path: root/Master/tlpkg/bin/tl-update-tlpdb

#!/usr/bin/env perl
# $Id$
# convert a set of tlpsrc files (given as arguments) into tlpdb
# 
# Copyright 2007-2018 Norbert Preining
# This file is licensed under the GNU General Public License version 2
# or any later version.

BEGIN {
  $^W = 1;
  chomp ($mydir = `dirname $0`);
  unshift (@INC, "$mydir/..");
}

use strict;

use TeXLive::TLConfig;
use TeXLive::TLPSRC;
use TeXLive::TLPOBJ;
use TeXLive::TLPDB;
use TeXLive::TLTREE;
use TeXLive::TLUtils;
use TeXLive::TeXCatalogue;
use Getopt::Long;
use Pod::Usage;
use File::Path;

my $opt_catalogue = "";
chomp (my $opt_master = `cd $::mydir/../.. && pwd`);
my $help = 0;
my $opt_dry_run = 0;
my $opt_fix_reverse_revisions = 0;  # needs svn commit
my $opt_fromfiles = 0;
my $opt_fromgit = 0;
my $opt_fromgitsvn = 0;
my $opt_no_commit = 0; # do/don't commit the changes
my $opt_no_revision_check = 0;
my $opt_nobinsplit = 0;
my $opt_output;
my $opt_save_anyway = 0;
my $opt_tlpsrc_from_cmdline = 0;
my $opt_w32warning;
my $opt_keep_revisions = 0;

TeXLive::TLUtils::process_logging_options();
GetOptions(
    "catalogue=s"                => \$opt_catalogue,
    "dry-run|n",                 => \$opt_dry_run,
    "fix-reverse-revisions!"     => \$opt_fix_reverse_revisions,
    "keep-revisions"             => \$opt_keep_revisions,
    "from-files"                 => \$opt_fromfiles,
    "from-git"                   => \$opt_fromgit,
    "from-gitsvn"                => \$opt_fromgitsvn,
    "master=s"                   => \$opt_master,
    "no-bin-split!"              => \$opt_nobinsplit,
    "no-commit!"                 => \$opt_no_commit,
    "no-reverse-revision-check!" => \$opt_no_revision_check,
    "output=s"                   => \$opt_output,
    "save-anyway!"               => \$opt_save_anyway,
    "tlpsrc-from-cmdline"        => \$opt_tlpsrc_from_cmdline,
    "with-w32-pattern-warning"   => \$opt_w32warning,
    "help|?" => \$help) or pod2usage(1);
pod2usage(-exitstatus => 0, -verbose => 2) if $help;

my $progname = TeXLive::TLUtils::basename($0);

exit (&main ());


sub main {
  if (! -d $opt_master) {
    die "$progname: Master $opt_master not a directory, goodbye.\n";
  }

  if ($opt_fromfiles + $opt_fromgit + $opt_fromgitsvn > 1) {
    die "$progname: only one option --from[lines|git|gitsvn] can be given.\n";
  }

  $opt_catalogue = "/home/httpd/html/catalogue"
    if ! $opt_catalogue;

  if ($opt_dry_run) {
    $opt_no_commit = 1;
    $opt_no_revision_check = 1;
    $opt_output = "/tmp/tlut" if ! $opt_output;
  }
  
  our $tlc = undef;
  if (-r $opt_catalogue) {
    info ("$progname: reading TeX Catalogue $opt_catalogue ...\n");
    $tlc = TeXLive::TeXCatalogue->new ("location" => $opt_catalogue);
  } else {
    tlwarn("$progname: reusing Catalogue data, since source is not readable: "
           . "$opt_catalogue\n");
  }

  # Create our working tlpdb with Catalogue data.
  my ($newtlpdb,$pkgcount) = create_tlpdb($tlc);

  my $oldtlpdb = undef;
  my $OLDTLPDB = "$opt_master/$TeXLive::TLConfig::InfraLocation/texlive.tlpdb";
  if (-r $OLDTLPDB) {
    $oldtlpdb = TeXLive::TLPDB->new ("root" => $opt_master);
  }
  if (!defined($oldtlpdb)) {
    tlwarn ("$progname: No revision check, no old tlpdb: $OLDTLPDB\n");
    finish ($newtlpdb, $pkgcount);
  }

