File Coverage

blib/lib/Makefile/Parser.pm

Criterion	Covered	Total	%
statement	236	606	38.9
branch	90	240	37.5
condition	34	58	58.6
subroutine	31	43	72.0
pod	10	10	100.0
total	401	957	41.9

package Makefile::Parser;

use strict;
use warnings;

use File::Spec;
use Cwd qw/ realpath /;
use List::MoreUtils qw( uniq pairwise ) ;
use Text::Balanced qw( gen_extract_tagged );
#use Smart::Comments;

#our $Debug = 0;
our $Strict = 0;
our $VERSION = '0.215';
our $Error;
our $Runtime = undef;

# usage: $class->new;
sub new {
    my $proto = shift;
    my $class = ref $proto || $proto;
    my $self = bless {
        _vars    => {},       # all the definitions of variables
        _tars    => undef,    # all the targets
        _default => undef,    # default target
        _depends => {},       # all the dependencies
        _imps    => [],       # targets in implicit rules
    }, $class;
    return $self;
}

my $extract_interp_1 = gen_extract_tagged('\$[(]', '[)]', '');
my $extract_interp_2 = gen_extract_tagged('\$[{]', '[}]', '');

sub _extract_interp {
    my ($res) = $extract_interp_1->($_[0]);
    if (!$res) {
        ($res) = $extract_interp_2->($_[0]);
    }
    $res;
}

# usage: $obj->parse($filename);
sub parse {
    my ($self, $file, $vars) = @_;
    $file ||= 'Makefile';
    my %init_vars = %$vars if $vars;

    $self->{_file} = $file;
    $self->{_vars} = {
        MAKE  => $0,
        CC    => 'cc',
        SHELL => 'sh',
        %init_vars,
    };
    undef $self->{_tars};
    undef $self->{_default};
    $self->{_depends} = {};
    $self->{_imps} = [];

    my $rvars = $self->{_vars};
    my $in;
    unless (open $in, $file) {
        $Error = "Cannot open $file for reading: $!";
        return undef;
    }

    my $state = 'S_IDLE';
    my ($var, $value, $tar_name, $tar, $colon_type, $depends, $cmd);
    my @cmds;
    my %tars;
    #%$rvars = ();
    my $first_tar = 1;
    while (<$in>) {
        next if /^\s*#/ and $state ne 'S_IN_VAL';
        next if /^\s*$/ and $state ne 'S_IN_VAL';
        #$tar_name = '' unless defined $var;
        #warn "(tar: $tar_name) Switching to tate $state with $_";
        #warn $state if $state ne 'S_IDLE';
        chomp;
        #if (/TEST_VERBOSE=/) {
            #### line: $_
            #### state: $state
        #}

        # expand the value of use-defined variables:
        #s/\$[\{\(](\w+)[\}\)]/exists $rvars->{$1} ? $rvars->{$1} : $&/ge;
        $_ = $self->_process_refs($_);

        if (($state eq 'S_IDLE' or $state eq 'S_CMD') and /^(\w+) \s* :?= \s* (.*)$/xo) {
            $var = $1;
            $value = $2;
            #warn "matched $var = $value\n";
            if ($value =~ m{\\\s*$}) {
                $value .= "\n";
                $state = 'S_IN_VAL' ;
            } else {
                $value =~ s/#.*//m;
                $rvars->{$var} = $value;
                ### variable: $var
                ### value: $value
                $state = 'S_IDLE';
            }
            #warn "$1 * $2 * $3";

        } elsif ($state eq 'S_IN_VAL') {
            #warn $1;
            my $line = $_;
            #warn "adding value line $line\n";
            $value .= "$line\n";
            if ($line !~ m{\\\s*$}) {
                $state = 'S_IDLE' ;
                #warn "Processing value '$value'\n";
                $value =~ s/[ \t]*\\\n[ \t]*/ /sg;
                $value =~ s/#.*//smg;
                #warn "Finale value '$value'\n";
                $value =~ s/\n//gs;
                $value =~ s/^\s+|\s+$//gs;
                $rvars->{$var} = $value;
                #warn "$var <=> $value\n";
            }

