#include <iostream>
+#include <fstream>
#include <iterator>
-#include <vector>
#include <algorithm>
-
-#include <sqlite3.h>
+#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;
-
- 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
+template<class ISTREAM, 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]) );
- }
- } 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 );
+ FileData *data = new FileData();
+ in >> (*data);
+ files.insert( data );
+ } while( ! in.eof() );
+}
- // Now divide the two sets into three sets (new, deleted and updated )
- FileDataPtrCmp cmp;
+template<class SET>
+void partition_sets( const SET ¤t, const SET &old,
+ SET &added, SET &common, SET &deleted ) {
+ FileDataNameCmp cmp;
- file_set new_set;
- set_difference( current.begin(), current.end(),
- backed_up.begin(), backed_up.end(),
- inserter( new_set, new_set.begin() ),
+ 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 );
- // backed_up should *definitely* be the first set here
- file_set updated;
- set_difference( backed_up.begin(), backed_up.end(),
- current.begin(), current.end(),
- inserter( updated, updated.begin() ),
+ 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" );
+ 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_list;
+ file_set backup_set;
- // backup all new files
- copy( new_set.begin(), new_set.end(), inserter( backup_list, backup_list.begin() ) );
+ // backup all added 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 ) {
+ // 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_list.insert( *i );
+ backup_set.insert( *i );
}
}
- // Now, sort the backup_list by filesize and build a list of up to SIZE
+ // 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 );
- // Now, sort the non-backed-up list my last_backup_date and back-fill
+ file_set final_set;
+ unsigned long long bytes_available = 4700000000ULL; // 4.3 GBytes
- // Remove deleted files from the database.
- const char *delete_sql = "delete from filedata where filename = :filename";
- sqlite3_stmt *ppStmt;
+ unsigned long long block_size = 512ULL;
- sqlite3_prepare( db, delete_sql, -1, &ppStmt, NULL );
- assert( NULL != ppStmt );
+ bool complete = true;
- for( file_set::iterator i = deleted.begin(); i != deleted.end(); ++i ) {
- const string &name = (*i)->getFileName();
+ // 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;
+ }
+ ++i;
+ }
- rc = sqlite3_bind_text( ppStmt, 1, name.data(), name.size(), SQLITE_TRANSIENT );
- assert( SQLITE_OK == rc );
+ // 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;
+ }
+ }
- rc = sqlite3_step( ppStmt );
- assert( SQLITE_OK == rc );
+ // 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 );
}
- rc = sqlite3_finalize( ppStmt );
- assert( SQLITE_OK == rc );
- // Now, use the current set to update values in the database. You should copy
- // last_backup_date dates from the backed_up set first. This will get all of
- // the latest permissions but include the last_backup_date.
+ // Write the 'current' list to the dbfile
+ ofstream dbout( "test.db" );
+ copy( current.begin(), current.end(), ostream_iterator<FileData*>( dbout, "" ) );
- // Now, update the last_backup_date for all of the files that are in the list
+ // 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; }
projects / backups / .git / blobdiff