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

#!/usr/bin/env perl
# $Id$
# Copyright 2007-2023 Norbert Preining
# This file is licensed under the GNU General Public License version 2
# or any later version.
# 
# Convert a TL tree and *.tlpsrc files into a (possibly updated) texlive.tlpdb.

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

use strict; use warnings;
use experimental qw(smartmatch); # what the heck, see ~~ below

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 = "";
my $opt_catalogue_compare = "";
my $opt_catalogue_dump = "";
my $opt_commit_max = 100;
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_keep_revisions = 0;
chomp(my $opt_master = `cd $::mydir/../.. && pwd`);
my $opt_no_binsplit = 0;
my $opt_no_commit = 0; # do/don't commit the changes
my $opt_no_revision_check = undef; # set depending on options, if not explicit.
my $opt_output;
my $opt_save_anyway = 0;
my $opt_tlpsrc_from_cmdline = 0;
my $opt_w32warning;
my $help = 0;

TeXLive::TLUtils::process_logging_options();
GetOptions(
    "catalogue=s"                => \$opt_catalogue,
    "catalogue-compare=s"        => \$opt_catalogue_compare,
    "catalogue-dump=s"           => \$opt_catalogue_dump,
    "commit-max=i"               => \$opt_commit_max,
    "dry-run|n",                 => \$opt_dry_run,
    "fix-reverse-revisions!"     => \$opt_fix_reverse_revisions,
    "from-files"                 => \$opt_fromfiles,
    "from-git"                   => \$opt_fromgit,
    "from-gitsvn"                => \$opt_fromgitsvn,
    "keep-revisions"             => \$opt_keep_revisions,
    "master=s"                   => \$opt_master,
    "no-bin-split!"              => \$opt_no_binsplit,
    "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,
    "w32-pattern-warning!"       => \$opt_w32warning,
    "help|?" => \$help) or pod2usage(1);
pod2usage(-exitstatus => 0, -verbose => 2) if $help;

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

exit(&main());


sub main {
  if (! -d $opt_master) {
    die "$prg: Master $opt_master not a directory, goodbye.\n";
  }
  if ($opt_fromfiles + $opt_fromgit + $opt_fromgitsvn > 1) {
    die "$prg: only one option of --from-[lines|git|gitsvn] can be given.\n";
  }

  if (! $opt_catalogue) {
    $opt_catalogue = "/home/httpd/html/catalogue";
  }

  if ($opt_catalogue_compare) {
    $opt_no_revision_check = 0 unless defined $opt_no_revision_check;
  }
  
  if ($opt_dry_run) {
    $opt_no_commit = 1;
    $opt_fix_reverse_revisions = 0;
    $opt_no_revision_check = 1 unless defined $opt_no_revision_check;
    $opt_output = "/tmp/tlut" if ! $opt_output;
  }
  
  our $tlc = undef;
  if (-r $opt_catalogue) {
    if (-s $opt_catalogue_dump) { # if dump file exists, use it.
      info("$prg: reading TeX Catalogue dump $opt_catalogue_dump ...\n");
      require $opt_catalogue_dump;
    } else {
      info("$prg: reading TeX Catalogue $opt_catalogue ...\n");
      $tlc = TeXLive::TeXCatalogue->new("location" => $opt_catalogue);
      if ($opt_catalogue_dump) {
        # Way to dump the catalogue and then use it instead of reparsing,
        # to save time when developing/debugging.
        # Maybe someday have the energy to merge this with the same dumping
        # code in TLUtils as a common function.
        require Data::Dumper;
        $Data::Dumper::Indent = 1;
        $Data::Dumper::Sortkeys = 1;  # stable output
        $Data::Dumper::Purity = 1; # recursive structures must be safe
        info("$prg: dumping TeX Catalogue to $opt_catalogue_dump ...\n");
        my $catdump = ">$opt_catalogue_dump";
        open (my $fh, $catdump) || die "open($catdump) failed: $!\n";
        print $fh Data::Dumper->Dump([$tlc], [qw(::tlc)]);
        close ($fh) || die "close($catdump) failed: $!\n";
      }
    }
  } else {
    tlwarn("$prg: reusing Catalogue data, since source is not readable: "
           . "$opt_catalogue\n");
    if ($opt_catalogue_compare) {
      tlwarn("$prg: unsetting --catalogue-compare since no Catalogue\n");
      $opt_catalogue_compare = "";
    }
  }

