--- trunk/bin/BackupPC_updatedb 2005/06/23 09:47:59 6 +++ trunk/bin/BackupPC_updatedb 2005/08/20 11:24:55 44 @@ -1,5 +1,4 @@ #!/usr/local/bin/perl -w -$| = 1; use strict; use DBI; @@ -8,315 +7,381 @@ use BackupPC::View; use Data::Dumper; use Getopt::Std; +use Time::HiRes qw/time/; +use File::Pid; +use POSIX qw/strftime/; use constant BPC_FTYPE_DIR => 5; +my $debug = 0; +$|=1; + +my $pidfile = new File::Pid; + +if (my $pid = $pidfile->running ) { + die "$0 already running: $pid\n"; +} elsif ($pidfile->pid ne $$) { + $pidfile->remove; + $pidfile = new File::Pid; +} +$pidfile->write; +print STDERR "$0 using pid ",$pidfile->pid," file ",$pidfile->file,"\n"; + +my $t_fmt = '%Y-%m-%d %H:%M:%S'; my $hosts; my $bpc = BackupPC::Lib->new || die; my %Conf = $bpc->Conf(); my $TopDir = $bpc->TopDir(); -my @beenThere = (); +my $beenThere = {}; -print Dumper(\%Conf); +my $dsn = "dbi:SQLite:dbname=$TopDir/$Conf{SearchDB}"; -my $dbh = DBI->connect("dbi:SQLite:dbname=$TopDir/$Conf{SearchDB}", "", "", { RaiseError => 1, AutoCommit => 0 }); +my $dbh = DBI->connect($dsn, "", "", { RaiseError => 1, AutoCommit => 0 }); my %opt; -if ( !getopts("cdu", \%opt ) ) { +if ( !getopts("cdm:v:", \%opt ) ) { print STDERR <do( - q{ - create table hosts - ( ID INTEGER PRIMARY KEY, - name VARCHAR(30) NOT NULL, - IP VARCHAR(20) NOT NULL + $dbh->do(qq{ + create table hosts ( + ID INTEGER PRIMARY KEY, + name VARCHAR(30) NOT NULL, + IP VARCHAR(15) ); - } - ); + }); - $dbh->do( - q{ - create table shares - ( ID INTEGER PRIMARY KEY, - hostID INTEGER NOT NULL, - name VARCHAR(30) NOT NULL, - share VARCHAR(200) NOT NULL, - localpath VARCHAR(200) + $dbh->do(qq{ + create table shares ( + ID INTEGER PRIMARY KEY, + hostID INTEGER NOT NULL references hosts(id), + name VARCHAR(30) NOT NULL, + share VARCHAR(200) NOT NULL, + localpath VARCHAR(200) ); - } - ); + }); - $dbh->do( - q{ - create table backups - ( hostID INTEGER NOT NULL, - num INTEGER NOT NULL, - date DATE, - type CHAR(1), - PRIMARY KEY(hostID, num) + $dbh->do(qq{ + create table backups ( + hostID INTEGER NOT NULL references hosts(id), + num INTEGER NOT NULL, + date DATE, + type CHAR(1), + PRIMARY KEY(hostID, num) ); - } - ); + }); - $dbh->do( - q{ - create table dvds - ( ID INTEGER PRIMARY KEY, - num INTEGER NOT NULL, - name VARCHAR(255) NOT NULL, - mjesto VARCHAR(255) + $dbh->do(qq{ + create table dvds ( + ID INTEGER PRIMARY KEY, + num INTEGER NOT NULL, + name VARCHAR(255) NOT NULL, + mjesto VARCHAR(255) ); - } - ); + }); - $dbh->do( - q{ - create table files - ( ID INTEGER NOT NULL PRIMARY KEY, - shareID INTEGER NOT NULL, - backupNum INTEGER NOT NULL, - name VARCHAR(255) NOT NULL, - path VARCHAR(255) NOT NULL, - fullpath VARCHAR(255) NOT NULL, - date TIMESTAMP NOT NULL, - type INTEGER NOT NULL, - size INTEGER NOT NULL, - dvdid INTEGER + $dbh->do(qq{ + create table files ( + ID INTEGER NOT NULL PRIMARY KEY, + shareID INTEGER NOT NULL references shares(id), + backupNum INTEGER NOT NULL references backups(num), + name VARCHAR(255) NOT NULL, + path VARCHAR(255) NOT NULL, + fullpath VARCHAR(255) NOT NULL, + date TIMESTAMP NOT NULL, + type INTEGER NOT NULL, + size INTEGER NOT NULL, + dvdid INTEGER references dvds(id) ); - } - ); - print "done\n"; - } - -if ($opt{d}) - { - print("deleting db first...\n"); - - $dbh->do( - q{ DELETE FROM hosts; } - ); - $dbh->do( - q{ DELETE FROM shares; } - ); - $dbh->do( - q{ DELETE FROM files;} - ); - $dbh->do( - q{ DELETE FROM dvds;} - ); - $dbh->do( - q{ DELETE FROM backups; } - ); - } + }); + + print "creating indexes...