The Perl Toolchain Summit needs more sponsors. If your company depends on Perl, please support this very important event.
#!/usr/bin/perl
###############################################################################
#
# This file copyright (c) 2001-2011 Randy J. Ray, all rights reserved
#
# See "LICENSE AND COPYRIGHT" in the documentation for licensing and
# redistribution terms.
#
###############################################################################
#
#   Description:    Simple tool to turn a Perl routine and the support data
#                   into the simple XML representation that RPC::XML::Server
#                   understands.
#
#   Functions:      read_external
#                   write_file
#
#   Libraries:      Config
#                   Getopt::Long
#                   IO::File
#                   File::Spec
#
#   Global Consts:  $VERSION
#                   $cmd
#
#   Environment:    None.
#
###############################################################################

use 5.006001;
use strict;
use warnings;
use vars qw($USAGE $VERSION);
use subs qw(read_from_file read_from_opts read_external write_file);

use Config;
use Carp 'croak';
use Getopt::Long;
use File::Spec;

my ($cmd, %opts, $ofh, %attrs);

$VERSION = '1.15';
($cmd = $0) =~ s{.*/}{};
$USAGE = "$cmd [ --options ]

Where:

--help        Generate this message.

--name        Specifies the external (published) name of the method.
--namespace   Specify an explicit namespace for the method to be created in
--type        Specify whether this defines a PROCEDURE, a METHOD or a
                FUNCTION (case-free)
--version     Gives the version that should be attached to the method.
--hidden      Takes no value; if passed, flags the method as hidden.
--signature   Specifies one method signature. May be specified more than once.
--helptext    Provides the help string.
--helpfile    Gives the name of a file from which the help-text is read.
--code        Gives the name of the file from which to read the code.
--output      Name of the file to write the resulting XML to.

--base        If passed, this is used as a base-name from which to derive all
              the other information. The file <base>.base must exist and be
              readable. That file will provide the information for the method,
              some of which may point to other files to be read. When done, the
              output is written to <base>.xpl.

              If --base is specified, all other options are ignored, and any
              missing information (such as no signatures, etc.) will cause an
              error.
";

GetOptions(\%opts,
           qw(help
              base=s
              name=s namespace=s type=s version=s hidden signature=s@ helptext=s
              helpfile=s code=s
              output=s))
    or croak "$USAGE\n\nStopped";

if ($opts{help})
{
    print $USAGE;
    exit 0;
}

# First we start by getting all our data. Once that's all in place, then the
# generation of the file is simple.
if ($opts{base})
{
    read_from_file($opts{base});

    $ofh = "$opts{base}.xpl";
}
else
{
    read_from_opts();

    if ($opts{output})
    {
        $ofh = $opts{output};
    }
    else
    {
        $ofh = \*STDOUT;
    }
}

write_file(
    $ofh,
    {
        name      => $attrs{name},
        namespace => $attrs{namespace},
        type      => $attrs{type},
        version   => $attrs{version},
        hidden    => $attrs{hidden},
        code      => $attrs{codetxt},
        help      => $attrs{helptxt},
        sigs      => $attrs{siglist},
    }
);

exit 0;

###############################################################################
#
#   Sub Name:       read_from_file
#
#   Description:    Read method data from the given *.base file
#
#   Arguments:      NAME      IN/OUT  TYPE      DESCRIPTION
#                   $file     in      scalar    File to read, minus the ".base"
#
#   Globals:        %attrs
#
#   Returns:        Success:    void
#                   Failure:    croaks
#
###############################################################################
sub read_from_file
{
    my $file = shift;

    my ($volume, $path) = File::Spec->splitpath($file);
    $path ||= q{.};

    $attrs{type}      = 'm'; # Default the type to 'm'ethod.
    $attrs{codetxt}   = {};
    $attrs{siglist}   = [];
    $attrs{namespace} = q{};
    $attrs{hidden}    = 0;
    $attrs{version}   = q{};


    my @lines;
    if (open my $fh, '<', "$file.base")
    {
        @lines = <$fh>;
        close $fh or croak "Error closing $file.base: $!\nStopped";
    }
    else
    {
        croak "Error opening $file.base for reading: $!\nStopped";
    }

    for my $line (@lines)
    {
        chomp $line;

