#! /usr/bin/perl
use strict;
use warnings;
use File::Basename qw(&dirname);
use File::Find;
use Data::Dumper;
use Statistics::Basic::StdDev;
use List::Util;

$|=1;
sub readfile {
  my($fname)=@_;
  local *F;
  open F,"$fname" or die $fname;
  local $/=undef();
  defined(my $r=<F>) or die $fname;
  close F or die $fname;
  return $r;
}

my %R;
if (($ARGV[0]||"") eq "-r") {
  shift;
  my $r=readfile shift;
  $R{$1}=1 while $r=~s/^(.*?)\n//;
  die $r if $r;
}

my %D;
# ==> build/Cadence-1.0.0-0.12.20200504git5787908.fc33.src.rpm.dtneeded <==
# /usr/lib/debug/.dwz/Cadence isdwzcommon 479079 NA
# /usr/bin/cadence-jackmeter /usr/lib/debug/.dwz/Cadence 717792 1329040
# /usr/bin/cadence-xycontroller /usr/lib/debug/.dwz/Cadence 1763616 2369832
# ==> build/CVector-1.0.3.1-21.fc33.src.rpm.dtneeded <==
# /usr/lib64/libCVector-1.0.3.so.2.0.0 nodwzcommon 28088 28896
for my $dtneededfn (glob "build/*.dtneeded") {
  next if %R && !$R{($dtneededfn=~m{^build/(.*?)-[^-]*-[^-]*$})[0]."-debuginfo"};
  local *F;
  open F,$dtneededfn or die "$dtneededfn: $1";
  local $_;
  while (<F>) {
    chomp;
    my($fn,$dwzcommon,$dwzsize,$dtsize)=/^(\S+) (\S+) (\S+) (\S+)$/ or die "$dtneededfn: $_";
    die if $dwzcommon eq "isdwzcommon" && $dtsize ne "NA";
    my $ref=\$D{$fn};
    if ($$ref) {
      die if $dwzcommon eq "isdwzcommon";
      $$ref=[];
    } else {
      $$ref=[$dwzcommon,$dwzsize,$dtsize];
    }
  }
  close F or die "close $dtneededfn: $1";
}

chdir "dtneeded.out" or die "dtneeded.out: $!";

my %F;
my %SONAME;
my @DEBUG;
my %SYMLINK;
find {
  "no_chdir"=>1,
  "wanted"=>sub {
    die $File::Find::dir if $File::Find::dir=~m{/$};
    my $binfn=$File::Find::name;
    if (-l $binfn) {
      my $target=readlink $binfn or die $binfn;
      $binfn=~s{^[.]}{};
      my $final;
      if ($target=~m{^/}) {
	$final=$target;
#warn "$binfn,target=rel=$$ref\n";
      } else {
	my $base=dirname(".$binfn");
	my $abs=File::Spec->rel2abs($target,$base);
	$final="/".File::Spec->abs2rel($abs);
#warn "$binfn,base=$base,target=$target,abs=$abs,rel=$$ref\n";
      }
      1 while $final=~s{/[^/]+/[.][.]/}{/};
      1 while $final=~s{/[^/]+/[.][.]$}{};
      if ($final=~m{/[.][.]}) {
	warn "$binfn,target=$target,final=$final\n";
	return;
      }
      my $ref=\$SYMLINK{$binfn};
      die if $$ref;
      $$ref=$final;
      return;
    }
    return if !-f $binfn;
    die $binfn if $binfn!~m{^[.]/};
    my $bin=readfile $binfn;
    $binfn=~s{^[.]}{} or die;
    my $rpath=($bin=~m{^.*\Q(R\E(?:UN)?\QPATH)\E\s*Library r(?:un)?path: \Q[\E(.*)\Q]\E$}m)[0];
    if ($rpath) {
      my $dirname=$File::Find::dir;
      die if $dirname!~s{^[.]/}{/};
      $rpath=~s{\$ORIGIN}{$dirname}g;
    }
    my $soname=($bin=~m{^.*\Q(SONAME)\E\s*Library soname: \Q[\E(.*)\Q]\E$}m)[0];
    my $h={
      "binfn"=>$binfn,
      "needed"=>[$bin=~m{^\s*0x0000000000000001\s*\Q(NEEDED)\E\s*\QShared library: [\E(.*)\Q]\E$}gm],
    };
    $h->{"rpath"}=$rpath if $rpath;
    $F{$binfn}=$h;
#warn "$binfn=".Dumper($h)."\n" if $binfn=~/libc.so.6/;
#    my $total=keys %F; warn "$total...\n" if 0==$total%1000;
    if ($soname) {
      my $sonamefn=$File::Find::dir."/".$soname;
      $sonamefn=~s{^[.]/}{/} or die $sonamefn;
      my $ref=\$SONAME{$sonamefn};
#      warn "soname=<$sonamefn> <$binfn> vs. <".$$ref->{"binfn"}.">\n" if $$ref;
      $$ref=$h;
    }
    push @DEBUG,$binfn if $bin=~m{ \Q(DEBUG) \E };
  },
},".";

