[ViewVC] Annotation of: cvs/Net-FCP/bin/fmd

#!/opt/bin/perl

system "echo $$ >fmd.pid";

=head1 fmd - the freenet mass downloader

Fmd is at a very early stage of development (and very hackish, too), as I
am learning the basics of freenet myself.

However, I use it in production, and since it verifies everything it
decodes etc., it seems to be quite safe to use, that is, if you know how
to debug perl :)

=head2 FEATURES

 - decoding is done "in place", i.e. all non-checkblock blocks are
   stored in-place and will not be moved, only checkblocks will
   be moved to their final position in the file before decoding.
 - high resistance against failures of fred or fmd
 - extremely persistent retry behaviour - there is no such thing
   as a permanent failure.
 - handles hundreds of simultaneous downloads with grace.

=head2 ENVIRONMENT

Set FMD_HOME to a directory (default ~/fmd) where fmd will store it's
files. It will not store your porn files or other freenet data outside
that directory.

The subdirectory db will contain a database (soon to go away), while tmp
contains queue files and partial splitfiles. All finished files will be
moved to the done subdir.

FREDHOST and FREDPORT do the obvious. Fix these docs if you disagree.

Also edit the fmd executable for the number of threads and other
non-useful constants. The default number (250) works for me, probably not
for you.

=head2 COMMANDS

=over 4

=item CHK@... (space or "/") filename

Just pasting a CHK and a filename sperated by one or more spaces or
slashes will add files to the queue. The line can have leading garbage,
i.e. you can paste full uris.

=item <attachment>...</attachment>

=item <attach>...</attach>

=item <attached>...</attached>

Will add frosts mentally deranged pseudo-xml format, soon to be replaced
by something even more horrible.

=item l

List all jobs by number.

=item <number> (optionally trailing command)

Selects the job with the given number for further comamnds that require a current job.

=item s

Show the current job. Bot useful right now.

=item k

Kill the current job. This will keep the temp. file around. Sorry. (But it
could be used to reconstruct the download.. hmm..)

=item pri<number>

Set the current job priority to <pri>. The default is 1. A job with higher
priority will (on average) get more requests.

The useful range is probably 0..100, but be careful and limit yourself to
small values (<10), otherwise your other downloads might starve!

=item q

Kill the command prompt. Yupp.

=back

=cut

use strict;

use Net::FCP;
use Storable;
use Time::HiRes;
use Coro 3.0;
use Coro::Channel;
use Coro::Event;
use Coro::Handle;
use Coro::Signal;
use Coro::Timer;
use Coro::AIO;
use List::Util;
use Digest::SHA1;
use Algorithm::FEC;
use Digest::SHA1;
use Net::FCP::Util;
use POSIX ();

Coro::AIO::max_poll_time 0.01;
Coro::AIO::min_parallel 1;
Coro::AIO::max_parallel 2;

$|=1;

our $MAX_TXN = 250; # use max. this many transactions in parallel
our @HTL = (5,12,20);
our $VERIFY_CHK = 1; # verify all blocks, again and again (only useful to debugging).
our $FMD_HOME = $ENV{FMD_HOME} || "$ENV{HOME}/fmd";

our $FCP = new Net::FCP;

defined $FMD_HOME
   or die "you currently must define FMD_HOME to a persistent directory";

mkdir $FMD_HOME, 0700;

our $QUEUE_HOME = "$FMD_HOME/tmp";
mkdir $QUEUE_HOME, 0700;
our $DONE_HOME = "$FMD_HOME/done";
mkdir $DONE_HOME, 0700;

our %job;

sub push_key {
   my ($key, $title) = @_;

   for my $job (values %job) {
      if ($job->{p}{key} eq $key) {
         warn "job $job->{id} already works on this, not adding";
         return;
      }
   }

   my $job = job->new_from_key ($key, $title);
   warn "added as job $job->{id}\n";
   $job;
}

sub cmdline {
   my ($i, $o) = @_;

   my $cmd = async {
      my $job;
      while (print $o "> " and defined ($_ = <$i>)) {
         chomp;
         if (/<attach(?:ment|ed|)>(.*) \* (CHK[^<]+)<\/attach/) {
            $job = push_key $2, $1;
         } elsif (/(CHK\@[a-zA-Z0-9,~\-]{54})[\/ ]+(.*?)\s*$/) {
            $job = push_key $1, $2;
         } elsif (s/^(\d+)//) {
            if ($job = $job{$1}) {
               print $o delete $job->{log};
               if ($job->{input}) {
               }
            }
            redo;
         } elsif (/^q/) {
            close $i;
            close $o;
         } elsif (/^l/) {
            for my $job (sort { $a->{id} <=> $b->{id} } values %job) {
               print $o "$_ $job->{id}: $job->{p}{key} $job->{p}{title} $job->{status}\n";
            }
#         } elsif (/^pri\s*(\d+)/) {
#            if ($job) {
#               $job->{p}{pri} = $1;
#               $job->save;
#            }
         } elsif (/^s$/) {
            print $o $job->show if $job;
         } elsif (/^k$/) {
            $job->kill if $job;
         } elsif (/^p(.*)/) {
            print $o +(join " ", eval $1), "\n";
         } elsif (/\S/) {
            #
         } else {
            print $o "?\n";
         }
         for my $job (sort { $a->{id} <=> $b->{id} } values %job) {
            if ($job->{_input}) {
                print $o "> $job->{id} $job->{title} $job->{status}\n";
            }
         }
      }
   };
}

package job;

use strict;

use Coro;
use Coro::AIO;
use Fcntl;
use IO::Handle;
use Array::Heap2;

our $count = 0;

sub new {
   my $class = shift;

   my $self = bless { @_ }, $class;

   $self->{p}{title} !~ /\//
      or die "$self->{p}{title}: malformed key\n";

