[p5sagit/Module-Metadata.git] / lib / Module / Metadata.pm

# -*- mode: cperl; tab-width: 8; indent-tabs-mode: nil; basic-offset: 2 -*-
# vim:ts=8:sw=2:et:sta:sts=2
package Module::Metadata;

# stolen from Module::Build::Version and ::Base - this is perl licensed code,
# copyright them.

# This module provides routines to gather information about
# perl modules (assuming this may be expanded in the distant
# parrot future to look at other types of modules).

use strict;
use vars qw($VERSION);
$VERSION = '0.36_04';
$VERSION = eval $VERSION;

use File::Spec;
use IO::File;
use Module::Metadata::Version;
BEGIN {
  if ($INC{'Log/Contextual.pm'}) {
    Log::Contextual->import('log_info');
  } else {
    *log_info = sub (&) { warn $_[0]->() };
  }
}
use File::Find qw(find);

my $V_NUM_REGEXP = qr{v?[0-9._]+};  # crudely, a v-string or decimal

my $PKG_REGEXP  = qr{   # match a package declaration
  ^[\s\{;]*             # intro chars on a line
  package               # the word 'package'
  \s+                   # whitespace
  ([\w:]+)              # a package name
  \s*                   # optional whitespace
  ($V_NUM_REGEXP)?        # optional version number
  \s*                   # optional whitesapce
  ;                     # semicolon line terminator
}x;

my $VARNAME_REGEXP = qr{ # match fully-qualified VERSION name
  ([\$*])         # sigil - $ or *
  (
    (             # optional leading package name
      (?:::|\')?  # possibly starting like just :: (Ì  la $::VERSION)
      (?:\w+(?:::|\'))*  # Foo::Bar:: ...
    )?
    VERSION
  )\b
}x;

my $VERS_REGEXP = qr{ # match a VERSION definition
  (?:
    \(\s*$VARNAME_REGEXP\s*\) # with parens
  |
    $VARNAME_REGEXP           # without parens
  )
  \s*
  =[^=~]  # = but not ==, nor =~
}x;


sub new_from_file {
  my $class    = shift;
  my $filename = File::Spec->rel2abs( shift );

  return undef unless defined( $filename ) && -f $filename;
  return $class->_init(undef, $filename, @_);
}

sub new_from_module {
  my $class   = shift;
  my $module  = shift;
  my %props   = @_;

  $props{inc} ||= \@INC;
  my $filename = $class->find_module_by_name( $module, $props{inc} );
  return undef unless defined( $filename ) && -f $filename;
  return $class->_init($module, $filename, %props);
}

{
  
  my $compare_versions = sub {
    my ($v1, $op, $v2) = @_;
    $v1 = Module::Metadata::Version->new($v1)
      unless UNIVERSAL::isa($v1,'Module::Metadata::Version');
  
    my $eval_str = "\$v1 $op \$v2";
    my $result   = eval $eval_str;
    log_info { "error comparing versions: '$eval_str' $@" } if $@;
  
    return $result;
  };

  my $normalize_version = sub {
    my ($version) = @_;
    if ( $version =~ /[=<>!,]/ ) { # logic, not just version
      # take as is without modification
    }
    elsif ( ref $version eq 'version' ||
            ref $version eq 'Module::Metadata::Version' ) { # version objects
      $version = $version->is_qv ? $version->normal : $version->stringify;
    }
    elsif ( $version =~ /^[^v][^.]*\.[^.]+\./ ) { # no leading v, multiple dots
      # normalize string tuples without "v": "1.2.3" -> "v1.2.3"
      $version = "v$version";
    }
    else {
      # leave alone
    }
    return $version;
  };

  # separate out some of the conflict resolution logic

  my $resolve_module_versions = sub {
    my $packages = shift;
  
    my( $file, $version );
    my $err = '';
      foreach my $p ( @$packages ) {
        if ( defined( $p->{version} ) ) {
  	if ( defined( $version ) ) {
   	  if ( $compare_versions->( $version, '!=', $p->{version} ) ) {
  	    $err .= "  $p->{file} ($p->{version})\n";
  	  } else {
  	    # same version declared multiple times, ignore
  	  }
  	} else {
  	  $file    = $p->{file};
  	  $version = $p->{version};
  	}
        }
        $file ||= $p->{file} if defined( $p->{file} );
      }
  
    if ( $err ) {
      $err = "  $file ($version)\n" . $err;
    }
  
    my %result = (
      file    => $file,
      version => $version,
      err     => $err
    );
  
    return \%result;
  };

  sub package_versions_from_directory {
    my ( $class, $dir, $files ) = @_;

    my @files;

    if ( $files ) {
      @files = @$files;
    } else {
      find( {
        wanted => sub {
          push @files, $_ if -f $_ && /\.pm$/;
        },
        no_chdir => 1,
      }, $dir );
    }

    # First, we enumerate all packages & versions,
    # separating into primary & alternative candidates
    my( %prime, %alt );
    foreach my $file (@files) {
      my $mapped_filename = File::Spec->abs2rel( $file, $dir );
      my @path = split( /\//, $mapped_filename );
      (my $prime_package = join( '::', @path )) =~ s/\.pm$//;
  
      my $pm_info = $class->new_from_file( $file );
  
      foreach my $package ( $pm_info->packages_inside ) {
        next if $package eq 'main';  # main can appear numerous times, ignore
        next if $package eq 'DB';    # special debugging package, ignore
        next if grep /^_/, split( /::/, $package ); # private package, ignore
  
        my $version = $pm_info->version( $package );
  
        if ( $package eq $prime_package ) {
          if ( exists( $prime{$package} ) ) {
            # M::B::ModuleInfo will handle this conflict
            die "Unexpected conflict in '$package'; multiple versions found.\n";
          } else {
            $prime{$package}{file} = $mapped_filename;
            $prime{$package}{version} = $version if defined( $version );
          }
        } else {
          push( @{$alt{$package}}, {
                                    file    => $mapped_filename,
                                    version => $version,
                                   } );
        }
      }
    }
  
