#!/usr/bin/env perl
# $Id$
# Public domain.  Originally written 2005, Karl Berry.
# 
# Return information given a TL package name (i.e., a tlpsrc file name).
# We require local copies of CTAN and the TeX Catalogue.

exit (&main ());

sub main {
  $CTAN = $ENV{"CTAN"} || "/home/ftp/tex-archive";
  $TMPDIR = $ENV{"TMPDIR"} || "/tmp";
  $CATALOGUE = $ENV{"TEX_CATALOGUE"} || "/home/httpd/html/catalogue/entries";
  -d "$CATALOGUE/k"
    || die "$0: TEX_CATALOGUE ($CATALOGUE) must point to entries/ subdir"
           . " of a TeX Catalogue checkout.\n";
  
  # erroneous or problematic tds files (when new, tell RobinF and author)
  $erroneous_tds = join ("|",
    qw(countriesofeurope engpron gost hacm he-she imakeidx imtekda 
       mathdesign spanish sttools titleps),
    );
  
  # Heiko's tds files (don't propagate through CTAN fast enough).
  $oberdiek_tds = "hyperref|oberdiek";

  # special packages from latex-tds project; used by prepare()
  $amslatex_tds_pkgs = "xamscls|xamsmath|xamsrefs";   # from amslatex.tds
  $latex_tds_pkgs  = $amslatex_tds_pkgs;
  $latex_tds_pkgs .= "|cyrillic|graphics|latex|psnfss|tools"; # latex
  $latex_tds_pkgs .= "|latex-tds|amsfonts|knuth";                   # other
  $latex_tds_dir = "$CTAN/macros/latex/contrib/latex-tds";
  # (amstex handled from the original.)
  
  # the CJK material is split into several packages.
  $cjk_pkgs = "bin-(cjk|ttf)utils|dnp|(|garuda-|norasi-)c90";

  if ($ARGV[0] eq "--ctan-dir") {
    my $output = &find_ctan_dir ($ARGV[1]);
    print "$output\n";
    return $output ? 0 : 1;

  } elsif ($ARGV[0] eq "--prepare") {
    my $output = &prepare ($ARGV[1]);
    print "$output\n";
    return $output ? 0 : 1;

  } elsif ($ARGV[0] eq "--ctan-root") {
    print "$CTAN\n";
    return 0;

  } else {
    die "Usage: $0 --ctan-dir PKGNAME\n" 
      . "   or:  --prepare PKGNAME\n"
      . "   or:  --ctan-root\n"
      . "(not \"@ARGV\").\n";
  }
}

# 
# Return 0 for success if we find a path (and print it on stdout), else
# return 1.
# 
sub find_ctan_dir {
  my ($pkgname) = @_;
  return 1 unless $pkgname;  
  
  # use explicit catalogue name if given in tlpsrc.
  my $me = &tlpsrc_find_catalogue ($pkgname);
  if (! $me) {
    if ($pkgname =~ /^($cjk_pkgs)$/) {
      # ctan bundles these packages together,
      # but we use separate tlp names.
      $me = "cjk";  # likewise CJK
    } elsif ($pkgname eq "mkgrkindex") {
      $me = "greek-makeindex";  # use directory name
    } else {
      $me = $pkgname;  # usually the same
    }
  }

  # The CTAN path to a package is sometimes stored in the TeX Catalogue,
  # but old entries don't have it, etc.  Still, we want to use it if present.
  my $ctan_dir = &catalogue_find_ctan_path ($me);
  if (! $ctan_dir) {
    # fall back on many special cases
    (my $me_wo_babel = $me) =~ s,^babel-,,;  # remove when Catalogue ready
    (my $me_wo_bin = $me) =~ s,^bin-,,;
    (my $me_wo_powerdot = $me) =~ s,^powerdot-,,;
    (my $me_wo_type1 = $me) =~ s,-type1$,,;