\n"; + + foreach my $index (qw( + hosts_name + backups_hostID + backups_num + shares_hostID + shares_name + files_shareID + files_path + files_name + files_date + files_size + )) { + my ($table,$col) = split(/_/, $index); + $dbh->do(qq{ create index $index on $table($col) }); + } + + +} + +if ($opt{d}) { + print "deleting "; + foreach my $table (qw(hosts shares files dvds backups)) { + print "$table "; + $dbh->do(qq{ DELETE FROM $table }); + } + print " done...\n"; +} + +if ($opt{v}) { + print "Debug level at $opt{v}\n"; + $debug = $opt{v}; +} #################################INSERT VALUES############################# # get hosts -$hosts = $bpc->HostInfoRead("localhost"); -my $host_key; +$hosts = $bpc->HostInfoRead(); my $hostID; my $shareID; -foreach $host_key (keys %{$hosts}) -{ - my $hostname = $hosts->{$host_key}->{'host'}; - my $backups; - my $sql; - - $sql = q{ SELECT ID FROM hosts WHERE name=? }; - my $st = $dbh->prepare($sql); - $st->bind_param(1,$hosts->{$host_key}->{'host'}); - $st->execute(); - my $tmp = $st->fetchrow_hashref(); - if ($tmp->{'ID'} ne "") - { - $hostID = $tmp->{'ID'}; - } - else - { - $sql = q{ INSERT INTO hosts ( ID, name, IP) VALUES (NULL,"}. - $hosts->{$host_key}->{'host'}."\", \"". - $hosts->{$host_key}->{'ip'}."\");"; - - $dbh->do($sql); - $hostID = $dbh->func('last_insert_rowid'); - - } - $st->finish(); - print("processing host ".$hosts->{$host_key}->{'host'}.":\n"); + +my $sth; + +$sth->{insert_hosts} = $dbh->prepare(qq{ +INSERT INTO hosts (name, IP) VALUES (?,?) +}); + +$sth->{hosts_by_name} = $dbh->prepare(qq{ +SELECT ID FROM hosts WHERE name=? +}); + +$sth->{backups_broj} = $dbh->prepare(qq{ +SELECT COUNT(*) +FROM backups +WHERE hostID=? AND num=? +}); + +$sth->{insert_backups} = $dbh->prepare(qq{ +INSERT INTO backups (hostID, num, date, type) +VALUES (?,?,?,?) +}); + +$sth->{insert_files} = $dbh->prepare(qq{ +INSERT INTO files + (shareID, backupNum, name, path, fullpath, date, type, size) + VALUES (?,?,?,?,?,?,?,?) +}); + +foreach my $host_key (keys %{$hosts}) { + + my $hostname = $hosts->{$host_key}->{'host'} || die "can't find host for $host_key"; + + $sth->{hosts_by_name}->execute($hosts->{$host_key}->{'host'}); + + unless (($hostID) = $sth->{hosts_by_name}->fetchrow_array()) { + $sth->{insert_hosts}->execute( + $hosts->{$host_key}->{'host'}, + $hosts->{$host_key}->{'ip'} + ); + + $hostID = $dbh->func('last_insert_rowid'); + } + + print("host ".$hosts->{$host_key}->{'host'}.": "); - # get backups for a host - my @backups = $bpc->BackupInfoRead($hostname); - foreach my $backup (@backups) - { - my $backupNum = $backup->{'num'}; - my @backupShares = (); + # get backups for a host + my @backups = $bpc->BackupInfoRead($hostname); + print scalar @backups, " increments\n"; + + my $inc_nr = 0; + + foreach my $backup (@backups) { + + $inc_nr++; + last if ($opt{m} && $inc_nr > $opt{m}); + + my $backupNum = $backup->{'num'}; + my @backupShares = (); + + print $hosts->{$host_key}->{'host'}, + "\t#$backupNum\t", $backup->{type} || '?', " ", + $backup->{nFilesNew} || '?', "/", $backup->{nFiles} || '?', + " files\n"; + + $sth->{backups_broj}->execute($hostID, $backupNum); + my ($broj) = $sth->{backups_broj}->fetchrow_array(); + next if ($broj > 0); + + my $files = BackupPC::View->new($bpc, $hostname, \@backups, 1); + foreach my $share ($files->shareList($backupNum)) { + + my $t = time(); + + print strftime($t_fmt,localtime())," ", $share; + $shareID = getShareID($share, $hostID, $hostname); + + my ($f, $nf, $d, $nd) = recurseDir($bpc, $hostname, $files, $backupNum, $share, "", $shareID); + printf(" %d/%d files %d/%d dirs [%.2f/s]\n", + $nf, $f, $nd, $d, + ( ($f+$d) / ((time() - $t) || 1) ) + ); + $dbh->commit(); + } + + $sth->{insert_backups}->execute( + $hostID, + $backupNum, + $backup->{'endTime'}, + $backup->{'type'} + ); + $dbh->commit(); - - if ($opt{u}) - { - my $sql = q{ - SELECT COUNT(*) AS broj - FROM backups - WHERE hostID=? AND - num=? - }; - - my $st = $dbh->prepare($sql); - $st->bind_param(1,$hostID); - $st->bind_param(2,$backupNum); - $st->execute(); - my $tmp = $st->fetchrow_hashref(); - $st->finish(); - if ($tmp->{'broj'} > 0) - { - next; - } } - - print "\tprocessing backup no. $backupNum..."; - my $sql = - q{ - INSERT INTO backups (hostID, num, date, type) - VALUES - (}.$hostID.",". $backupNum.q{, }.$backup->{'endTime'}.",\"". $backup->{'type'}.q{"); - }; - $dbh->do($sql); - - my $files = BackupPC::View->new($bpc, $hostname, \@backups); - @backupShares = $files->shareList($backupNum); - foreach my $share (@backupShares) - { - my @flattenedFiles = (); - clearBeenThereCache(); - print "\n\t\tprocessing share ".$share."..."; - $shareID = getShareID($share, $hostID, $hostname); - - @flattenedFiles = recurseDir($bpc, $hostname, \@backups, $backupNum, $share, ""); - print "done\n"; - print "\t\tinserting data into db..."; - foreach my $file (@flattenedFiles) - { - $dbh->do("INSERT INTO files(ID, shareID, backupNum, name, path, fullpath, date, type, size) VALUES " - ."( NULL, $shareID, " - .$backupNum.", \"" - .$file->{'fileName'}."\", \"" - .$file->{'relPath'}. "\", \"" - .$file->{'fullPath'}."\", " - .$file->{'mtime'}.", " - .$file->{'type'}.", " - .$file->{'size'}.")" - ); - } - print "done\n"; - } - } - print "done.\n"; } +undef $sth; $dbh->commit(); $dbh->disconnect(); +$pidfile->remove; -sub haveBeenThere -{ - my ($where) = @_; - - foreach my $st (@beenThere) - { - if ($where eq $st) - { - return 1; - } - } - - push(@beenThere, $where); - return 0; -} +sub getShareID() { + + my ($share, $hostID, $hostname) = @_; + + $sth->{share_id} ||= $dbh->prepare(qq{ + SELECT ID FROM shares WHERE hostID=? AND name=? + }); -sub clearBeenThereCache -{ - @beenThere = (); + $sth->{share_id}->execute($hostID,$share); + + my ($id) = $sth->{share_id}->fetchrow_array(); + + return $id if (defined($id)); + + $sth->{insert_share} ||= $dbh->prepare(qq{ + INSERT INTO shares + (hostID,name,share,localpath) + VALUES (?,?,?,?) + }); + + my $drop_down = $hostname . '/' . $share; + $drop_down =~ s#//+#/#g; + + $sth->{insert_share}->execute($hostID,$share, $drop_down ,undef); + return $dbh->func('last_insert_rowid'); } -sub getShareID() - { - my ($share, $hostID, $hostname) = @_; - my $tmp; - - my $st = $dbh -> prepare(" SELECT shares.ID AS ID FROM shares WHERE hostID=? AND name=?"); - $st -> execute($hostID,$share); - my $tmp = $st->fetchrow_hashref(); - $st->finish(); - if ($tmp) - { - return $tmp->{'ID'}; - } - my $sql = - q{ - INSERT INTO shares(ID,hostID,name,share,localpath) - VALUES - (NULL,}. - "$hostID,\"$share\",\"//$hostname$share\",NULL);"; - $dbh->do($sql); - return $dbh->func('last_insert_rowid'); - } +sub found_in_db { + + my ($shareID,undef,$name,$path,undef,$date,undef,$size) = @_; + + $sth->{file_in_db} ||= $dbh->prepare(qq{ + SELECT count(*) FROM files + WHERE shareID = ? and + path = ? and + name = ? and + date = ? and + size = ? + }); + + my @param = ($shareID,$path,$name,$date,$size); + $sth->{file_in_db}->execute(@param); + my ($rows) = $sth->{file_in_db}->fetchrow_array(); +# print