        # Skip blanks and comments
        next if ($line =~ /^\s*(?:#.*)?$/);

        # I'm using a horrendous if-else cascade to avoid moving the required
        # version of Perl to 5.012 just for the "when" construct.
        ## no critic (ProhibitCascadingIfElse)
        if ($line =~ /^name:\s+([\w.]+)$/i)
        {
            $attrs{name} = $1;
        }
        elsif ($line =~ /^namespace:\s+([\w.]+)$/i)
        {
            $attrs{namespace} = $1;
        }
        elsif ($line =~ /^type:\s+(\S+)$/i)
        {
            $attrs{type} = substr lc $1, 0, 1;
        }
        elsif ($line =~ /^version:\s+(\S+)$/i)
        {
            $attrs{version} = $1;
        }
        elsif ($line =~ /^signature:\s+\b(.*)$/i)
        {
            push @{$attrs{siglist}}, $1;
        }
        elsif ($line =~ /^hidden:\s+(no|yes)/i)
        {
            $attrs{hidden} = (lc $1 eq 'yes') ? 1 : 0;
        }
        elsif ($line =~ /^helpfile:\s+(.*)/i)
        {
            $attrs{helptxt} =
                read_external(File::Spec->catpath($volume, $path, $1));
        }
        elsif ($line =~ /^codefile(?:\[(.*)\])?:\s+(.*)/i)
        {
            $attrs{codetxt}->{$1 || 'perl'} =
                read_external(File::Spec->catpath($volume, $path, $2));
        }
    }
    if (! keys %{$attrs{codetxt}})
    {
        croak "Error: no code specified in $opts{base}.base, stopped";
    }
    if (! @{$attrs{siglist}})
    {
        croak "Error: no signatures found in $opts{base}.base, stopped";
    }

    return;
}

###############################################################################
#
#   Sub Name:       read_from_opts
#
#   Description:    Read method data from the command-line options
#
#   Arguments:      None.
#
#   Globals:        %opts
#                   %attrs
#
#   Returns:        Success:    void
#                   Failure:    croaks
#
###############################################################################
sub read_from_opts
{
    $attrs{siglist} = [];

    if ($opts{name})
    {
        $attrs{name} = $opts{name};
    }
    else
    {
        croak 'No name was specified for the published routine, stopped';
    }

    $attrs{namespace} = $opts{namespace} || q{};
    $attrs{type}      = $opts{type}      || 'm';
    $attrs{hidden}    = $opts{hidden}    || 0;
    $attrs{version}   = $opts{version}   || q{};

    if ($opts{signature})
    {
        for my $val (@{$opts{signature}})
        {
            $val =~ s/:/ /g;
            push @{$attrs{siglist}}, $val;
        }
    }
    else
    {
        croak "At least one signature must be specified for $attrs{name}, " .
            'stopped';
    }

    if ($opts{helptext})
    {
        $attrs{helptxt} = \"$opts{helptext}\n";
    }
    elsif ($opts{helpfile})
    {
        $attrs{helptxt} = read_external($opts{helpfile});
    }
    else
    {
        $attrs{helptxt} = \q{};
    }

    if ($opts{code})
    {
        $attrs{codetxt}->{perl} = read_external($opts{code});
    }
    else
    {
        $attrs{codetxt}->{perl} = do { local $/ = undef; <> };
    }

    return;
}

###############################################################################
#
#   Sub Name:       read_external
#
#   Description:    Simple snippet to read in an external file and return the
#                   results as a ref-to-scalar
#
#   Arguments:      NAME      IN/OUT  TYPE      DESCRIPTION
#                   $file     in      scalar    File to open and read
#
#   Returns:        Success:    scalar ref
#                   Failure:    dies
#
###############################################################################
sub read_external
{
    my $file = shift;
    my ($fh, $content);

    if (! open $fh, '<', $file)
    {
        croak "Cannot open file $file for reading: $!, stopped";
    }
    else
    {
        $content = do { local $/ = undef; <$fh> };
        close $fh or
            croak "Error closing $file: $!, stopped";
    }

    return \$content;
}

###############################################################################
#
#   Sub Name:       write_file
#
#   Description:    Write the XML file that will describe a publishable method
#
#   Arguments:      NAME      IN/OUT  TYPE      DESCRIPTION
#                   $fh       in      IO        Filehandle to write to
#                   $args     in      hashref   Hashref of arguments
#
#   Globals:        $cmd
#                   $VERSION
#
#   Environment:    None.
#
#   Returns:        void
#
###############################################################################
sub write_file
{
    my ($fh, $args) = @_;