        } elsif (($state eq 'S_IDLE' or $state eq 'S_CMD') and /^(\.\w+) (\.\w+) \s* (::?)\s*$/xo) {
            $_ = "%$2 $3 %$1\n";
            #warn $_;
            redo;

        } elsif (($state eq 'S_IDLE' or $state eq 'S_CMD') and /^(\S[^:]*) (::?) \s* (.*)$/xo) {
            $tar_name = $1;
            $colon_type = $2;
            $depends = $3;
            $tar_name =~ s/^\s+|\s+$//g;

            my $cmd;
            if ($depends =~ s/;(.*)//) {
                $cmd = $1;
            }

            # Ignore .SUFFIXES currently:
            next if $tar_name eq '.SUFFIXES';

            #warn "Adding target $tar_name...\n";
            $tar = Makefile::Target->new($tar_name, $colon_type);
            if (my $old_tars = $tars{$tar_name}) {
                if ($colon_type eq ':') {
                    $tar->add_prereq($old_tars->[0]->prereqs);
                    if (my @cmd = $old_tars->[0]->commands) {
                        $tar->add_command(@cmd);
                    }
                    @$old_tars = $tar;
                } else {
                    push @$old_tars, $tar;
                }
            } else {
                $tars{$tar_name} = [$tar];
            }
            if ($tar_name =~ m/%/) {
                push @{$self->{_imps}}, $tar_name;
            }
            if ($first_tar) {
                $self->{_default} = $tar;
                $first_tar = 0;
            }
            if ($depends =~ s/\s+\\$//o) {
                $state = 'S_IN_DEPENDS';
            } else {
                $depends =~ s/\^\\$/\\/;
                $state = 'S_CMD';
            }
            my @depends = split /\s+/, $depends;
            map { $self->{_depends}->{$_} = 1 } @depends;
            $tar->add_depend(@depends);
            $tar->add_command($cmd) if defined $cmd;
        }
        elsif ($state eq 'S_IN_DEPENDS' and /^\s+ (.*)$/xo) {
            $depends = $1;
            if ($depends !~ s/\s+\\$//o) {
                $depends =~ s/\^\\$/\\/;
                my @depends = split /\s+/, $depends;
                map { $self->{_depends}->{$_} = 1 } @depends;
                $tar->add_depend(@depends);
                $state = 'S_CMD';
            }
        }
        elsif ($state eq 'S_CMD' and /^\s+(.*)/o) {
            $cmd = $1;
            if ($cmd =~ s/\s+\\$//o) {
                $state = 'S_IN_CMD';
            } else {
                $tar->add_command($cmd);
            }
        }
        elsif ($state eq 'S_IN_CMD' and /^\s+(.*)/o) {
            $cmd .= " $1";
            if ($cmd !~ s/\s+\\$//o) {
                $tar->add_command($cmd);
                $state = 'S_CMD';
            }
        }
        elsif ($Strict) {
            $Error = "syntax error: line $.: $_\n";
            return undef;
        } else {
            warn "I dunno how to do with it: $_\n";
        }
    }
    $self->{_tars} = \%tars;
    $self->post_parse;
    #warn Data::Dumper->Dump([\%tars], ['TARGETS']);
    close $in;
    return $self;
}

sub post_parse {
    my $self = shift;
    my $rdepends = $self->{_depends};
    my $rimps = $self->{_imps};
    for (keys %$rdepends) {
        next if /%/;
        #warn "Trying to match implicit rules one by one against $_...\n";
        $self->solve_imp($_);
    }
    for (@$rimps) {
        delete $self->{_tars}->{$_};
    }
}

sub solve_imp {
    my ($self, $depend) = @_;
    my $rimps = $self->{_imps};
    for my $imp (@$rimps) {
        my $obj = $self->target($imp);
        die "Rules for $imp not found" unless $obj and ref $obj;
        my $regex = quotemeta($imp);
        $regex =~ s/\\%/(.+)/;  # `%' can match any nonempty substring
        #warn "Processing regex $regex...\n";
        if ($depend =~ m/^$regex$/) {
            #warn "Succeeded to match $imp against $depend!\n";
            my $matched_part = $1;
            my $tar = Makefile::Target->new($depend, $obj->colon_type);
            my $dep;
            my @deps = map {
                s/%/$matched_part/;
                $self->{_depends}->{$_} = 1;
                #warn "Recursively solving dependent gole $_...\n";
                $self->solve_imp($_);
                $dep = $_;
                $_
            } $obj->depends;
            $tar->add_depend(@deps);
            my @cmds = map {
                s/\$
                s/\$\*/$matched_part/g;
                $_
            } $obj->commands;
            $tar->add_command(@cmds);
            $self->{_tars}->{$depend} = [$tar];
        }
    }
}

