# parrot future to look at other types of modules).
use strict;
-use vars qw($VERSION);
-$VERSION = '1.000007';
+use warnings;
+
+our $VERSION = '1.000019';
$VERSION = eval $VERSION;
+use Carp qw/croak/;
use File::Spec;
use IO::File;
use version 0.87;
my $V_NUM_REGEXP = qr{v?[0-9._]+}; # crudely, a v-string or decimal
+my $PKG_FIRST_WORD_REGEXP = qr{ # the FIRST word in a package name
+ [a-zA-Z_] # the first word CANNOT start with a digit
+ (?:
+ [\w']? # can contain letters, digits, _, or ticks
+ \w # But, NO multi-ticks or trailing ticks
+ )*
+}x;
+
+my $PKG_ADDL_WORD_REGEXP = qr{ # the 2nd+ word in a package name
+ \w # the 2nd+ word CAN start with digits
+ (?:
+ [\w']? # and can contain letters or ticks
+ \w # But, NO multi-ticks or trailing ticks
+ )*
+}x;
+
+my $PKG_NAME_REGEXP = qr{ # match a package name
+ (?: :: )? # a pkg name can start with aristotle
+ $PKG_FIRST_WORD_REGEXP # a package word
+ (?:
+ (?: :: )+ ### aristotle (allow one or many times)
+ $PKG_ADDL_WORD_REGEXP ### a package word
+ )* # ^ zero, one or many times
+ (?:
+ :: # allow trailing aristotle
+ )?
+}x;
+
my $PKG_REGEXP = qr{ # match a package declaration
^[\s\{;]* # intro chars on a line
package # the word 'package'
\s+ # whitespace
- ([\w:]+) # a package name
+ ($PKG_NAME_REGEXP) # a package name
\s* # optional whitespace
($V_NUM_REGEXP)? # optional version number
\s* # optional whitesapce
$VARNAME_REGEXP # without parens
)
\s*
- =[^=~] # = but not ==, nor =~
+ =[^=~>] # = but not ==, nor =~, nor =>
}x;
-
sub new_from_file {
my $class = shift;
my $filename = File::Spec->rel2abs( shift );
}
{
-
+
my $compare_versions = sub {
my ($v1, $op, $v2) = @_;
$v1 = version->new($v1)
unless UNIVERSAL::isa($v1,'version');
-
+
my $eval_str = "\$v1 $op \$v2";
my $result = eval $eval_str;
log_info { "error comparing versions: '$eval_str' $@" } if $@;
-
+
return $result;
};
my $resolve_module_versions = sub {
my $packages = shift;
-
+
my( $file, $version );
my $err = '';
foreach my $p ( @$packages ) {
}
$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 provides {
my $class = shift;
- die "provides() requires key/value pairs \n" if @_ % 2;
+ croak "provides() requires key/value pairs \n" if @_ % 2;
my %args = @_;
- die "provides() takes only one of 'dir' or 'files'\n"
+ croak "provides() takes only one of 'dir' or 'files'\n"
if $args{dir} && $args{files};
+ croak "provides() requires a 'version' argument"
+ unless defined $args{version};
+
+ croak "provides() does not support version '$args{version}' metadata"
+ unless grep { $args{version} eq $_ } qw/1.4 2/;
+
$args{prefix} = 'lib' unless defined $args{prefix};
my $p;
$p = $class->package_versions_from_directory($args{dir});
}
else {
- die "provides() requires 'files' to be an array reference\n"
+ croak "provides() requires 'files' to be an array reference\n"
unless ref $args{files} eq 'ARRAY';
$p = $class->package_versions_from_directory($args{files});
}
# separating into primary & alternative candidates
my( %prime, %alt );
foreach my $file (@files) {
- my $mapped_filename = File::Spec->abs2rel( $file, $dir );
+ my $mapped_filename = File::Spec::Unix->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 );
-
+