STDERR ( $rows ? '+' : '-' ), join(" ",@param), "\n"; + return $rows; +} #################################################### # recursing through filesystem structure and # # and returning flattened files list # #################################################### -sub recurseDir -{ - my ($bpc, $hostname, $backups, $backupNo, $share, $dir) = @_; - my @ret = (); - my $files = BackupPC::View->new($bpc, $hostname, $backups); - my $filesInBackup = $files->dirAttrib($backupNo, $share, $dir); - my $file_key = ""; +sub recurseDir($$$$$$$$) { - - # first, add all the entries in current directory - foreach $file_key (keys %{$filesInBackup}) - { - push(@ret, { - 'fileName' => $file_key, - 'relPath' => $filesInBackup->{$file_key}->{'relPath'}, - 'fullPath' => $filesInBackup->{$file_key}->{'fullPath'}, - 'sharePath'=> $filesInBackup->{$file_key}->{'sharePathM'}, - 'size' => $filesInBackup->{$file_key}->{'size'}, - 'mtime' => $filesInBackup->{$file_key}->{'mtime'}, - 'type' => $filesInBackup->{$file_key}->{'type'} - }); - } - - # then, recurse thru subfolders - foreach my $fold (@ret) - { - if ($fold->{'type'} == BPC_FTYPE_DIR && - haveBeenThere($fold->{'relPath'}) != 1 - ) - { - - push(@ret, - recurseDir($bpc, $hostname, $backups, $backupNo, $share, $fold->{'relPath'}) - ); - } - } - return @ret; + my ($bpc, $hostname, $files, $backupNum, $share, $dir, $shareID) = @_; + + print STDERR "\nrecurse($hostname,$backupNum,$share,$dir,$shareID)\n" if ($debug >= 1); + + my ($nr_files, $new_files, $nr_dirs, $new_dirs) = (0,0,0,0); + + { # scope + my @stack; + + my $filesInBackup = $files->dirAttrib($backupNum, $share, $dir); + + # first, add all the entries in current directory + foreach my $path_key (keys %{$filesInBackup}) { + my @data = ( + $shareID, + $backupNum, + $path_key, + $filesInBackup->{$path_key}->{'relPath'}, + $filesInBackup->{$path_key}->{'fullPath'}, + # $filesInBackup->{$path_key}->{'sharePathM'}, + $filesInBackup->{$path_key}->{'mtime'}, + $filesInBackup->{$path_key}->{'type'}, + $filesInBackup->{$path_key}->{'size'} + ); + + my $key = join(" ", ( + $shareID, + $dir, + $path_key, + $filesInBackup->{$path_key}->{'mtime'}, + $filesInBackup->{$path_key}->{'size'} + )); + + + if (! $beenThere->{$key} && ! found_in_db(@data)) { + print STDERR "# key: $key [", $beenThere->{$key},"]" if ($debug >= 2); + $sth->{'insert_files'}->execute(@data); + if ($filesInBackup->{$path_key}->{'type'} == BPC_FTYPE_DIR) { + $new_dirs++; + print STDERR " dir\n" if ($debug >= 2); + } else { + $new_files++; + print STDERR " file\n" if ($debug >= 2); + } + } + $beenThere->{$key}++; + + if ($filesInBackup->{$path_key}->{'type'} == BPC_FTYPE_DIR) { + $nr_dirs++; + + my $full_path = $dir . '/' . $path_key; + push @stack, $full_path; + print STDERR "### store to stack: $full_path\n" if ($debug >= 3); + +# my ($f,$nf,$d,$nd) = recurseDir($bpc, $hostname, $backups, $backupNum, $share, $path_key, $shareID) unless ($beenThere->{$key}); +# +# $nr_files += $f; +# $new_files += $nf; +# $nr_dirs += $d; +# $new_dirs += $nd; + + } else { + $nr_files++; + } + } + + print STDERR "## STACK ",join(", ", @stack),"\n" if ($debug >= 2); + + while ( my $dir = shift @stack ) { + my ($f,$nf,$d,$nd) = recurseDir($bpc, $hostname, $files, $backupNum, $share, $dir, $shareID); + print STDERR "# $dir f: $f nf: $nf d: $d nd: $nd\n" if ($debug >= 1); + $nr_files += $f; + $new_files += $nf; + $nr_dirs += $d; + $new_dirs += $nd; + } + } + + return ($nr_files, $new_files, $nr_dirs, $new_dirs); }