   $self->{p}{title} =~ s/\s+$//;
   $self->{p}{title} =~ s/\n/_/g;

   $self->{id} = ++$count;
   $self->{job} ||= "$QUEUE_HOME/" . Time::HiRes::time . ":$count.j";

   $job{$self->{id}} = $self;

   $self->save;
   $self->start;
   $self;
}

sub new_from_key {
   my ($class, $key, $title) = @_;
   $class->new (p => { key => $key, title => $title, state => "examine" });
}

sub new_from_file {
   my ($class, $path) = @_;
   $class->new (job => $path, p => Storable::retrieve $path);
}

sub save {
   my ($self) = @_;
   Storable::nstore $self->{p}, "$self->{job}~";
   rename "$self->{job}~", $self->{job};
}

sub clean {
   my ($self) = @_;

   delete $job{$self->{id}};
   $self->save;
   rename $self->{job}, "$DONE_HOME/$self->{p}{title}.job";
   unlink $self->{job};
}

sub kill {
   my ($self) = @_;

   $self->clean;
   $self->{coro}->cancel;
}

our @queue;
our $queue_change = new Coro::Signal;
our $queue_alloc = 0;

async {
   for (;;) {
      while (@queue
             and (($queue[0][0] > 10 and $queue_alloc < $MAX_TXN)
               or ($queue[0][0] >  1 and $queue_alloc < $MAX_TXN - 3)
               or $queue_alloc < $MAX_TXN - 5)) {
         (pop_heap @queue)->[1]->send;
         $queue_alloc++;
         Coro::Timer::sleep 0.05;
      }
      $queue_change->wait;
   }
};

sub txn_begin {
   my ($pri) = @_;
   my $sig = new Coro::Signal;

   #warn "txn_begin $pri\n";#d#
   push_heap @queue, [$pri, $sig];
   $queue_change->send;
   $sig->wait;
}

sub txn_end {
   $queue_alloc--;
   $queue_change->send;
}

sub txn_client_get {
   my %arg = @_;

   txn_begin $arg{pri};
   $FCP->txn_client_get ($arg{uri}, $arg{htl})->cb (unblock_sub {
      txn_end;

      $arg{cb}->(@_);
   });
}

sub fetch_uri {
   my ($pri, $uri) = @_;

   for(my $count = 1; ; $count += 0.3) {
      for my $htl (@HTL) {
         txn_begin time + $htl + $count;
         my ($meta, $data) = eval { @{ $FCP->client_get ($uri, $htl) } };
         txn_end;
         if ($@) {
            if (UNIVERSAL::isa ($@, Net::FCP::Exception::)) {
               if ($@->type ("data_not_found")
                   || $@->type ("route_not_found")) {
                  next;
               }
               if ($@->type ("short_data")) {
                  warn "(short_data, redo)\n";
                  redo;
               }
               die;
            }
         }
         if (defined $data) {
            return ($meta, $data);
         }
      }
   }

   die;
}

sub log {
   my ($self, $text) = @_;
   my $time = POSIX::strftime "%H:%M:%S", localtime time;
   warn "$time $self->{id},$self->{p}{pri}: $text\n";
   $self->{log} .= "$time $text\n";
}

sub feedback {
   my ($self, $prompt) = @_;
   $self->{input} = [$Coro::current, $prompt];
   Coro::schedule;
}

sub show {
   my ($self) = @_;

   "ID: $self->{id}\n"
      . "Title: $self->{p}{title}\n"
      . "Blocks#: " . @{$self->{p}{blk}} . "\n"
      . "Blocks: " . (join "", map {
         $_->{done} ? "+" : "-"
      } @{$self->{p}{blk}}) . "\n" .
   ""
}

our $id;

sub MAXSEG (){ 128*1024*1024 }
sub MINSEG (){   6* 128*1024 }

sub blocksize($) {
   return
        $_[0] >= 64*1024*1024 ? 1024*1024
      : $_[0] >= 32*1024*1024 ?  512*1024
      : $_[0] >=    1024*1024 ?  256*1024
                              :  128*1024;
}

sub start {
   my ($self) = @_;

   $self->{p}{pri} ||= 1;
   $self->{p}{state} ||= "examine";

   $self->{job} =~ /^(.*)\.j$/ or die "$self->{job}: missing .j";
   $self->{file} = "$1.d";
   $self->{fh} = aio_open $self->{file}, O_RDWR|O_CREAT, 0600
      or die "$self->{file}: $!";

   $self->{status} = "starting";
   $self->{coro} = async {
      for(;;) {
         my ($state, @args) = ref $self->{p}{state} ? @{$self->{p}{state}} : $self->{p}{state};
         my $next = eval { $self->can ("state_$state")->($self, @args) };
         if ($@) {
            $self->log ($@);
            $next = $self->feedback ("continue with state: ");
         }
         $self->log ($self->{status} = "STATE CHANGE: ". join ", ", ref $next ? @$next : $next);
         $self->{p}{state} = $next;
         $self->save;
      }
   };
}

sub state_finish {
   my ($self, $save) = @_;

   if ($save) {
      aio_fsync $self->{fh};
      close $self->{fh};

      aio_unlink "$DONE_HOME/$self->{p}{title}";
      aio_link $self->{file}, "$DONE_HOME/$self->{p}{title}"
         and die "link: $self->{file} => $DONE_HOME/$self->{p}{title}: $!";
      system "sync";
   }
   $self->clean;

   aio_unlink $self->{file};

   $self->{status} = "finished";
   $self->feedback ("finished");
   terminate;
}

sub state_examine {
   my ($self) = @_;
   my $p = $self->{p};

   $self->{status} = "initial fetch";

   for (;;) {
      $self->log ("fetching $p->{key} (=$p->{title})");