sub var {
    my ($self, $var) = @_;
    $self->parse if !defined $self->{_file};
    return $self->{_vars}->{$var};
}

sub vars {
    my $self = shift;
    $self->parse if !defined $self->{_file};
    return keys %{$self->{_vars}};
}

sub target {
    my ($self, $tar_name) = @_;
    $self->parse if !defined $self->{_file};
    return $self->{_default} if !defined $tar_name;
    my $tars = $self->{_tars}->{$tar_name};
    $tars ||= [];
    wantarray ? @$tars : $tars->[0];
}

sub targets {
    my $self = shift;
    $self->parse if !defined $self->{_file};
    return map { @$_ } values %{$self->{_tars}};
}

sub roots {
    my $self = shift;
    $self->parse if !defined $self->{_file};
    my %depends = %{$self->{_depends}};
    my %tars = %{$self->{_tars}};
    my @roots = ();
    my ($key, $val);
    while (($key, $val) = each %tars) {
        #next if $key =~ m/%/;
        next if $depends{$key};
        push @roots, $key;
    }
    return @roots;
}

sub error {
    return $Error;
}

sub _solve_refs_in_tokens ($$) {
    my ($self, $tokens) = @_;
    return '' if !$tokens;
    my $rvars = $self->{_vars};
    my @new_tokens;
    for my $token (@$tokens) {
        if ($token =~ /^\$[{(](.*)[)}]$/) {
            my $s = $1;
            if ($s =~ /^([-\w]+)\s+(.*)$/) {
                my $res = $self->_process_func_ref($1, $2);
                if (defined $res) {
                    push @new_tokens, $res;
                    next;
                }
            } elsif ($s =~ /^(\S+?):(\S+?)=(\S+)$/) {
                my ($var, $from, $to) = ($1, $2, $3);
                my $res = $self->_process_func_ref(
                    'patsubst', "\%$from,\%$to,\$($var)"
                );
                if (defined $res) {
                    push @new_tokens, $res;
                    next;
                }
            }
            if (exists $rvars->{$s}) {
                push @new_tokens, $rvars->{$s};
                next;
            } else {
                # FIXME: undefined var == ''
                #push @new_tokens, '';
                #next;
            }
        } elsif ($token =~ /^\$[@<|]$/) {
            # currently do nothing with the automatic vars
        } elsif ($token =~ /^\$\$$/) {
            push @new_tokens, '$';
            next;
        } elsif ($token =~ /^\$(.)$/) {
            if (exists $rvars->{$1}) {
                push @new_tokens, $rvars->{$1};
                next;
            } else {
                # FIXME: undef var == ''
                # push @new_tokens, '';
                # next;
            }
            ### found single-letter variable: $1
            ### value: $rvars->{$1}
            ### token: $token
        }
        push @new_tokens, $token;
    }
    ### retval: join '', @$tokens
    return join '', @new_tokens;
}

sub _process_refs {
    my ($self, $s) = @_;
    my @tokens = '';
    while (1) {
        if ($s =~ /\G[^\$]+/gc) {
            $tokens[-1] .= $&;
        } elsif (my $res = _extract_interp($s)) {
            push @tokens, $res, '';
        } elsif ($s =~ /\G\$./gc) {
            push @tokens, $&, '';
        } elsif ($s =~ /\G./gc) {
            $tokens[-1] .= $&;
        } else {
            last;
        }
    }
    ### tokens: @tokens
    return $self->_solve_refs_in_tokens(\@tokens);
}

sub _pat2re ($@) {
    my ($pat, $capture) = @_;
    $pat = quotemeta $pat;
    if ($capture) {
        $pat =~ s/\\\%/(\\S*)/g;
    } else {
        $pat =~ s/\\\%/\\S*/g;
    }
    $pat;
}