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

  my $oldtlpdb = undef;
  my $OLDTLPDB = "$opt_master/$DatabaseLocation";
  if (-s $OLDTLPDB) {
    $oldtlpdb = TeXLive::TLPDB->new("root" => $opt_master);
  }
  if (!defined($oldtlpdb)) {
    tlwarn("$prg: no revision check since no old tlpdb: $OLDTLPDB\n");
    finish($newtlpdb, $pkgcount);
  }

  if ($opt_no_revision_check) {
    if ($opt_keep_revisions) {
      copy_revisions_over($oldtlpdb, $newtlpdb);
      info("$prg: keeping old revisions due to --keep-revisions\n");
    } else {
     info("$prg: not checking revisions due to --no-reverse-revision-check\n");
    }
    finish($newtlpdb, $pkgcount);
  }

  # compare_and_fix_tlpdbs returns true if any reversed revisions were
  # found. If the option --fix-reverse-revisions (etc.) was specified,
  # it also tried to fix them, by committing changes.
  my $buggy_tlpdb = 0;
  if (compare_and_fix_tlpdbs($oldtlpdb, $newtlpdb)) {
    $buggy_tlpdb = 1;
    #
    # There have been reversed revisions that have already been warned about
    # and that have been (hopefully) fixed by compare_and_fix_tlpdbs.
    #
    # If we are running from svn (--from-files is not given),
    # and we tried to fix revisions (--fix-reverse-revisions is given),
    # and we are committing the changes (--no-commit is not given),
    # then we recreate the tlpdb, else we will report as buggy.
    #
    if (!$opt_fromfiles && $opt_fix_reverse_revisions && !$opt_no_commit) {
      info("$prg: regenerating tlpdb after fixing\n");
      ($newtlpdb,$pkgcount) = create_tlpdb($tlc);
      #
      # now recheck, but do NOT try to check in changes again, since it
      # already failed: forcibly unset $opt_fix_reverse_revisions.
      info("$prg: unsetting --fix-reverse-revisions for second check\n");
      $opt_fix_reverse_revisions = 0;
      #
      # Also do not do catalogue checks, since the Catalogue comparison
      # tree, which has old data, has not changed, so they would just
      # fail again. (We should save and restore these values, or pass
      # them as arguments, instead of just eradicating them, but in
      # practice we do not use them again.)
      info("$prg: unsetting --catalogue-compare for second check\n");
      $opt_catalogue_compare = "";
      # 
      if (compare_and_fix_tlpdbs($oldtlpdb, $newtlpdb)) {
        tlwarn("$prg: *** OOPS, should not happen!\n"
                . "  After one iteration of compare_tlpdbs\n"
                . "  the new tlpdb still has changes;\n"
                . "  check output above and fix!\n");
        # $buggy_tlpdb is already set to 1
      } else {
        # fixing succeeded
        $buggy_tlpdb = 0;
      }
    }
  }
  
  # $newtlpdb points to the right one
  if ($buggy_tlpdb) {
    tlwarn("$prg: the newly created tlpdb seems to be buggy.\n");
    if ($opt_output) {
      tlwarn("$prg: saving it anyway due to --output (to $opt_output)\n");
      $buggy_tlpdb = 0;
    } elsif ($opt_save_anyway) {
      tlwarn("$prg: saving it anyway due to --save-anyway\n");
      $buggy_tlpdb = 0;
    } else {
      tlwarn("$prg: not saving it, use --save-anyway or --output "
             . "if you want it\n");
    }
  }
  if (!$buggy_tlpdb) {
    finish($newtlpdb, $pkgcount);
  } else {
    # return error to the shell script, it is evaluated in cron.tl!
    return 1;
  }
}


# Set the revision of each package in NEWTLPDB
# to its revision in OLDTLPDB (if the pkg exists in OLDTLPDB).
# 
sub copy_revisions_over {
  my ($oldtlpdb,$newtlpdb) = @_;
  for my $pkg ($newtlpdb->list_packages) {
    my $oldtlp = $oldtlpdb->get_package($pkg);
    if (defined($oldtlp)) {
      my $oldrev = $oldtlp->revision;
      my $newrev = $newtlpdb->get_package($pkg)->revision;
      debug("pkg $pkg: changing revision from $newrev to $oldrev\n");
      $newtlpdb->get_package($pkg)->revision($oldtlp->revision);
    }
  }
}