    # Then we iterate over all the packages found above, identifying conflicts
    # and selecting the "best" candidate for recording the file & version
    # for each package.
    foreach my $package ( keys( %alt ) ) {
      my $result = $resolve_module_versions->( $alt{$package} );
  
      if ( exists( $prime{$package} ) ) { # primary package selected
  
        if ( $result->{err} ) {
  	# Use the selected primary package, but there are conflicting
  	# errors among multiple alternative packages that need to be
  	# reported
          log_info {
  	    "Found conflicting versions for package '$package'\n" .
  	    "  $prime{$package}{file} ($prime{$package}{version})\n" .
  	    $result->{err}
          };
  
        } elsif ( defined( $result->{version} ) ) {
  	# There is a primary package selected, and exactly one
  	# alternative package
  
  	if ( exists( $prime{$package}{version} ) &&
  	     defined( $prime{$package}{version} ) ) {
  	  # Unless the version of the primary package agrees with the
  	  # version of the alternative package, report a conflict
  	  if ( $compare_versions->(
                 $prime{$package}{version}, '!=', $result->{version}
               )
             ) {

            log_info {
              "Found conflicting versions for package '$package'\n" .
  	      "  $prime{$package}{file} ($prime{$package}{version})\n" .
  	      "  $result->{file} ($result->{version})\n"
            };
  	  }
  
  	} else {
  	  # The prime package selected has no version so, we choose to
  	  # use any alternative package that does have a version
  	  $prime{$package}{file}    = $result->{file};
  	  $prime{$package}{version} = $result->{version};
  	}
  
        } else {
  	# no alt package found with a version, but we have a prime
  	# package so we use it whether it has a version or not
        }
  
      } else { # No primary package was selected, use the best alternative
  
        if ( $result->{err} ) {
          log_info {
            "Found conflicting versions for package '$package'\n" .
  	    $result->{err}
          };
        }
  
        # Despite possible conflicting versions, we choose to record
        # something rather than nothing
        $prime{$package}{file}    = $result->{file};
        $prime{$package}{version} = $result->{version}
  	  if defined( $result->{version} );
      }
    }
  
    # Normalize versions.  Can't use exists() here because of bug in YAML::Node.
    # XXX "bug in YAML::Node" comment seems irrelvant -- dagolden, 2009-05-18
    for (grep defined $_->{version}, values %prime) {
      $_->{version} = $normalize_version->( $_->{version} );
    }
  
    return \%prime;
  }
} 
  

sub _init {
  my $class    = shift;
  my $module   = shift;
  my $filename = shift;
  my %props = @_;

  my( %valid_props, @valid_props );
  @valid_props = qw( collect_pod inc );
  @valid_props{@valid_props} = delete( @props{@valid_props} );
  warn "Unknown properties: @{[keys %props]}\n" if scalar( %props );

  my %data = (
    module       => $module,
    filename     => $filename,
    version      => undef,
    packages     => [],
    versions     => {},
    pod          => {},
    pod_headings => [],
    collect_pod  => 0,

    %valid_props,
  );

  my $self = bless(\%data, $class);

  $self->_parse_file();

  unless($self->{module} and length($self->{module})) {
    my ($v, $d, $f) = File::Spec->splitpath($self->{filename});
    if($f =~ /\.pm$/) {
      $f =~ s/\..+$//;
      my @candidates = grep /$f$/, @{$self->{packages}};
      $self->{module} = shift(@candidates); # punt
    }
    else {
      if(grep /main/, @{$self->{packages}}) {
        $self->{module} = 'main';
      }
      else {
        $self->{module} = $self->{packages}[0] || '';
      }
    }
  }

  $self->{version} = $self->{versions}{$self->{module}}
      if defined( $self->{module} );

  return $self;
}

# class method
sub _do_find_module {
  my $class   = shift;
  my $module  = shift || die 'find_module_by_name() requires a package name';
  my $dirs    = shift || \@INC;

  my $file = File::Spec->catfile(split( /::/, $module));
  foreach my $dir ( @$dirs ) {
    my $testfile = File::Spec->catfile($dir, $file);
    return [ File::Spec->rel2abs( $testfile ), $dir ]
	if -e $testfile and !-d _;  # For stuff like ExtUtils::xsubpp
    return [ File::Spec->rel2abs( "$testfile.pm" ), $dir ]
	if -e "$testfile.pm";
  }
  return;
}

# class method
sub find_module_by_name {
  my $found = shift()->_do_find_module(@_) or return;
  return $found->[0];
}

# class method
sub find_module_dir_by_name {
  my $found = shift()->_do_find_module(@_) or return;
  return $found->[1];
}


# given a line of perl code, attempt to parse it if it looks like a
# $VERSION assignment, returning sigil, full name, & package name
sub _parse_version_expression {
  my $self = shift;
  my $line = shift;

  my( $sig, $var, $pkg );
  if ( $line =~ $VERS_REGEXP ) {
    ( $sig, $var, $pkg ) = $2 ? ( $1, $2, $3 ) : ( $4, $5, $6 );
    if ( $pkg ) {
      $pkg = ($pkg eq '::') ? 'main' : $pkg;
      $pkg =~ s/::$//;
    }
  }

  return ( $sig, $var, $pkg );
}

sub _parse_file {
  my $self = shift;

  my $filename = $self->{filename};
  my $fh = IO::File->new( $filename )
    or die( "Can't open '$filename': $!" );

