path: root/Master/tlpkg/bin/tlpkg-ctan-check

#!/usr/bin/env perl
# $Id$
# Public domain.  Originally written 2005, Karl Berry.
# Rewritten for new infrastructure 2007, Norbert Preining.
# Check if a package in TL has any changes on CTAN.  Just the beginnings.

BEGIN {
  chomp ($mydir = `dirname $0`);  # we are in Master/tlpkg/bin
  unshift (@INC, "$mydir/..");
}

use TeXLive::TLConfig;
use TeXLive::TLPOBJ;
use TeXLive::TLPDB;

use File::Basename;

my $tlpdb;
my $TL;

my @WorkingTLP = qw(
        12many
        a0poster a2ping aastex abbr abc abstract abstyles accfonts achemso
          acmtrans acronym 
          active-conf addlines adrconv ae aguplus aichej AkkTeX algorithm2e
          algorithmicx algorithms
          alnumsec altfont ametsoc ams amsaddr amsldoc-it amsrefs
          amsthdoc-it animate ANUfinalexam anyfontsize apacite appendix apl
          arabi arabtex arabxetex archaic arcs arev
          armenian arphic arrayjob arydshln Asana-Math ascii assignment astro
          attachfile aurical authoraftertitle authorindex auto-pst-pdf automata
          auncial-new
        babelbib barcodes bbcard bbold
          beamer beamer-contrib beamer-tut-pt begriff bera betababel
          bezos bghyphen bibhtml bibleref bibtopic bibtopicprefix
          bibunits bigfoot
          binomexp bitfield blacklettert1 blindtext blockdraw_mp boldtensors
          bookest booklet bophook
          boxhandler bpchem breakurl bullcntr bussproofs bytefield 
        captcont caption carlisle catechis carolmin-ps cb
          cbcoptic ccaption cclicenses cdpbundl cellspace
          changebar changepage changes chappg
          chem-journal chemcompounds chemscheme
          chemstyle cherokee chessboard chessfss circ
          citeref cjhebrew classicthesis
          clefval cleveref
          clock clrscode cmap cmastro cmdstring cmdtrack cmll colorinfo
          colortbl colorwav commath compact compactbib complexity comprehensive
          computational-complexity confproc constants
          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 cweb-latex cyrillic
        dancers dashbox datatool
          dateiliste datetime dcpic decimal delimtxt digiconfigs din1505
          dinat dingbat directory disser dk-bib dlfltxb doi doipubmed
          dotarrow dottex
          doublestroke dpfloat dprogress draftcopy
          draftwatermark dramatist dtxgallery
          dtxtut duerer-latex dvdcoll dviincl dvipsconfig dyntree
        easy ebong ebsthesis eCards ecc ecclesiastic economic
          ecv ed edmac ednotes eemeir egameps
          egplot eijkhout elatex
          ellipsis elpres elsevier elsevier-bib emulateapj endfloat endheads
          engrec enumitem envbig environ epigrafica epiolmec eplain epsdice
          epsf epsincl epslatex-fr epspdfconversion epstopdf
          eqlist eqparbox errata es-tex-faq
          ESIEEcv esdiff esint esint-type1 eskd eskdx eso-pic esvect etaremune
          ethiop ethiop-t1 etoolbox euenc
          eulervm euproposal euro europecv eurosans everypage exam
          examdesign examplep exercise expl3 expressg extarrows exteps
          extpfeil extract extsizes
        faktor fancybox fancyhdr fancynum fancyref fancytooltips fax
          fbs featpost
          feyn fge filecontents fink fixme
          flagderiv flippdf float floatrow flowfram fltpoint
          fmtcount fnbreak
          fncychap foekfont fontch fontspec fonttable footbib forloop formular
          fouridx fourier fouriernc
          fp fpl frankenstein free-math-font-survey
          frenchle fribrief frletter frontespizio fullpict functan
        gaceta galois gastex gatech-thesis gb4e gcard gcite
          genmisc genmpage germbib
          gfsartemisia gfsbodoni gfscomplutum gfsdidot gfsneohellenic
          gfsporson gfssolomos
          ginpenc glhyph glossaries gmdoc gmeometric
          gmiflink gmutils gmverb gnuplottex gost graphics
          graphicx-psmin greenpoint
          grnumalt gu GuIT guitar guitbeamer
        HA-prosper hands harmony hep hepthesis hepunits hhtensor histogr hitec
          hrlatex hvfloat hwkatakana
          hyperref hyperxmp hyphen-base hyphen-basque
          hyphen-french hyphen-german
          hyphen-greek hyphen-italian hyphen-latin hyphen-mongolian
          hyphen-norwegian hyphen-ukenglish
          hyphen-usorbian hyphenex hyplain
        ibygrk icsv IEEEconf IEEEtran ifplatform ifxetex
          ijmart ijqc image-gallery
          inlinebib interactiveworkbook inversepath invoice
          iopart-num isodoc isorot isotope itnumpar iwona
        japanese jeopardy jknapltx jneurosci jpsj jurabib juramisc jurarsp
        kalender kerkis kerntest keystroke koma-script kpfonts
        l2picfaq l2tabu-english l2tabu-spanish
          labbook labelcas labels lastpage latex latex-tds latexmp
