup->getDocumentProperties();
        if ( !xDocProps.is() )
        {
            std::cerr << "No Document Properties!" << std::endl;
            return 1;
        }

        // Attributes
        rtl::OUString Title        = xDocProps->getTitle();
        rtl::OUString Subject      = xDocProps->getSubject();
        rtl::OUString Description  = xDocProps->getDescription();

        rtl::OUString Generator    = xDocProps->getGenerator();
        rtl::OUString Author       = xDocProps->getAuthor();

        util::DateTime CreationDate     = xDocProps->getCreationDate();
        rtl::OUString ModifiedBy        = xDocProps->getModifiedBy();
        util::DateTime ModificationDate = xDocProps->getModificationDate();

        uno::Sequence< ::rtl::OUString > Keywords = xDocProps->getKeywords();
        lang::Locale Language = xDocProps->getLanguage();

        rtl::OUString PrintedBy  = xDocProps->getPrintedBy();
        util::DateTime PrintDate = xDocProps->getPrintDate();

        rtl::OUString TemplateName  = xDocProps->getTemplateName();
        rtl::OUString TemplateURL   = xDocProps->getTemplateURL();
        util::DateTime TemplateDate = xDocProps->getTemplateDate();

        rtl::OUString AutoloadURL   = xDocProps->getAutoloadURL();
        sal_Int32 AutoloadSecs      = xDocProps->getAutoloadSecs();
        rtl::OUString DefaultTarget = xDocProps->getDefaultTarget();

        uno::Sequence< beans::NamedValue > DocumentStatistics = 
xDocProps->getDocumentStatistics();
        sal_Int32   nPageCount,
                    nTableCount,
                    nImageCount,
                    nObjectCount,
                    nParagraphCount,
                    nWordCount,
                    nCharacterCount;
         nPageCount = nTableCount = nImageCount = nObjectCount =
            nParagraphCount = nWordCount = nCharacterCount = 0;

        for ( sal_Int32 n = 0; n < DocumentStatistics.getLength(); n++ )
        {
            rtl::OUString sName = DocumentStatistics[n].Name;
            uno::Any aValue     = DocumentStatistics[n].Value;

            if ( sName.equalsAsciiL( RTL_CONSTASCII_STRINGPARAM( "PageCount" ) 
) )
                 aValue >>= nPageCount;
            else if ( sName.equalsAsciiL( RTL_CONSTASCII_STRINGPARAM( 
"TableCount" ) ) )
                 aValue >>= nTableCount;
            else if ( sName.equalsAsciiL( RTL_CONSTASCII_STRINGPARAM( 
"ImageCount" ) ) )
                 aValue >>= nImageCount;
            else if ( sName.equalsAsciiL( RTL_CONSTASCII_STRINGPARAM( 
"ObjectCount" ) ) )
                 aValue >>= nObjectCount;
            else if ( sName.equalsAsciiL( RTL_CONSTASCII_STRINGPARAM( 
"ParagraphCount" ) ) )
                 aValue >>= nParagraphCount;
            else if ( sName.equalsAsciiL( RTL_CONSTASCII_STRINGPARAM( 
"WordCount" ) ) )
                 aValue >>= nWordCount;
            else if ( sName.equalsAsciiL( RTL_CONSTASCII_STRINGPARAM( 
"CharacterCount" ) ) )
                 aValue >>= nCharacterCount;
        }

        std::cout << "PageCount       " << nPageCount       << std::endl;
        std::cout << "TableCount      " << nTableCount      << std::endl;
        std::cout << "ImageCount      " << nImageCount      << std::endl;
        std::cout << "ObjectCount     " << nObjectCount     << std::endl;
        std::cout << "ParagraphCount  " << nParagraphCount  << std::endl;
        std::cout << "WordCount       " << nWordCount       << std::endl;
        std::cout << "CharacterCount  " << nCharacterCount  << std::endl;

        sal_Int16 EditingCycles   = xDocProps->getEditingCycles();
        sal_Int32 EditingDuration = xDocProps->getEditingDuration();

        std::cout << "Title :        " << U2C( Title )        <<  std::endl;
        std::cout << "Subject :      " << U2C( Subject )      <<  std::endl;
        std::cout << "Description :  " << U2C( Description )  <<  std::endl;
        std::cout << "Generator :    " << U2C( Generator )    <<  std::endl;
        std::cout << "Author :       " << U2C( Author )       <<  std::endl;
        std::cout << "ModifiedBy :   " << U2C( ModifiedBy )   <<  std::endl;
        std::cout << "PrintedBy :    " << U2C( PrintedBy )    <<  std::endl;
        std::cout << "TemplateName : " << U2C( TemplateName ) <<  std::endl;
        std::cout << "TemplateURL :  " << U2C( TemplateURL )  <<  std::endl;
        std::cout << "AutoloadURL :  " << U2C( AutoloadURL )  <<  std::endl;

    }
    catch ( uno::RuntimeException& e )
    {
        ret = 1;
        OSL_ASSERT ( false );
    }
    catch ( uno::Exception& e )
    {
        ret = 1;
        OSL_ASSERT ( false );
    }
    return ret;
}
// kate: indent-mode cstyle; space-indent on; indent-width 4;

pgpvD6TnZrLgh.pgp
Description: PGP signature