  $self->_parse_fh($fh);
}

sub _parse_fh {
  my ($self, $fh) = @_;

  my( $in_pod, $seen_end, $need_vers ) = ( 0, 0, 0 );
  my( @pkgs, %vers, %pod, @pod );
  my $pkg = 'main';
  my $pod_sect = '';
  my $pod_data = '';

  while (defined( my $line = <$fh> )) {
    my $line_num = $.;

    chomp( $line );
    next if $line =~ /^\s*#/;

    $in_pod = ($line =~ /^=(?!cut)/) ? 1 : ($line =~ /^=cut/) ? 0 : $in_pod;

    # Would be nice if we could also check $in_string or something too
    last if !$in_pod && $line =~ /^__(?:DATA|END)__$/;

    if ( $in_pod || $line =~ /^=cut/ ) {

      if ( $line =~ /^=head\d\s+(.+)\s*$/ ) {
	push( @pod, $1 );
	if ( $self->{collect_pod} && length( $pod_data ) ) {
          $pod{$pod_sect} = $pod_data;
          $pod_data = '';
        }
	$pod_sect = $1;


      } elsif ( $self->{collect_pod} ) {
	$pod_data .= "$line\n";

      }

    } else {

      $pod_sect = '';
      $pod_data = '';

      # parse $line to see if it's a $VERSION declaration
      my( $vers_sig, $vers_fullname, $vers_pkg ) =
	  $self->_parse_version_expression( $line );

      if ( $line =~ $PKG_REGEXP ) {
        $pkg = $1;
        push( @pkgs, $pkg ) unless grep( $pkg eq $_, @pkgs );
        $vers{$pkg} = (defined $2 ? $2 : undef)  unless exists( $vers{$pkg} );
        $need_vers = defined $2 ? 0 : 1;

      # VERSION defined with full package spec, i.e. $Module::VERSION
      } elsif ( $vers_fullname && $vers_pkg ) {
	push( @pkgs, $vers_pkg ) unless grep( $vers_pkg eq $_, @pkgs );
	$need_vers = 0 if $vers_pkg eq $pkg;

	unless ( defined $vers{$vers_pkg} && length $vers{$vers_pkg} ) {
	  $vers{$vers_pkg} =
	    $self->_evaluate_version_line( $vers_sig, $vers_fullname, $line );
	} else {
	  # Warn unless the user is using the "$VERSION = eval
	  # $VERSION" idiom (though there are probably other idioms
	  # that we should watch out for...)
	  warn <<"EOM" unless $line =~ /=\s*eval/;
Package '$vers_pkg' already declared with version '$vers{$vers_pkg}',
ignoring subsequent declaration on line $line_num.
EOM
	}

      # first non-comment line in undeclared package main is VERSION
      } elsif ( !exists($vers{main}) && $pkg eq 'main' && $vers_fullname ) {
	$need_vers = 0;
	my $v =
	  $self->_evaluate_version_line( $vers_sig, $vers_fullname, $line );
	$vers{$pkg} = $v;
	push( @pkgs, 'main' );

      # first non-comment line in undeclared package defines package main
      } elsif ( !exists($vers{main}) && $pkg eq 'main' && $line =~ /\w+/ ) {
	$need_vers = 1;
	$vers{main} = '';
	push( @pkgs, 'main' );

      # only keep if this is the first $VERSION seen
      } elsif ( $vers_fullname && $need_vers ) {
	$need_vers = 0;
	my $v =
	  $self->_evaluate_version_line( $vers_sig, $vers_fullname, $line );


	unless ( defined $vers{$pkg} && length $vers{$pkg} ) {
	  $vers{$pkg} = $v;
	} else {
	  warn <<"EOM";
Package '$pkg' already declared with version '$vers{$pkg}'
ignoring new version '$v' on line $line_num.
EOM
	}

      }

    }

  }

  if ( $self->{collect_pod} && length($pod_data) ) {
    $pod{$pod_sect} = $pod_data;
  }

  $self->{versions} = \%vers;
  $self->{packages} = \@pkgs;
  $self->{pod} = \%pod;
  $self->{pod_headings} = \@pod;
}

{
my $pn = 0;
sub _evaluate_version_line {
  my $self = shift;
  my( $sigil, $var, $line ) = @_;

  # Some of this code came from the ExtUtils:: hierarchy.

  # We compile into $vsub because 'use version' would cause
  # compiletime/runtime issues with local()
  my $vsub;
  $pn++; # everybody gets their own package
  my $eval = qq{BEGIN { q#  Hide from _packages_inside()
    #; package Module::Metadata::_version::p$pn;
    use Module::Metadata::Version;
    no strict;

    local $sigil$var;
    \$$var=undef;
      \$vsub = sub {
        $line;
        \$$var
      };
  }};

  local $^W;
  # Try to get the $VERSION
  eval $eval;
  # some modules say $VERSION = $Foo::Bar::VERSION, but Foo::Bar isn't
  # installed, so we need to hunt in ./lib for it
  if ( $@ =~ /Can't locate/ && -d 'lib' ) {
    local @INC = ('lib',@INC);
    eval $eval;
  }
  warn "Error evaling version line '$eval' in $self->{filename}: $@\n"
    if $@;
  (ref($vsub) eq 'CODE') or
    die "failed to build version sub for $self->{filename}";
  my $result = eval { $vsub->() };
  die "Could not get version from $self->{filename} by executing:\n$eval\n\nThe fatal error was: $@\n"
    if $@;

