Script for downloading and adjusting pattern files.

1 files changed, 116 insertions, 0 deletions

diff --git a/install-hyphenation b/install-hyphenation
new file mode 100755
index 00000000..7ee90602
--- /dev/null
+++ b/install-hyphenation
@@ -0,0 +1,116 @@
+#!/usr/bin/env perl
+
+use POSIX;
+use File::Basename;
+
+sub getpattendir () {
+   # 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: $!";
+   while (<MF>) {
+      if (/^\s*prefix\s*=\s*(.*)\s*$/) {
+         $prefix = $1;
+     }        
+   }
+
+   close MF;
+   
+   if ($prefix eq "") {
+      die "prefix not defined in $mf";
+   }
+   
+   return "$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";
+
+}
+
+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.
+
+   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;
+        }
+      }
+   }
+   
+   close IN;
+   close OUT;
+
+   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;
+   }
+   
+   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;
+   }
+}