sub _text2words ($) {
    my ($text) = @_;
    $text =~ s/^\s+|\s+$//g;
    split /\s+/, $text;
}

sub _check_numeric ($$$$) {
    my ($self, $func, $order, $n) = @_;
    if ($n !~ /^\d+$/) {
        warn $self->{_file}, ":$.: ",
            "*** non-numeric $order argument to `$func' function: '$n'.  Stop.\n";
        exit(2);
    }
}

sub _check_greater_than ($$$$$) {
    my ($self, $func, $order, $n, $value) = @_;
    if ($n <= $value) {
        warn $self->{_file}, ":$.: *** $order argument to `$func' function must be greater than $value.  Stop.\n";
        exit(2);
   }
}

sub _trim ($@) {
    for (@_) {
        s/^\s+|\s+$//g;
    }
}

sub _split_args($$$$) {
    my ($self, $func, $s, $m, $n) = @_;
    $n ||= $m;
    my @tokens = '';
    my @args;
    ### $n
    while (@args <= $n) {
        ### split args: @args
        ### split tokens: @tokens
        if ($s =~ /\G\s+/gc) {
            push @tokens, $&, '';
        }
        elsif ($s =~ /\G[^\$,]+/gc) {
            $tokens[-1] .= $&;
        }
        elsif ($s =~ /\G,/gc) {
            if (@args < $n - 1) {
                push @args, [grep { $_ ne '' } @tokens];
                @tokens = '';
            } else {
                $tokens[-1] .= $&;
            }
        }
        elsif (my $res = _extract_interp($s)) {
            push @tokens, $res, '';
        }
        elsif ($s =~ /\G\$./gc) {
            push @tokens, $&, '';
        }
        elsif ($s =~ /\G./gc) {
            $tokens[-1] .= $&;
        }
        else {
            if (@args <= $n - 1) {
                push @args, [grep { $_ ne '' } @tokens];
            }
            last if @args >= $m and @args <= $n;
            warn $self->{_file}, ":$.: ",
            "*** insufficient number of arguments (",
            scalar(@args), ") to function `$func'.  Stop.\n";
            exit(2);
        }
    }
    return @args;
}

sub _trim_tokens ($) {
    my $tokens = shift;
    return if !@$tokens;
    if ($tokens->[0] =~ /^\s+$/) {
        shift @$tokens;
    }
    return if !@$tokens;
    if ($tokens->[-1] =~ /^\s+$/) {
        pop @$tokens;
    }
}

sub _process_func_ref ($$$) {
    my ($self, $name, $args) = @_;
    #### process func ref: $name
    $name = $self->_process_refs($name);
    my @args;
    my $nargs = scalar(@args);
    if ($name eq 'subst') {
        my @args = $self->_split_args($name, $args, 3);
        map { $_ = $self->_solve_refs_in_tokens($_) } @args;
        ### arguments: @args
        my ($from, $to, $text) = @args;
        $from = quotemeta($from);
        $text =~ s/$from/$to/g;
        return $text;
    }
    if ($name eq 'patsubst') {
        my @args = $self->_split_args($name, $args, 3);
        map { $_ = $self->_solve_refs_in_tokens($_) } @args;
        my ($pattern, $replacement, $text) = @args;
        my $re = _pat2re($pattern, 1);
        $replacement =~ s/\%/\${1}/g;
        $replacement = qq("$replacement");
        #### pattern: $re
        #### replacement: $replacement
        #### text: $text
        my $code = "s/^$re\$/$replacement/e";
        #### code: $code
        my @words = _text2words($text);
        map { eval $code; } @words;
        return join ' ', grep { $_ ne '' } @words;
    }
    if ($name eq 'strip') {
        my @args = $self->_split_args($name, $args, 1);
        map { $_ = $self->_solve_refs_in_tokens($_) } @args;
        my ($string) = @args;
        $string =~ s/^\s+|\s+$//g;
        $string =~ s/\s+/ /g;
        return $string;
    }
    if ($name eq 'findstring') {
        my @args = $self->_split_args($name, $args, 2);
        map { $_ = $self->_solve_refs_in_tokens($_) } @args;
        my ($find, $in) = @args;
        if (index($in, $find) >= 0) {
            return $find;
        } else {
            return '';
        }
        my ($patterns, $text) = @args;
        my @regexes = map { _pat2re($_) }
            split /\s+/, $patterns;
        ### regexes: @regexes
        my $regex = join '|', map { "(?:$_)" } @regexes;
        ### regex: $regex
        my @words = _text2words($text);
        return join ' ', grep /^$regex$/, @words;