  # Activestate apparently creates custom versions like '1.23_45_01', which
  # cause M::B::Version to think it's an invalid alpha.  So check for that
  # and strip them
  my $num_dots = () = $result =~ m{\.}g;
  my $num_unders = () = $result =~ m{_}g;
  if ( substr($result,0,1) ne 'v' && $num_dots < 2 && $num_unders > 1 ) {
    $result =~ s{_}{}g;
  }

  # Bless it into our own version class
  eval { $result = Module::Metadata::Version->new($result) };
  die "Version '$result' from $self->{filename} does not appear to be valid:\n$eval\n\nThe fatal error was: $@\n"
    if $@;

  return $result;
}
}


############################################################

# accessors
sub name            { $_[0]->{module}           }

sub filename        { $_[0]->{filename}         }
sub packages_inside { @{$_[0]->{packages}}      }
sub pod_inside      { @{$_[0]->{pod_headings}}  }
sub contains_pod    { $#{$_[0]->{pod_headings}} }

sub version {
    my $self = shift;
    my $mod  = shift || $self->{module};
    my $vers;
    if ( defined( $mod ) && length( $mod ) &&
	 exists( $self->{versions}{$mod} ) ) {
	return $self->{versions}{$mod};
    } else {
	return undef;
    }
}

sub pod {
    my $self = shift;
    my $sect = shift;
    if ( defined( $sect ) && length( $sect ) &&
	 exists( $self->{pod}{$sect} ) ) {
	return $self->{pod}{$sect};
    } else {
	return undef;
    }
}

1;

=head1 NAME

Module::Metadata - Gather package and POD information from perl module files

=head1 DESCRIPTION

=over 4

=item new_from_file($filename, collect_pod => 1)

Construct a C<ModuleInfo> object given the path to a file. Takes an optional
argument C<collect_pod> which is a boolean that determines whether
POD data is collected and stored for reference. POD data is not
collected by default. POD headings are always collected.

=item new_from_module($module, collect_pod => 1, inc => \@dirs)

Construct a C<ModuleInfo> object given a module or package name. In addition
to accepting the C<collect_pod> argument as described above, this
method accepts a C<inc> argument which is a reference to an array of
of directories to search for the module. If none are given, the
default is @INC.

=item name()

Returns the name of the package represented by this module. If there
are more than one packages, it makes a best guess based on the
filename. If it's a script (i.e. not a *.pm) the package name is
'main'.

=item version($package)

Returns the version as defined by the $VERSION variable for the
package as returned by the C<name> method if no arguments are
given. If given the name of a package it will attempt to return the
version of that package if it is specified in the file.

=item filename()

Returns the absolute path to the file.

=item packages_inside()

Returns a list of packages.

=item pod_inside()

Returns a list of POD sections.

=item contains_pod()

Returns true if there is any POD in the file.

=item pod($section)

Returns the POD data in the given section.

=item find_module_by_name($module, \@dirs)

Returns the path to a module given the module or package name. A list
of directories can be passed in as an optional parameter, otherwise
@INC is searched.

Can be called as either an object or a class method.

=item find_module_dir_by_name($module, \@dirs)

Returns the entry in C<@dirs> (or C<@INC> by default) that contains
the module C<$module>. A list of directories can be passed in as an
optional parameter, otherwise @INC is searched.

Can be called as either an object or a class method.

=item package_versions_from_directory($dir, \@files?)

Scans C<$dir> for .pm files (unless C<@files> is given, in which case looks
for those files in C<$dir> - and reads each file for packages and versions,
returning a hashref of the form:

  {
    'Package::Name' => {
      version => '0.123',
      file => 'Package/Name.pm'
    },
    'OtherPackage::Name' => ...
  }

=item log_info (internal)

Used internally to perform logging; imported from Log::Contextual if
Log::Contextual has already been loaded, otherwise simply calls warn.

=back

=head1 AUTHOR

Ken Williams <kwilliams@cpan.org>, Randy W. Sims <RandyS@ThePierianSpring.org>

Released as Module::Metadata by Matt S Trout (mst) <mst@shadowcat.co.uk> with
assistance from David Golden (xdg) <dagolden@cpan.org>

=head1 COPYRIGHT

Copyright (c) 2001-2006 Ken Williams.  All rights reserved.

This library is free software; you can redistribute it and/or
modify it under the same terms as Perl itself.

=head1 SEE ALSO

perl(1), L<Module::Build::ModuleInfo>(3)

