NAME ExtUtils::PerlPP - A Perl Preprocessor SYNOPSIS use ExtUtils::PerlPP; my $config = { 'version' => $VERSION, 'driver' => $DRIVER }; # The long and winding road ... my $self = ExtUtils::PerlPP->new(); $self->{'in_fh'} = IO::File->new('file.PL', 'r'); $self->{'out_fh'} = IO::File->new('file', 'w'); $self->{'config'} = $self->parse(); # And now a short cut for the same: ppp('file.PL', 'file', $config); DESCRIPTION Perl's installation suite, ExtUtils::MakeMaker, contains a mechanism for installing preparsed files, so-called *PL* files: If the MakeMaker utility detects files with the extension `.PL' then these files are executed by *make*, usually creating a file of the same name, except the `.PL' extension. Writing these PL files is usually always the same, for example a typical `.PL' file might look like this: my $script = <<'SCRIPT'; ... # True file following here SCRIPT # Modify variable $script, depending on configuration, local # site or whatever ... if (!open(FILE, ">file") || !(print FILE $script) || !close(FILE)) { die "Cannot write file: $!"; } But in essence, what else is this than a Perl preprocessor? Traditionally you have to write such a Perl preprocessor for yourself all the time, although I have found that they always do the same, for example: - Fix defaults, for example installation paths. - Including or excluding code sections. It is a matter of taste whether one likes to see if ($] < 5.003) { # Thirty lines of code following here ... } else { # A single line of code ... } when already using Perl 5.005. I don't. This module is dedicated to simplify such tasks. In short, you can use it like this: Create a new preprocessor You start with creating an instance of *ExtUtils::PerlPP* by calling the *new* constructor: my $ppp = ExtUtils::PerlPP->new(%attr); The constructor accepts a list of attributes, including the following: in_fh The input file, any kind of IO object, for example an instance of IO::File or IO::Scalar. More general: It can be any object that offers a *getline* method. A scalar value (to be distinguished from an IO::Scalar instance!) will be interpreted as a file name that the method opens for you. out_fh The output file; another IO object or any other object that offers a *print* method. A scalar value is accepted as output file name. config A hash ref of preprocessor variables. In other words $ppp->{'config'}->{'var'} = 1; is what `-Dvar=val' is for the C preprocessor. Similarly you can compare delete $ppp->{'config'}; with `-Uvar'. See the section on "Macro replacements" below. Unlike C, variables may be arbitrarily complex, in particular you can use hash or array refs as values. Surprisingly you may pass a scalar value again: In that case the file of the same name evaluated and the result is used as a configuration hash. In other words $ppp->{'config'} = "myapp.cfg"; is similar to $ppp->{'config'} = do "myapp.cfg"; Such config files can easily be created using the *Data::Dumper* module. the Data::Dumper(3) manpage. no_config_default If a variable name is used, but no such attribute is present in the *config* hash, then by default the variable is looked up in the `$Config' from the *Config* module. This behaviour is suppressed, if you set *no_config_default* to a TRUE value. the Config(3) manpage. no_makedirs By default directories are created silently if required. For example, if you pass a value of `/usr/local/foo/bar' as output file and only `/usr/local' exists, then the subdirectory `foo' will be created. The option *no_makedirs* suppresses this behaviour. Running the preprocessor This is done by executing $ppp->parse(); A Perl exception will be thrown in case of errors, thus the complete use might look like this: eval { $ppp->parse(); }; if ($@) { print "An error occurred: $@\n" } Using the frontend Most applications won't call the *new* or *parse* methods directly, but rather do a use