path: root/Build/tools/tpm-ctan-check

#!/usr/bin/env perl
# $Id$
# Public domain.  Originally written 2005, Karl Berry.
# Check if a package in TL has any changes on CTAN.  Just the beginnings.

use File::Basename;

# euenc
my @WorkingTPM = qw(
        12many
        a0poster aastex abbr abc abstract abstyles accfonts achemso
          acmtrans acronym 
          active-conf adrconv aguplus algorithm2e algorithmicx algorithms
          alnumsec altfont ams amsldoc-it amsrefs
          amsthdoc-it animate apacite appendix apl
          arabi arabtex arabxetex archaic arcs arev
          armenian arrayjob arydshln ascii assignment attachfile
          aurical authorindex auto-pst-pdf automata
        babelbib barcodes
          beamer beamer-contrib beamer-tut-pt begriff bera betababel
          bezos bghyphen bibhtml bibleref bibtopic bibunits bigfoot
          binomexp bitfield blacklettert1 blindtext boldtensors
          bookest booklet bophook
          boxhandler breakurl bullcntr bussproofs bytefield 
        caption carlisle cbcoptic ccaption cdpbundl cellspace
          changebar chappg chem-journal chemcompounds chemscheme
          chemstyle cherokee circ
          citeref cjhebrew classicthesis
          clefval cleveref
          clock clrscode cmap cmdstring cmdtrack cmll colorinfo
          colortbl colorwav commath compactbib complexity comprehensive
          computational-complexity cooking cool coollist coolstr cooltooltips
          coordsys courier-scaled courseoutline coursepaper coverpage
          crop crossreference csquotes csvtools ctable
          cursolatex cursor cuisine
          currvita curve curve2e curves
          custom-bib cyrillic
        dancers dashbox datatool
          dateiliste datetime dcpic decimal delimtxt digiconfigs
          dinat dingbat directory disser dk-bib dlfltxb doi doipubmed dottex
          doublestroke dpfloat
          draftwatermark dramatist dtxgallery
          dtxtut dvdcoll dvipsconfig dyntree
        ebong ebsthesis eCards economic ecv edmac ednotes eemeir egameps
          egplot eijkhout elatex
          ellipsis elpres elsevier elsevier-bib emulateapj endfloat endheads
          engrec enumitem envbig epiolmec eplain epsdice
          epsf epslatex-fr eqlist eqparbox es-tex-faq
          ESIEEcv esint esint-type1 eskd eskdx eso-pic etaremune
          ethiop ethiop-t1 eulervm euro europecv eurosans everypage exam
          examdesign examplep exercise expl3 expressg extarrows exteps
          extpfeil extract extsizes
        faktor fancyhdr fancynum fancyref fancytooltips fax featpost
          fge filecontents fink fixme
          flagderiv flippdf float floatflt floatrow flowfram fltpoint
          fmtcount fnbreak
          fncychap foekfont fontch fontspec footbib forloop formular
          fourier fouriernc
          fp fpl frankenstein free-math-font-survey
          frenchle fribrief fullpict functan
        galois gastex gatech-thesis gcard genmisc genmpage germbib ginpenc
          glossaries gmdoc gmeometric
          gmiflink gmutils gmverb gnuplottex gost graphics
          graphicx-psmin greenpoint
          grnumalt GuIT
        HA-prosper hands harmony hepthesis hhtensor histogr hitec
          hrlatex hvfloat
          hyperref hyperxmp hyphen-base hyphen-basque
          hyphen-french hyphen-german
          hyphen-greek hyphen-italian hyphen-latin
          hyphen-norwegian hyphen-ukenglish
          hyphen-usorbian hyplain
        ibygrk icsv IEEEconf IEEEtran ifxetex ijqc inlinebib inversepath
          invoice
          iopart-num isodoc isorot isotope itnumpar iwona
        japanese jeopardy jneurosci jpsj jurabib juramisc
        kalender kerkis kerntest keystroke koma-script kpfonts
        l2picfaq l2tabu-english labelcas labels latex latex-tds latexmp
          layouts lcd lcg leading leaflet ledmac leftidx lewis lfb lgreek lh
          limap linearA lineno lipsum listings listliketab lkproof lm
          logpap lsc lshort-bulgarian lshort-dutch lshort-english
          lshort-finnish lshort-french lshort-german lshort-italian
          lshort-japanese lshort-korean lshort-mongolian
          lshort-polish lshort-portuguese lshort-slovak
          lshort-spanish lshort-thai lshort-turkish lshort-ukr
          lshort-vietnamese ltablex ltabptch ltxindex ltxmisc