# Compare OLDTLPDB and NEWTLPDB, returning whether any packages needed
# special fixing. Within the list of packages that differ, the
# problematic ones are those where either (a) files were added or
# removed, or (b) NEWTLPDB has Catalogue data changes compared to the
# --catalogue-compare tree, but the revision did not increase.
# 
# If a package has been removed or added (or not changed at all) that is
# fine for here, as it will be dealt with in tl-update-containers.
#
sub compare_and_fix_tlpdbs {
  my ($oldtlpdb,$newtlpdb) = @_;
  my %packages_needing_fixup;
  
  my %cmp = TeXLive::TLUtils::compare_tlpdbs($oldtlpdb, $newtlpdb);
  if (exists $cmp{'different_packages'}) {
    for my $p (sort keys %{$cmp{'different_packages'}}) {
      if (defined($cmp{'different_packages'}->{$p}->{'added'}) ||
          defined($cmp{'different_packages'}->{$p}->{'removed'}) ||
          defined($cmp{'different_packages'}->{$p}->{'fmttrigger'})) {
        if (!defined($cmp{'different_packages'}->{$p}->{'revision'})) {
          # no revision change reported, but files or fmttriggers have
          # been added or removed, try fixing it
          $packages_needing_fixup{$p} = 1;
        } else {
          # revision change has been reported, if the revision is increasing
          # that is fine, otherwise try fixing it
          my ($rA, $rB) = split(':', 
                            $cmp{'different_packages'}->{$p}->{'revision'});
          if ($rA >= $rB) {
            $packages_needing_fixup{$p} = 1;
          }
        }
      } else {
        # there have been no changed filelists, but we still check that
        # revision is not going backward for some other strange reason.
        if (defined($cmp{'different_packages'}->{$p}->{'revision'})) {
          # revision change has been reported, if the revision is increasing
          # that is fine, otherwise try fixing it
          my ($rA, $rB) = split(':', 
                            $cmp{'different_packages'}->{$p}->{'revision'});
          # since we only register 'revision' key when there is a
          # change, we have either rA < rB or rA > rB
          if ($rA > $rB) {
            tlwarn("$prg: BIG WARNING for $p: file lists didn't change "
                   . "but revision is going backward ($rA > $rB)!\n");
            $packages_needing_fixup{$p} = 1;
          } # if strange revision change
        }   # no revision changes
      }     # not (added or removed)
    }       # loop through differing packages
  } # no different_packages

  # if requested, also may need to fix up packages with catalogue changs.
  if ($opt_catalogue_compare) {
    my %cat_fixup = tlpdb_catalogue_compare($newtlpdb, $opt_catalogue_compare);
    info("$prg: ", 0 + keys %cat_fixup, " packages with Catalogue changes: ",
         join(" ", sort keys %cat_fixup), "\n") if (keys %cat_fixup);
    # hash slice to merge %cat_fixup into %packages_needing_fixup;
    # all the values are 1, so that doesn't matter.
    @packages_needing_fixup{ keys %cat_fixup } = values %cat_fixup;
  }

  my $do_commit = 0;
  my @svn_changed_files = ();

  for my $p (sort keys %packages_needing_fixup) {
    tlwarn("$prg: package $p has different files but "
           . "same/decreasing revision number",
           $opt_fix_reverse_revisions ? "\n" 
                        : " [skipping fix due to no fix_reverse_revisions]\n");
    if ($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 might be 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($tlp)) {
          tlwarn("$prg: Very strange, $p is said to have changed files,\n");
          tlwarn("but it is not found in new tlpdb;\n");
          tlwarn("skipping $p.\n");
          next;
        }
        my @allf = $tlp->all_files;
        if ($#allf < 0) {
          tlwarn("$prg: no files found in $p, and no tlpsrc file present.\n");
          tlwarn("don't know how to fix decreasing version in this case;\n");
          tlwarn("skipping $p.\n");
          next;
        } 
        # still here, so use the first file in the list.
        $fixfile = "$opt_master/";
        if ($tlp->relocated) {
          $fixfile .= "$RelocTree/";
        }
        $fixfile .= $allf[0];
      }
      tlwarn("$prg: trying to fix it by running:\n");
      my $time = time();
      my $cmd = "svn propset texlive:force_incr_after_delete $time $fixfile";
      $do_commit = 1;
      tlwarn("  $cmd\n");
      TeXLive::TLUtils::xsystem($cmd);
      push @svn_changed_files, $fixfile;
    }
  }

  if (@svn_changed_files > $opt_commit_max) {
    tlwarn(<<TOO_MUCH_CHANGED);
$prg: more than $opt_commit_max files have been changed ($#svn_changed_files+1).
This probably means a bug. Forcing --no-commit.

