texunexpand

#!/usr/bin/perl
#
#
# Explodes a LaTeX file obtained from texexpand -x, and then compares
# the resulting files with the files existing on the current machines,
# removing the unmodified ones.
#
# Useful when collaborating with a remote author that is not particularly
# familiar with version control and/or basic tools like tar.
#
# (c) Roberto Di Cosmo <roberto@dicosmo.org> 2003-2011
# Distributend under the GPL Licence, V2 or later
#

# SUBROUTINES

#
# get the file name from a path
#
sub fnFromPath { local($path) = @_;
    $path=~s-[./]*.*[/]--;
    return $path;
}

#
# get the long path name from $_
#
sub getOutPath{
     $_=~s/^.*% [*]*FULLNAME://;
     $OUTPATH=$_; 
}

#
# switch output file
#
sub getOutpathSwitchOutFile{
     # get the new filename
     &getOutPath();
     # do the actual switch
     &switchOutFile();
}

#
# switch output file without getting the new file name ...
#
sub switchOutFile{local($ouname);
     # save the current output file on the stack
     $filestack[$filetop++] = $OUT;
     # open the new output file
     $OUT=&fnFromPath($OUTPATH);
     $outname=$OUT; chop $outname;
     if (-f $outname) {print STDERR "Warning: $outname already exists... copying to .teximport\n";
                   if (!(-e "./.teximport")) {system("/bin/mkdir .teximport");};
                   $OUT = "./.teximport/$OUT";
      };
     open($OUT,">".$OUT);
     print STDERR "x $OUT";
     # also update the filename/filepath cross reference table
     $paths{$OUT}=$OUTPATH;
}


#
# get the CWD of the original archive here ...
#
$_=<>;
if (!($_=~/% [*]*CWD is:/)) {die "not in texexpand -x format!";};
$_=~ s/% [*]*CWD is://;
chop;
$origcwd=$_;

#
# check the TOPLEVEL directive, and if it is a stdin, then output on stdout
#
$_=<>;
if (!($_=~/% [*]*TOPLEVEL/)) {die "not in texexpand -x format!";};
if (/STDIN/) {print STDERR "Warning: this file was generated from stdin... output to stdin.tex\n\n\n";
              $OUT="stdin.tex\n"; $OUTPATH="stdin.tex\n";
             &switchOutFile();}
else {
      #
      # open the main file corresponding to the first FULLNAME directive ...
      #
      $_=<>;
      if (!($_=~/% [*]*FULLNAME:/)) {die "not in texexpand -x format!";};
      $filetop=0;
      &getOutpathSwitchOutFile();
     };

#
# start the proper explode loop
#
while (<>) {
if (/% [*]*FULLNAME:/){
                       # restore characters preceding this directive
                       $tmp=$_;
                       $tmp=~s/% [*]*FULLNAME:.*$//; chop $tmp; print $OUT $tmp;
                       # get the new short file name
                       &getOutPath();
                       # restore the original input/include line before switching output file
                       $_=<>;
                       if (!(/% [*]*EXPORTBEGS:/)) {die "Not in texexpand -x format!";};
                       ~s/% [*]*EXPORTBEGS: //; chop; if (!(/\\bibliography/)) {print $OUT $_;}; #do not mess with bibliography ...
                       # now switch otput file ... 
                       &switchOutFile();
                       next;
 }
if (/% [*]*EXPORTENDS:/){#close current output file and resume previous one ...
                         $OLDOUT=$OUT;
                         close($OUT);
                         $OUT=$filestack[--$filetop];
                         #print "Closing $OLDOUT because of $_, popping to $OUT\n";
                         next;
 }
if (/% [*]*TOPLEVEL/){#close current output file and resume previous one ...
                      close($OUT);
                      if ($filetop> 1) {print STDERR "Warning: improper nesting at line \n";};
                      $filetop=0;
                      $_=<>;
                      if (!($_=~/% [*]*FULLNAME:/)) {die "not in texexpand -x format!";};
                      # now switch otput file ... 
                      print STDERR "---------\n";
                      &getOutpathSwitchOutFile();
                      next;
 }
if (/% [*]*EXPANDEDINCLUDEONLY/){ print $OUT "\\includeonly{"; next;};
if (/% [*]*ENDEXPANDEDINCLUDEONLY/){ print $OUT "}"; next;};
if (/% [*]*INCLUDEONLYFILE:/){$_=~s/% [*]*INCLUDEONLYFILE://; chop; print $OUT $_,","; next;};
if (/% [*]*EXPANDEDBIBSTYLE: /){ ~s/% [*]*EXPANDEDBIBSTYLE: //; print $OUT $_; next;};
if (/% [*]*EXPANDEDBIBFILE: /){
                               ~s/% [*]*EXPANDEDBIBFILE: //; print $OUT $_; # restore \bibliography line
			       # restore characters preceding this directive
			       $tmp=$_;
			       $tmp=~s/% [*]*EXPANDEDBIBFILE:.*$//; chop $tmp; print $OUT $tmp;
			       # restore the original input/include line before switching output file
			       $_=<>;
			       if (!(/% [*]*EXPORTBEGS:/)) {die "Not in texexpand -x format!";};
			       ~s/% [*]*EXPORTBEGS: \\bibliography{([^}]*)}/\1/; $OUTPATH=$_; # extract the name of the .bbl file
			       # now switch otput file ... 
			       &switchOutFile();
                               next;};
#
# Otherwise, output the current line
#
print $OUT $_;
};

#
# close the main file
#
close($OUT);


#
# Now check for modifications ...
#
open(MLOG,"> .mlog");
foreach (keys %paths){$original = $paths{$_}; $new=$_; chop $new; chop $original;
    $original=~s/^[ 	]*[.]/$origcwd/;
    if ($new eq $original) {print "U ",$new," *not* erased: it is the current file\n";next;};
    system("diff $new $original > /dev/null 2>&1"); $result = $?;
    if ($result == 0) {print "U ",$new," erased.\n"; unlink($new);}
    elsif ($result == 256) {print "M ", $new,"\n"; print MLOG $new, " - ", $original,"\n";}
       else {print STDERR "? ", $new, $original," is not available!\n";};
};
close(MLOG);