    # Might need to open a FH here, and keep it open for a while.
    ## no critic (RequireBriefOpen)

    if (! ref $fh)
    {
        if (! open my $newfh, '>', $fh)
        {
            croak "Error opening $fh for writing: $!, stopped";
        }
        else
        {
            $fh = $newfh;
        }
    }

    my $date = scalar localtime;
    my %typemap = (
        'm' => 'method',
        p   => 'procedure',
        f   => 'function',
    );
    my $tag  = "$typemap{$args->{type}}def";

    # Armor against XML confusion
    foreach (qw(name namespace version help))
    {
        $args->{$_} =~ s/&/&amp;/g;
        $args->{$_} =~ s/</&lt;/g;
        $args->{$_} =~ s/>/&gt;/g;
    }
    for (keys %{$args->{code}})
    {
        if (($_ eq 'perl') and (index(${$args->{code}->{$_}}, ']]>') == -1) and
            (index(${$args->{code}->{$_}}, '__END__') == -1))
        {
            ${$args->{code}->{$_}} =
                "<![CDATA[\n$Config{startperl}\n${$args->{code}->{$_}}\n" .
                "__END__\n]]>";
        }
        else
        {
            ${$args->{code}->{$_}} =~ s/&/&amp;/g;
            ${$args->{code}->{$_}} =~ s/</&lt;/g;
            ${$args->{code}->{$_}} =~ s/>/&gt;/g;
        }
    }

    print {$fh} <<"EO_HDR";
<?xml version="1.0" encoding="iso-8859-1"?>
<!DOCTYPE $tag SYSTEM "rpc-method.dtd">
<!--
    Generated automatically by $cmd $VERSION on $date

    Any changes made here will be lost.
-->
<$tag>
EO_HDR

    print {$fh} "<name>$args->{name}</name>\n";
    if ($args->{namespace})
    {
        print {$fh} "<namespace>$args->{namespace}</namespace>\n";
    }
    if ($args->{version})
    {
        print {$fh} "<version>$args->{version}</version>\n";
    }
    if ($args->{hidden})
    {
        print {$fh} "<hidden />\n";
    }
    print {$fh} map { "<signature>$_</signature>\n" } @{$args->{sigs}};
    if ($args->{help})
    {
        print {$fh} "<help>\n${$args->{help}}</help>\n";
    }
    for (sort keys %{$args->{code}})
    {
        print {$fh} qq{<code language="$_">\n${$args->{code}->{$_}}</code>\n};
    }

    print {$fh} "</$tag>\n";

    return;
}

__END__

=head1 NAME

make_method - Turn Perl code into an XML description for RPC::XML::Server

=head1 SYNOPSIS

    make_method --name=system.identification --helptext='System ID string'
        --signature=string --code=ident.pl --output=ident.xpl

    make_method --base=methods/identification

=head1 DESCRIPTION

This is a simple tool to create the XML descriptive files for specifying
methods to be published by an B<RPC::XML::Server>-based server.

If a server is written such that the methods it exports (or I<publishes>) are
a part of the running code, then there is no need for this tool. However, in
cases where the server may be separate and distinct from the code (such as an
Apache-based RPC server), specifying the routines and filling in the
supporting information can be cumbersome.

One solution that the B<RPC::XML::Server> package offers is the means to load
publishable code from an external file. The file is in a simple XML dialect
that clearly delinates the externally-visible name, the method signatures, the
help text and the code itself. These files may be created manually, or this
tool may be used as an aide.

=head1 REQUIRED ARGUMENTS

There are no required arguments, but if there are not sufficient options passed
you will be told by an error message.

=head1 OPTIONS

The tool recognizes the following options:

=over 4

=item --help

Prints a short summary of the options.

=item --name=STRING

Specifies the published name of the method being encoded. This is the name by
which it will be visible to clients of the server.

=item --namespace=STRING

Specifies a namespace that the code of the method will be evaluated in,
when the XPL file is loaded by a server instance.