      ($p->{meta}, $p->{data}) = fetch_uri 100, "freenet:$p->{key}";
      $self->save;
      #use PApp::Util; print STDERR PApp::Util::dumpval [keys %{$meta->{document}[0]{split_file}}];
      $self->log ("type $p->{meta}{document}[0]{info}{format}");

      if (my $splitfile = $p->{meta}{document}[0]{split_file}) {
         return "splitfile";
      } elsif ((defined $p->{data}) and (length $p->{data})) {
         syswrite $self->{fh}, delete $p->{data};
         aio_fsync $self->{fh};
         return ["finish", 1];
      }

      $self->log ("EMPTY, retrying in an hour");
      Coro::Timer::sleep 3600;
   }

}

sub state_splitfile {
   my ($self) = @_;
   my $p = $self->{p};

   my $splitfile = $p->{meta}{document}[0]{split_file};
   my $filesize = hex $splitfile->{size};

   if ($splitfile->{algo_name} eq "OnionFEC_a_1_2") {
      my $data_packets = hex $splitfile->{block_count};
      my $check_packets = hex $splitfile->{check_block_count};

      my $blk = ($p->{blk} ||= []);

      unless (@$blk) {
         for (1..$data_packets) {
            push @$blk, {
               uri => $splitfile->{block}{sprintf "%x", $_},
            };
         }
         for (1..$check_packets) {
            push @$blk, {
               uri => $splitfile->{check_block}{sprintf "%x", $_},
            };
         }
      }

      my @segments;
      my $segments = 0;

      {
         # that is a horrible algorithm :(, these freenet freaks are... java-disabled
         # hardcoding lots of magic parameters is soo dumb.
         my $size = $filesize;
         my $offset = 0;
         my $offset2 = ($filesize & ~(1024*1024-1)) + 1024*1024; # leave enough space after last data block
         my $idx  = 0;
         my $idx2 = $data_packets;
         my @redundandy = (0,1,2); # maybe OnionFAC_a_1_2 means 1/2 redundancy(?)

         while ($size > 0) {
            my $segsize = $size >= MAXSEG ? MAXSEG : $size <= MINSEG ? MINSEG : $size;
            my $blksize = blocksize $segsize;
            my $seg    = 
               {
                  id      => $segments++,
                  todo    => int (($segsize + $blksize - 1) / $blksize),
                  done    => 0,
                  blk     => [],
                  blksize => $blksize,
               };

            push @segments, $seg;
            $size -= $segsize;

            while ($segsize > 0) {
               push @{$seg->{blk}}, $idx;
               for ($blk->[$idx++]) {
                  $_->{offset} = $offset;
                  #$_->{size}   = $blksize > $segsize ? $segsize : $blksize; # WRONG
                  $_->{size}   = $blksize;
                  $_->{seg}    = $seg;
               }

               $segsize -= $blksize;
               $offset  += $blksize;

               if (($redundandy[0] += $redundandy[1]) >= $redundandy[2]) {
                  $redundandy[0] -= $redundandy[2];

                  push @{$seg->{blk}}, $idx2;
                  for ($blk->[$idx2++]) {
                     $_->{offset} = $offset2;
                     $_->{size}   = $blksize;
                     $_->{seg}    = $seg;
                  }

                  $offset2 += $blksize;
               }
            }
         }

         $idx == $data_packets
            or die "$self->{id}/$p->{tile} $self->{job}\nidx $idx != data_packets $data_packets";
         $idx2 == $data_packets + $check_packets
            or die "$self->{id}/$p->{tile} $self->{job}\nidx2 $idx2 != data_packets $data_packets + check_packets $check_packets";
      }

      for (@$blk) {
         ++$_->{seg}{done} if $_->{done};
         delete $_->{htl};
      }

      my $fail = 0;
      my $sig = new Coro::Signal;

      $self->{status} = "splitfile fetch (" .  @$blk . " blocks)";

      my @txn;

      for (;;) {
         for my $id (0 .. $#$blk) {
            my $blk = $blk->[$id];

            next if $txn[$id] || $blk->{done} || $blk->{seg}{todo} <= $blk->{seg}{done};

            my $htl = $HTL[$blk->{htl}++ % @HTL];
            my $pri = int time + $htl / 20 * 7200 * rand;

            #warn $self->{id} . ", GET<$htl, $pri>\n";#d#
            $txn[$id] ||= txn_client_get pri => $pri, uri => $blk->{uri}, htl => $htl, cb => sub {
               undef $txn[$id];

               my $seg = $blk->{seg};

               my ($meta, $data) = eval { @{ $_[0]->result } };

               if (defined $data) {
                  $blk->{size} == length $data
                     or die sprintf "block $id expected size %d, got %d\n", $blk->{size}, length $data;

                  (length $data) == (aio_write $self->{fh}, $blk->{offset}, (length $data), $data, 0)
                     or die "unable to write chunk to disk, not setting valid flag";
                  aio_fsync $self->{fh};

                  $blk->{done} = 1;
                  $blk->{meta} = $meta->{raw} if length $meta->{raw};#d#
                  $seg->{done}++;
                  $self->save;

                  $::htl_sum += $htl;
                  $::htl_cnt++;

                  $self->log (sprintf "got block $seg->{id}.$id %d ($seg->{done}/$seg->{todo}) at htl $htl (%f) and pri $pri",
                                       length $data, $::htl_sum / $::htl_cnt);
               } else {
                  if ($@) {
                     if ($@->type ("data_not_found")) {
                        # nop
                     } elsif ($@->type ("network_error")) {
                        $self->log ("$@, retrying in 1s");
                        CORE::sleep 1;
                     } else {
                        $self->log ("$@");
                     }
                  }
                  ++$fail;
               }
               $self->{status} = "splitfile fetch ($seg->{done}/$seg->{todo}, $fail failed)";