    }
    if ($name eq 'filter') {
        my @args = $self->_split_args($name, $args, 2);
        map { $_ = $self->_solve_refs_in_tokens($_) } @args;
        my ($patterns, $text) = @args;
        my @regexes = map { _pat2re($_) }
            split /\s+/, $patterns;
        ### regexes: @regexes
        my $regex = join '|', map { "(?:$_)" } @regexes;
        ### regex: $regex
        my @words = _text2words($text);
        return join ' ', grep /^$regex$/, @words;
    }
    if ($name eq 'filter-out') {
        my @args = $self->_split_args($name, $args, 2);
        map { $_ = $self->_solve_refs_in_tokens($_) } @args;
        my ($patterns, $text) = @args;
        my @regexes = map { _pat2re($_) }
            split /\s+/, $patterns;
        ### regexes: @regexes
        my $regex = join '|', map { "(?:$_)" } @regexes;
        ### regex: $regex
        my @words = _text2words($text);
        return join ' ', grep !/^$regex$/, @words;
    }
    if ($name eq 'sort') {
        my @args = $self->_split_args($name, $args, 1);
        map { $_ = $self->_solve_refs_in_tokens($_) } @args;
        my ($list) = @args;
        _trim($list);
        return join ' ', uniq sort split /\s+/, $list;
    }
    if ($name eq 'words') {
        my @args = $self->_split_args($name, $args, 1);
        map { $_ = $self->_solve_refs_in_tokens($_) } @args;
        my ($text) = @args;
        my @words = _text2words($text);
        return scalar(@words);
    }
    if ($name eq 'word') {
        my @args = $self->_split_args($name, $args, 2);
        map { $_ = $self->_solve_refs_in_tokens($_) } @args;
        my ($n, $text) = @args;
        _trim($n);
        $self->_check_numeric('word', 'first', $n);
        $self->_check_greater_than('word', 'first', $n, 0);
        my @words = _text2words($text);
        return $n > @words ? '' : $words[$n - 1];
    }
    if ($name eq 'wordlist') {
        my @args = $self->_split_args($name, $args, 3);
        map { $_ = $self->_solve_refs_in_tokens($_) } @args;
        my ($s, $e, $text) = @args;
        _trim($s, $e, $text);
        $self->_check_numeric('wordlist', 'first', $s);
        $self->_check_numeric('wordlist', 'second', $e);
        $self->_check_greater_than('wordlist', 'first', $s, 0);
        $self->_check_greater_than('wordlist', 'second', $s, -1);
        my @words = _text2words($text);
        if ($s > $e || $s > @words || $e == 0) {
            return '';
        }
        $e = @words if $e > @words;
        return join ' ', @words[$s-1..$e-1];
    }
    if ($name eq 'firstword') {
        my @args = $self->_split_args($name, $args, 1);
        map { $_ = $self->_solve_refs_in_tokens($_) } @args;
        my ($text) = @args;
        my @words = _text2words($text);
        return @words > 0 ? $words[0] : '';
    }
    if ($name eq 'lastword') {
        my @args = $self->_split_args($name, $args, 1);
        map { $_ = $self->_solve_refs_in_tokens($_) } @args;
        my ($text) = @args;
        my @words = _text2words($text);
        return @words > 0 ? $words[-1] : '';
    }
    if ($name eq 'dir') {
        my @args = $self->_split_args($name, $args, 1);
        map { $_ = $self->_solve_refs_in_tokens($_) } @args;
        my ($text) = @args;
        my @names = _text2words($text);
        return join ' ', map { /.*\// ? $& : './' } @names;
    }
    if ($name eq 'notdir') {
        my @args = $self->_split_args($name, $args, 1);
        map { $_ = $self->_solve_refs_in_tokens($_) } @args;
        my ($text) = @args;
        my @names = _text2words($text);
        return join ' ', map { s/.*\///; $_ } @names;
    }
    if ($name eq 'suffix') {
        my @args = $self->_split_args($name, $args, 1);
        map { $_ = $self->_solve_refs_in_tokens($_) } @args;
        my ($text) = @args;
        my @names = _text2words($text);
        my $s = join ' ', map { /.*(\..*)/ ? $1 : '' } @names;
        $s =~ s/\s+$//g;
        return $s;
    }
    if ($name eq 'basename') {
        my @args = $self->_split_args($name, $args, 1);
        map { $_ = $self->_solve_refs_in_tokens($_) } @args;
        my ($text) = @args;
        my @names = _text2words($text);
        my $s = join ' ', map { /(.*)\./ ? $1 : $_ } @names;
        $s =~ s/\s+$//g;