    for my $dir (
      "macros/latex/contrib/$me",		# most everything
      "macros/latex/contrib/babel-contrib/$me_wo_babel", # babel-*
      "macros/latex/contrib/biblatex/$me",	# authoryear-icomp-tt      
      "macros/latex/contrib/biblatex-contrib/$me", # biblatex-*
      "macros/latex/contrib/gene/$me",		# eqname
      "macros/latex/contrib/\L$me",		# HA-prosper
      "macros/latex/contrib/powerdot/contrib/$me_wo_powerdot", # powerdot-doc-vn
      "macros/latex/exptl/$me",			# semioneside
      "macros/latex/required/$me",		# babel
      "macros/xetex/latex/$me",			# xetex
      "macros/xetex/generic/$me",		# harvardkyoto
      "macros/plain/$me",			# plnfss
      "macros/plain/contrib/$me",		# timetable
      "macros/luatex/latex/$me",		# luabidi
      "macros/generic/$me",			# abbr
      "macros/generic/diagrams/$me",		# circ
      "macros/context/contrib/$me",		# context
      "support/$me",				# thumbpdf
      "support/$me_wo_bin",			# bin-thumbpdf
      "language/vietnamese/$me/unpacked",	# vntex
      "language/polish/$me",			# cc-pl
      "language/hyphenation/$me",		# bghyphen
      "language/hebrew/$me",			# cjhebrew
      "language/greek/package-babel/$me",	# ibycus-babel
      "language/devanagari/$me",		# velthuis
      "language/croatian/$me",			# hrlatex
      "language/coptic/$me",			# cbcoptic
      "language/japanese/$me",			# e.g., jfontmaps
      "info/spanish/$me",			# guia-bibtex
      "info/symbols/$me",			# comprehensive
      "info/math/voss/$me",			# mathmode
      "info/challenges/$me",			# AroBend
      "info/bibtex/$me",			# tamethebeast
      "info/$me",				# Type1fonts
      "help/$me",				# es-tex-faq
      "graphics/$me",				# sparklines
      "macros/pdftex/$me",			# mptopdf
      "graphics/metapost/contrib/macros/$me",	# mpattern
      "graphics/pictex/addon/$me",		# autoarea
      "graphics/pstricks/contrib/pedigree/$me",	# pst-pdgr
      "graphics/pstricks/contrib/$me",		# pstricks-add
      "fonts/$me",				# MnSymbol
      "fonts/utilities/$me",			# accfonts
      "fonts/ps-type1/$me",			# cm-super
      "fonts/ps-type1/$me_wo_type1",		# esint-type1
      "fonts/greek/$me",			# lfb
      "fonts/gothic/$me",			# blacklettert1
      "fonts/cyrillic/$me/texmf",		# lh
      "fonts/chess/$me",			# skaknew
      "biblio/bibtex/contrib/$me",		# dk-bib
                 ) {
      if (-d "$CTAN/$dir") {
        $ctan_dir = $dir;
        last;
      }
    }

    # names totally dissimilar
    $ctan_dir = "fonts/fourier-GUT" if $me eq "fourier";
    $ctan_dir = "graphics/pdftex" if $me eq "pdftex-def";
    $ctan_dir = "info/biblio" if $me eq "beebe";
    $ctan_dir = "info/epslatex/french" if $me eq "epslatex-fr";
    $ctan_dir = "info/impatient/fr" if $me eq "impatient-fr";
    $ctan_dir = "info/italian/amsldoc" if $me eq "amsldoc-it";
    $ctan_dir = "info/italian/amsthdoc" if $me eq "amsthdoc-it";
    $ctan_dir = "info/l2tabu/italian" if $me eq "l2tabu-it";
    $ctan_dir = "info/latex2e-help-texinfo/spanish" if $me eq "latex2e-help-texinfo-spanish";
    $ctan_dir = "info/lshort/chinese" if $me eq "lshort-chinese";
    $ctan_dir = "info/tex-references" if $me eq "tex-refs";
    $ctan_dir = "info/translations/vn" if $me eq "ntheorem-vn";
    $ctan_dir = "language/armenian/armtex" if $me eq "armenian";
    $ctan_dir = "language/basque" if $me eq "hyphen-basque";
    $ctan_dir = "language/hungarian/babel" if $me eq "magyar";
    $ctan_dir = "language/hyphenation/dehyph" if $me eq "hyphen-german";
    $ctan_dir = "language/hyphenation/elhyphen" if $me eq "hyphen-greek";
    $ctan_dir = "macros/generic" if $me eq "genmisc";
    $ctan_dir = "macros/latex/contrib/misc" if $me eq "ltxmisc";
    $ctan_dir = "macros/latex/contrib/t2/cyrplain" if $me eq "cyrplain";

