diff options
| -rwxr-xr-x | install-hyphenation | 221 |
1 files changed, 129 insertions, 92 deletions
diff --git a/install-hyphenation b/install-hyphenation index 7ee90602..6cea93a6 100755 --- a/install-hyphenation +++ b/install-hyphenation @@ -1,116 +1,153 @@ #!/usr/bin/env perl - use POSIX; use File::Basename; +use Net::FTP; + +$commondir = "language/hyph-utf8/tex/generic/hyph-utf8/patterns/txt"; + +# You may want do adjust these values. + +$host = "ctan.org"; +$dir = "/tex-archive/$commondir"; + +#$host = "mirrors.dotsrc.org"; +#$dir = "/ctan/$commondir"; -sub getpattendir () { +#$host = "ftp.dante.de"; +#$dir = "/pub/tex/$commondir"; + +$help = 0; +if (@ARGV == 0) { + $help = 1; +} else { + foreach $arg (@ARGV) { + if ($arg eq '-?' || $arg eq '--?' || $arg eq '-h' || + $arg eq '--h' || $arg eq '-help' || $arg eq '--help') { + $arg = 1; + } + } +} + +if ($help) { + print "Usage: $0 <lang1> [<lang2> ...]\n\n"; + print <<EOT; +Download and install hyphenation patterns from CTAN for different languages. +Languages are specified as defined by ISO 639-1 ("en" for English etc.). + +If there are multiple pattern files for a language (and so the filename is not +"hyph-<lang>.pat.txt"), you must specify the respective part of the filename, +e. g. "en-gb" or "en-us". In this case, the extra part ("-gb" or "-us") is +automatically removed. + +If you are not sure, simply specify the language code ("en" in this example); +you will then given a list of existing pattern files. +EOT +} else { # Extract pattern directory from Makefile, in the same directory as # this script. A complete interpretation of the Makefiles would be # too complicated, so only ${prefix} is read, which is (hopefully) # defined as constant. - - my $mf = (dirname $0) . "/Makefile", $prefix = ""; - open MF, $mf or die "Cannot open $mf for reading: $!"; + $prefix = ""; + $makefile = (dirname $0) . "/Makefile"; + open MF, $makefile or die "Cannot open $makefile: $!"; while (<MF>) { if (/^\s*prefix\s*=\s*(.*)\s*$/) { $prefix = $1; - } + } } - close MF; - + if ($prefix eq "") { - die "prefix not defined in $mf"; + die "Prefix not defined in $makefile."; } - - return "$prefix/lib/dillo/hyphenation"; -} + $patterdir = "$prefix/lib/dillo/hyphenation"; -sub printhelp { - print " Syntax: install-hyphenation <lang1> [<lang2> ...]\n"; - print "\n"; - print " If there are multiple pattern files for a language, and so the file name\n"; - print " is not \"hyph-<lang>.tex\", you can specify the origin by using a colon:\n"; - print " \"<src-lang>:<dest-lang>\", eg. \"de-1996:de\". If both are identical, the\n"; - print " colon can be committed; \"ru\" is equivalent to \"ru:ru\".\n"; - -} + if (!-e $patterdir) { + mkdir $patterdir or die "Cannot create directory $patterdir: $!"; + print "Created $patterdir.\n"; + } -sub trpatfile { - # Extract the data needed by dillo from the TeX file, based on two - # rules: - # - # 1. Extract the argument of the "\pattern" sequence, i. e. - # anything between "\pattern{" and the next "}". - # 2. As an exception, comments are always preserved. + # Connect to CTAN FTP server, change to the directory where the + # patterns lie, and read files list (which may be useful later). + $ftp = Net::FTP->new($host,Timeout=>240) + or die "Cannot connect to $host: $!"; + $ftp->login() or die "Cannot login: $!"; + $ftp->cwd($dir) or die "Cannot change to directory $dir: $!"; + @files = $ftp->ls or die "Cannot read directory: $!"; - my $url = $_[0], $in = $_[1], $out = $_[2]; - - open IN, $in or die die "Cannot open $in for reading: $!"; - open OUT, "> $out" or die "Cannot open $out for writing: $!"; - - printf OUT "%% The original file was downloaded from\n"; - printf OUT "%%\n"; - printf OUT "%% $url\n"; - printf OUT "%%\n"; - printf OUT "%% and automatically translated into this format. The original comments,\n"; - printf OUT "%% including the original copyright notice, are preserved.