r265: * It turns out it's possible to set up make variables that are specific to

[debhelper.git] / dh_compress

diff --git a/dh_compress b/dh_compress
index e1f8ed643525f9102e87a652b60515ea71337713..847bcb6d73f49d31089b2e49a221114c758d5e5a 100755 (executable)
--- a/dh_compress
+++ b/dh_compress
@@ -1,40 +1,115 @@
-#!/bin/sh -e
+#!/usr/bin/perl -w
 #
 # Compresses files and makes sure that symlinks pointing to the 
 # compressed files get fixed.
 
-PATH=debian:$PATH:/usr/lib/debhelper
-source dh_lib
-
-# The config file is a sh script that outputs the files to be compressed
-# (typically using find).
-if [ -f debian/compress ]; then
-       files=`sh debian/compress 2>/dev/null`
-else
-       # By default fall back on what the policy manual says to compress.
-       files=`
-               find debian/tmp/usr/info debian/tmp/usr/man \
-                       debian/tmp/usr/X11*/man -type f 2>/dev/null ;
-               find debian/tmp/usr/doc -type f -size +4k \
-                       ! -name "*.htm*" ! -name "*.gif" \
-                       ! -name "debian/tmp/usr/doc/*/copyright" 2>/dev/null
-       `
-fi
-
-if [ "$files" ]; then
-       # This is just a cosmetic fix.
-       files=`echo $files | tr "\n" " "`       
-
-       doit "gzip -9 $files" || true
-fi
-
-# Fix up symlinks that were pointing to the uncompressed files.
-for file in `find debian/tmp -type l`; do
-       DIRECTORY=`expr $file : "\(.*\)/[^/]*"`
-       NAME=`expr $file : ".*/\([^/]*\)"`
-       LINKVAL=`ls -l $DIRECTORY/$NAME | awk '{ print $11;}'`
-       if [ ! -e $DIRECTORY/$LINKVAL -a -f $DIRECTORY/$LINKVAL.gz ]; then
-               doit "rm $DIRECTORY/$NAME"
-               doit "ln -s $LINKVAL.gz $DIRECTORY/$NAME.gz"
-       fi
-done
+use Cwd;
+BEGIN { push @INC, "debian", "/usr/share/debhelper" }
+use Dh_Lib;
+init();
+
+foreach $PACKAGE (@{$dh{DOPACKAGES}}) {
+       $TMP=tmpdir($PACKAGE);
+       $compress=pkgfile($PACKAGE,"compress");
+
+       # Run the file name gathering commands from within the directory
+       # structure that will be effected.
+       $olddir=getcwd();
+       verbose_print("cd $TMP");
+       chdir($TMP) || error("Can't cd to $TMP: $!");
+
+       # Figure out what files to compress.
+       @files=();
+       # First of all, deal with any files specified right on the command line.
+       if (($PACKAGE eq $dh{FIRSTPACKAGE} || $dh{PARAMS_ALL}) && @ARGV) {
+               push @files,#ARGV;
+       }
+       if ($compress) {
+               # The config file is a sh script that outputs the files to be compressed
+               # (typically using find).
+               push @files, split(/\n/,`sh $olddir/$compress 2>/dev/null`);
+       }
+       else {
+               # By default, fall back to what the policy manual says to compress.
+               # Note that all the excludes of odd things like _z are because
+               # gzip refuses to compress such files, assumming they are zip files.
+               # I looked at the gzip source to get the complete list of such
+               # extentions: ".gz", ".z", ".taz", ".tgz", "-gz", "-z", "_z"
+               
+               push @files, split(/\n/,`
+                       find usr/info usr/share/info usr/man usr/share/man usr/X11*/man -type f ! -name "*.gz" 2>/dev/null || true;
+                       find usr/doc usr/share/doc -type f \\( -size +4k -or -name "changelog*" \\) \\
+                               \\( -name changelog.html -or ! -name "*.htm*" \\) \\
+                               ! -name "*.gif" ! -iname "*.gz" \\
+                               ! -iname "*.taz" ! -iname "*.tgz" ! -iname "*.z" \\
+                               ! -iname "*-gz" ! -iname "*-z" ! -iname "*_z" \\
+                               ! -name "copyright" 2>/dev/null || true
+               `);
+       }
+
+       # Exclude files from compression.
+       if (@files && defined($dh{EXCLUDE}) && $dh{EXCLUDE}) {
+               @new=();
+               foreach (@files) {
+                       $ok=1;
+                       foreach $x (@{$dh{EXCLUDE}}) {
+                               if (/\Q$x\E/) {
+                                       $ok='';
+                                       last;
+                               }
+                       }
+                       push @new,$_ if $ok;
+               }
+               @files=@new;
+       }
+       
+       # Look for files with hard links. If we are going to compress both,
+       # we can preserve the hard link across the compression and save
+       # space in the end.
+       my @f=();
+       my %hardlinks;
+       foreach (@files) {
+               ($dev, $inode, undef, $nlink)=stat($_);
+               if ($nlink > 1) {
+                       if (! $seen{"$inode.$dev"}) {
+                               $seen{"$inode.$dev"}=$_;
+                               push @f, $_;
+                       }
+                       else {
+                               # This is a hardlink.
+                               $hardlinks{$_}=$seen{"$inode.$dev"};
+                       }
+               }
+               else {
+                       push @f, $_;
+               }
+       }
+
+       if (@f) {
+               xargs(\@f,"gzip","-9f");
+       }
+       
+       # Now change over any files we can that used to be hard links so
+       # they are again.
+       foreach (keys %hardlinks) {
+               # Remove old file.
+               doit("rm","-f","$_");
+               # Make new hardlink.
+               doit("ln","$hardlinks{$_}.gz","$_.gz");
+       }
+
+       verbose_print("cd $olddir");
+       chdir($olddir);
+
+       # Fix up symlinks that were pointing to the uncompressed files.
+       open (FIND,"find $TMP -type l |");
+       while (<FIND>) {
+               chomp;
+               ($directory)=m:(.*)/:;
+               $linkval=readlink($_);
+               if (! -e "$directory/$linkval" && -e "$directory/$linkval.gz") {
+                       doit("rm","-f",$_);
+                       doit("ln","-sf","$linkval.gz","$_.gz");
+               }
+       }
+}