+ $prime_package = $package if lc($prime_package) eq lc($package);
if ( $package eq $prime_package ) {
if ( exists( $prime{$package} ) ) {
- die "Unexpected conflict in '$package'; multiple versions found.\n";
+ croak "Unexpected conflict in '$package'; multiple versions found.\n";
} else {
+ $mapped_filename = "$package.pm" if lc("$package.pm") eq lc($mapped_filename);
$prime{$package}{file} = $mapped_filename;
$prime{$package}{version} = $version if defined( $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
" $prime{$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
" $result->{file} ($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};
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
+ # XXX "bug in YAML::Node" comment seems irrelevant -- dagolden, 2009-05-18
for (grep defined $_->{version}, values %prime) {
$_->{version} = $normalize_version->( $_->{version} );
}
-
+
return \%prime;
}
-}
-
+}
+
sub _init {
my $class = shift;
# class method
sub _do_find_module {
my $class = shift;
- my $module = shift || die 'find_module_by_name() requires a package name';
+ my $module = shift || croak 'find_module_by_name() requires a package name';
my $dirs = shift || \@INC;
my $file = File::Spec->catfile(split( /::/, $module));
my $line = shift;
my( $sig, $var, $pkg );
- if ( $line =~ $VERS_REGEXP ) {
+ if ( $line =~ /$VERS_REGEXP/o ) {
( $sig, $var, $pkg ) = $2 ? ( $1, $2, $3 ) : ( $4, $5, $6 );
if ( $pkg ) {
$pkg = ($pkg eq '::') ? 'main' : $pkg;
my $filename = $self->{filename};
my $fh = IO::File->new( $filename )
- or die( "Can't open '$filename': $!" );
+ or croak( "Can't open '$filename': $!" );
+
+ $self->_handle_bom($fh, $filename);
$self->_parse_fh($fh);
}
+# Look for a UTF-8/UTF-16BE/UTF-16LE BOM at the beginning of the stream.
+# If there's one, then skip it and set the :encoding layer appropriately.
+sub _handle_bom {
+ my ($self, $fh, $filename) = @_;
+
+ my $pos = $fh->getpos;
+ return unless defined $pos;
+
+ my $buf = ' ' x 2;
+ my $count = $fh->read( $buf, length $buf );
+ return unless defined $count and $count >= 2;
+
+ my $encoding;
+ if ( $buf eq "\x{FE}\x{FF}" ) {
+ $encoding = 'UTF-16BE';
+ } elsif ( $buf eq "\x{FF}\x{FE}" ) {
+ $encoding = 'UTF-16LE';
+ } elsif ( $buf eq "\x{EF}\x{BB}" ) {
+ $buf = ' ';
+ $count = $fh->read( $buf, length $buf );
+ if ( defined $count and $count >= 1 and $buf eq "\x{BF}" ) {
+ $encoding = 'UTF-8';
+ }
+ }
+
+ if ( defined $encoding ) {
+ if ( "$]" >= 5.008 ) {
+ # $fh->binmode requires perl 5.10
+ binmode( $fh, ":encoding($encoding)" );
+ }
+ } else {
+ $fh->setpos($pos)
+ or croak( sprintf "Can't reset position to the top of '$filename'" );
+ }
+
+ return $encoding;
+}
+
sub _parse_fh {
my ($self, $fh) = @_;
my $pkg = 'main';
my $pod_sect = '';
my $pod_data = '';
+ my $in_end = 0;
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 && $line =~ /^__(?:DATA|END)__$/;
+ # From toke.c : any line that begins by "=X", where X is an alphabetic
+ # character, introduces a POD segment.
+ my $is_cut;
+ if ( $line =~ /^=([a-zA-Z].*)/ ) {
+ my $cmd = $1;
+ # Then it goes back to Perl code for "=cutX" where X is a non-alphabetic
+ # character (which includes the newline, but here we chomped it away).