        mafr magyar mailing makebox makecell makecirc makeglos
          makeplot maple marginnote marvosym mathdots mathexam
          mathmode maybemath
          mcaption mceinleger mdwtools memoir MemoirChapStyles mentis
          menu metaobj metaplot metatex metauml mff mfpic mftinc mhchem
          mhs microtype minipage-marginpar minitoc minutes mkpattern
          mla-paper mlist mnsymbol
          moderncv moreverb movie15 ms msg multicap munich muthesis
          mwcls mwrite
        nag natbib nath nature ncclatex ncctools newlfm nih noitcrul
          nomencl nomentbl ntgclass ntheorem-vn numline numname numprint
        oberdiek ocr-latex octavo ofs ogham oldstyle onlyamsmath opcit
          ordinalpt osa
        papercdcase papertex parallel paresse parrun pauldoc
          pdf-trans pdfcprot
          pdfpages pdfscreen pdfsync pdftex-def perception perltex permute
          pgf phaistos philosophersimprint
          pict2e pictex pictex2 pittetd placeins plari plates play
          poemscol polynom polynomial polytable powerdot ppr-prv
          pracjourn preprint
          prettyref preview probsoln procIAGssymp program protocol pseudocode
          psfrag psnfss pspicture pst-2dplot pst-3d pst-3dplot pst-bar
          pst-barcode pst-blur pst-dbicons pst-eps pst-eucl pst-fill
          pst-fr3d pst-fractal pst-func
          pst-geo pst-grad pst-infixplot pst-labo
          pst-lens pst-math pst-optexp pst-optic
          pst-osci pst-pdf pst-pdgr pst-qtree
          pst-slpe pst-uml pst-vue3d
          pstricks pstricks-add
        qcm qstest qtree
        randbild rccol rcsinfo refstyle regcount register resume robustcommand
          robustindex rsc rtkinenc rtklage
        sae sauerj sauterfonts savefnmark savesym savetrees scale
          scalebar schedule scientificpaper sciposter screenplay script
          sdrt sectionbox sectsty semantic semaphor seminar semioneside
          seqsplit setspace sf298 sffms sfg sgame shadbox shapepar
          shorttoc showdim showexpl showlabels sidecap sides siggraph
          simplecv
          SIstyle SIunits skaknew slideshow smalltableof smartref
          snapshot songbook sort-by-letters soul sparklines spie
          splines splitbib splitindex spotcolor sprite srcltx sseq
          ssqquote stack stage stdclsdv stdpage stellenbosch struktex
          sttools subeqn subeqnarray subfig subfigure subfloat substr sudoku
          sudokubundle sugconf supertabular svgcolor svn svn-multi svninfo
          swebib syntax synproof syntrace synttree
        t-angles Tabbing tableaux tabulary tabvar talk tamethebeast
          tengwarscript tensor teubner tex-gyre texilikecover texmate
          texpower texshade
          textcase textfit textopo textpath textpos thesis-titlepage-fhac
          thumb thumbpdf ticket titlefoot titlesec titling tocbibind
          tocloft todo tokenizer toolbox tools toptesi totpages tracking
          trajan trfsigns trivfloat
          trsym tugboat tugboat-plain twoup Type1fonts
          typedref typogrid
        ucthesis uhrzeit uiucthesis ulsy umich-thesis uml umlaute
          umthesis underlin undertilde units unitsdef unroman
          upmethodology upquote ushort
        vancouver variations varindex vector velthuis verse versions
          vhistory visualfaq vmargin volumes vpe
        wallpaper warning warpcol williams wordcount wordlike wrapfig
        xbase xcolor xdoc xetexref xfor xfrac xifthen xkeyval xltxtra
          xnewcommand xoptarg
          xq xtab xtcapts xunicode xyling xypic-tut-pt
          xytree
        york-thesis
    ); 
#abstyles not updated but too old to matter
#amscls amsltx2 amsmath - AMS packages come from tds zips, all very weird
#beebe needs more work
#euenc - xetex people don't keep ctan up to date
#logic/gn-logic14 conflicts with logic font
#lshort-russian # only zip files on CTAN
#malayalam # missing C sources on CTAN wrote author 5jun05.
#mathspic # inside pictex later
#mpattern # not up-to-date on CTAN wrote author 28may05.
#plnfss # only a zip file on CTAN
#slidenotes		# non-free: ALl rights reserved, nothing else!
#shortlst	# works, but license: no profit! needs catalogue update
#sae		# works, no profit license!
#xltxtra - xetex people don't keep ctan up to date