               $sig->send;
            };
         }

         for my $seg (@segments) {
            if ($seg->{done} >= $seg->{todo} && !$seg->{finished}) {

               $self->log ("segment done, cancelling segment $seg->{id}");
               for my $id (@{$seg->{blk}}) {
                  (delete $txn[$id])->cancel if $txn[$id];
               }

               my $verify;
               for my $id (@{$seg->{blk}}) {
                  my $blk = $blk->[$id];
                  if ($blk->{done}) {
                     aio_read $self->{fh}, $blk->{offset}, $blk->{size}, my $buf, 0;

                     my $k1 = Net::FCP::Util::extract_chk_hash $blk->{uri};
                     my $k2 = Net::FCP::Util::generate_chk_hash $blk->{meta}, $buf;

                     if ($k1 ne $k2) {
                        $verify .= "v";
                        #warn sprintf "$p->{title} block $id BROKEN (%s != %s)", (unpack "H*", $k1), (unpack "H*", $k2);
                        $blk->{done} = 0;
                        $seg->{done}--;
                        $self->save;
                     } else {
                        $verify .= "V";
                     }
                  }
               }
               $self->log ("verified segment $seg->{id}");
               $self->log ($verify);

               if ($seg->{done} >= $seg->{todo} && !$seg->{finished}) {
                  $self->log ("verified segment OK $seg->{id}");
                  $seg->{finished}++;
                  $segments--;
               } else {
                  $self->log ("verified segment NOT OK $seg->{id}");
               }
            }
         }

         last unless $segments;

         $sig->wait;
      }

      $self->log ("decoding < $self->{job} $self->{file} $filesize >");

      for my $seg (@segments) {
         my @part;
         my @idx;
         my @blk = map $blk->[$_], sort { $a <=> $b } @{$seg->{blk}};

         for my $id (0 .. $#blk) {
            my $blk = $blk[$id];
            next unless $blk->{done};

            push @part, [$self->{fh}, $blk->{offset}];
            push @idx, $id;

            last if @idx == $seg->{todo};
         }

         my $fec = new Algorithm::FEC
                          $seg->{todo},
                          scalar @blk,
                          $seg->{blksize};

         $fec->shuffle (\@part, \@idx);

         # now copy check blocks to their destination position
         for my $i (0 .. $#idx) {
            next if $idx[$i] == $i;

            my $src = $part[$i];
            $part[$i] = [$self->{fh}, $blk[$i]{offset}];
            $fec->copy ($src, $part[$i]);
         }

         $fec->set_decode_blocks (\@part, \@idx);
         $fec->decode;
      }

      my $sha1 = new Digest::SHA1;
      open my $dd, "-|", "head -c$filesize \Q$self->{file}\E"
         or die "DD: $!";
      #$dd = Coro::Handle::unblock $dd;
      $sha1->addfile ($dd);
      $sha1 = $sha1->hexdigest;

      if (exists $p->{meta}{document}[0]{info}{checksum}
          and $p->{meta}{document}[0]{info}{checksum} ne $sha1) {
         $self->log ("META: $p->{meta}{document}[0]{info}{checksum} and real checksum $sha1 for $filesize DIFFER");
#         $self->feedback ("CHECKSUM ERROR");
#         terminate;
      }

      truncate $self->{fh}, $filesize;
      sysseek $self->{fh}, 0, 0;

      return ["finish", 1];
   } else {
      $self->log ("splitfile algo '$splitfile->{algo_name}' unknown");
      $self->feedback ("algo unknown");
      terminate;
   }
}

package main;

$|=1;

async {
   for (<\Q$QUEUE_HOME\E/*.j>) {
      job->new_from_file ($_);
      print "J";
   }
   print "\n";
};

open my $stdin , "<&0" or die;
open my $stdout, ">&1" or die;
cmdline unblock $stdin, unblock $stdout;

&Coro::Event::loop;