=item --type=STRING

Specify the type for the resulting file. "Type" here refers to whether the
container tag used in the resulting XML will specify a B<procedure> or a
B<method>. The default is B<method>. The string is treated case-independant,
and only the first character (C<m> or C<p>) is actually regarded.

=item --version=STRING

Specify a version stamp for the code routine.

=item --hidden

If this is passe, the resulting file will include a tag that tells the server
daemon to not make the routine visible through any introspection interfaces.

=item --signature=STRING [ --signature=STRING ... ]

Specify one or more signatures for the method. Signatures should be the type
names as laid out in the documentation in L<RPC::XML|RPC::XML>, with the
elements separated by a colon. You may also separate them with spaces, if you
quote the argument. This option may be specified more than once, as some
methods may have several signatures.

=item --helptext=STRING

Specify the help text for the method as a simple string on the command line.
Not suited for terribly long help strings.

=item --helpfile=FILE

Read the help text for the method from the file specified.

=item --code=FILE

Read the actual code for the routine from the file specified. If this option is
not given, the code is read from the standard input file descriptor.

=item --output=FILE

Write the resulting XML representation to the specified file. If this option
is not given, then the output goes to the standard output file descriptor.

=item --base=NAME

This is a special, "all-in-one" option. If passed, all other options are
ignored.

The value is used as the base element for reading information from a file
named B<BASE>.base. This file will contain specification of the name, version,
hidden status, signatures and other method information. Each line of the file
should look like one of the following:

=over 4

=item B<Name: I<STRING>>

Specify the name of the routine being published. If this line does not appear,
then the value of the B<--base> argument with all directory elements removed
will be used.

=item B<Version: I<STRING>>

Provide a version stamp for the function. If no line matching this pattern is
present, no version tag will be written.

=item B<Hidden: I<STRING>>

If present, I<STRING> should be either C<yes> or C<no> (case not important).
If it is C<yes>, then the method is marked to be hidden from any introspection
API.

=item B<Signature: I<STRING>>

This line may appear more than once, and is treated cumulatively. Other
options override previous values if they appear more than once. The portion
following the C<Signature:> part is taken to be a published signature for the
method, with elements separated by whitespace. Each method must have at least
one signature, so a lack of any will cause an error.

=item B<Helpfile: I<STRING>>

Specifies the file from which to read the help text. It is not an error if
no help text is specified.

=item B<Codefile: I<STRING>>

Specifies the file from which to read the code. Code is assumed to be Perl,
and will be tagged as such in the resulting file.

=item B<Codefile[lang]: I<string>>

Specifies the file from which to read code, while also identifying the
language that the code is in. This allows for the creation of a B<XPL> file
that includes multiple language implementations of the given method or
procedure.

=back

Any other lines than the above patterns are ignored.

If no code has been read, then the tool will exit with an error message.

The output is written to B<BASE>.xpl, preserving the path information so that
the resulting file is right alongside the source files. This allows constructs
such as:

    make_method --base=methods/introspection

=back

=head1 FILE FORMAT AND DTD

The file format for these published routines is a very simple XML dialect.
This is less due to XML being an ideal format than it is the availability of
the parser, given that the B<RPC::XML::Server> class will already have the
parser code in core. Writing a completely new format would not have gained
anything.

The Document Type Declaration for the format can be summarized by:

    <!ELEMENT  proceduredef (name, namespace?, version?, hidden?,
                             signature+, help?, code)>
    <!ELEMENT  methoddef  (name, namespace?, version?, hidden?,
                           signature+, help?, code)>
    <!ELEMENT  functiondef (name, namespace?, version?, hidden?,
                            signature+, help?, code)>
    <!ELEMENT  name       (#PCDATA)>
    <!ELEMENT  namespace  (#PCDATA)>
    <!ELEMENT  version    (#PCDATA)>
    <!ELEMENT  hidden     EMPTY>
    <!ELEMENT  signature  (#PCDATA)>
    <!ELEMENT  help       (#PCDATA)>
    <!ELEMENT  code       (#PCDATA)>
    <!ATTLIST  code       language (#PCDATA)>

The file C<rpc-method.dtd> that comes with the distribution has some
commentary in addition to the actual specification.

