int docs_to_xml (char * path, Input_Obj_Type tMode )
{
int i = 0;
SingleList_t tFileList, *pFileList=&tFileList, tFileTypeFilter, *pFileTypeFilter = &tFileTypeFilter;
int xml_fd;
char *ext = NULL;
initList( pFileList );
setFileTypeFilter ( pFileTypeFilter );
#ifndef OLD_ZRT
if ( strcmp ( getext_( path ), "zip") == 0 )
{
get_file_list_inzip ( path , pFileList );
}
else
#endif
if ( tMode == zip_obj )
get_file_list_inzip( OBJECT_DEVICE_NAME, pFileList );
else if ( tMode == mail_obj )
get_file_list( path, pFileList, pFileTypeFilter );
else
return 0;
/*
printf ( "list of indexed docs\n" );
for ( i = 0; i < pFileList->count; i++)
printf ( "%s\n", pFileList->list[i] );
*/
///////////////////////
xml_fd = open_xml_( XML_PATH, tMode ); //FIXME const
for ( i = 0; i < pFileList->count; i++)
{
add_doc_to_xml ( xml_fd, pFileList->list[i], tMode );
}
close_xml_( xml_fd );
///////////////////////
freeList( pFileList );
freeList( pFileTypeFilter );
return 0;
}
int do_extract_text (char *input_file, char *output_file)
{
char *ext = NULL;
ext = (char *) getext_( input_file );
if (ext == NULL)
return 1;
if ( (strcasecmp( ext, "doc" ) == 0) || (strcasecmp( ext, "rtf" ) == 0) )
{
char *argv_catdoc [] = {"catdoc", input_file};
int argc_catdoc = PCHARSIZE( argv_catdoc );
catdoc_main((int)argc_catdoc, argv_catdoc ); //
//doc_to_text (input_file, output_file ); // old doc extractor. antiword
}
else if (( strcasecmp( ext, "txt" ) == 0) || ( strcasecmp( ext, "sh" ) == 0) || ( strcasecmp( ext, "html" ) == 0) || ( strcasecmp( ext, "c" ) == 0) )
{
output_file [ strlen ( output_file ) - 4 ] = '\0';
}
else if ( strcasecmp( ext, "docx" ) == 0)
{
docx_to_text( input_file, output_file );
}
else if ( strcasecmp( ext, "odt" ) == 0)
{
docx_to_text( input_file, output_file );
}
else if ( strcasecmp( ext, "pdf" ) == 0)
{
pdf_to_text( input_file, output_file );
}
else
{
free (ext);
return 1;
}
free (ext);
return 0;
}
int docs_to_xml (char * path, int doc_type)
{
int i = 0;
SingleList_t tFileList, *pFileList=&tFileList, tFileTypeFilter, *pFileTypeFilter = &tFileTypeFilter;
int xml_fd;
char *ext = NULL;
initList( pFileList );
setFileTypeFilter ( pFileTypeFilter );
#ifndef OLD_ZRT
if ( strcmp ( getext_( path ), "zip") == 0 )
{
get_file_list_inzip ( path , pFileList );
}
else
#endif
print_dir_tree (path);
get_file_list ( path, pFileList, pFileTypeFilter );
printf ( "list of indexed docs\n" );
for ( i = 0; i < pFileList->count; i++)
printf ( "%s\n", pFileList->list[i] );
///////////////////////
xml_fd = open_xml_( XML_PATH ); //FIXME const
for ( i = 0; i < pFileList->count; i++)
{
add_doc_to_xml ( xml_fd, pFileList->list[i], doc_type );
}
close_xml_( xml_fd );
///////////////////////
freeList( pFileList );
freeList( pFileTypeFilter );
return 0;
}