    # do last, for sake of subdirs above.
    $ctan_dir = "language/hyphenation"
      if $me =~ /^hyphen-/ && ! $ctan_dir;
  }

  # For mathspic, we cannot use the catalogue directory:
  # we want to ignore the dos/ subdir and just go with the perl/.
  if ($me eq "mathspic") {
    $ctan_dir .= "/perl";
  }                         
  
  # For cs, we need to unpack from multiple tarballs,
  # so copy the entire cstex directory.
  if ($pkgname eq "cs") {
    $ctan_dir = "$CTAN/macros/cstex/base";
  }                         
  
  # prepend ctan root if not an absolute dir (this happens when we make
  # a temp dir).
  $ctan_dir = "$CTAN/$ctan_dir" if $ctan_dir =~ m,^[^/],;
  return $ctan_dir;
}



# If the .tlpsrc file for ME has a catalogue entry, return it.
# Else return empty string.
# 
sub tlpsrc_find_catalogue {
  my ($me) = @_;
  
  chomp (my $mydir = `dirname $0`);
  chomp (my $tlpsrcdir = `cd $mydir/../tlpsrc && /bin/pwd`);
  my $tlpsrc = "$tlpsrcdir/$me.tlpsrc";
  if (! -r $tlpsrc) {
    warn "$0: no tlpsrc $tlpsrc\n";
    return "";
  }

  chomp (my $cat = `awk '\$1 == "catalogue" {print \$2}' $tlpsrc`);
  return $cat;
}



# look up ctan path for given package name in catalogue entry.
# xml is too hard to parse, so just look for the <ctan path...> entry.
# 
# Return the ctan path if found (without leading /), or undef.
# 
sub catalogue_find_ctan_path {
  my ($pkgname) = @_;
  
  # catalogue uses all-lowercase file/directory names.
  my $firstchar = substr (lc ($pkgname), 0, 1);
  my $catfile =  "$CATALOGUE/$firstchar/$pkgname.xml";
  return undef unless -r $catfile;

  # get the raw tag from the catalogue file.
  open (CATFILE, "<$catfile") || die "Cannot read $catfile: $!\nFix me!\n";
  while ($ctan_path = <CATFILE>) {
    last if $ctan_path =~ /<ctan /;
  }
  if (! $ctan_path) { # give up if the tag wasn't there.
    close CATFILE;
    return undef;
  } 
  # if we found the tag, read more lines until we get to the end of the value.
  while ($ctan_path !~ m,/>,) {
    $ctan_path .= <CATFILE>;
  }
  close CATFILE;
  $ctan_path =~ s/\n/ /g;
  ($ctan_path) = ($ctan_path =~ m,(<.*?/>),);
  return undef unless $ctan_path;  # in case it's not present at all

  # extract just the dir or file name, without options, etc.
  $ctan_path =~ m,path=(["'])/(.*?)\1,;
  $ctan_loc = $2;
  return undef unless $ctan_loc; # should never happen, but who knows

  # if the Catalogue lists the path as a single file, there are two
  # possibilities: (1) it really is a single file, e.g., texilikecover,
  # in which case we copy that file into a temp dir and return that temp
  # dir; (2) it is actually in its own directory, e.g., ifxetex, in
  # which case we return undef here and let the code above find it.
  # (The Catalogue maintainers do not consider this a problem.)
  # 
  if ($ctan_path =~ /file='true'/) {
    if ($ctan_path =~ m,/$pkgname/,) {  # pkg dir somewhere in path?
      return undef;
    } else {
      return $do_copy # are we called by prepare() or not? 
        ? &copy_to_tmpdir ($pkgname, "$CTAN/$ctan_loc")
        : "$CTAN/$ctan_loc";
    }
  }
  