chomp (my $mydir = `dirname $0`);
chomp ($mydir = `cd $mydir && pwd`);
chomp ($TL = `cd $mydir/../../Master && pwd`);
chdir ($TL) || die "chdir($TL) failed: $!";

exit (&main ());

sub main
{
  $OPT{"verbose"} = 1;
  
  if ($ARGV[0] eq "--list-not-treated-tpm") {
    @AllTPM = glob("{texmf-dist,texmf-doc,texmf}/tpm/*.tpm");
    my @nottreatedtpm=();
    foreach (@AllTPM) {
      my $b = basename($_,".tpm");
      my @foo = grep(/^$b$/, @WorkingTPM);
      if ($#foo < 0) { push @nottreatedtpm, $_ ; }
    }
    print "List of tpms which are not in ctan-tpm-check:\n";
    foreach (@nottreatedtpm) {
      print "$_\n";
    }
    exit 0;
  }
  # xx read FILES.last7days, determine CTAN changes, map to TL packages
  # xx alternatively, check all tpm's
  if ($ARGV[0] eq "--check") {
    $OPT{"verbose"} = 0;

    # xx but for now, just check the few packages we have worked with.
    # xx don't check babel, heiko patched it.
    @ARGV = @WorkingTPM;
  }

  my $errcount = 0;
  for my $tpm (@ARGV) {
    
    # look in our various hierarchies.
    my $found = "";
    for my $dir ("texmf-dist", "texmf-doc", "texmf") {
      my $try = "$dir/tpm/$tpm.tpm";
      if (-r $try) {
        $found = $try;
        last;
      }
    }
    
    if ($found) {
      print "checking $tpm..." if $OPT{"verbose"};
      $errcount += &do_tpm ($found);
    } else {
      warn "$0: cannot find tpm file $tpm\n";
      $errcount++;
    }
  }
  return $errcount;
}



# Return 1 if TPM needs updating, 0 if ok.
# 
sub do_tpm
{
  my ($tpm) = @_;
  my $needed = 0;
  
  my %tpm = &read_tpm ($tpm);
  return 1 unless keys %tpm;
  my @tpm_files = @{$tpm{"files"}};

  my @compared = ();
  for my $file (@tpm_files) {
#warn "checking file $file\n";
    my $tl_file = "$TL/$file";
    if (! -e $tl_file) {
      warn "$tl_file: TL file missing\n";
      next;
    }
    
    (my $basefile = $file) =~ s,^.*/,,;
    chomp (my @ctan_files = `find $tpm{CTANdir}/ -name $basefile`);
#warn "ctan find $basefile: @ctan_files\n";
    # the trailing / is so if we happen to hit a symlink on CTAN, it'll
    # go through.
    next if @ctan_files > 1; # if more than one file by same name, skip

    my $ctan_file = $ctan_files[0];
#warn "ctan file is $ctan_file\n";
    if (! -e $ctan_file) {
      # maybe it'll be there with a case change in the name
      chomp (@ctan_files = `find $tpm{CTANdir}/ -iname $basefile`);
      next if @ctan_files > 1; # if more than one file by same name, skip
      $ctan_file = $ctan_files[0];
      
      if (! -e $ctan_file) {
        # we generate lots of files, eg perlmacros.sty, so might skip.
        warn "$ctan_file: CTAN file missing\n"
        if $ctan_file && $ctan_file !~ /(cfg|dvi|sty|tex)$/;
        next;
      }
    }

    push (@compared, $basefile);
    if (&files_differ ($tl_file, $ctan_file)) {
      # we sometimes regenerate pdf's for TL, so we don't complain if
      # they are different.  On the other hand, we don't always
      # regenerate them, so might as well check first.
      next if $tl_file =~ /\.pdf$/;

      print "# $tpm\ndiff $ctan_file $tl_file\n";
      $needed = 1;
      last;
    }
  }
  
  # xx check ctan dir for new files
  
  if (@compared == 0) {
    warn "\n$tpm: no files to compare in $tpm{CTANdir}, fixme!\n";
    warn "(tpm_files = @tpm_files)\n";
  } elsif ($needed == 0) {
    print "ok, " . (@compared + 0) . " compared (@compared)\n"
      if $OPT{"verbose"};
  }

  return $needed;
}



# Return useful information from a TPM.
# 
sub read_tpm
{
  my ($tpm) = @_;
  my %ret;
  
  local *TPM;
  open (TPM, $tpm) || die "open($tpm) failed: $!\n";

  my @files = ();
  while (<TPM>) {
    if (m!</TPM:.*Files!) {  # might be on same line
      $collecting = 0;
    } elsif (/<TPM:.*Files/) {
      $collecting = 1;
      next;
    }
    if ($collecting) {
      chomp;
      next if m,plain/tugboat,;  # have to rely on latex tugboat,
                                 # plain tugboat README conflicts
      next if m,active-header-2006.tex,;  # ctan example conflicts with
                                          # dtx-generated file
      push (@files, $_) if $_ !~ /\.tpm$/;
    }
  }
  close (TPM) || warn "close($tpm) failed: $!\n";

  warn "$tpm: no files found\n" if ! @files;  # probably misparsed the tpm
  $ret{"files"} = \@files;

  (my $me = $tpm) =~ s,.*/,,;
  $me =~ s/\.tpm$//;
  chomp (my $ctan_dir = `$mydir/tlpkginfo --ctan-dir '$me'`);

  if (! $ctan_dir) {
    warn "$0: oops, no CTAN directory for $me, fix fix\n";
    return ();
  }
  $ret{"CTANdir"} = $ctan_dir;
  
  return %ret;
}



# 0 if files are the same, 1 if they are different.
# 
sub files_differ
{
  my ($tl_file,$ctan_file) = @_;
#warn "comparing $tl_file $ctan_file\n";
  
  my $different = 0;
  if (-s $tl_file != -s $ctan_file) {
    # sometimes what's checked into CTAN vs. TL differs in the eol
    # conventions.  Don't worry about that.
    if (system ("diff -q --strip-trailing-cr $tl_file $ctan_file >/dev/null")
        != 0) {
      $different = 1;
    }
  }

  return $different;
}