#include <iostream>
+#include <fstream>
#include <iterator>
-#include <string>
-#include <vector>
#include <algorithm>
+#include <cassert>
+#include <ctime>
+
+#include "filedata.hpp"
using namespace std;
-vector<string> split( const string &line, char c, int limit = -1 ) {
- string::size_type start = 0, end = 0;
+unsigned long long current_time() {
+ unsigned long long rc = 0;
+ time_t now_tt = time( 0 );
+ tm *now = localtime( &now_tt );
+ rc += ( now->tm_year + 1900ULL ) * 10000000000ULL;
+ rc += ( now->tm_mon + 1ULL ) * 100000000ULL;
+ rc += now->tm_mday * 1000000ULL;
+ rc += now->tm_hour * 10000ULL;
+ rc += now->tm_min * 100ULL;
+ rc += now->tm_sec;
+
+ return rc;
+}
+
+template<class I, class O>
+bool copy_until_full( I begin, I end, O out, unsigned long long &space ) {
+ const unsigned long long block_size = 0x200ULL;
+ bool complete = true;
+
+ I i = begin;
+ while( 0 != space && i != end ) {
+ unsigned long long size = (*i)->getFileSize();
+ unsigned long long blocks = size & ( ~(block_size-1) );
+ if( blocks < size ) blocks += block_size;
- vector<string> out;
- while( 0 != limit-- && end != line.size() ) {
- if( 0 == limit ) {
- end = line.size();
+ if( blocks <= space ) {
+ space -= blocks;
+ out = *i;
+ ++out;
} else {
- end = line.find( c, start );
- if( end == string::npos ) {
- end = line.size();
- }
+ // We missed a file that should be included so the backup is not complete
+ complete = false;
}
- out.push_back( line.substr( start, end-start ) );
- start = end + 1;
+ ++i;
}
- return out;
+ return complete;
}
-int main() {
- string file_string;
+template<class SET>
+void populate_set( istream &in, SET &files ) {
do {
- file_string.clear();
- for( int c = cin.get(); 0 != c && ! cin.eof(); c = cin.get() ) {
- file_string.push_back( c );
+ FileData *data = new FileData();
+ in >> data;
+ if( data->getFileName().size() ) {
+ files.insert( data );
+ } else {
+ delete data;
+ }
+ } while( ! in.eof() );
+}
+
+template<class SET>
+void partition_sets( const SET ¤t, const SET &old,
+ SET &added, SET &common, SET &deleted ) {
+ FileDataNameCmp cmp;
+
+ set_difference( current.begin(), current.end(),
+ old.begin(), old.end(),
+ inserter( added, added.begin() ),
+ cmp );
+
+ set_difference( old.begin(), old.end(),
+ current.begin(), current.end(),
+ inserter( deleted, deleted.begin() ),
+ cmp );
+
+ set_union( current.begin(), current.end(),
+ old.begin(), old.end(),
+ inserter( common, common.begin() ),
+ cmp );
+}
+
+int main() {
+ // Parse the list of current files on stdin
+ file_set current;
+ populate_set( cin, current );
+
+ file_set backed_up;
+ ifstream db( "test.db" );
+ if( db && db.good() ) {
+ populate_set( db, backed_up );
+ }
+
+ // Now divide the two sets into three sets (added, deleted and common )
+ file_set added, deleted, common;
+ partition_sets( current, backed_up, added, common, deleted );
+
+ { // This little block will copy the last_backup_date from the second set to the first
+ FileDataNameCmp cmp;
+
+ file_set common_with_dates;
+ set_union( backed_up.begin(), backed_up.end(),
+ current.begin(), current.end(),
+ inserter( common_with_dates, common_with_dates.begin() ),
+ cmp );
+
+ file_set::iterator i = common.begin(), j = common_with_dates.begin();
+ for( ; i != common.end(); ++i, ++j ) {
+ (*i)->setLastBackupDate( (*j)->getLastBackupDate() );
}
- if( 0 != file_string.size() ) {
- // Example entry
- // type perms user group size datemodified name (7 total)
- // f 0600 cnb cnb 424 20051015205340 ./.git/index
- vector<string> values = split( file_string, ' ', 7 );
+ }
+
+ // Now find the list of files to backup.
+ file_set backups;
+
+ // backup all added files
+ copy( added.begin(), added.end(), inserter( backups, backups.begin() ) );
+
+ // backup common files that have changed since the last backup date.
+ for( file_set::iterator i = common.begin(); i != common.end(); ++i ) {
+ if( (*i)->getLastBackupDate() < (*i)->getModifiedDate() ) {
+ backups.insert( *i );
}
- } while( ! cin.eof() );
+ }
+
+ // Now, sort the backups by filesize and build a list of up to SIZE
+ file_vector backups_s;
+ copy( backups.begin(), backups.end(), back_inserter( backups_s ) );
+
+ FileDataSizeCmp sizecmp;
+ sort( backups_s.begin(), backups_s.end(), sizecmp );
+
+ file_set final;
+ unsigned long long space = 0x107c00000ULL; // 4220 MBytes
+
+ insert_iterator<file_set> final_i( final, final.begin() );
+
+ // Copy files over until full or out of files
+ bool complete = copy_until_full( backups_s.rbegin(),
+ backups_s.rend(),
+ final_i,
+ space );
+
+ // Now, sort the non-backed-up list by last_backup_date and back-fill
+ if( 0 != space ) {
+ file_vector leftovers;
+ FileDataNameCmp cmp;
+ set_difference( current.begin(), current.end(),
+ final.begin(), final.end(),
+ back_inserter( leftovers ),
+ cmp );
+
+ FileDataLastBackupCmp lastbackupcmp;
+ sort( leftovers.begin(), leftovers.end(), lastbackupcmp );
+
+ copy_until_full( leftovers.begin(), leftovers.end(), final_i, space );
+ }
+
+ unsigned long long now = current_time();
+ for( file_set::iterator k = final.begin(); k != final.end(); ++k ) {
+ (*k)->setLastBackupDate( now );
+ }
+
+ // Write the 'current' list to the dbfile
+ ofstream dbout( "test.db" );
+ copy( current.begin(), current.end(), ostream_iterator<FileData*>( dbout, "" ) );
+
+ // Write the 'final' list to stdout
+ copy( final.begin(), final.end(), ostream_iterator<FileData*>( cout, "" ) );
+
+ if( ! complete ) { cerr << "incomplete" << endl; }
+
+ // Clean-up
+ for( file_set::iterator i = backed_up.begin(); i != backed_up.end(); ++i ) { delete *i; }
+ for( file_set::iterator i = current.begin(); i != current.end(); ++i ) { delete *i; }
}
projects / backups / .git / blobdiff