=cut
Commit	Line	Data
5ac756c6	1	# -- mode: cperl; tab-width: 8; indent-tabs-mode: nil; basic-offset: 2 --
	2	# vim:ts=8:sw=2:et:sta:sts=2
	3	package Module::Metadata;
	4
	5	# stolen from Module::Build::Version and ::Base - this is perl licensed code,
	6	# copyright them.
	7
	8	# This module provides routines to gather information about
	9	# perl modules (assuming this may be expanded in the distant
	10	# parrot future to look at other types of modules).
	11
	12	use strict;
	13	use vars qw($VERSION);
	14	$VERSION = '0.36_04';
	15	$VERSION = eval $VERSION;
	16
	17	use File::Spec;
	18	use IO::File;
	19	use Module::Metadata::Version;
3db27017	20	BEGIN {
	21	if ($INC{'Log/Contextual.pm'}) {
	22	Log::Contextual->import('log_info');
	23	} else {
e6ddd765	24	*log_info = sub (&) { warn $_[0]->() };
3db27017	25	}
3db27017	26	}
5ac756c6	27	use File::Find qw(find);
	28
	29	my $V_NUM_REGEXP = qr{v?[0-9._]+}; # crudely, a v-string or decimal
	30
	31	my $PKG_REGEXP = qr{ # match a package declaration
	32	^[\s\{;]* # intro chars on a line
	33	package # the word 'package'
	34	\s+ # whitespace
	35	([\w:]+) # a package name
	36	\s* # optional whitespace
	37	($V_NUM_REGEXP)? # optional version number
	38	\s* # optional whitesapce
	39	; # semicolon line terminator
	40	}x;
	41
	42	my $VARNAME_REGEXP = qr{ # match fully-qualified VERSION name
	43	([\$]) # sigil - $ or
	44	(
	45	( # optional leading package name
	46	(?:::\|\')? # possibly starting like just :: (Ì la $::VERSION)
	47	(?:\w+(?:::\|\'))* # Foo::Bar:: ...
	48	)?
	49	VERSION
	50	)\b
	51	}x;
	52
	53	my $VERS_REGEXP = qr{ # match a VERSION definition
	54	(?:
	55	\(\s$VARNAME_REGEXP\s\) # with parens
	56	\|
	57	$VARNAME_REGEXP # without parens
	58	)
	59	\s*
	60	=[^=~] # = but not ==, nor =~
	61	}x;
	62
	63
	64	sub new_from_file {
	65	my $class = shift;
	66	my $filename = File::Spec->rel2abs( shift );
	67
	68	return undef unless defined( $filename ) && -f $filename;
	69	return $class->_init(undef, $filename, @_);
	70	}
	71
	72	sub new_from_module {
	73	my $class = shift;
	74	my $module = shift;
	75	my %props = @_;
	76
	77	$props{inc} \|\|= \@INC;
	78	my $filename = $class->find_module_by_name( $module, $props{inc} );
	79	return undef unless defined( $filename ) && -f $filename;
	80	return $class->_init($module, $filename, %props);
	81	}
	82
	83	{
	84
	85	my $compare_versions = sub {
	86	my ($v1, $op, $v2) = @_;
	87	$v1 = Module::Metadata::Version->new($v1)
	88	unless UNIVERSAL::isa($v1,'Module::Metadata::Version');
	89
	90	my $eval_str = "\$v1 $op \$v2";
91	my $result = eval $eval_str;
92	log_info { "error comparing versions: '$eval_str' $@" } if $@;
93
94	return $result;
95	};
96
97	my $normalize_version = sub {
98	my ($version) = @_;
99	if ( $version =~ /[=<>!,]/ ) { # logic, not just version
100	# take as is without modification
101	}
102	elsif ( ref $version eq 'version' \|\|
103	ref $version eq 'Module::Metadata::Version' ) { # version objects
104	$version = $version->is_qv ? $version->normal : $version->stringify;
105	}
106	elsif ( $version =~ /^[^v][^.]*\.[^.]+\./ ) { # no leading v, multiple dots
107	# normalize string tuples without "v": "1.2.3" -> "v1.2.3"
108	$version = "v$version";
109	}
110	else {
111	# leave alone
112	}
113	return $version;
114	};
115
116	# separate out some of the conflict resolution logic
117
118	my $resolve_module_versions = sub {
119	my $packages = shift;
120
121	my( $file, $version );
122	my $err = '';
123	foreach my $p ( @$packages ) {
124	if ( defined( $p->{version} ) ) {
125	if ( defined( $version ) ) {
126	if ( $compare_versions->( $version, '!=', $p->{version} ) ) {
127	$err .= " $p->{file} ($p->{version})\n";
128	} else {
129	# same version declared multiple times, ignore
130	}
131	} else {
132	$file = $p->{file};
133	$version = $p->{version};
134	}
135	}
136	$file \|\|= $p->{file} if defined( $p->{file} );
137	}
138
139	if ( $err ) {
140	$err = " $file ($version)\n" . $err;
141	}
142
143	my %result = (
144	file => $file,
145	version => $version,
146	err => $err
147	);
148
149	return \%result;
150	};
151
152	sub package_versions_from_directory {
153	my ( $class, $dir, $files ) = @_;
154
155	my @files;
156
157	if ( $files ) {
158	@files = @$files;
159	} else {
160	find( {
161	wanted => sub {
162	push @files, $_ if -f $_ && /\.pm$/;
163	},
164	no_chdir => 1,
165	}, $dir );
166	}
167
168	# First, we enumerate all packages & versions,
169	# separating into primary & alternative candidates
170	my( %prime, %alt );
171	foreach my $file (@files) {
172	my $mapped_filename = File::Spec->abs2rel( $file, $dir );
173	my @path = split( /\//, $mapped_filename );
174	(my $prime_package = join( '::', @path )) =~ s/\.pm$//;
175
176	my $pm_info = $class->new_from_file( $file );
177
178	foreach my $package ( $pm_info->packages_inside ) {
179	next if $package eq 'main'; # main can appear numerous times, ignore
180	next if $package eq 'DB'; # special debugging package, ignore
181	next if grep /^_/, split( /::/, $package ); # private package, ignore
182
183	my $version = $pm_info->version( $package );
184
185	if ( $package eq $prime_package ) {
186	if ( exists( $prime{$package} ) ) {