+ $is_cut = $cmd =~ /^cut(?:[^a-zA-Z]|$)/;
+ $in_pod = !$is_cut;
+ }
- if ( $in_pod || $line =~ /^=cut/ ) {
+ if ( $in_pod ) {
- if ( $line =~ /^=head\d\s+(.+)\s*$/ ) {
+ if ( $line =~ /^=head[1-4]\s+(.+)\s*$/ ) {
push( @pod, $1 );
if ( $self->{collect_pod} && length( $pod_data ) ) {
$pod{$pod_sect} = $pod_data;
}
$pod_sect = $1;
-
} elsif ( $self->{collect_pod} ) {
$pod_data .= "$line\n";
}
- } else {
+ } elsif ( $is_cut ) {
+ if ( $self->{collect_pod} && length( $pod_data ) ) {
+ $pod{$pod_sect} = $pod_data;
+ $pod_data = '';
+ }
$pod_sect = '';
- $pod_data = '';
+
+ } else {
+
+ # Skip after __END__
+ next if $in_end;
+
+ # Skip comments in code
+ next if $line =~ /^\s*#/;
+
+ # Would be nice if we could also check $in_string or something too
+ if ($line eq '__END__') {
+ $in_end++;
+ next;
+ }
+ last if $line eq '__DATA__';
# parse $line to see if it's a $VERSION declaration
my( $vers_sig, $vers_fullname, $vers_pkg ) =
- $self->_parse_version_expression( $line );
+ ($line =~ /VERSION/)
+ ? $self->_parse_version_expression( $line )
+ : ();
- if ( $line =~ $PKG_REGEXP ) {
+ if ( $line =~ /$PKG_REGEXP/o ) {
$pkg = $1;
push( @pkgs, $pkg ) unless grep( $pkg eq $_, @pkgs );
- $vers{$pkg} = (defined $2 ? $2 : undef) unless exists( $vers{$pkg} );
+ $vers{$pkg} = $2 unless exists( $vers{$pkg} );
$need_vers = defined $2 ? 0 : 1;
# VERSION defined with full package spec, i.e. $Module::VERSION
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
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
}
}
# compiletime/runtime issues with local()
my $vsub;
$pn++; # everybody gets their own package
- my $eval = qq{BEGIN { q# Hide from _packages_inside()
+ my $eval = qq{BEGIN { my \$dummy = q# Hide from _packages_inside()
#; package Module::Metadata::_version::p$pn;
use version;
no strict;
+ no warnings;
\$vsub = sub {
local $sigil$var;
};
}};
+ $eval = $1 if $eval =~ m{^(.+)}s;
+
local $^W;
# Try to get the $VERSION
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}";
+ croak "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"
+ croak "Could not get version from $self->{filename} by executing:\n$eval\n\nThe fatal error was: $@\n"
if $@;
# Upgrade it into a version object
my $version = eval { _dwim_version($result) };
- die "Version '$result' from $self->{filename} does not appear to be valid:\n$eval\n\nThe fatal error was: $@\n"
+ croak "Version '$result' from $self->{filename} does not appear to be valid:\n$eval\n\nThe fatal error was: $@\n"
unless defined $version; # "0" is OK!
return $version;
last if defined $version;
}
- die $error unless defined $version;
+ croak $error unless defined $version;
return $version;
}
############################################################
# accessors
-sub name { $_[0]->{module} }
+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 filename { $_[0]->{filename} }
+sub packages_inside { @{$_[0]->{packages}} }
+sub pod_inside { @{$_[0]->{pod_headings}} }
+sub contains_pod { 0+@{$_[0]->{pod_headings}} }
sub version {
my $self = shift;
}
}
+sub is_indexable {
+ my ($self, $package) = @_;
+
+ my @indexable_packages = grep { $_ ne 'main' } $self->packages_inside;
+
+ # check for specific package, if provided
+ return !! grep { $_ eq $package } @indexable_packages if $package;
+
+ # otherwise, check for any indexable packages at all
+ return !! @indexable_packages;
+}
+
1;
=head1 NAME
my $version = $info->version;
# CPAN META 'provides' field for .pm files in a directory
- my $provides = Module::Metadata->provides(dir => 'lib');
+ my $provides = Module::Metadata->provides(
+ dir => 'lib', version => 2
+ );
=head1 DESCRIPTION
-This module provides a standard way to gather metadata about a .pm file
-without executing unsafe code.
+This module provides a standard way to gather metadata about a .pm file through
+(mostly) static analysis and (some) code execution. When determining the
+version of a module, the C<$VERSION> assignment is C<eval>ed, as is traditional
+in the CPAN toolchain.
