X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=lib%2FModule%2FMetadata.pm;h=d9924777bc59457cad92978b795c2291383ff169;hb=dd5a4b10a5e1a7e13fc31c6001ebb21628642c6a;hp=514dd1c16ea6593d39442dcbc6b38b43ab55ccf1;hpb=6290f67cbac16f68d0c2b536fa9d5457230e2e41;p=p5sagit%2FModule-Metadata.git diff --git a/lib/Module/Metadata.pm b/lib/Module/Metadata.pm index 514dd1c..d992477 100644 --- a/lib/Module/Metadata.pm +++ b/lib/Module/Metadata.pm @@ -11,9 +11,10 @@ package Module::Metadata; use strict; use vars qw($VERSION); -$VERSION = '1.000005'; +$VERSION = '1.000011'; $VERSION = eval $VERSION; +use Carp qw/croak/; use File::Spec; use IO::File; use version 0.87; @@ -36,7 +37,7 @@ my $PKG_REGEXP = qr{ # match a package declaration \s* # optional whitespace ($V_NUM_REGEXP)? # optional version number \s* # optional whitesapce - ; # semicolon line terminator + [;\{] # semicolon line terminator or block start (since 5.16) }x; my $VARNAME_REGEXP = qr{ # match fully-qualified VERSION name @@ -60,7 +61,6 @@ my $VERS_REGEXP = qr{ # match a VERSION definition =[^=~] # = but not ==, nor =~ }x; - sub new_from_file { my $class = shift; my $filename = File::Spec->rel2abs( shift ); @@ -93,16 +93,16 @@ sub new_from_module { } { - + 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; }; @@ -128,7 +128,7 @@ sub new_from_module { my $resolve_module_versions = sub { my $packages = shift; - + my( $file, $version ); my $err = ''; foreach my $p ( @$packages ) { @@ -146,20 +146,58 @@ sub new_from_module { } $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; + + croak "provides() requires key/value pairs \n" if @_ % 2; + my %args = @_; + + 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; + if ( $args{dir} ) { + $p = $class->package_versions_from_directory($args{dir}); + } + else { + croak "provides() requires 'files' to be an array reference\n" + unless ref $args{files} eq 'ARRAY'; + $p = $class->package_versions_from_directory($args{files}); + } + + # Now, fix up files with prefix + if ( length $args{prefix} ) { # check in case disabled with q{} + $args{prefix} =~ s{/$}{}; + for my $v ( values %$p ) { + $v->{file} = "$args{prefix}/$v->{file}"; + } + } + + return $p + } + sub package_versions_from_directory { my ( $class, $dir, $files ) = @_; @@ -180,23 +218,25 @@ sub new_from_module { # 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 ); } @@ -208,15 +248,15 @@ sub new_from_module { } } } - + # 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 @@ -226,11 +266,11 @@ sub new_from_module { " $prime{$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 @@ -246,28 +286,28 @@ sub new_from_module { " $result->{file} ($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}; @@ -275,17 +315,17 @@ sub new_from_module { 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; @@ -347,7 +387,7 @@ sub _init { # 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)); @@ -381,7 +421,7 @@ sub _parse_version_expression { 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; @@ -397,11 +437,51 @@ sub _parse_file { 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) = @_; @@ -415,16 +495,21 @@ sub _parse_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; @@ -432,25 +517,37 @@ sub _parse_fh { } $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 comments in code + next if $line =~ /^\s*#/; + + # Would be nice if we could also check $in_string or something too + last if $line =~ /^__(?:DATA|END)__$/; # 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 @@ -461,14 +558,6 @@ sub _parse_fh { 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 @@ -494,11 +583,6 @@ EOM 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 $line_num. -EOM } } @@ -554,15 +638,15 @@ sub _evaluate_version_line { warn "Error evaling version line '$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; @@ -620,7 +704,7 @@ sub _evaluate_version_line { last if defined $version; } - die $error unless defined $version; + croak $error unless defined $version; return $version; } @@ -673,9 +757,10 @@ Module::Metadata - Gather package and POD information from perl module files my $info = Module::Metadata->new_from_file( $file ); my $version = $info->version; - # information about a directory full of .pm files - my $provides = - Module::Metadata->package_versions_from_directory('lib'); + # CPAN META 'provides' field for .pm files in a directory + my $provides = Module::Metadata->provides( + dir => 'lib', version => 2 + ); =head1 DESCRIPTION @@ -690,27 +775,43 @@ without executing unsafe code. =item C<< new_from_file($filename, collect_pod => 1) >> -Construct a C