#while (my($src,$target)=each %SYMLINK) {
#  die "$src->$target" if exists $SYMLINK{$target};
#}

my $dwzsizeall=0;
my $dwzsizeduplall=0;
my $dtsizeall=0;
my @ratioall;
my @ratioduplall;
my $computed=0;
BINFN: for my $binfn (@DEBUG) {
#  warn "$binfn...\n".Dumper([sort @{$F{$binfn}{"needed"}}]);
  die $binfn if exists $SYMLINK{$binfn};
  my @l=$binfn;
  my %l=($binfn=>1);
  while (@l) {
    my $l=shift @l;
    my $h=$F{$l};
    for my $needed (@{$h->{"needed"}}) {
      my $found;
      if ($needed=~m{^/}) {
	$needed=$SYMLINK{$needed} while exists $SYMLINK{$needed};
	$found=$needed;
      } else {
#	die "$binfn: $l: $needed" if $needed=~m{/};
	my @rpath;
	my $rpath=$h->{"rpath"};
	push @rpath,split /:/,$rpath if $rpath;
	push @rpath,qw(/lib64 /usr/lib64);
	for my $rpath (@rpath) {
	  if ($rpath!~m{^/}) {
	    warn "$binfn: $l: $rpath";
	    next;
	  }
	  my $fn="$rpath/$needed";
	  $fn=$SYMLINK{$fn} while exists $SYMLINK{$fn};
	  next if !$SONAME{$fn};
	  $found=$fn;
	  last;
	}
	warn "$binfn: $l: $needed not found; rpath=".join(":",@rpath)."\n" if !$found;
      }
      push @l,$found if $found&&!$l{$found}++;
    }
  }
#warn Dumper $binfn,\%l;
  my $dwzsizetot=0;
  my $dwzsizedupltot=0;
  my $dtsizetot=0;
  my %dwzcommons;
  for my $l (keys(%l)) {
    my $ref=$D{$l};
    if (!defined $ref) {
      warn "$binfn: $l: missing\n";
      next BINFN;
    }
    if (0==@$ref) {
      warn "$binfn: $l: ambiguous\n";
      next BINFN;
    }
    my $dtsize=$ref->[2];
    die if !defined $dtsize;
    $dtsizetot+=$dtsize;
    my $dwzsize=$ref->[1];
    die if !defined $dwzsize;
    $dwzsizetot+=$dwzsize;
    $dwzsizedupltot+=$dwzsize;
    $computed++;
    my $dwzcommon=$ref->[0];
    next if $dwzcommon eq "nodwzcommon";
    die if $dwzcommon eq "isdwzcommon";
    my $duplicate=$dwzcommons{$dwzcommon}++;
    my $dwzcommonref=$D{$dwzcommon};
    die if !$dwzcommonref;
    die if $dwzcommonref->[0] ne "isdwzcommon";
    die if $dwzcommonref->[2] ne "NA";
    my $dwzcommonsize=$dwzcommonref->[1];
    $dwzsizetot+=$dwzcommonsize if !$duplicate;
    $dwzsizedupltot+=$dwzcommonsize;
  }
  print "$binfn: dwzsizetot=$dwzsizetot dtsizetot=$dtsizetot\n";
warn "$binfn: ".Dumper(\%dwzcommons);
  $dwzsizeall+=$dwzsizetot;
  $dwzsizeduplall+=$dwzsizedupltot;
  $dtsizeall+=$dtsizetot;
  push @ratioall    ,$dwzsizetot    /$dtsizetot;
  push @ratioduplall,$dwzsizedupltot/$dtsizetot;
#  warn "$binfn done\n".Dumper([sort keys(%l)]);
}
print "dwzsizeall    =$dwzsizeall"    ." dtsizeall=$dtsizeall =".$dwzsizeall    /$dtsizeall." avg=".List::Util::sum(@ratioall    )/@ratioall    ." stddev=".Statistics::Basic::stddev(\@ratioall    )."\n";
print "dwzsizeduplall=$dwzsizeduplall"." dtsizeall=$dtsizeall =".$dwzsizeduplall/$dtsizeall." avg=".List::Util::sum(@ratioduplall)/@ratioduplall." stddev=".Statistics::Basic::stddev(\@ratioduplall)."\n";
print "computed=$computed of DEBUG=".(0+@DEBUG)." =".$computed/@DEBUG."\n";
#print Dumper \@ratioall;