          layouts lcd lcg leading leaflet ledmac leftidx lewis lfb lgreek lh
          libertine limap linearA lineno linguex
          lipsum listings listliketab lkproof lm
          logpap lps 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 lxfonts
        mafr magyar mailing makebox makecell makecirc makeglos
          makeplot maple marginnote marvosym mathdots mathexam
          mathmode maybemath
          mcaption mceinleger mciteplus mdwtools memexsupp
          memoir MemoirChapStyles mentis
          menu metaobj metaplot metatex metauml mff mfpic mftinc mh mhchem
          mhs microtype minipage-marginpar minitoc minutes mkjobtexmf mkpattern
          mla-paper mlist mnsymbol
          moderncv mongolian-babel moreverb movie15 mp3d ms msg multicap
          multido munich muthesis mwcls mwrite
        nag namespc natbib nath nature ncclatex ncctools newlfm nih noitcrul
          nomencl nomentbl nostarch notes2bib
          ntgclass ntheorem-vn numline numname numprint
        oberdiek ocr-latex octavo ofs ogham oldstyle onlyamsmath opcit
          ordinalpt
        papercdcase papertex parallel paresse parrun pauldoc pbox
          pdf-forms-tutorial-de pdf-forms-tutorial-en pdf-trans pdfcprot
          pdfcrop
          pdfpages pdfscreen pdfsync pdftex-def perception perltex permute
          pgf pgf-soroban pgfplots phaistos philex philosophersimprint
          pict2e pictex pictex2 pinlabel pittetd placeins plari plates play
          poemscol polski polynom polynomial polytable powerdot ppr-prv
          pracjourn preprint
          prettyref preview probsoln procIAGssymp program protocol pseudocode
          psfrag psfrag-italian psnfss pspicture
          pst-2dplot pst-3d pst-3dplot pst-asr pst-bar
          pst-barcode pst-blur pst-circ pst-cox pst-dbicons pst-diffraction
          pst-eps pst-eucl pst-fill
          pst-fr3d pst-fractal pst-func
          pst-geo pst-grad pst-infixplot pst-jtree pst-labo
          pst-lens pst-light3d pst-math pst-ob3d pst-optexp pst-optic
          pst-osci pst-pad pst-pdf pst-pdgr pst-qtree
          pst-solide3d pst-soroban pst-spectra
          pst-slpe pst-stru pst-uml pst-vue3d
          pstricks pstricks-add
          pxfonts
        qcm qstest quotmark qtree
        r_und_s randbild rccol rcs rcsinfo recipecard
          refstyle regcount register resume robustcommand
          robustindex rsc rtkinenc rtklage
        sae sagetex sauerj sauterfonts savefnmark savesym savetrees scale
          scalebar schedule scientificpaper sciposter screenplay script
          sdrt sectionbox sectsty semantic semaphor seminar semioneside
          seqsplit setspace sf298 sffms sfg sfmath sgame shadbox shapepar
          shipunov shorttoc show2e showdim showexpl showlabels
          SIstyle SIunits 
          si sidecap sides siggraph simplecv simplewick siunitx
          skak skaknew slideshow smalltableof smartref
          snapshot songbook sort-by-letters soul sparklines spie
          splines splitbib splitindex spotcolor sprite srcltx sseq
          ssqquote stack stage staves stdclsdv stdpage stellenbosch
          stringstrings struktex
          sttools stubs subdepth subeqn subeqnarray
          subfig subfigure subfloat substr sudoku sudokubundle sugconf 
	  supertabular susy svgcolor svn svn-multi svninfo
          swebib syntax synproof syntrace synttree
        t-angles Tabbing tableaux tablists tablor tabulary tabvar talk 
	  tamethebeast
          ted templates-fenn templates-sommer tengwarscript
          tensor teubner tex-gyre texilikecover texmate
          texpower texshade
          textcase textfit textopo textpath textpos thesis-titlepage-fhac
          thinsp thmtools thumb thumbpdf thuthesis ticket titlefoot titlesec
          titling tocbibind
          tocloft todo tokenizer toolbox tools toptesi totpages tracking
          trajan translator tree-dvips trfsigns trivfloat
          trsym tugboat tugboat-plain turnstile twoup txfonts Type1fonts
          typedref typogrid
        uebungsblatt ucthesis uhrzeit uiucthesis ulsy umich-thesis uml umlaute
          umthesis umtypewriter underlin undertilde units unitsdef unroman
          upmethodology upquote ushort urwthesis
        vancouver variations varindex varsfromjobname
         vector velthuis venn verse versions
          vhistory visualfaq vmargin vntex volumes vpe vxu
        wallpaper warning warpcol williams wordcount wordlike wrapfig
        xargs xbase xcolor xdoc xecyr xetexref xfor xfrac xgreek xifthen
          xkeyval xltxtra xnewcommand xoptarg
          xq xskak xtab xtcapts xunicode xyling xypic-tut-pt xytree
        york-thesis
        ziffer
    ); 
#abstyles not updated but too old to matter
#amscls amsltx2 amsmath - AMS packages come from tds zips, all very weird
#beebe needs more work
#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