        return $s;
    }
    if ($name eq 'addsuffix') {
        my @args = $self->_split_args($name, $args, 2);
        map { $_ = $self->_solve_refs_in_tokens($_) } @args;
        my ($suffix, $text) = @args;
        #_trim($suffix);
        my @names = _text2words($text);
        return join ' ', map { $_ . $suffix } @names;
    }
    if ($name eq 'addprefix') {
        my @args = $self->_split_args($name, $args, 2);
        map { $_ = $self->_solve_refs_in_tokens($_) } @args;
        my ($suffix, $text) = @args;
        #_trim($suffix);
        my @names = _text2words($text);
        return join ' ', map { $suffix . $_ } @names;
    }
    if ($name eq 'join') {
        my @args = $self->_split_args($name, $args, 2);
        map { $_ = $self->_solve_refs_in_tokens($_) } @args;
        my ($list_1, $list_2) = @args;
        my @list_1 = _text2words($list_1);
        my @list_2 = _text2words($list_2);
        return join ' ', pairwise {
            no warnings 'uninitialized';
            $a . $b
        } @list_1, @list_2;
    }
    if ($name eq 'wildcard') {
        my @args = $self->_split_args($name, $args, 1);
        map { $_ = $self->_solve_refs_in_tokens($_) } @args;
        my ($pattern) = @args;
        return join ' ', grep { -e $_ } glob $pattern;
    }
    if ($name eq 'realpath') {
        no warnings 'uninitialized';
        my @args = $self->_split_args($name, $args, 1);
        map { $_ = $self->_solve_refs_in_tokens($_) } @args;
        my ($text) = @args;
        my @names = _text2words($text);
        return join ' ', map { realpath($_) } @names;
    }
    if ($name eq 'abspath') {
        my @args = $self->_split_args($name, $args, 1);
        map { $_ = $self->_solve_refs_in_tokens($_) } @args;
        my ($text) = @args;
        my @names = _text2words($text);
        my @paths = map { File::Spec->rel2abs($_) } @names;
        for my $path (@paths) {
            my @f = split '/', $path;
            my @new_f;
            for (@f) {
                if ($_ eq '..') {
                    pop @new_f;
                } else {
                    push @new_f, $_;
                }
            }
            $path = join '/', @new_f;
        }
        return join ' ', @paths;
    }
    if ($name eq 'shell') {
        my @args = $self->_split_args($name, $args, 1);
        map { $_ = $self->_solve_refs_in_tokens($_) } @args;
        my ($cmd) = @args;
        my $output = `$cmd`;
        $output =~ s/(?:\r?\n)+$//g;
        $output =~ s/\r?\n/ /g;
        return $output;
    }
    if ($name eq 'if') {
        my @args = $self->_split_args($name, $args, 2, 3);
        #map { $_ = $self->_solve_refs_in_tokens($_) } @args;
        my ($condition, $then_part, $else_part) = @args;
        _trim_tokens($condition);
        $condition = $self->_solve_refs_in_tokens($condition);
        return $condition eq '' ?
                    $self->_solve_refs_in_tokens($else_part)
               :
                    $self->_solve_refs_in_tokens($then_part);
    }
    if ($name eq 'or') {
        my @args = $self->_split_args($name, $args, 1, 1000_000_000);
        #map { $_ = $self->_solve_refs_in_tokens($_) } @args;
        for my $arg (@args) {
            _trim_tokens($arg);
            my $value = $self->_solve_refs_in_tokens($arg);
            return $value if $value ne '';
        }
        return '';
    }
    if ($name eq 'and') {
        my @args = $self->_split_args($name, $args, 1, 1000_000_000);
        #map { $_ = $self->_solve_refs_in_tokens($_) } @args;
        ### arguments for 'and': @args
        my $value;
        for my $arg (@args) {
            _trim_tokens($arg);
            $value = $self->_solve_refs_in_tokens($arg);
            return '' if $value eq '';
        }
        return $value;
    }
    if ($name eq 'foreach') {
        my @args = $self->_split_args($name, $args, 3);
        my ($var, $list, $text) = @args;
        $var = $self->_solve_refs_in_tokens($var);
        $list = $self->_solve_refs_in_tokens($list);
        my @words = _text2words($list);
        # save the original status of $var
        my $rvars = $self->{_vars};
        my $not_exist = !exists $rvars->{$var};
        my $old_val = $rvars->{$var};

