#include <iostream>
+#include <fstream>
#include <iterator>
-#include <string>
-#include <vector>
#include <algorithm>
+#include <cassert>
+
+#include "filedata.hpp"
using namespace std;
-vector<string> split( const string &line, char c, int limit = -1 ) {
- string::size_type start = 0, end = 0;
+template<class ISTREAM, class SET>
+void populate_set( ISTREAM &in, SET &files ) {
+ do {
+ FileData *data = new FileData();
+ in >> (*data);
+ files.insert( data );
+ } while( ! in.eof() );
+}
- vector<string> out;
- while( 0 != limit-- && end != line.size() ) {
- if( 0 == limit ) {
- end = line.size();
- } else {
- end = line.find( c, start );
- if( end == string::npos ) {
- end = line.size();
- }
- }
- out.push_back( line.substr( start, end-start ) );
- start = end + 1;
- }
- return out;
+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() {
- string file_string;
- do {
- file_string.clear();
- for( int c = cin.get(); 0 != c && ! cin.eof(); c = cin.get() ) {
- file_string.push_back( c );
+ // Parse the list of current files on stdin
+ file_set current;
+ populate_set( cin, current );
+
+ file_set backed_up;
+ ifstream db( "test.db" );
+ 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 updated_mirror;
+ set_union( current.begin(), current.end(),
+ backed_up.begin(), backed_up.end(),
+ inserter( updated_mirror, updated_mirror.begin() ),
+ cmp );
+
+ // TODO Now we need to copy the last_backup_date from
+ file_set::iterator i = common.begin(), j = updated_mirror.begin();
+ for( ; i != common.end(); ++i, ++j ) {
+ (*i)->setLastBackupDate( (*j)->getLastBackupDate() );
+ }
+ }
+
+ // Now find the list of files to backup.
+ file_set backup_set;
+
+ // backup all added files
+ copy( added.begin(), added.end(), inserter( backup_set, backup_set.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() ) {
+ backup_set.insert( *i );
+ }
+ }
+
+ // Now, sort the backup_set by filesize and build a list of up to SIZE
+ file_vector backups_bysize;
+ copy( backup_set.begin(), backup_set.end(), back_inserter( backups_bysize ) );
+ FileDataSizeCmp sizecmp;
+ sort( backups_bysize.begin(), backups_bysize.end(), sizecmp );
+
+ file_set final_set;
+ unsigned long long bytes_available = 4700000000ULL; // 4.3 GBytes
+
+ unsigned long long block_size = 512ULL;
+
+ bool complete = true;
+
+ // Copy files over until full or out of files
+ file_vector::reverse_iterator i = backups_bysize.rbegin();
+ while( 0 != bytes_available && i != backups_bysize.rend() ) {
+ unsigned long long size = (*i)->getFileSize();
+ unsigned long long blocks = size & ( ~(block_size-1) );
+ if( blocks < size ) blocks += block_size;
+ if( blocks <= bytes_available ) {
+ bytes_available -= blocks;
+ final_set.insert( *i );
+ } else {
+ // We missed a file that should be included so the backup is not complete
+ complete = false;
}
- 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 );
+ ++i;
+ }
+
+ // Now, sort the non-backed-up list by last_backup_date and back-fill
+ if( 0 != bytes_available ) {
+ file_vector leftovers;
+ FileDataNameCmp cmp;
+ set_difference( current.begin(), current.end(),
+ final_set.begin(), final_set.end(),
+ back_inserter( leftovers ),
+ cmp );
+
+ FileDataLastBackupCmp lastbackupcmp;
+ sort( leftovers.begin(), leftovers.end(), lastbackupcmp );
+
+ // Copy files over until full or out of files
+ file_vector::const_iterator j = leftovers.begin();
+ while( 0 != bytes_available && j != leftovers.end() ) {
+ unsigned long long size = (*j)->getFileSize();
+ unsigned long long blocks = size & ( ~(block_size-1) );
+
+ if( blocks < size ) blocks += block_size;
+
+ if( blocks <= bytes_available ) {
+ bytes_available -= blocks;
+ final_set.insert( *j );
+ }
+ ++j;
}
- } while( ! cin.eof() );
+ }
+
+ // TODO Get 'now' from time clock
+ unsigned long long now = 20051019211200ULL;
+ for( file_set::iterator k = final_set.begin(); k != final_set.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_set' list to stdout
+ copy( final_set.begin(), final_set.end(), ostream_iterator<FileData*>( cout, "" ) );
+
+ // If ! complete then write a flag to /tmp
+ 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; }
}