187	# M::B::ModuleInfo will handle this conflict
188	die "Unexpected conflict in '$package'; multiple versions found.\n";
189	} else {
190	$prime{$package}{file} = $mapped_filename;
191	$prime{$package}{version} = $version if defined( $version );
192	}
193	} else {
194	push( @{$alt{$package}}, {
195	file => $mapped_filename,
196	version => $version,
197	} );
198	}
199	}
200	}
201
202	# Then we iterate over all the packages found above, identifying conflicts
203	# and selecting the "best" candidate for recording the file & version
204	# for each package.
205	foreach my $package ( keys( %alt ) ) {
206	my $result = $resolve_module_versions->( $alt{$package} );
207
208	if ( exists( $prime{$package} ) ) { # primary package selected
209
210	if ( $result->{err} ) {
211	# Use the selected primary package, but there are conflicting
212	# errors among multiple alternative packages that need to be
213	# reported
214	log_info {
215	"Found conflicting versions for package '$package'\n" .
216	" $prime{$package}{file} ($prime{$package}{version})\n" .
217	$result->{err}
218	};
219
220	} elsif ( defined( $result->{version} ) ) {
221	# There is a primary package selected, and exactly one
222	# alternative package
223
224	if ( exists( $prime{$package}{version} ) &&
225	defined( $prime{$package}{version} ) ) {
226	# Unless the version of the primary package agrees with the
227	# version of the alternative package, report a conflict
228	if ( $compare_versions->(
229	$prime{$package}{version}, '!=', $result->{version}
230	)
231	) {
232
233	log_info {
234	"Found conflicting versions for package '$package'\n" .
235	" $prime{$package}{file} ($prime{$package}{version})\n" .
236	" $result->{file} ($result->{version})\n"
237	};
238	}
239
240	} else {
241	# The prime package selected has no version so, we choose to
242	# use any alternative package that does have a version
243	$prime{$package}{file} = $result->{file};
244	$prime{$package}{version} = $result->{version};
245	}
246
247	} else {
248	# no alt package found with a version, but we have a prime
249	# package so we use it whether it has a version or not
250	}
251
252	} else { # No primary package was selected, use the best alternative
253
254	if ( $result->{err} ) {
255	log_info {
256	"Found conflicting versions for package '$package'\n" .
257	$result->{err}
258	};
259	}
260
261	# Despite possible conflicting versions, we choose to record
262	# something rather than nothing
263	$prime{$package}{file} = $result->{file};
264	$prime{$package}{version} = $result->{version}
265	if defined( $result->{version} );
266	}
267	}
268
269	# Normalize versions. Can't use exists() here because of bug in YAML::Node.
270	# XXX "bug in YAML::Node" comment seems irrelvant -- dagolden, 2009-05-18
271	for (grep defined $_->{version}, values %prime) {
272	$_->{version} = $normalize_version->( $_->{version} );
273	}
274
275	return \%prime;
276	}
277	}
278
279
280	sub _init {
281	my $class = shift;
282	my $module = shift;
283	my $filename = shift;
284	my %props = @_;
285
286	my( %valid_props, @valid_props );
287	@valid_props = qw( collect_pod inc );
288	@valid_props{@valid_props} = delete( @props{@valid_props} );
289	warn "Unknown properties: @{[keys %props]}\n" if scalar( %props );
290
291	my %data = (
292	module => $module,
293	filename => $filename,
294	version => undef,
295	packages => [],
296	versions => {},
297	pod => {},
298	pod_headings => [],
299	collect_pod => 0,
300
301	%valid_props,
302	);
303
304	my $self = bless(\%data, $class);
305
306	$self->_parse_file();
307
308	unless($self->{module} and length($self->{module})) {
309	my ($v, $d, $f) = File::Spec->splitpath($self->{filename});
310	if($f =~ /\.pm$/) {
311	$f =~ s/\..+$//;
312	my @candidates = grep /$f$/, @{$self->{packages}};
313	$self->{module} = shift(@candidates); # punt
314	}
315	else {
316	if(grep /main/, @{$self->{packages}}) {
317	$self->{module} = 'main';
318	}
319	else {
320	$self->{module} = $self->{packages}[0] \|\| '';
321	}
322	}
323	}
324
325	$self->{version} = $self->{versions}{$self->{module}}
326	if defined( $self->{module} );
327
328	return $self;
329	}
330
331	# class method
332	sub _do_find_module {
333	my $class = shift;
334	my $module = shift \|\| die 'find_module_by_name() requires a package name';
335	my $dirs = shift \|\| \@INC;
336
337	my $file = File::Spec->catfile(split( /::/, $module));
338	foreach my $dir ( @$dirs ) {
339	my $testfile = File::Spec->catfile($dir, $file);
340	return [ File::Spec->rel2abs( $testfile ), $dir ]
341	if -e $testfile and !-d _; # For stuff like ExtUtils::xsubpp
342	return [ File::Spec->rel2abs( "$testfile.pm" ), $dir ]
343	if -e "$testfile.pm";
344	}
345	return;
346	}
347
348	# class method
349	sub find_module_by_name {
350	my $found = shift()->_do_find_module(@_) or return;
351	return $found->[0];
352	}
353
354	# class method
355	sub find_module_dir_by_name {
356	my $found = shift()->_do_find_module(@_) or return;
357	return $found->[1];
358	}
359
360
361	# given a line of perl code, attempt to parse it if it looks like a
362	# $VERSION assignment, returning sigil, full name, & package name
363	sub _parse_version_expression {