If it's actually correct, rerun and set --commit-max=999 (or whatever
value); probably easiest to do by editing cron.tl and rerunning.
TOO_MUCH_CHANGED
    $opt_no_commit = 1;
  }
  if ($opt_no_commit) {
    if ($do_commit) {
      tlwarn("$prg: tlpsrc files have been changed but not committed.\n");
      tlwarn("$prg: svn status output follows:\n");
      my $r = `svn status @svn_changed_files 2>&1`;
      tlwarn("$r\n");
    }
    $do_commit = 0;
  }
  if ($do_commit) {
    tlwarn("$prg: committing changes to tlpsrc files:\n");
    TeXLive::TLUtils::xsystem("svn commit "
      . "-m'($prg) force increased revision after Catalogue change"
      . " or file removal' "
      . "@svn_changed_files");
  }
  
  return (keys %packages_needing_fixup >= 1);
}


# Return a TLPDB object based on $opt_master, using passed $TLC for the
# Catalogue source and various $opt_... globals for all else.
#
sub create_tlpdb {
  my ($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 (!$opt_tlpsrc_from_cmdline) {
      debug("$prg: generating tlpdb in /tmp/tlut from tlpsrc args\n");
      ddebug("$prg: @ARGV\n");
      $opt_tlpsrc_from_cmdline = 1;
    }
  }
  if (!$opt_tlpsrc_from_cmdline) {
    # if not otherwise given, use all tlpsrc files.
    chdir($tlpsrc_dir) || die "$prg: chdir($tlpsrc_dir) failed: $!";
    @ARGV = glob("*.tlpsrc");
  }
  
  info("$prg: 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("$prg: 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 ($opt_no_binsplit) {
      # - split: texlive.infra but NO OTHER PACKAGE containing \.
      # - NO split: 00texlive
      # packages which have already a .SOMETHING extension or the 
      # 00texlive.something package are NOT split, never!
      if (($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) {
            ddebug("Package with revision number 0 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);
}


# Compare Catalogue data (and nothing else) in MASTER_TLPDB to the tlpdb
# found in CMP_TLPDB_STR. Return hash (only keys matter, values are 1)
# of packages that so differ.
# 
# We do not worry about packages found in one tlpdb and not the other;
# those will get dealt with elsewhere. Our goal here is just to compare
# the Catalogue data in packages common to the two tlpdbs.
# 
# In practice, CMP_TLPDB is tlnet (or tlpretest). This whole check is
# necessary because we intentionally do not (normally) update packages
# in tlnet when only Catalogue data has changed; that can easily lead to
# extra downloads for a single real update of the package, since the
# metadata often needs tweaking after an upload.
# 
# On the other hand, we do want to get current Catalogue data into TL
# sometimes, or it would grow stale forever; so that's what this
# comparison is for. We enable it once a month or so from cron.tl.
# 
# We do update Catalogue data in the texlive.tlpdb checked into the
# repository (under Master/) nightly, but that is not what users
# (typically) use.
# 
# This was not an issue before the switch to versioned containers in
# November 2019 and the TL'20 release; before then, we recreated all
# packages at every yearly release, so got all Catalogue updates at that
# time. But nowadays we don't ever blindly recreate all packages.
# 
sub tlpdb_catalogue_compare {
  my ($master_tlpdb,$cmp_tlpdb_str) = @_;
  my %ret;
  
  my $cmp_tlpdb = TeXLive::TLPDB->new("root" => $cmp_tlpdb_str);
  die "$prg: could not get comparison tlpdb from: $cmp_tlpdb_str\n"
    if ! defined $cmp_tlpdb;  

  for my $p ($master_tlpdb->list_packages()) {
    my $tlpA = $master_tlpdb->get_package($p);
    my $tlpB = $cmp_tlpdb->get_package($p);
    
    next unless defined $tlpA && defined $tlpB;
    
    # The Master tree isn't relocatable, so for the comparison (tlnet) tree,
    # must swap RELOC marker for actual tree so we can compare strings.
    $tlpB->replace_reloc_prefix();
    
    if (! &tlpobj_catalogue_equal($tlpA, $tlpB)) {
      debug("$prg: Catalogue changes in package: $p\n");
      $ret{$p} = 1;
    }
  }
  debug("$prg: ", 0 + keys %ret, " packages with Catalogue changes\n");

  return %ret;
}

# Return true if TLPOBJ arguments TLPA and TLPB have the same
# Catalogue-related data. Includes short and long descriptions,
# catalogue-* keys (cataloguedata), and tags of docfiles (docfiledata).
# 
sub tlpobj_catalogue_equal {
  my ($tlpA,$tlpB) = @_;
  my $ret = 0;
  
  my $pkg = $tlpA->name; # just for debugging output
  ddebug("$pkg: comparing catalogue info:\n");
  
