#include <iostream>
+#include <fstream>
#include <iterator>
#include <algorithm>
#include <cassert>
-
-#include <sqlite3.h>
+#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;
-}
-
-// Callback function for getting files from the database
-int populate_set( void *files_v, int, char **vals, char ** ) {
- file_set *files = reinterpret_cast<file_set*>( files_v );
- files->insert( new FileData( vals[0][0],
- vals[1],
- vals[2],
- vals[3],
- atoi( vals[4] ),
- atoi( vals[5] ),
- vals[6]) );
- return 0;
+ return complete;
}
-int main() {
- string file_string;
-
- file_set current;
-
- // Parse the list of files on stdin
+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 );
- }
- 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> vals = split( file_string, ' ', 7 );
- current.insert( new FileData( vals[0][0],
- vals[1],
- vals[2],
- vals[3],
- atoi( vals[4].c_str() ),
- atoi( vals[5].c_str() ),
- vals[6]) );
+ FileData *data = new FileData();
+ in >> data;
+ if( data->getFileName().size() ) {
+ files.insert( data );
+ } else {
+ delete data;
}
- } while( ! cin.eof() );
-
- // Get the list of previously backed up files from the database.
- sqlite3 *db;
-
- const char *dbname = "test.db";
- int rc = sqlite3_open( dbname, &db );
- assert( SQLITE_OK == rc );
-
- char *sqliteErrMsg = 0;
- file_set backed_up;
- rc = sqlite3_exec( db, "select * from filedata;", populate_set, &backed_up, &sqliteErrMsg );
- assert( SQLITE_OK == rc );
-
- rc = sqlite3_close( db );
- assert( SQLITE_OK == rc );
+ } while( ! in.eof() );
+}
- // Now divide the two sets into three sets (new, deleted and updated )
+template<class SET>
+void partition_sets( const SET ¤t, const SET &old,
+ SET &added, SET &common, SET &old_common, SET &deleted ) {
FileDataNameCmp cmp;
- file_set added;
- set_difference( current.begin(), current.end(),
- backed_up.begin(), backed_up.end(),
+ set_difference( current.begin(), current.end(),
+ old.begin(), old.end(),
inserter( added, added.begin() ),
cmp );
- file_set deleted;
- set_difference( backed_up.begin(), backed_up.end(),
- current.begin(), current.end(),
+ set_difference( old.begin(), old.end(),
+ current.begin(), current.end(),
inserter( deleted, deleted.begin() ),
cmp );
- file_set updated;
- set_union( current.begin(), current.end(),
- backed_up.begin(), backed_up.end(),
- inserter( updated, updated.begin() ),
- cmp );
-
- { // This little block will copy the last_backup_date from the second set to the first
- 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 = updated.begin(), j = updated_mirror.begin();
- for( ; i != updated.end(); ++i, ++j ) {
- (*i)->setLastBackupDate( (*j)->getLastBackupDate() );
- }
+ set_union( current.begin(), current.end(),
+ old.begin(), old.end(),
+ inserter( common, common.begin() ),
+ cmp );
+
+ set_union( old.begin(), old.end(),
+ common.begin(), common.end(),
+ inserter( old_common, old_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, old_common;
+ partition_sets( current, backed_up, added, common, old_common, deleted );
+
// Now find the list of files to backup.
- file_set backup_set;
+ file_set backups;
- // backup all new files
- copy( added.begin(), added.end(), inserter( backup_set, backup_set.begin() ) );
+ // backup all added files
+ copy( added.begin(), added.end(), inserter( backups, backups.begin() ) );
- // backup already backed-up files that have changed since the last backup date.
- for( file_set::iterator i = updated.begin(); i != updated.end(); ++i ) {
- if( (*i)->getLastBackupDate() < (*i)->getModifiedDate() ) {
- backup_set.insert( *i );
- }
+ // Backup files that have been modified
+ file_set::iterator i = common.begin(), j = old_common.begin();
+ for( ; i != common.end(); ++i, ++j ) {
+ (*i)->setLastBackupDate( (*j)->getLastBackupDate() );
+
+ if( needs_backup( *j, *i ) ) backups.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 ) );
+ // Now, sort the backups by filesize and build a list that'll fit on a DVD
+ file_vector backups_s;
+ copy( backups.begin(), backups.end(), back_inserter( backups_s ) );
+
FileDataSizeCmp sizecmp;
- sort( backups_bysize.begin(), backups_bysize.end(), sizecmp );
+ sort( backups_s.begin(), backups_s.end(), sizecmp );
- file_set final_set;
- unsigned long long bytes_available = 4700000000ULL; // 4.3 GBytes
+ file_set final;
+ unsigned long long space = 0x107c00000ULL; // 4220 MBytes
- unsigned long long block_size = 512ULL;
+ insert_iterator<file_set> final_i( final, final.begin() );
// 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 );
- }
- ++i;
- }
+ 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 != bytes_available ) {
+ if( 0 != space ) {
file_vector leftovers;
- set_difference( current.begin(), current.end(),
- final_set.begin(), final_set.end(),
+ 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 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;
- }
+ copy_until_full( leftovers.begin(), leftovers.end(), final_i, space );
}
- unsigned long long now = 20051019211200ULL;
- for( file_set::iterator k = final_set.begin(); k != final_set.end(); ++k ) {
+ unsigned long long now = current_time();
+ for( file_set::iterator k = final.begin(); k != final.end(); ++k ) {
(*k)->setLastBackupDate( now );
}
- // Clean out the database
- rc = sqlite3_exec( db, "delete from filedata;", NULL, NULL, &sqliteErrMsg );
- assert( SQLITE_OK == rc );
- rc = sqlite3_exec( db, "vacuum;", NULL, NULL, &sqliteErrMsg );
- assert( SQLITE_OK == rc );
+ // 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; }