  if ($opt_no_revision_check) {
    if ($opt_keep_revisions) {
      move_revisions_over($oldtlpdb, $newtlpdb);
      info ("$progname: Keeping old revisions due to --keep-revisions\n");
    } else {
      info ("$progname: Not checking revisions due to --no-reverse-revision-check\n");
    }
    finish ($newtlpdb, $pkgcount);
  }
  # compare_tlpdbs returns true if some reversed revisions have been found.
  # If in addition the option --fix-reverse-revisions is given it tries
  # to fix that, by committing changes.
  my $buggy_tlpdb = 0;
  if (compare_and_fix_tlpdbs($oldtlpdb, $newtlpdb)) {
    $buggy_tlpdb = 1;
    # there have been reversed revision that have already been warned about
    # and that have been (hopefully) fixed by compared_tlpdbs if
    # --fix-reverse-revisions has been given
    #
    # if we are running from svn (no --from-files is given) and we try
    # to fix revisions (--fix-reverse-revisions is given), and we do
    # commit the changes (--no_commit is not given), then we recreate the tlpdb
    #
    if (!$opt_fromfiles && $opt_fix_reverse_revisions && !$opt_no_commit) {
      info ("$progname: Regenerating tlpdb after fixing\n");
      ($newtlpdb, $pkgcount) = create_tlpdb($tlc);
      #
      # now recheck, but do NOT try to check in changes again by unsetting
      # $opt_fix_reverse_revisions
      $opt_fix_reverse_revisions = 0;
      if (compare_and_fix_tlpdbs($oldtlpdb, $newtlpdb)) {
        tlwarn ("$progname: *** oops, should not happen!\n"
                . "  After one iteration of compare_tlpdbs\n"
                . "  the new tlpdb is still going backwards in time;\n"
                . "  check output above and fix!\n");
        # $buggy_tlpdb is already set to 1
      } else {
        # fixing succeeded
        $buggy_tlpdb = 0;
      }
    }
  }
  # new $newtlpdb points to the right one
  if ($buggy_tlpdb) {
    tlwarn("$progname: the newly created tlpdb seems to be buggy.\n");
    if ($opt_output) {
      tlwarn("$progname: saving it anyway due to --output given\n");
      $buggy_tlpdb = 0;
    } elsif ($opt_save_anyway) {
      tlwarn("$progname: saving it anyway due to --save-anyway\n");
      $buggy_tlpdb = 0;
    } else {
      tlwarn("$progname: not saving it, you might use --save-anyway\n");
    }
  }
  if (!$buggy_tlpdb) {
    finish ($newtlpdb, $pkgcount);
  } else {
    # return error to the shell script, it is evaluated in cron.tl!
    return 1;
  }
}


sub move_revisions_over {
  my ($oldtlpdb, $newtlpdb) = @_;
  for my $pkg ($newtlpdb->list_packages) {
    my $oldtlp = $oldtlpdb->get_package($pkg);
    if (defined($oldtlp)) {
      $newtlpdb->get_package($pkg)->revision($oldtlp->revision);
    }
  }
}


sub compare_and_fix_tlpdbs {
  my ($oldtlpdb, $newtlpdb) = @_;

  my %ret = TeXLive::TLUtils::compare_tlpdbs($oldtlpdb, $newtlpdb);

  # we only care for differences in packages. If a packages has been
  # removed or added that is fine for here as this will be dealt with
  # in tl-update-containers

  # within the list of packages that differ the problematic ones are
  # where there are either files added or removed, but the revision
  # did not increase

  my %packages_that_need_fixup;

  if (defined($ret{'different_packages'})) {
    for my $p (keys %{$ret{'different_packages'}}) {
      if (defined($ret{'different_packages'}->{$p}->{'added'}) ||
          defined($ret{'different_packages'}->{$p}->{'removed'})) {
        if (!defined($ret{'different_packages'}->{$p}->{'revision'})) {
          # no revision change reported, but files have been added
          # or removed, try fixing it
          $packages_that_need_fixup{$p} = 1;
        } else {
          # revision change has been reported, if the revision is increasing
          # that is fine, otherwise try fixing it
          my ($rA, $rB) = split (':', 
                            $ret{'different_packages'}->{$p}->{'revision'});
          if ($rA >= $rB) {
            $packages_that_need_fixup{$p} = 1;
          }
        }
      } else {
        # there have been no changed filelists, but we still check that
        # revision is not going backward out of some strange other reason
        if (defined($ret{'different_packages'}->{$p}->{'revision'})) {
          # revision change has been reported, if the revision is increasing
          # that is fine, otherwise try fixing it
          my ($rA, $rB) = split (':', 
                            $ret{'different_packages'}->{$p}->{'revision'});
          if ($rA > $rB) {
            tlwarn("$progname: big warning for $p: file lists didn't change but "
                   . "revision is going backward! Very strange!\n");
            $packages_that_need_fixup{$p} = 1;
          }
        }
      }
    }
  }
   