  # The shortdesc and longdesc usually come from the Catalogue,
  # but sometimes are in the tlpsrc.
  if (! &suredef($tlpA->shortdesc) eq &suredef($tlpB->shortdesc)) {
    # Maybe someday combine this silly code to avoid redundancy,
    # and show strings (hashes) for all.
    debug("$pkg:  shortdesc now: ", &suredef($tlpA->shortdesc), "\n",
          " " x length($pkg),
            "    vs. compare db: ", &suredef($tlpB->shortdesc), "\n");
    
  } elsif (! &suredef($tlpA->longdesc) eq &suredef($tlpB->longdesc)) {
    debug("$pkg:  longdesc now: ", &suredef($tlpA->longdesc), "\n",
          " " x length($pkg),
            "   vs. compare db: ", &suredef($tlpB->longdesc), "\n");

  } elsif (! &suredef($tlpA->catalogue) eq &suredef($tlpB->catalogue)) {
    debug("$pkg: catalogue value changed (", &suredef($tlpA->catalogue),
          " -> ", &suredef($tlpB->catalogue), ")\n");

  } elsif (! &equal_hashes($tlpA->cataloguedata, $tlpB->cataloguedata)) {
    debug("$pkg: ", debug_hash_str("cataloguedata now", $tlpA->cataloguedata),
      " " x (length($pkg) + 7),
                    debug_hash_str("vs. compare db", $tlpB->cataloguedata));

  } elsif (! &equal_hashes($tlpA->docfiledata, $tlpB->docfiledata)) {
    debug("$pkg: docfiledata changed\n");
  
  } else {
    $ret = 1; # seems there were no Catalogue changes.
  }
  
  return $ret;
}


# Return ARG if it is defined, or empty string if ARG is undef.
# It is irritating to have to do this, but it seems the best available
# option to avoid uninitialized warnings. We could use the pragma, but
# cleaner to program the checks. 
# 
sub suredef {
  return defined $_[0] ? $_[0] : "";
}


# Return true if both H1 and H2 are references to hashes
# with identical contents -- same keys, same values, recursively.
# (Or are both undef.)
# 
# Only checks strings and hash references, not lists or other types.
# 
sub equal_hashes {
  my ($h1,$h2) = @_;

  # if neither is defined, that's ok.
  return 1 if !defined $h1 && !defined $h2;
  
  # both must be hash references.
  if (! (&is_hash_ref($h1) && &is_hash_ref($h2))) {
    debug(" not both hash ref: $h1 vs. $h2\n");
    return 0;
  }

  # must have the same keys (smartmatch operator).
  if (! (%$h1 ~~ %$h2)) {
    debug(" keys not equal:", join("|", keys %$h1), " vs. ",
           join("|", keys %$h2), "\n");
    return 0;
  }
  
  # if they have no keys, they are the same (we won't go through the loop).
  my $ret = (keys %$h1 == 0);
  #
  for my $k (keys %$h1) {
    #debug(" checking key $k\n");
    
    # ignore catalogue-date differences. We will eliminate the field
    # entirely at some point, but there are so many packages that differ
    # only in catalogue-date, no need to do it all at once.
    next if $k eq "date"; 
    
    my $aval = $h1->{$k};
    my $bval = $h2->{$k};
    
    if (!defined $aval && !defined $bval) {
      # ok if both are undefined.
      $ret = 1;
      next;
    } else {
      # else both have to be defined.
      #debug("  for $k, aval=$aval, bval=$bval\n");
      return 0 unless (defined $aval && defined $bval);
    }

    #debug(" comparing key $k: $aval vs. $bval\n");
    if (&is_hash_ref($aval)) {
      # if both are hashes compare, else must be not equal.
      $ret = &is_hash_ref($bval) ? &equal_hashes($aval, $bval) : 0;
    } else {
      # we don't have arrays, non-strings, etc., so this is enough.
      $ret = &suredef($aval) eq &suredef($bval);
    }

    last if $ret == 0; # as soon as something is not equal, we're done
  }
  
  #debug("equal_hashes returns: $ret\n");
  return $ret;
}

# Return true if arg H is a reference to a hash.
# 
sub is_hash_ref {
  my ($h) = @_;
  return defined $h && ref $h && $h =~ /HASH/;
}


sub finish {
  my ($tlpdb,$pkgcount) = @_;
  if ($opt_output) {
    if (open(OUT, ">$opt_output")) {
      $tlpdb->writeout(\*OUT);
      info("$prg: output $pkgcount packages to $opt_output.\n");
    } else {
      tldie("$prg: open(>$opt_output) failed: $!");
    }
  } else {
    $tlpdb->save;
    info("$prg: saved $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<--catalogue-compare> I<tlpdb>