Revision:	1.16
Committed:	Mon Dec 4 13:47:28 2006 UTC (19 years, 9 months ago) by root
Branch:	MAIN
Changes since 1.15:	+6 -8 lines
Log Message:	* empty log message *
#	User	Rev	Content
1	root	1.1	#!/opt/bin/perl
2
3	root	1.14	system "echo $$ >fmd.pid";
4
5	root	1.1	=head1 fmd - the freenet mass downloader
6
7			Fmd is at a very early stage of development (and very hackish, too), as I
8			am learning the basics of freenet myself.
9
10			However, I use it in production, and since it verifies everything it
11			decodes etc., it seems to be quite safe to use, that is, if you know how
12			to debug perl :)
13
14			=head2 FEATURES
15
16			- decoding is done "in place", i.e. all non-checkblock blocks are
17			stored in-place and will not be moved, only checkblocks will
18			be moved to their final position in the file before decoding.
19			- high resistance against failures of fred or fmd
20			- extremely persistent retry behaviour - there is no such thing
21			as a permanent failure.
22			- handles hundreds of simultaneous downloads with grace.
23
24			=head2 ENVIRONMENT
25
26			Set FMD_HOME to a directory (default ~/fmd) where fmd will store it's
27			files. It will not store your porn files or other freenet data outside
28			that directory.
29
30			The subdirectory db will contain a database (soon to go away), while tmp
31			contains queue files and partial splitfiles. All finished files will be
32			moved to the done subdir.
33
34			FREDHOST and FREDPORT do the obvious. Fix these docs if you disagree.
35
36			Also edit the fmd executable for the number of threads and other
37	root	1.14	non-useful constants. The default number (250) works for me, probably not
38	root	1.1	for you.
39
40			=head2 COMMANDS
41
42			=over 4
43
44			=item CHK@... (space or "/") filename
45
46			Just pasting a CHK and a filename sperated by one or more spaces or
47			slashes will add files to the queue. The line can have leading garbage,
48			i.e. you can paste full uris.
49
50			=item <attachment>...</attachment>
51
52			=item <attach>...</attach>
53
54			=item <attached>...</attached>
55
56			Will add frosts mentally deranged pseudo-xml format, soon to be replaced
57			by something even more horrible.
58
59			=item l
60
61			List all jobs by number.
62
63			=item <number> (optionally trailing command)
64
65			Selects the job with the given number for further comamnds that require a current job.
66
67			=item s
68
69			Show the current job. Bot useful right now.
70
71			=item k
72
73			Kill the current job. This will keep the temp. file around. Sorry. (But it
74			could be used to reconstruct the download.. hmm..)
75
76			=item pri<number>
77
78	root	1.2	Set the current job priority to <pri>. The default is 1. A job with higher
79			priority will (on average) get more requests.
80	root	1.1
81	root	1.2	The useful range is probably 0..100, but be careful and limit yourself to
82			small values (<10), otherwise your other downloads might starve!
83	root	1.1
84			=item q
85
86			Kill the command prompt. Yupp.
87
88			=back
89
90			=cut
91
92	root	1.10	use strict;
93
94	root	1.1	use Net::FCP;
95			use Storable;
96			use Time::HiRes;
97	root	1.15	use Coro 3.0;
98			use Coro::Channel;
99	root	1.1	use Coro::Event;
100			use Coro::Handle;
101			use Coro::Signal;
102			use Coro::Timer;
103	root	1.9	use Coro::AIO;
104	root	1.1	use List::Util;
105			use Digest::SHA1;
106			use Algorithm::FEC;
107			use Digest::SHA1;
108			use Net::FCP::Util;
109			use POSIX ();
110
111	root	1.9	Coro::AIO::max_poll_time 0.01;
112			Coro::AIO::min_parallel 1;
113			Coro::AIO::max_parallel 2;
114
115	root	1.1	$\|=1;
116
117	root	1.14	our $MAX_TXN = 250; # use max. this many transactions in parallel
118	root	1.9	our @HTL = (5,12,20);
119	root	1.1	our $VERIFY_CHK = 1; # verify all blocks, again and again (only useful to debugging).
120			our $FMD_HOME = $ENV{FMD_HOME} \|\| "$ENV{HOME}/fmd";
121
122			our $FCP = new Net::FCP;
123
124			defined $FMD_HOME
125			or die "you currently must define FMD_HOME to a persistent directory";
126
127			mkdir $FMD_HOME, 0700;
128
129			our $QUEUE_HOME = "$FMD_HOME/tmp";
130			mkdir $QUEUE_HOME, 0700;
131			our $DONE_HOME = "$FMD_HOME/done";
132			mkdir $DONE_HOME, 0700;
133
134			our %job;
135
136			sub push_key {
137			my ($key, $title) = @_;
138
139			for my $job (values %job) {
140			if ($job->{p}{key} eq $key) {
141			warn "job $job->{id} already works on this, not adding";
142			return;
143			}
144			}
145
146			my $job = job->new_from_key ($key, $title);
147			warn "added as job $job->{id}\n";
148			$job;
149			}
150
151			sub cmdline {
152			my ($i, $o) = @_;
153
154			my $cmd = async {
155			my $job;
156			while (print $o "> " and defined ($_ = <$i>)) {
157			chomp;
158			if (/<attach(?:ment\|ed\|)>(.) \ (CHK[^<]+)<\/attach/) {
159			$job = push_key $2, $1;
160	root	1.7	} elsif (/(CHK\@[a-zA-Z0-9,~\-]{54})[\/ ]+(.?)\s$/) {
161	root	1.1	$job = push_key $1, $2;
162			} elsif (s/^(\d+)//) {
163			if ($job = $job{$1}) {
164			print $o delete $job->{log};
165			if ($job->{input}) {