exit (&main ());

sub main
{
  # get the Master dir
  chomp (my $Master = `cd $mydir/../.. && pwd`);
  $TL = $Master;
  # First we load the current texlive.tlpdb
  $tlpdb = TeXLive::TLPDB->new ("root" => "$Master");
  die("Cannot find tlpdb in $Master!\n") unless defined($tlpdb);

  $OPT{"verbose"} = 1;
  
  if ($ARGV[0] eq "--list-not-treated") {
    @AllTLP = $tlpdb->list_packages;
    my @nottreatedtlp=();
    foreach my $b (@AllTLP) {
      my @foo = grep(/^$b$/, @WorkingTLP);
      if ($#foo < 0) { push @nottreatedtlp, $b ; }
    }
    print "List of tlps which are not in ctan-check:\n";
    foreach (@nottreatedtlp) {
      print "$_\n";
    }
    exit 0;
  }
  # xx read FILES.last7days, determine CTAN changes, map to TL packages
  # xx alternatively, check all tlp's, etc. ...
  if ($ARGV[0] eq "--check") {
    $OPT{"verbose"} = 0;

    # xx but for now, just check the few packages we have worked with.
    @ARGV = @WorkingTLP;
  }

  my $errcount = 0;
  for my $tlp (@ARGV) {
    print "checking $tlp..." if $OPT{"verbose"};
    $errcount += &do_tlp ($tlp);
  }
  return $errcount;
}



# Return 1 if TPM needs updating, 0 if ok.
# 
sub do_tlp
{
  my ($tlpn) = @_;
  my $needed = 0;
  
  my $tlp = $tlpdb->get_package($tlpn);
  if (not(defined($tlp))) {
    warn("Cannot get package $tlpn, strange!\n");
    return 1;
  }

  chomp (my $ctan_dir = `$mydir/tlpkginfo --ctan-dir '$tlpn'`);
  if (! $ctan_dir) {
    warn "$0: oops, no CTAN directory for $tlpn, fix fix\n";
    return ();
  }
  my @tpm_files = ();
  push @tpm_files, $tlp->runfiles;
  push @tpm_files, $tlp->docfiles;
  push @tpm_files, $tlp->srcfiles;
  # we don't push bin files, should we?
  
  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 $ctan_dir/ -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 $ctan_dir/ -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 $ctan_dir, fixme!\n";
    warn "(tpm_files = @tpm_files)\n";
  } elsif ($needed == 0) {
    print "ok, " . (@compared + 0) . " compared (@compared)\n"
      if $OPT{"verbose"};
  }

  return $needed;
}



# 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;
}