364	my $self = shift;
365	my $line = shift;
366
367	my( $sig, $var, $pkg );
368	if ( $line =~ $VERS_REGEXP ) {
369	( $sig, $var, $pkg ) = $2 ? ( $1, $2, $3 ) : ( $4, $5, $6 );
370	if ( $pkg ) {
371	$pkg = ($pkg eq '::') ? 'main' : $pkg;
372	$pkg =~ s/::$//;
373	}
374	}
375
376	return ( $sig, $var, $pkg );
377	}
378
379	sub _parse_file {
380	my $self = shift;
381
382	my $filename = $self->{filename};
383	my $fh = IO::File->new( $filename )
384	or die( "Can't open '$filename': $!" );
385
386	$self->_parse_fh($fh);
387	}
388
389	sub _parse_fh {
390	my ($self, $fh) = @_;
391
392	my( $in_pod, $seen_end, $need_vers ) = ( 0, 0, 0 );
393	my( @pkgs, %vers, %pod, @pod );
394	my $pkg = 'main';
395	my $pod_sect = '';
396	my $pod_data = '';
397
398	while (defined( my $line = <$fh> )) {
399	my $line_num = $.;
400
401	chomp( $line );
402	next if $line =~ /^\s*#/;
403
404	$in_pod = ($line =~ /^=(?!cut)/) ? 1 : ($line =~ /^=cut/) ? 0 : $in_pod;
405
406	# Would be nice if we could also check $in_string or something too
407	last if !$in_pod && $line =~ /^__(?:DATA\|END)__$/;
408
409	if ( $in_pod \|\| $line =~ /^=cut/ ) {
410
411	if ( $line =~ /^=head\d\s+(.+)\s*$/ ) {
412	push( @pod, $1 );
413	if ( $self->{collect_pod} && length( $pod_data ) ) {
414	$pod{$pod_sect} = $pod_data;
415	$pod_data = '';
416	}
417	$pod_sect = $1;
418
419
420	} elsif ( $self->{collect_pod} ) {
421	$pod_data .= "$line\n";
422
423	}
424
425	} else {
426
427	$pod_sect = '';
428	$pod_data = '';
429
430	# parse $line to see if it's a $VERSION declaration
431	my( $vers_sig, $vers_fullname, $vers_pkg ) =
432	$self->_parse_version_expression( $line );
433
434	if ( $line =~ $PKG_REGEXP ) {
435	$pkg = $1;
436	push( @pkgs, $pkg ) unless grep( $pkg eq $_, @pkgs );
437	$vers{$pkg} = (defined $2 ? $2 : undef) unless exists( $vers{$pkg} );
438	$need_vers = defined $2 ? 0 : 1;
439
440	# VERSION defined with full package spec, i.e. $Module::VERSION
441	} elsif ( $vers_fullname && $vers_pkg ) {
442	push( @pkgs, $vers_pkg ) unless grep( $vers_pkg eq $_, @pkgs );
443	$need_vers = 0 if $vers_pkg eq $pkg;
444
445	unless ( defined $vers{$vers_pkg} && length $vers{$vers_pkg} ) {
446	$vers{$vers_pkg} =
447	$self->_evaluate_version_line( $vers_sig, $vers_fullname, $line );
448	} else {
449	# Warn unless the user is using the "$VERSION = eval
450	# $VERSION" idiom (though there are probably other idioms
451	# that we should watch out for...)
452	warn <<"EOM" unless $line =~ /=\s*eval/;
453	Package '$vers_pkg' already declared with version '$vers{$vers_pkg}',
454	ignoring subsequent declaration on line $line_num.
455	EOM
456	}
457
458	# first non-comment line in undeclared package main is VERSION
459	} elsif ( !exists($vers{main}) && $pkg eq 'main' && $vers_fullname ) {
460	$need_vers = 0;
461	my $v =
462	$self->_evaluate_version_line( $vers_sig, $vers_fullname, $line );
463	$vers{$pkg} = $v;
464	push( @pkgs, 'main' );
465
466	# first non-comment line in undeclared package defines package main
467	} elsif ( !exists($vers{main}) && $pkg eq 'main' && $line =~ /\w+/ ) {
468	$need_vers = 1;
469	$vers{main} = '';
470	push( @pkgs, 'main' );
471
472	# only keep if this is the first $VERSION seen
473	} elsif ( $vers_fullname && $need_vers ) {
474	$need_vers = 0;
475	my $v =
476	$self->_evaluate_version_line( $vers_sig, $vers_fullname, $line );
477
478
479	unless ( defined $vers{$pkg} && length $vers{$pkg} ) {
480	$vers{$pkg} = $v;
481	} else {
482	warn <<"EOM";
483	Package '$pkg' already declared with version '$vers{$pkg}'
484	ignoring new version '$v' on line $line_num.
485	EOM
486	}
487
488	}
489
490	}
491
492	}
493
494	if ( $self->{collect_pod} && length($pod_data) ) {
495	$pod{$pod_sect} = $pod_data;
496	}
497
498	$self->{versions} = \%vers;
499	$self->{packages} = \@pkgs;
500	$self->{pod} = \%pod;
501	$self->{pod_headings} = \@pod;
502	}
503
504	{
505	my $pn = 0;
506	sub _evaluate_version_line {
507	my $self = shift;
508	my( $sigil, $var, $line ) = @_;
509
510	# Some of this code came from the ExtUtils:: hierarchy.
511
512	# We compile into $vsub because 'use version' would cause
513	# compiletime/runtime issues with local()
514	my $vsub;
515	$pn++; # everybody gets their own package
516	my $eval = qq{BEGIN { q# Hide from _packages_inside()
517	#; package Module::Metadata::_version::p$pn;
518	use Module::Metadata::Version;
519	no strict;
520
521	local $sigil$var;
522	\$$var=undef;
523	\$vsub = sub {
524	$line;
525	\$$var
526	};
527	}};
528
529	local $^W;
530	# Try to get the $VERSION
531	eval $eval;
532	# some modules say $VERSION = $Foo::Bar::VERSION, but Foo::Bar isn't
533	# installed, so we need to hunt in ./lib for it
534	if ( $@ =~ /Can't locate/ && -d 'lib' ) {
535	local @INC = ('lib',@INC);
536	eval $eval;
537	}
538	warn "Error evaling version line '$eval' in $self->{filename}: $@\n"
539	if $@;
540	(ref($vsub) eq 'CODE') or
541	die "failed to build version sub for $self->{filename}";
542	my $result = eval { $vsub->() };
543	die "Could not get version from $self->{filename} by executing:\n$eval\n\nThe fatal error was: $@\n"
544	if $@;
545
546	# Activestate apparently creates custom versions like '1.23_45_01', which