166			}
167			}
168			redo;
169			} elsif (/^q/) {
170			close $i;
171			close $o;
172			} elsif (/^l/) {
173			for my $job (sort { $a->{id} <=> $b->{id} } values %job) {
174			print $o "$_ $job->{id}: $job->{p}{key} $job->{p}{title} $job->{status}\n";
175			}
176	root	1.16	# } elsif (/^pri\s*(\d+)/) {
177			# if ($job) {
178			# $job->{p}{pri} = $1;
179			# $job->save;
180			# }
181	root	1.1	} elsif (/^s$/) {
182			print $o $job->show if $job;
183			} elsif (/^k$/) {
184			$job->kill if $job;
185	root	1.14	} elsif (/^p(.*)/) {
186	root	1.16	print $o +(join " ", eval $1), "\n";
187	root	1.1	} elsif (/\S/) {
188			#
189			} else {
190			print $o "?\n";
191			}
192			for my $job (sort { $a->{id} <=> $b->{id} } values %job) {
193			if ($job->{_input}) {
194			print $o "> $job->{id} $job->{title} $job->{status}\n";
195			}
196			}
197			}
198			};
199			}
200
201			package job;
202
203	root	1.10	use strict;
204
205	root	1.1	use Coro;
206	root	1.9	use Coro::AIO;
207	root	1.1	use Fcntl;
208			use IO::Handle;
209	root	1.6	use Array::Heap2;
210	root	1.1
211	root	1.14	our $count = 0;
212	root	1.1
213			sub new {
214			my $class = shift;
215
216			my $self = bless { @_ }, $class;
217
218	root	1.9	$self->{p}{title} !~ /\//
219			or die "$self->{p}{title}: malformed key\n";
220
221	root	1.13	$self->{p}{title} =~ s/\s+$//;
222			$self->{p}{title} =~ s/\n/_/g;
223
224	root	1.1	$self->{id} = ++$count;
225			$self->{job} \|\|= "$QUEUE_HOME/" . Time::HiRes::time . ":$count.j";
226
227			$job{$self->{id}} = $self;
228
229			$self->save;
230			$self->start;
231			$self;
232			}
233
234			sub new_from_key {
235			my ($class, $key, $title) = @_;
236			$class->new (p => { key => $key, title => $title, state => "examine" });
237			}
238
239			sub new_from_file {
240			my ($class, $path) = @_;
241			$class->new (job => $path, p => Storable::retrieve $path);
242			}
243
244			sub save {
245			my ($self) = @_;
246			Storable::nstore $self->{p}, "$self->{job}~";
247	root	1.12	rename "$self->{job}~", $self->{job};
248	root	1.1	}
249
250			sub clean {
251			my ($self) = @_;
252
253			delete $job{$self->{id}};
254			$self->save;
255	root	1.12	rename $self->{job}, "$DONE_HOME/$self->{p}{title}.job";
256			unlink $self->{job};
257	root	1.1	}
258
259			sub kill {
260			my ($self) = @_;
261
262			$self->clean;
263			$self->{coro}->cancel;
264			}
265
266	root	1.14	our @queue;
267			our $queue_change = new Coro::Signal;
268			our $queue_alloc = 0;
269	root	1.1
270			async {
271			for (;;) {
272			while (@queue
273			and (($queue[0][0] > 10 and $queue_alloc < $MAX_TXN)
274			or ($queue[0][0] > 1 and $queue_alloc < $MAX_TXN - 3)
275			or $queue_alloc < $MAX_TXN - 5)) {
276			(pop_heap @queue)->[1]->send;
277			$queue_alloc++;
278	root	1.4	Coro::Timer::sleep 0.05;
279	root	1.1	}
280			$queue_change->wait;
281			}
282			};
283
284			sub txn_begin {
285			my ($pri) = @_;
286			my $sig = new Coro::Signal;
287
288			#warn "txn_begin $pri\n";#d#
289			push_heap @queue, [$pri, $sig];
290			$queue_change->send;
291			$sig->wait;
292			}
293
294			sub txn_end {
295			$queue_alloc--;
296			$queue_change->send;
297	root	1.15	}
298
299			sub txn_client_get {
300			my %arg = @_;
301
302			txn_begin $arg{pri};
303			$FCP->txn_client_get ($arg{uri}, $arg{htl})->cb (unblock_sub {
304			txn_end;
305
306			$arg{cb}->(@_);
307			});
308	root	1.1	}
309
310			sub fetch_uri {
311			my ($pri, $uri) = @_;
312
313			for(my $count = 1; ; $count += 0.3) {
314			for my $htl (@HTL) {
315	root	1.4	txn_begin time + $htl + $count;
316	root	1.1	my ($meta, $data) = eval { @{ $FCP->client_get ($uri, $htl) } };
317			txn_end;
318			if ($@) {
319			if (UNIVERSAL::isa ($@, Net::FCP::Exception::)) {
320			if ($@->type ("data_not_found")
321			\|\| $@->type ("route_not_found")) {
322			next;
323			}
324			if ($@->type ("short_data")) {
325			warn "(short_data, redo)\n";
326			redo;
327			}
328			die;
329			}
330			}
331			if (defined $data) {
332			return ($meta, $data);
333			}
334			}
335			}
336
337			die;
338			}
339
340			sub log {
341			my ($self, $text) = @_;
342			my $time = POSIX::strftime "%H:%M:%S", localtime time;
343	root	1.3	warn "$time $self->{id},$self->{p}{pri}: $text\n";
344	root	1.1	$self->{log} .= "$time $text\n";
345			}
346
347			sub feedback {
348			my ($self, $prompt) = @_;
349			$self->{input} = [$Coro::current, $prompt];
350			Coro::schedule;
351			}
352
353			sub show {
354			my ($self) = @_;
355
356			"ID: $self->{id}\n"
357			. "Title: $self->{p}{title}\n"
358			. "Blocks#: " . @{$self->{p}{blk}} . "\n"
359			. "Blocks: " . (join "", map {
360			$_->{done} ? "+" : "-"
361			} @{$self->{p}{blk}}) . "\n" .
362	root	1.11	""
363	root	1.1	}
364
365	root	1.14	our $id;
366	root	1.1
367			sub MAXSEG (){ 12810241024 }
368			sub MINSEG (){ 6* 128*1024 }
369
370			sub blocksize($) {
371			return
372			$_[0] >= 6410241024 ? 1024*1024
373			: $_[0] >= 3210241024 ? 512*1024
374			: $_[0] >= 10241024 ? 2561024
375			: 128*1024;
376			}
377
378			sub start {
379			my ($self) = @_;
380
381			$self->{p}{pri} \|\|= 1;