A file is (for now) limited to one definition. This is started by the one of
the opening tags C<E<lt>methoddefE<gt>>, C<E<lt>functiondefE<gt>> or
C<E<lt>proceduredefE<gt>>. This is followed by exactly one C<E<lt>nameE<gt>>
container specifying the method name, an optional version stamp, an optional
hide-from-introspection flag, one or more C<E<lt>signatureE<gt>> containers
specifying signatures, an optional C<E<lt>helpE<gt>> container with the help
text, then the C<E<lt>codeE<gt>> container with the actual program code. All
text should use entity encoding for the symbols:

    & C<&amp;> (ampersand)
    E<lt> C<&lt;>  (less-than)
    E<gt> C<&gt;>  (greater-than)

The parsing process within the server class will decode the entities. To make
things easier, the tool scans all text elements and encodes the above entities
before writing the file.

=head2 The Specification of Code

This is not I<"Programming 101">, nor is it I<"Perl for the Somewhat Dim">.
The code that is passed in via one of the C<*.xpl> files gets passed to
C<eval> with next to no modification (see below). Thus, badly-written or
malicious code can very well wreak havoc on your server. This is not the fault
of the server code. The price of the flexibility this system offers is the
responsibility on the part of the developer to ensure that the code is tested
and safe.

Code itself is treated as verbatim as possible. Some edits may occur on the
server-side, as it make the code suitable for creating an anonymous subroutine
from. The B<make_method> tool will attempt to use a C<CDATA> section to embed
the code within the XML document, so that there is no need to encode entities
or such. This allows for the resulting F<*.xpl> files to be syntax-testable
with C<perl -cx>. You can aid this by ensuring that the code does not contain
either of the two following character sequences:

    ]]>

    __DATA__

The first is the C<CDATA> terminator. If it occurs naturally in the code, it
would trigger the end-of-section in the parser. The second is the familiar
Perl token, which is inserted so that the remainder of the XML document does
not clutter up the Perl parser.

=head1 EXAMPLES

The B<RPC::XML> distribution comes with a number of default methods in a
subdirectory called (cryptically enough) C<methods>. Each of these is
expressed as a set of (C<*.base>, C<*.code>, C<*.help>) files. The Makefile.PL
file configures the resulting Makefile such that these are used to create
C<*.xpl> files using this tool, and then install them.

=head1 DIAGNOSTICS

Most problems come out in the form of error messages followed by an abrupt
exit.

=head1 EXIT STATUS

The tool exits with a status of 0 upon success, and 255 otherwise.

=head1 CAVEATS

I don't much like this approach to specifying the methods, but I liked my
other ideas even less.

=head1 BUGS

Please report any bugs or feature requests to
C<bug-rpc-xml at rt.cpan.org>, or through the web interface at
L<http://rt.cpan.org/NoAuth/ReportBug.html?Queue=RPC-XML>. I will be
notified, and then you'll automatically be notified of progress on
your bug as I make changes.

=head1 SUPPORT

=over 4

=item * RT: CPAN's request tracker

L<http://rt.cpan.org/NoAuth/Bugs.html?Dist=RPC-XML>

=item * AnnoCPAN: Annotated CPAN documentation

L<http://annocpan.org/dist/RPC-XML>

=item * CPAN Ratings

L<http://cpanratings.perl.org/d/RPC-XML>

=item * Search CPAN

L<http://search.cpan.org/dist/RPC-XML>

=item * Source code on GitHub

L<http://github.com/rjray/rpc-xml>

=back

=head1 LICENSE AND COPYRIGHT

This module and the code within are released under the terms of the Artistic
License 2.0
(http://www.opensource.org/licenses/artistic-license-2.0.php). This code may
be redistributed under either the Artistic License or the GNU Lesser General
Public License (LGPL) version 2.1
(http://www.opensource.org/licenses/lgpl-2.1.php).

=head1 SEE ALSO

L<RPC::XML|RPC::XML>, L<RPC::XML::Server|RPC::XML::Server>

=head1 CREDITS

The B<XML-RPC> standard is Copyright (c) 1998-2001, UserLand Software, Inc.
See <http://www.xmlrpc.com> for more information about the B<XML-RPC>
specification.

=head1 AUTHOR

Randy J. Ray <rjray@blackperl.com>

=cut