  # not a single file, so use the regular ctan location.  
  return $ctan_loc;
}

# 
# Create a tmpdir with the tds-ready tree unpacked if any,
# else return the normal ctan dir.
#
sub prepare {
  my ($pkg) = @_;
  return undef unless $pkg;

  # find the real ctan dir and return it if our tds is erroneous
  $do_copy = 1; # ask find_catalogue() to copy single files, too
  my $ctan_loc = &find_ctan_dir ($pkg);
  return $ctan_loc if $pkg =~ /^($erroneous_tds)$/;

  # tds path is usually in ctan/install...
  my $tds_path = "$ctan_loc.tds.zip";
  $tds_path =~ s,^$CTAN,$CTAN/install,;

  # ...except for heiko, when running on tug.org...
  if ($pkg =~ /^($oberdiek_tds)$/) {
    $tds_path = "/home/ftp/tex/$pkg/$pkg.tds.zip";
    # will be ignored below if it doesn't exist.

  # ...and except for context, and others...
  } elsif ($pkg eq "context") {
    $tds_path = "/home/ftp/mirror/www.pragma-ade.com/context/current/cont-tmf.zip";

  } elsif ($pkg eq "hfoldsty") {
    $tds_path = glob ("$CTAN/fonts/hfoldsty/hfoldsty-v*.zip");

  # ...and except for files in latex-tds...
  } elsif ($pkg =~ /^($latex_tds_pkgs)$/) {
    if ($pkg =~ /^($amslatex_tds_pkgs)$/) {
      $tds_path = "$latex_tds_dir/amslatex.tds.zip" ;
    } else {
      $tds_path = "$latex_tds_dir/$pkg.tds.zip";
    }
    # and more latex exceptions to the latex exceptions:
    $tds_path = "$latex_tds_dir/source.tds.zip" if $pkg eq "latex-tds";
    $tds_path = "$latex_tds_dir/base.tds.zip" if $pkg eq "latex";
  }

  return $ctan_loc unless (-s $tds_path);
  
  # now we have a tds, so unzip it in a tmpdir and be sure it's readable.
  my $tmpdir = &copy_to_tmpdir ($pkg);
  system ("unzip -q $tds_path -d $tmpdir");
  system ("chmod -R a+rX $tmpdir");
  
  # put a flag for ctan2tds' donormal() to work
  system ("echo $tds_path >$tmpdir/TDS_READY");

  # problems with amslatex packages: they are all bundled together,
  # we now have to remove the unwanted ones.
  if ($pkg =~ /^($amslatex_tds_pkgs)$/) {
    chdir ($tmpdir) || die "chdir($tmpdir) failed: $!";
    my @deldir = ();
    # List of subsubdirectories, at the package level.
    chomp (my @dirlist = `find */*/* -type d | sort -u`);
    for my $dir (@dirlist) {
      next if $dir =~ m!$pkg$!;  # keep matching directories
      next if $pkg eq "amscls" && $dir =~ m,bst/ams$,;  # ams{alpha,plain}.bst
      push (@deldir, $dir);
    }
    system ("rm -rf @deldir");       # discard the rest
  }

  return $tmpdir;
}

# copy file to temp dir and return that for ctan2tl to use.
#
sub copy_to_tmpdir {
  my ($pkgname,$src) = @_;
  
  my $pkgdir = "$TMPDIR/tl.$pkgname";
  system ("rm -rf $pkgdir");
  mkdir ($pkgdir, 0777) || die "mkdir($pkgdir) failed: $!";

  # copying single file $src ...
  system ("cp -p '$src' '$pkgdir/'")
    if $src;

  # special cases
  if ($pkgname eq "cc-pl") {
    # there's no real ctan dir for this package, just a zip
    system ("cd $pkgdir && unzip -q cc-pl.zip && rm cc-pl.zip");
  }
  
  return $pkgdir;
}