382			$self->{p}{state} \|\|= "examine";
383
384	root	1.9	$self->{job} =~ /^(.*)\.j$/ or die "$self->{job}: missing .j";
385			$self->{file} = "$1.d";
386			$self->{fh} = aio_open $self->{file}, O_RDWR\|O_CREAT, 0600
387	root	1.1	or die "$self->{file}: $!";
388
389			$self->{status} = "starting";
390			$self->{coro} = async {
391			for(;;) {
392			my ($state, @args) = ref $self->{p}{state} ? @{$self->{p}{state}} : $self->{p}{state};
393			my $next = eval { $self->can ("state_$state")->($self, @args) };
394			if ($@) {
395			$self->log ($@);
396			$next = $self->feedback ("continue with state: ");
397			}
398	root	1.10	$self->log ($self->{status} = "STATE CHANGE: ". join ", ", ref $next ? @$next : $next);
399	root	1.1	$self->{p}{state} = $next;
400			$self->save;
401			}
402			};
403			}
404
405			sub state_finish {
406			my ($self, $save) = @_;
407
408			if ($save) {
409	root	1.10	aio_fsync $self->{fh};
410	root	1.1	close $self->{fh};
411
412	root	1.10	aio_unlink "$DONE_HOME/$self->{p}{title}";
413			aio_link $self->{file}, "$DONE_HOME/$self->{p}{title}"
414			and die "link: $self->{file} => $DONE_HOME/$self->{p}{title}: $!";
415	root	1.1	system "sync";
416			}
417			$self->clean;
418
419	root	1.10	aio_unlink $self->{file};
420	root	1.1
421			$self->{status} = "finished";
422			$self->feedback ("finished");
423			terminate;
424			}
425
426			sub state_examine {
427			my ($self) = @_;
428			my $p = $self->{p};
429
430			$self->{status} = "initial fetch";
431
432			for (;;) {
433			$self->log ("fetching $p->{key} (=$p->{title})");
434
435			($p->{meta}, $p->{data}) = fetch_uri 100, "freenet:$p->{key}";
436			$self->save;
437			#use PApp::Util; print STDERR PApp::Util::dumpval [keys %{$meta->{document}[0]{split_file}}];
438			$self->log ("type $p->{meta}{document}[0]{info}{format}");
439
440			if (my $splitfile = $p->{meta}{document}[0]{split_file}) {
441			return "splitfile";
442			} elsif ((defined $p->{data}) and (length $p->{data})) {
443	root	1.8	syswrite $self->{fh}, delete $p->{data};
444	root	1.12	aio_fsync $self->{fh};
445	root	1.1	return ["finish", 1];
446			}
447
448			$self->log ("EMPTY, retrying in an hour");
449			Coro::Timer::sleep 3600;
450			}
451
452			}
453
454			sub state_splitfile {
455			my ($self) = @_;
456			my $p = $self->{p};
457
458			my $splitfile = $p->{meta}{document}[0]{split_file};
459			my $filesize = hex $splitfile->{size};
460
461			if ($splitfile->{algo_name} eq "OnionFEC_a_1_2") {
462			my $data_packets = hex $splitfile->{block_count};
463			my $check_packets = hex $splitfile->{check_block_count};
464
465			my $blk = ($p->{blk} \|\|= []);
466
467			unless (@$blk) {
468			for (1..$data_packets) {
469			push @$blk, {
470			uri => $splitfile->{block}{sprintf "%x", $_},
471			};
472			}
473			for (1..$check_packets) {
474			push @$blk, {
475			uri => $splitfile->{check_block}{sprintf "%x", $_},
476			};
477			}
478			}
479
480			my @segments;
481			my $segments = 0;
482
483			{
484			# that is a horrible algorithm :(, these freenet freaks are... java-disabled
485			# hardcoding lots of magic parameters is soo dumb.
486			my $size = $filesize;
487			my $offset = 0;
488			my $offset2 = ($filesize & ~(10241024-1)) + 10241024; # leave enough space after last data block
489			my $idx = 0;
490			my $idx2 = $data_packets;
491			my @redundandy = (0,1,2); # maybe OnionFAC_a_1_2 means 1/2 redundancy(?)
492
493			while ($size > 0) {
494			my $segsize = $size >= MAXSEG ? MAXSEG : $size <= MINSEG ? MINSEG : $size;
495			my $blksize = blocksize $segsize;
496			my $seg =
497			{
498			id => $segments++,
499			todo => int (($segsize + $blksize - 1) / $blksize),
500			done => 0,
501			blk => [],
502			blksize => $blksize,
503			};
504
505			push @segments, $seg;
506			$size -= $segsize;
507
508			while ($segsize > 0) {
509			push @{$seg->{blk}}, $idx;
510			for ($blk->[$idx++]) {
511			$_->{offset} = $offset;
512			#$_->{size} = $blksize > $segsize ? $segsize : $blksize; # WRONG
513			$_->{size} = $blksize;
514			$_->{seg} = $seg;
515			}
516
517			$segsize -= $blksize;
518			$offset += $blksize;
519
520			if (($redundandy[0] += $redundandy[1]) >= $redundandy[2]) {
521			$redundandy[0] -= $redundandy[2];
522
523			push @{$seg->{blk}}, $idx2;
524			for ($blk->[$idx2++]) {
525			$_->{offset} = $offset2;
526			$_->{size} = $blksize;
527			$_->{seg} = $seg;
528			}
529
530			$offset2 += $blksize;
531			}
532			}
533			}
534
535			$idx == $data_packets
536			or die "$self->{id}/$p->{tile} $self->{job}\nidx $idx != data_packets $data_packets";
537			$idx2 == $data_packets + $check_packets
538			or die "$self->{id}/$p->{tile} $self->{job}\nidx2 $idx2 != data_packets $data_packets + check_packets $check_packets";
539			}
540
541			for (@$blk) {
542			++$_->{seg}{done} if $_->{done};
543			delete $_->{htl};
544			}
545
546			my $fail = 0;
547			my $sig = new Coro::Signal;
548
549			$self->{status} = "splitfile fetch (" . @$blk . " blocks)";
550
551			my @txn;
552
553			for (;;) {
554			for my $id (0 .. $#$blk) {
555			my $blk = $blk->[$id];
