/**
* Open a DFXML file and import its hash elements.
*/
hashdb_exporter_t (const std::string& _dfxml_outfilename)
: dfxml_outfilename(_dfxml_outfilename),
x(_dfxml_outfilename, false) {
// start with dfxml tag
x.push("dfxml");
// add creator information
x.add_DFXML_creator(PACKAGE_NAME, PACKAGE_VERSION, "svn not tracked", command_line);
}
void report_status(dfxml_writer& x) const {
// x.push("hashdb_status");
x.xmlout("hashdb_path", hashdb_dir);
x.xmlout("file_mode", file_mode_type_to_string(file_mode_type));
hash_store->report_status(x);
hash_duplicates_store->report_status(x);
bloom1->report_status(x, 1);
bloom2->report_status(x, 2);
// x.pop();
}
void report_remove_changes(dfxml_writer& x) const {
x.push("hashdb_changes");
x.xmlout("hashes_removed", hashes_removed);
x.xmlout("hashes_not_removed_invalid_file_offset", hashes_not_removed_invalid_file_offset);
x.xmlout("hashes_not_removed_wrong_hash_block_size", hashes_not_removed_wrong_hash_block_size);
x.xmlout("hashes_not_removed_wrong_hashdigest_type", hashes_not_removed_wrong_hashdigest_type);
x.xmlout("hashes_not_removed_no_hash", hashes_not_removed_no_hash);
x.xmlout("hashes_not_removed_different_source", hashes_not_removed_different_source);
x.pop();
}
// generate reports to dfxml
void report_insert_changes(dfxml_writer& x) const {
x.push("hashdb_changes");
x.xmlout("hashes_inserted", hashes_inserted);
x.xmlout("hashes_not_inserted_invalid_file_offset", hashes_not_inserted_invalid_file_offset);
x.xmlout("hashes_not_inserted_wrong_hash_block_size", hashes_not_inserted_wrong_hash_block_size);
x.xmlout("hashes_not_inserted_wrong_hashdigest_type", hashes_not_inserted_wrong_hashdigest_type);
x.xmlout("hashes_not_inserted_exceeds_max_duplicates", hashes_not_inserted_exceeds_max_duplicates);
x.xmlout("hashes_not_inserted_duplicate_source", hashes_not_inserted_duplicate_source);
x.pop();
}
void do_export(const hashdb_db_manager_t& hashdb_in) {
const size_t hash_block_size = hashdb_in.hashdb_settings.hash_block_size;
hashdb_db_manager_t::hashdb_iterator_t it = hashdb_in.begin();
hashdb_db_manager_t::hashdb_iterator_t end = hashdb_in.end();
while (it != end) {
// get the hashdb element associated with the iterator
hashdb_element_t hashdb_element = *it;
// get md5 from first
md5_t md5 = hashdb_element.first;
// get offset, repository_name, and filename from second
uint64_t file_offset = hashdb_element.second.file_offset;
std::string repository_name = hashdb_element.second.repository_name;
std::string filename = hashdb_element.second.filename;
// start the fileobject tag
x.push("fileobject");
// write the repository name tag
x.xmlout("repository_name", repository_name);
// write the filename tag
x.xmlout("filename", std::string(filename));
// start the byte_run tag with its file_offset attribute
std::stringstream ss;
ss << "file_offset='" << file_offset << "' len='" << hash_block_size << "'";
x.push("byte_run", ss.str());
// write the hashdigest
std::string hashdigest = md5.hexdigest();
x.xmlout("hashdigest", hashdigest, "type='MD5'", false);
// close the byte_run tag
x.pop();
// close the fileobject tag
x.pop();
// move to next iterator
++it;
}
}
~hashdb_exporter_t() {
// close dfxml_writer
x.add_rusage();
x.pop();
x.close();
}
static void dfxml_create(dfxml_writer &xreport,const std::string &command_line,const BulkExtractor_Phase1::Config &cfg)
{
xreport.push("dfxml","xmloutputversion='1.0'");
xreport.push("metadata",
"\n xmlns='http://afflib.org/bulk_extractor/' "
"\n xmlns:xsi='http://www.w3.org/2001/XMLSchema-instance' "
"\n xmlns:dc='http://purl.org/dc/elements/1.1/'" );
xreport.xmlout("dc:type","Feature Extraction","",false);
xreport.pop();
xreport.add_DFXML_creator(PACKAGE_NAME,PACKAGE_VERSION,svn_revision_clean(),command_line);
xreport.push("configuration");
xreport.xmlout("threads",cfg.num_threads);
xreport.xmlout("pagesize",cfg.opt_pagesize);
xreport.xmlout("marginsize",cfg.opt_marginsize);
xreport.push("scanners");
/* Generate a list of the scanners in use */
std::vector<std::string> ev;
be13::plugin::get_enabled_scanners(ev);
for(std::vector<std::string>::const_iterator it=ev.begin();it!=ev.end();it++){
xreport.xmlout("scanner",(*it));
}
xreport.pop(); // scanners
xreport.pop(); // configuration
}