\n"; - printf OUT "%%\n"; - printf OUT "%% -----------------------------------------------------------------------------\n"; - printf OUT "%%\n"; - - $inpatterns = 0; - while (<IN>) { - if (/^%/) { - # Adopt all comments - print OUT; - } elsif (!$inpatterns) { - if (/\\patterns\s*{/) { - $inpatterns = 1; - } - } else { - if (/}/) { - $inpatterns = 0; - } else { - print OUT; - } + # Finally, read pattern files. + foreach $arg (@ARGV) { + if ($arg =~ /^([a-z]+)-.*$/) { + # More files per language, e. g. "en-gb". + $lang = $1; + } else { + # One file per language, e. g. "ru". + $lang = $arg; } - } - - close IN; - close OUT; + + # First, donwload the pattern file to a temporary file. + $tmppat = tmpnam(); + if ($ftp->get ("hyph-$arg.pat.txt", $tmppat)) { + printf ("Successfully downloaded pattern file for \"$arg\".\n"); + + # Search for a licence file. (Only a warning, when it does + # not exist.) + $tmplic = tmpnam(); + $licfound = 0; + if ($ftp->get ("hyph-$arg.lic.txt", $tmplic)) { + $licfound = 1; + } else { + print "Warning: Cannot download license file for \"$arg\": $!\n"; + } + + # Combine both, licence and pattern, to the final pattern + # file. + $outfile = "$patterdir/$lang.pat"; + open OUT, "> $outfile" or die "Cannot open $outfile: $!"; + + if ($licfound) { + print OUT + "% Licence from ftp://$host$dir/hyph-$arg.lic.txt\n"; + print OUT "%\n"; + open IN, $tmplic or die "Cannot open $tmplic: $!"; + while (<IN>) { + # Each line from the licence file must be a comment. + if (!/^%/) { + print OUT "% "; + } + print OUT; + } + close IN; + unlink $tmplic; - printf "Wrote pattern file: $out\n"; -} - -sub dlpatfile { - my $lang = $_[0], $destdir = $_[1], $lang1, $lang2; - if ($lang =~ /(.*):(.*)/) { - $lang1 = $1; - $lang2 = $2; - } else { - $lang1 = $lang2 = $lang; + print OUT "%\n"; + } + + print OUT "% Patterns from ftp://$host$dir/hyph-$arg.pat.txt\n"; + print OUT "%\n"; + open IN, $tmppat or die "Cannot open $tmppat: $!"; + while (<IN>) { + print OUT; + } + close IN; + unlink $tmppat; + + close OUT; + } else { + # Not found. If a single language was specified (e. g. "en"), + # search for possibilities. + print "Error: Cannot download pattern file for \"$arg\": $!\n"; + if ($lang eq $arg) { + print "Try one of these:\n"; + foreach(@files) { + if (/^hyph-($lang-.*)\.pat\.txt$/) { + print " $1\n"; + } + } + } + } } - my $url = "ftp://ftp.mpi-sb.mpg.de/pub/tex/mirror/ftp.dante.de/pub/tex/language/hyph-utf8/tex/generic/hyph-utf8/patterns/tex/hyph-$lang1.tex"; - my $tmp = tmpnam(); - if (system("wget -O $tmp $url") == 0) { - my $out = "$destdir/$lang2.pat"; - trpatfile $url, $tmp, $out; - unlink $tmp; - } else { - print "Error downloading patterns for language $lang1. See messages above for details.\n" - } -} - -$destdir = getpattendir; - -foreach $lang (@ARGV) { - if ($lang eq '-?' || $lang eq '--?' || $lang eq '-h' || - $lang eq '--h' || $lang eq '-help' || $lang eq '--help') { - printhelp; - } else { - dlpatfile $lang, $destdir; - } + $ftp->quit; } |