By default, packages for which the only change is in the Catalogue data
are not marked as updated and written with new revisions. This option
checks the Catalogue data against that in I<tlpdb> (typically tlnet or
tlpretest), and forces the package to be updated if there are
differences. We enable this from time to time to re-sync the package
information even if the package contents have not changed.

As a consequence, after this has been invoked the first time, the same
packages will continue to be found as different until I<tlpdb> is
updated. Beware of confusion when testing/developing.

This implies C<--reverse-revision-check>, since the same mechanism is
used to force the new version numbers.

=item B<--catalogue-dump> I<file>

This is for debugging and development. If specified, and I<file> is
nonempty, it is <C>require</C>d, instead of reading the XML files from
I<Catalogue_dir> from C<--catalogue>. If I<file> is empty or does not
exist, the XML tree is read, and then dumped (with L<Data::Dumper>) to
I<file>.

The idea is to specify this, the dump gets written, and then subsequent
runs will use it, which is much faster than reading the XML. Don't
forget to delete the file when done, though, since there is no check for
staleness.

Also, don't use this if you are actually working on the Catalogue
reading routines (L<TeXLive::TeXCatalogue>), since they won't get
invoked at all if the dump is read.

=item B<--commit-max> I<integer>

If more than I<integer> files have been changed, do not do any commits
(like C<--no-commit>). The default is 100; that is, if more than 100
files have been changed, no commit will be done and a warning issued. If
by some strange chance more than 100 files actually need to be
committed, you can override with this option (probably best by editing
C<cron.tl>, if it was part of the nightly build).

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

If a package with revision number going backward is found, this option
forces a dummy change (C<svn propset>) to the package's tlpsrc (or
other) file to artificially increase its version number.

=item B<--from-files>

=item B<--from-git>

=item B<--from-gitsvn>

Use C<find>, C<git>, and git-over-svn, respectively, instead of C<svn
-status> to create the list of files. This amounts to calling the
respective function in L<TeXLive::TLTree> to initialize the tree of
files. By default, Subversion is used.

With C<--from-files>, all the revision numbers will be 1.

=item B<--keep-revisions>

If B<--no-reverse-revision-check> is also specified, this option copies
revision numbers from the existing tlpdb to the new one. Normally this
dos not happen.

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

The I<Master/dir> directory must point to a valid repository of TeX
Live's Master directory (Subversion by default; depends on the
C<--from...> options given). 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-fix-reverse-revisions>, plus C<--no-reverse-revision-check> (if
not otherwise specified) plus output to 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
current tlpdb (see below). This option disables this check.

=item B<--output> I<filename>

By default, C<tl-update-tlpdb> saves/updates the tlpdb in the default
place, except that with C<-n>, the default is C</tmp/tlut>. This option
specifies the place where the output should go to.

=item B<--save-anyway>

If the newly created tlpdb is considered buggy due to (unfixed) 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
for testing, or if 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<--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 a TeX Live directory hierarchy, along with
TeX Live source package files (tlpsrc) files, into a TeX Live package
database (tlpdb).

By default, it looks at Subversion output to determine the files in the
current tree to place into the tlpdb. With the C<--from-files> option,
it reads the filesystem with C<find>, and analogously for C<--from-git>
and C<--from-gitsvn>.

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., has a smaller revision number in the new
tlpdb). This can happen, for example, when a file is removed and no
other changes to the package made. In this case it does an C<svn
propset> on the package's tlpsrc (or other) file to ensure that
revisions are always increasing. The check is omitted if
C<--no-reverse-revision-check> is specified, and the property changes
are not committed unless C<--fix-reverse-revisions> is given.

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

=head1 SEE ALSO

The modules in C<Master/tlpkg/TeXLive/> (L<TeXLive::TLConfig> and
the rest), and the other scripts in C<Master/tlpg/bin/> (especially
C<tl-update-containers>), the documentation in C<Master/tlpkg/doc/>, etc.

=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: #