  my $commitit = 0;

  my @svn_changed_files = ();

  for my $p (keys %packages_that_need_fixup) {
    tlwarn("$progname: package $p has different files but same/decreasing revision number\n");
    if ($opt_fix_reverse_revisions) {
      my $fixfile = "$opt_master/tlpkg/tlpsrc/$p.tlpsrc";
      # make sure that we actually fix a file of the right package
      if (! -r $fixfile) {
        # we re quite probably trying to fix a .ARCH package with decreasing
        # revision number, but since we do NOT include the .tlpsrc file
        # in the computation of the .ARCH packages, we have to find a file
        # in the packages
        my $tlp = $newtlpdb->get_package($p);
        if (!defined($tlp)) {
          tlwarn("Very strange, $p is mentioned as changed files,\n");
          tlwarn("but it is not found in new tlppdb.\n");
          tlwarn("Not trying to fix anything, don't know how!\n");
          next;
        }
        my @allf = $tlp->all_files;
        if ($#allf < 0) {
          tlwarn("No files found in $p, and no tlpsrc file present.\n");
          tlwarn("Don't know how to fix decreasing version in this case.\n");
          next;
        } 
        # still here, so use the first file in the list.
        $fixfile = "$opt_master/";
        if ($tlp->relocated) {
          $fixfile .= "$RelocTree/";
        }
        $fixfile .= $allf[0];
      }
      tlwarn("$progname: trying to fix it by running:\n");
      my $time = time();
      my $cmd = "svn propset texlive:force_incr_after_delete $time $fixfile";
      $commitit = 1;
      tlwarn("  $cmd\n");
      TeXLive::TLUtils::xsystem($cmd);
      push @svn_changed_files, $fixfile;
    }
  }
  if ($opt_no_commit) {
    if ($commitit) {
      tlwarn("$progname: tlpsrc files have been changed but not committed.\n");
      tlwarn("$progname: svn status output follows:\n");
      my $r = `svn status @svn_changed_files 2>&1`;
      tlwarn($r . "\n");
    }
    $commitit = 0;
  }
  if ($commitit) {
    tlwarn("$progname: committing changes to tlpsrc files:\n");
    TeXLive::TLUtils::xsystem("svn commit "
      . "-m'($progname) force increased revision after removal of files' "
      . "@svn_changed_files");
  }
  if (keys %packages_that_need_fixup) {
    return 1;
  } else {
    return 0;
  }
}


sub create_tlpdb {
  my ($tlc) = @_;
  
  if (defined($opt_w32warning)) {
    $::tlpsrc_pattern_warn_win = 1;
  }
  my $tlpsrc_dir = "$opt_master/tlpkg/tlpsrc";

  if (@ARGV) {
    # if we have cmd line arguments act as if --tlpsrc-from-cmdline was given.
    if (!$opt_tlpsrc_from_cmdline) {
      debug ("$progname: generating tlpdb in /tmp/tlut for @ARGV.\n");
      $opt_tlpsrc_from_cmdline = 1;
    }
  }
  if (!$opt_tlpsrc_from_cmdline) {
    # if not otherwise given, use all tlpsrc files.
    chdir ($tlpsrc_dir) || die "$progname: chdir($tlpsrc_dir) failed: $!";
    @ARGV = glob ("*.tlpsrc");
  }
  
  info ("$progname: reading Master $opt_master ...\n");
  my $tltree = TeXLive::TLTREE->new ("svnroot" => $opt_master);
  if ($opt_fromfiles) {
    $tltree->init_from_files;
  } elsif ($opt_fromgit) {
    $tltree->init_from_git;
  } elsif ($opt_fromgitsvn) {
    $tltree->init_from_gitsvn;
  } else {
    $tltree->init_from_svn;
  }

  info ("$progname: updating from tlpsrc ...\n");
  my $tldb = TeXLive::TLPDB->new;
  my $src_count = 0;
  my $found_texlive_installation = 0;
  foreach my $f (@ARGV) {
    next if ($f =~ m/^00texlive\.autopatterns/);
    $src_count++;

    my $tlsrc = new TeXLive::TLPSRC;
    $tlsrc->from_file ($f);
    my $tlp = $tlsrc->make_tlpobj ($tltree, $opt_master);