ExtUtils::PerlPP; ppp('infile', 'outfile', 'configfile'); This is equivalent to my $parser = ExtUtils::PerlPP->new('in_fh' => 'infile', 'out_fh' => 'outfile', 'config' => 'configfile'); $parser->parse(); In order to be easily used within Makefiles, the ppp frontend can read from @ARGV. That is, you can use the module like this: perl -MExtUtils::PerlPP -e ppp from the commandline. Macro replacements The primary use of preprocessor variables (aka attributes of `$ppp- '{'config'}>) is replacing patterns in the stream written to the output file. With `$c = $ppp-'{'config'}> in mind the typical patterns and their replacements are: ~~a~~ $c->{'a'} ~~b~~ $c->{'b'} ~~a->b~~ $c->{'a'}->{'b'} ~~a->e~~ $c->{'a'}->{'e'} ~~a->1~~ $c->{'a'}->[1] ~~a->1->b~~ $c->{'a'}->[1]->{'b'} I hope the idea is obvious. Real world examples might be: my $config_file = "~~etc_dir~~/configuration"; my $VERSION = "~~version~~"; Preprocessor variables need not be scalar values: If a variable contains a code ref, then the module will execute &$var($ppp, $text); and replace the pattern with the result. `$text' is the pattern being replaced, for example, if `$ppp-'{'config'}->{'bar'}> has the value `\&foo', then `~~bar~~' will be replaced with the result of foo($ppp, "bar"); Arguments are not yet supported. Creating macros When talking about code refs, we need a possibility to create them. The best possibility is creating them within the input file, as in ~&foo&~ my($self, $text) = @_; $text x 2; ~&&~ This example is mainly equivalent to $ppp->{'config'}->{'foo'} = sub { my($self, $text) = @_; $text x 2; }; The `~&var&~' definition must start at the beginning of a line, much like the C preprocessor. The end pattern ~&&~ may appear at any point, but the remaining line will be ignored. Conditional output The next application of a preprocessor is conditional output, as in an #ifdef var ... #endif segment. This can be done with ~#if#~ ... ~#elsif#~ ... ~#else#~ ... ~#endif#~ `' is handled as follows: First it is subject to the usual pattern replacements and then it is evaluated as a Perl expression returning a TRUE or FALSE value. Examples: ~#if#~ "~~a~~" is TRUE, if and only if $ppp->{'config'}->{'a'} is TRUE. Currently conditionals must start at the beginning of a line and expressions must not exceed a single line. Nesting conditions is possible. Embedding into MakeMaker For using the preprocessor from within MakeMaker, I propose the following: First of all you create a config file from within Makefile.PL. For example the *libnet* suite creates a file `libnet.cfg' and the *SNMP::Monitor* and *Cisco::Conf* modules create a file `configuration'. The *Data::Dumper* module will aid you in that task. the Data::Dumper(3) manpage. Then you add the following to your Makefile.PL, I assume the name `myapp.cnf' for the config file: package MY; sub processPL { my($self) = shift; return "" unless $self->{PL_FILES}; my(@m, $from, $to); foreach $from (sort keys %{$self->{PL_FILES}}) { $to = $self->{PL_FILES}->{$from}; push @m, " all :: $self->{PL_FILES}->{$plfile} $self->{NOECHO}\$(NOOP) $self->{PL_FILES}->{$plfile} :: $plfile \$(PERL) -I\$(INST_ARCHLIB) -I\$(INST_LIB) \ -I\$(PERL_ARCHLIB) -I\$(PERL_LIB) \ -MExtUtils::PerlPP -e 'ppp($from, $to, \"myapp.cnf\")' "; } join "", @m; } Next you create your template files under their usual names, but add an extension `.PL'. The MakeMaker utility will automatically detect these files for you and add appropriate rules to the Makefile it generates. AUTHOR AND COPYRIGHT This module is Copyright (C) 1998 by Jochen Wiedmann Am Eisteich 9 72555 Metzingen Germany Email: joe@ispsoft.de Phone: +49 7123 14887 All rights reserved. You may distribute this module under the terms of either the GNU General Public License or the Artistic License, as specified in the Perl README file. SEE ALSO the ExtUtils::MakeMaker(3) manpage, the Data::Dumper(3) manpage