object given the path to a file. Takes an -optional argument C 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 object given the path to a file. Returns +undef if the filename does not exist. + +C 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, $filename, collect_pod => 1) >> This works just like C, except that a handle can be provided -as the first argument. Note that there is no validation to confirm that the -handle is a handle or something that can act like one. Passing something that -isn't a handle will cause a exception when trying to read from it. The -C argument is mandatory or undef will be returned. +as the first argument. + +Note that there is no validation to confirm that the handle is a handle or +something that can act like one. Passing something that isn't a handle will +cause a exception when trying to read from it. The C argument is +mandatory or undef will be returned. + +You are responsible for setting the decoding layers on C<$handle> if +required. =item C<< new_from_module($module, collect_pod => 1, inc => \@dirs) >> -Construct a C object given a module or package name. In addition -to accepting the C argument as described above, this -method accepts a C argument which is a reference to an array of -of directories to search for the module. If none are given, the -default is @INC. Returns undef if the module cannot be found. +Constructs a C object given a module or package name. +Returns undef if the module cannot be found. + +In addition to accepting the C argument as described above, +this method accepts a C argument which is a reference to an array of +directories to search for the module. If none are given, the default is +@INC. + +If the file that contains the module begins by an UTF-8, UTF-16BE or +UTF-16LE byte-order mark, then it is skipped before processing, and the +content of the file is also decoded appropriately starting from perl 5.8. =item C<< find_module_by_name($module, \@dirs) >> @@ -728,6 +829,55 @@ optional parameter, otherwise @INC is searched. Can be called as either an object or a class method. +=item C<< provides( %options ) >> + +This is a convenience wrapper around C +to generate a CPAN META C data structure. It takes key/value +pairs. Valid option keys include: + +=over + +=item version B<(required)> + +Specifies which version of the L should be used as +the format of the C 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 changes. + +The C option is required. If it is omitted or if +an unsupported version is given, then C will throw an error. + +=item dir + +Directory to search recursively for F<.pm> files. May not be specified with +C. + +=item files + +Array reference of files to examine. May not be specified with C. + +=item prefix + +String to prepend to the C field of the resulting output. This defaults +to F, which is the common case for most CPAN distributions with their +F<.pm> files in F. This option ensures the META information has the +correct relative path even when the C or C arguments are +absolute or have relative paths from a location other than the distribution +root. + +=back + +For example, given C of 'lib' and C of 'lib', the return value +is a hashref of the form: + + { + 'Package::Name' => { + version => '0.123', + file => 'lib/Package/Name.pm' + }, + 'OtherPackage::Name' => ... + } + =item C<< package_versions_from_directory($dir, \@files?) >> Scans C<$dir> for .pm files (unless C<@files> is given, in which case looks @@ -742,6 +892,14 @@ returning a hashref of the form: 'OtherPackage::Name' => ... } +The C and C
packages are always omitted, as are any "private" +packages that have leading underscores in the namespace (e.g. +C) + +Note that the file path is relative to C<$dir> if that is specified. +This B be used directly for CPAN META C. See +the C method instead. + =item C<< log_info (internal) >> Used internally to perform logging; imported from Log::Contextual if @@ -773,7 +931,10 @@ Returns the absolute path to the file. =item C<< packages_inside() >> -Returns a list of packages. +Returns a list of packages. Note: this is a raw list of packages +discovered (or assumed, in the case of C
). It is not +filtered for C, C
or private packages the way the +C method does. =item C<< pod_inside() >> @@ -797,7 +958,7 @@ Original code from Module::Build::ModuleInfo by Ken Williams Released as Module::Metadata by Matt S Trout (mst) with assistance from David Golden (xdg) . -=head1 COPYRIGHT +=head1 COPYRIGHT & LICENSE Original code Copyright (c) 2001-2011 Ken Williams. Additional code Copyright (c) 2010-2011 Matt Trout and David Golden.