    (my $base_f = $f) =~ s/\.tlpsrc$//;
    $base_f = TeXLive::TLUtils::basename ($base_f);
    warn "$f: package name " . $tlp->name . " does not match filename\n"
      if $tlp->name ne $base_f;
      
    if ($tlc) {
      # merge TeX Catalogue information
      $tlp->update_from_catalogue ($tlc);
    }
   
    # split bin packages off unless requested otherwise (no reason to so
    # request these days).
    unless ($opt_nobinsplit) {
      # - split: texlive.infra but NO OTHER PACKAGE containing \.
      # - NO split: 00texlive
      # packages which have already a .SOMETHING extension or the 
      # 00texlive.something package are NOT split, never!
      if (($tlp->name !~ m/\./ || $tlp->name =~ m/texlive\.infra/)
          && $tlp->name !~ m/^00texlive/) {
        my @binobjs = $tlp->split_bin_package;
        foreach my $binobj (@binobjs) {
          # we recompute the revision numbers of the binary packages
          # since that might be much smaller, and we don't want to 
          # update every package only because one binary for one arch
          # was changed
          #
          # we do NOT use the tlpsrc file for computing the file revision
          # as this is actually wrong ..
          $binobj->recompute_revision ($tltree);
          if ($binobj->revision == 0) {
            debug("Packages with revision number 0 are not added: " .
                  $binobj->name . "\n");
          } else {
            $tldb->add_tlpobj ($binobj);
          }
        }
        # we have to recompute the revision of the package itself 
        # since the binfiles have been removed and that might lower the
        # revision number
        $tlp->recompute_revision ($tltree, $tltree->file_svn_lastrevision("tlpkg/tlpsrc/$f"));
      }
    }
    if ($tlp->name eq "00texlive.installation") {
      $found_texlive_installation = 1;
    }
    if ($tlp->name eq "00texlive.config") {
      # defaults for these options are from %TeXLive::TLConfig::TLPDBConfigs
      # but can be overriden by putting them into this source package
      my %cfgs;
      # first read the default settings
      for my $k (keys %TeXLive::TLConfig::TLPDBConfigs) {
        $cfgs{$k} = $TeXLive::TLConfig::TLPDBConfigs{$k};
      }
      # next parse the ones set in the tlpsrc
      for my $dep ($tlp->depends) {
        my ($k,$v) = split('/',$dep,2);
        if (defined($k)) {
          if (defined($cfgs{$k})) {
            # override setting
            $cfgs{$k} = $v;
          } else {
            die("Unknown key in 00texlive.config.tlpsrc: $dep");
          }
        }
      }
      my @bar;
      for my $k (sort keys %cfgs) {
        push @bar, "$k/" . $cfgs{$k};
      }
      # add the latest revision
      push @bar, "revision/".$tltree->revision;
      $tlp->depends (@bar);
    }
    $tldb->add_tlpobj ($tlp);
  }
  $tldb->root ($opt_master);

  #
  # we set several options in 00texlive.installation only in two cases:
  # 1) we are updating from *all* tlpsrc file, so no cmd line options
  # 2) we are updating from cmd line tlpsrc file, but 00texlive.installation
  #    is contained in the list

  if (!$opt_tlpsrc_from_cmdline || $found_texlive_installation) {
    # do set the option_archs from the dirs in Master/bin
    my @archs = $tltree->architectures;
    $tldb->setting("available_architectures", sort @archs);

    # reset options to what we have by default
    $tldb->add_default_options();
  }

  return($tldb, $src_count);
}



sub finish {
  my ($tlpdb,$pkgcount) = @_;
  if ($opt_output) {
    if (open(OUT, ">$opt_output")) {
      $tlpdb->writeout(\*OUT);
      info ("$progname: wrote $pkgcount packages to $opt_output.\n");
    } else {
      tldie ("$progname: open(>$opt_output) failed: $!");
    }
  } else {
    $tlpdb->save;
    info ("$progname: wrote $pkgcount packages to " . $tlpdb->root . ".\n");
  }
  exit (0);
}

__END__

=head1 NAME

tl-update-tlpdb - generate/update a TeX Live database from tlpsrc files

=head1 SYNOPSIS

tl-update-tlpdb [OPTION]... [TLPSRC]...

