#include <iostream>
+#include <iterator>
#include <algorithm>
+#include <cassert>
+
+#include <sqlite3.h>
+
+#include "filedata.hpp"
using namespace std;
+vector<string> split( const string &line, char c, int limit = -1 ) {
+ string::size_type start = 0, end = 0;
+
+ 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;
+}
+
+// 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;
+}
+
int main() {
string file_string;
+
+ file_set current;
+
+ // Parse the list of files on stdin
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() ) {
- cout << file_string << endl;
+ // 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]) );
}
} 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 );
+
+ // Now divide the two sets into three sets (new, deleted and updated )
+ FileDataNameCmp cmp;
+
+ file_set added;
+ set_difference( current.begin(), current.end(),
+ backed_up.begin(), backed_up.end(),
+ inserter( added, added.begin() ),
+ cmp );
+
+ file_set deleted;
+ set_difference( backed_up.begin(), backed_up.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() );
+ }
+ }
+
+ // Now find the list of files to backup.
+ file_set backup_set;
+
+ // backup all new files
+ copy( added.begin(), added.end(), inserter( backup_set, backup_set.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 );
+ }
+ }
+
+ // 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;
+
+ // 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;
+ }
+
+ // Now, sort the non-backed-up list by last_backup_date and back-fill
+ if( 0 != bytes_available ) {
+ file_vector leftovers;
+ 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;
+ }
+ }
+
+ unsigned long long now = 20051019211200ULL;
+ for( file_set::iterator k = final_set.begin(); k != final_set.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 );
+
+ // 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