556
557			next if $txn[$id] \|\| $blk->{done} \|\| $blk->{seg}{todo} <= $blk->{seg}{done};
558
559			my $htl = $HTL[$blk->{htl}++ % @HTL];
560	root	1.5	my $pri = int time + $htl / 20 * 7200 * rand;
561	root	1.1
562	root	1.5	#warn $self->{id} . ", GET<$htl, $pri>\n";#d#
563	root	1.15	$txn[$id] \|\|= txn_client_get pri => $pri, uri => $blk->{uri}, htl => $htl, cb => sub {
564	root	1.1	undef $txn[$id];
565
566			my $seg = $blk->{seg};
567
568			my ($meta, $data) = eval { @{ $_[0]->result } };
569
570			if (defined $data) {
571			$blk->{size} == length $data
572			or die sprintf "block $id expected size %d, got %d\n", $blk->{size}, length $data;
573
574	root	1.9	(length $data) == (aio_write $self->{fh}, $blk->{offset}, (length $data), $data, 0)
575	root	1.1	or die "unable to write chunk to disk, not setting valid flag";
576	root	1.9	aio_fsync $self->{fh};
577	root	1.1
578			$blk->{done} = 1;
579			$blk->{meta} = $meta->{raw} if length $meta->{raw};#d#
580			$seg->{done}++;
581			$self->save;
582
583			$::htl_sum += $htl;
584			$::htl_cnt++;
585
586	root	1.4	$self->log (sprintf "got block $seg->{id}.$id %d ($seg->{done}/$seg->{todo}) at htl $htl (%f) and pri $pri",
587			length $data, $::htl_sum / $::htl_cnt);
588	root	1.1	} else {
589			if ($@) {
590			if ($@->type ("data_not_found")) {
591			# nop
592			} elsif ($@->type ("network_error")) {
593			$self->log ("$@, retrying in 1s");
594			CORE::sleep 1;
595			} else {
596			$self->log ("$@");
597			}
598			}
599			++$fail;
600			}
601			$self->{status} = "splitfile fetch ($seg->{done}/$seg->{todo}, $fail failed)";
602
603			$sig->send;
604	root	1.15	};
605	root	1.1	}
606
607			for my $seg (@segments) {
608			if ($seg->{done} >= $seg->{todo} && !$seg->{finished}) {
609
610			$self->log ("segment done, cancelling segment $seg->{id}");
611			for my $id (@{$seg->{blk}}) {
612			(delete $txn[$id])->cancel if $txn[$id];
613			}
614
615	root	1.10	my $verify;
616	root	1.1	for my $id (@{$seg->{blk}}) {
617			my $blk = $blk->[$id];
618			if ($blk->{done}) {
619	root	1.10	aio_read $self->{fh}, $blk->{offset}, $blk->{size}, my $buf, 0;
620	root	1.1
621			my $k1 = Net::FCP::Util::extract_chk_hash $blk->{uri};
622			my $k2 = Net::FCP::Util::generate_chk_hash $blk->{meta}, $buf;
623
624			if ($k1 ne $k2) {
625	root	1.10	$verify .= "v";
626	root	1.1	#warn sprintf "$p->{title} block $id BROKEN (%s != %s)", (unpack "H", $k1), (unpack "H", $k2);
627			$blk->{done} = 0;
628			$seg->{done}--;
629			$self->save;
630			} else {
631	root	1.10	$verify .= "V";
632	root	1.1	}
633			}
634			}
635	root	1.10	$self->log ("verified segment $seg->{id}");
636			$self->log ($verify);
637	root	1.1
638			if ($seg->{done} >= $seg->{todo} && !$seg->{finished}) {
639			$self->log ("verified segment OK $seg->{id}");
640			$seg->{finished}++;
641			$segments--;
642			} else {
643			$self->log ("verified segment NOT OK $seg->{id}");
644			}
645			}
646			}
647
648			last unless $segments;
649
650			$sig->wait;
651			}
652
653			$self->log ("decoding < $self->{job} $self->{file} $filesize >");
654
655			for my $seg (@segments) {
656			my @part;
657			my @idx;
658			my @blk = map $blk->[$_], sort { $a <=> $b } @{$seg->{blk}};
659
660			for my $id (0 .. $#blk) {
661			my $blk = $blk[$id];
662			next unless $blk->{done};
663
664			push @part, [$self->{fh}, $blk->{offset}];
665			push @idx, $id;
666
667			last if @idx == $seg->{todo};
668			}
669
670			my $fec = new Algorithm::FEC
671			$seg->{todo},
672			scalar @blk,
673			$seg->{blksize};
674
675			$fec->shuffle (\@part, \@idx);
676
677			# now copy check blocks to their destination position
678			for my $i (0 .. $#idx) {
679			next if $idx[$i] == $i;
680
681			my $src = $part[$i];
682			$part[$i] = [$self->{fh}, $blk[$i]{offset}];
683			$fec->copy ($src, $part[$i]);
684			}
685
686			$fec->set_decode_blocks (\@part, \@idx);
687			$fec->decode;
688			}
689
690			my $sha1 = new Digest::SHA1;
691			open my $dd, "-\|", "head -c$filesize \Q$self->{file}\E"
692			or die "DD: $!";
693			#$dd = Coro::Handle::unblock $dd;
694			$sha1->addfile ($dd);
695			$sha1 = $sha1->hexdigest;
696
697			if (exists $p->{meta}{document}[0]{info}{checksum}
698			and $p->{meta}{document}[0]{info}{checksum} ne $sha1) {
699			$self->log ("META: $p->{meta}{document}[0]{info}{checksum} and real checksum $sha1 for $filesize DIFFER");
700	root	1.9	# $self->feedback ("CHECKSUM ERROR");
701			# terminate;
702	root	1.1	}
703
704			truncate $self->{fh}, $filesize;
705			sysseek $self->{fh}, 0, 0;
706
707			return ["finish", 1];
708			} else {
709			$self->log ("splitfile algo '$splitfile->{algo_name}' unknown");
710			$self->feedback ("algo unknown");
711			terminate;
712			}
713			}
714
715			package main;
716
717			$\|=1;
718
719	root	1.13	async {
720			for (<\Q$QUEUE_HOME\E/*.j>) {
721			job->new_from_file ($_);
722			print "J";
723			}
724			print "\n";
725			};
726	root	1.1
727			open my $stdin , "<&0" or die;
728			open my $stdout, ">&1" or die;
729			cmdline unblock $stdin, unblock $stdout;
730
731			&Coro::Event::loop;
732