        my @results;
        for my $word (@words) {
            $rvars->{$var} = $word;
            #warn "$word";
            push @results, $self->_solve_refs_in_tokens($text);
        }

        # restore the original status of $var
        if ($not_exist) {
            delete $rvars->{$var};
        } else {
            $rvars->{$var} = $old_val;
        }

        return join ' ', @results;
    }
    if ($name eq 'error') {
        my ($text) = $self->_split_args($name, $args, 1);
        $text = $self->_solve_refs_in_tokens($text);
        warn $self->{_file}, ":$.: *** $text.  Stop.\n";
        exit(2) if $Runtime;
        return '';
    }
    if ($name eq 'warning') {
        my ($text) = $self->_split_args($name, $args, 1);
        $text = $self->_solve_refs_in_tokens($text);
        warn $self->{_file}, ":$.: $text\n";
        return '';
    }
    if ($name eq 'info') {
        my ($text) = $self->_split_args($name, $args, 1);
        $text = $self->_solve_refs_in_tokens($text);
        print "$text\n";
        return '';
    }

    return undef;
}

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

package Makefile::Target;

use overload
    '""'  => sub { shift->name },
    'cmp' => sub { my ($a,$b) = @_; "$a" cmp "$b" },
    'eq'  => sub { my ($a,$b) = @_; "$a" eq  "$b" },
    'lt'  => sub { my ($a,$b) = @_; "$a" lt  "$b" };

# usage: $class->new($name, $colon_type)
sub new {
    my $class = shift;
    my $self = {
        _name => shift,
        _colon_type => shift,
        _commands  => [],
        _depends => [],
    };
    return bless $self, $class;
}

sub name {
    return shift->{_name};
}

sub colon_type {
    return shift->{_colon_type};
}

sub prereqs {
    return @{shift->{_depends}};
}

*depends = \&prereqs;

sub add_prereq {
    push @{shift->{_depends}}, @_;
}

*add_depend = \&add_prereq;

sub commands {
    return @{shift->{_commands}};
}

sub add_command {
    my $self = shift;
    my @cmds = @_;
    my $name = $self->name;
    if ($name !~ m/%/) {
        map { s/\$\@/$self->{_name}/g } @cmds;
    }
    push @{$self->{_commands}}, @cmds;
}

sub run_commands {
    my $self = shift;
    my @cmd = $self->commands;
    for my $cmd (@cmd) {
        my ($quiet, $continue);
        while (1) {
            if ($cmd =~ s/^\s*\@//) {
                $quiet = 1;
            } elsif ($cmd =~ s/^\s*-//) {
                $continue = 1;
            } else {
                last;
            }
        }
        $cmd =~ s/^\s+|\s+$//gs;
        next if $cmd =~ /^$/;
        print "$cmd\n" unless $quiet;
        # currently only 'sh' is specified
        system('/bin/sh', '-c', $cmd);
        if ($? != 0 && !$continue) {
            die "$cmd returns nonzero status value: $?\n";
        }
    }
}

1;
__END__