=head1 OPTIONS

=over 4

=item B<--catalogue> I<Catalogue_dir>

The I<Catalogue_file> must point to a valid TeX Catalogue checkout; the
default is C</home/httpd/html/catalogue>.  If the directory is readable,
the generated tlpobj files will contain information gathered from the
TeX Catalogue.  To disable this, specify a nonexistent filename, e.g.,
C<--catalogue=/nonesuch>.

=item B<--fix-reverse-revisions>

If a package with revisions going backward is found, this option makes
C<tl-update-tlpdb> try to check in a dummy change to the package's
tlpsrc file to artificially increase its version number.

=item B<--from-files>

Use C<find> instead of C<svn -status> to create the list of files.

=item B<--master> I<Master_dir>

The location I<Master/dir> must point to a valid svn repository of TeX
Live's Master directory.  Defaults to C<../..> of C<$0>.

=item B<-n>, B<--dry-run>

Don't make permanent changes: same as C<--no-commit> and
C<--no-reverse-revision-check>, plus output to the C</tmp/tlut> (if no
C<--output> file is given) rather than updating the current tlpdb.

=item B<--no-bin-split>

Do NOT split files with binaries into several packages (one for the arch
independent and one for each of the architectures with only the respective
binaries).  There is no reason to use this option.

=item B<--no-commit>

If changes have been made to the tlpsrc files (because reversed
revisions have been found and C<--fix-reverse-revisions> is enabled)
these changes will by default be committed using C<svn commit> on the
tlpsrc directory.  With this option the commit step is skipped.

=item B<--no-reverse-revision-check>

By default, C<tl-update-tlpdb> checks that in the generated tlpdb the
revision numbers of packages are not going backward compared to the
previous version. This option disables this check.

=item B<--output=filename>

By default, C<tl-update-tlpdb> saves/updates the tlpdb in the default
place.  This option specifies the place where the output should go to.

=item B<--save-anyway>

If the newly created tlpdb is in considered buggy due to reversed
revisions it will normally not be saved. This switch allows saving it
anyway. It can be useful in combination with the C<--from-files> option
where the revision numbers will be adjusted afterwards by an external
program.

=item B<--tlpsrc-from-cmdline>

By default, C<tl-update-tlpdb> recreates the tlpdb from all the tlpsrc
files present in the current tree. If this option is given, the tlpsrc
file(s), or package names, are given on the command line. If you're
trying to debug the conversion of one particular package, this is the
option to use (along with C<-n> and C<--catalogue=/nonesuch>, most
likely). Indeed, if any non-option arguments are given, the program
assumes you meant to specify this option.

=item B<--with-w32-pattern-warning>

Normally, this program issues a warning for patterns without a matching
file, with the exception that it does I<not> warn for binpatterns
related to Windows. Giving this option makes it issue all warnings,
i.e., also those where a W32-related binpattern does not hit.

=back

The standard options C<-q>, C<-v>, and C<-logfile>=I<file> are also
accepted; see the C<process_logging_options> function in
L<TeXLive::TLUtils> for details.

=head1 DESCRIPTION

C<tl-update-tlpdb> converts TeX Live source package files (tlpsrc) files
into a TeX Live package database (tlpdb).

In addition if at the same location there is already a TLPDB, it
compares the revision number of packages and checks if a package is
going back in time (i.e., having a smaller revision number in the new
tlpdb).  This can happen when a file is removed and no other changes
made.  In this case it commits a dummy change to the package's tlpsrc
file to make sure that revisions are always increasing.

If a TeX Catalogue hierarchy is specified, additional information from
the Catalogue is merged into the database (see L<TeXLive::TLPOBJ> for
the specifics).

=head1 AUTHORS AND COPYRIGHT

This script and its documentation were written for the TeX Live
distribution (L<http://tug.org/texlive>) and both are licensed under the
GNU General Public License Version 2 or later.

=cut

### Local Variables:
### perl-indent-level: 2
### tab-width: 2
### indent-tabs-mode: nil
### End:
# vim:set tabstop=2 expandtab: #