=head1 USAGE
=item C<< new_from_file($filename, collect_pod => 1) >>
-Construct a C<Module::Metadata> object given the path to a file. Takes an
-optional argument C<collect_pod> which is a boolean that determines whether POD
-data is collected and stored for reference. POD data is not collected by
-default. POD headings are always collected. Returns undef if the filename
-does not exist.
+Constructs a C<Module::Metadata> object given the path to a file. Returns
+undef if the filename does not exist.
+
+C<collect_pod> is a optional boolean argument that determines whether POD
+data is collected and stored for reference. POD data is not collected by
+default. POD headings are always collected.
+
+If the file begins by an UTF-8, UTF-16BE or UTF-16LE byte-order mark, then
+it is skipped before processing, and the content of the file is also decoded
+appropriately starting from perl 5.8.
=item C<< new_from_handle($handle, $filename, collect_pod => 1) >>
This works just like C<new_from_file>, except that a handle can be provided
-as the first argument. Note that there is no validation to confirm that the
-handle is a handle or something that can act like one. Passing something that
-isn't a handle will cause a exception when trying to read from it. The
-C<filename> argument is mandatory or undef will be returned.
+as the first argument.
+
+Note that there is no validation to confirm that the handle is a handle or
+something that can act like one. Passing something that isn't a handle will
+cause a exception when trying to read from it. The C<filename> argument is
+mandatory or undef will be returned.
+
+You are responsible for setting the decoding layers on C<$handle> if
+required.
=item C<< new_from_module($module, collect_pod => 1, inc => \@dirs) >>
-Construct a C<Module::Metadata> object given a module or package name. In addition
-to accepting the C<collect_pod> argument as described above, this
-method accepts a C<inc> argument which is a reference to an array of
-of directories to search for the module. If none are given, the
-default is @INC. Returns undef if the module cannot be found.
+Constructs a C<Module::Metadata> object given a module or package name.
+Returns undef if the module cannot be found.
+
+In addition to accepting the C<collect_pod> argument as described above,
+this method accepts a C<inc> argument which is a reference to an array of
+directories to search for the module. If none are given, the default is
+@INC.
+
+If the file that contains the module begins by an UTF-8, UTF-16BE or
+UTF-16LE byte-order mark, then it is skipped before processing, and the
+content of the file is also decoded appropriately starting from perl 5.8.
=item C<< find_module_by_name($module, \@dirs) >>
=over
+=item version B<(required)>
+
+Specifies which version of the L<CPAN::Meta::Spec> should be used as
+the format of the C<provides> output. Currently only '1.4' and '2'
+are supported (and their format is identical). This may change in
+the future as the definition of C<provides> changes.
+
+The C<version> option is required. If it is omitted or if
+an unsupported version is given, then C<provides> will throw an error.
+
=item dir
Directory to search recursively for F<.pm> files. May not be specified with
=item C<< 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
+is more than one package, it makes a best guess based on the
filename. If it's a script (i.e. not a *.pm) the package name is
'main'.
Returns a list of packages. Note: this is a raw list of packages
discovered (or assumed, in the case of C<main>). It is not
filtered for C<DB>, C<main> or private packages the way the
-C<provides> method does.
+C<provides> method does. Invalid package names are not returned,
+for example "Foo:Bar". Strange but valid package names are
+returned, for example "Foo::Bar::", and are left up to the caller
+on how to handle.
=item C<< pod_inside() >>
Returns the POD data in the given section.
+=item C<< is_indexable($package) >> or C<< is_indexable() >>
+
+Returns a boolean indicating whether the package (if provided) or any package
+(otherwise) is eligible for indexing by PAUSE, the Perl Authors Upload Server.
+Note This only checks for valid C<package> declarations, and does not take any
+ownership information into account.
+
=back
=head1 AUTHOR
Released as Module::Metadata by Matt S Trout (mst) <mst@shadowcat.co.uk> with
assistance from David Golden (xdg) <dagolden@cpan.org>.
-=head1 COPYRIGHT
+=head1 COPYRIGHT & LICENSE
Original code Copyright (c) 2001-2011 Ken Williams.
Additional code Copyright (c) 2010-2011 Matt Trout and David Golden.
projects / p5sagit/Module-Metadata.git / blobdiff