547	# cause M::B::Version to think it's an invalid alpha. So check for that
548	# and strip them
549	my $num_dots = () = $result =~ m{\.}g;
550	my $num_unders = () = $result =~ m{_}g;
551	if ( substr($result,0,1) ne 'v' && $num_dots < 2 && $num_unders > 1 ) {
552	$result =~ s{_}{}g;
553	}
554
555	# Bless it into our own version class
556	eval { $result = Module::Metadata::Version->new($result) };
557	die "Version '$result' from $self->{filename} does not appear to be valid:\n$eval\n\nThe fatal error was: $@\n"
558	if $@;
559
560	return $result;
561	}
562	}
563
564
565	############################################################
566
567	# accessors
568	sub name { $_[0]->{module} }
569
570	sub filename { $_[0]->{filename} }
571	sub packages_inside { @{$_[0]->{packages}} }
572	sub pod_inside { @{$_[0]->{pod_headings}} }
573	sub contains_pod { $#{$_[0]->{pod_headings}} }
574
575	sub version {
576	my $self = shift;
577	my $mod = shift \|\| $self->{module};
578	my $vers;
579	if ( defined( $mod ) && length( $mod ) &&
580	exists( $self->{versions}{$mod} ) ) {
581	return $self->{versions}{$mod};
582	} else {
583	return undef;
584	}
585	}
586
587	sub pod {
588	my $self = shift;
589	my $sect = shift;
590	if ( defined( $sect ) && length( $sect ) &&
591	exists( $self->{pod}{$sect} ) ) {
592	return $self->{pod}{$sect};
593	} else {
594	return undef;
595	}
596	}
597
598	1;
599
5ac756c6	600	=head1 NAME
5ac756c6	601
2c11e51d	602	Module::Metadata - Gather package and POD information from perl module files
5ac756c6	603
	604	=head1 DESCRIPTION
	605
	606	=over 4
	607
	608	=item new_from_file($filename, collect_pod => 1)
	609
	610	Construct a C<ModuleInfo> object given the path to a file. Takes an optional
	611	argument C<collect_pod> which is a boolean that determines whether
	612	POD data is collected and stored for reference. POD data is not
	613	collected by default. POD headings are always collected.
	614
	615	=item new_from_module($module, collect_pod => 1, inc => \@dirs)
	616
	617	Construct a C<ModuleInfo> object given a module or package name. In addition
	618	to accepting the C<collect_pod> argument as described above, this
	619	method accepts a C<inc> argument which is a reference to an array of
	620	of directories to search for the module. If none are given, the
	621	default is @INC.
	622
	623	=item name()
	624
	625	Returns the name of the package represented by this module. If there
	626	are more than one packages, it makes a best guess based on the
	627	filename. If it's a script (i.e. not a *.pm) the package name is
	628	'main'.
	629
	630	=item version($package)
	631
	632	Returns the version as defined by the $VERSION variable for the
	633	package as returned by the C<name> method if no arguments are
	634	given. If given the name of a package it will attempt to return the
	635	version of that package if it is specified in the file.
	636
	637	=item filename()
	638
	639	Returns the absolute path to the file.
	640
	641	=item packages_inside()
	642
	643	Returns a list of packages.
	644
	645	=item pod_inside()
	646
	647	Returns a list of POD sections.
	648
	649	=item contains_pod()
	650
	651	Returns true if there is any POD in the file.
	652
	653	=item pod($section)
	654
	655	Returns the POD data in the given section.
	656
	657	=item find_module_by_name($module, \@dirs)
	658
	659	Returns the path to a module given the module or package name. A list
	660	of directories can be passed in as an optional parameter, otherwise
	661	@INC is searched.
	662
	663	Can be called as either an object or a class method.
	664
	665	=item find_module_dir_by_name($module, \@dirs)
	666
667	Returns the entry in C<@dirs> (or C<@INC> by default) that contains
668	the module C<$module>. A list of directories can be passed in as an
669	optional parameter, otherwise @INC is searched.
670
671	Can be called as either an object or a class method.
672
2c11e51d	673	=item package_versions_from_directory($dir, \@files?)
	674
	675	Scans C<$dir> for .pm files (unless C<@files> is given, in which case looks
	676	for those files in C<$dir> - and reads each file for packages and versions,
	677	returning a hashref of the form:
5ac756c6	678
2c11e51d	679	{
	680	'Package::Name' => {
	681	version => '0.123',
	682	file => 'Package/Name.pm'
	683	},
	684	'OtherPackage::Name' => ...
	685	}
	686
	687	=item log_info (internal)
	688
	689	Used internally to perform logging; imported from Log::Contextual if
	690	Log::Contextual has already been loaded, otherwise simply calls warn.
	691
	692	=back
5ac756c6	693
	694	=head1 AUTHOR
	695
	696	Ken Williams <kwilliams@cpan.org>, Randy W. Sims <RandyS@ThePierianSpring.org>
	697
2c11e51d	698	Released as Module::Metadata by Matt S Trout (mst) <mst@shadowcat.co.uk> with
2c11e51d	699	assistance from David Golden (xdg) <dagolden@cpan.org>
5ac756c6	700
	701	=head1 COPYRIGHT
	702
	703	Copyright (c) 2001-2006 Ken Williams. All rights reserved.
	704
	705	This library is free software; you can redistribute it and/or
	706	modify it under the same terms as Perl itself.
	707
5ac756c6	708	=head1 SEE ALSO
5ac756c6	709
2c11e51d	710	perl(1), L<Module::Build::ModuleInfo>(3)
5ac756c6	711
	712	=cut
	713