X-Git-Url: http://www.chiark.greenend.org.uk/ucgi/~ian/git?p=appendix-a6.git;a=blobdiff_plain;f=normalise;h=0af8f19623d021aa3d9c34f912fc3be0584fc908;hp=ca5ecf7975742a29bb521de392b4295f68dabb3d;hb=5fc4bc0993f2f2dc97c43217b3c0bfc51cdf9415;hpb=a11536b183032286d08363afa8f0d1fbd124b979

diff --git a/normalise b/normalise
index ca5ecf7..0af8f19 100755
--- a/normalise
+++ b/normalise
@@ -1,27 +1,92 @@
 #!/usr/bin/perl -w
+#
+# usage:
+#  normalise [NORM-OPTIONS...] [--] INPUT-FILES...
+#
+# NORM-OPTIONS are
+#   +OPTNAME[=OPTVAL]           Election option
+#   CAND=[DESCRIPTION]
+#   +CAND+OPTNAME[=OPTVAL]      Candidate option
+#   --                          End of options to normalise
+#   -...                        Reserved for future options to normalise
 
 use strict;
 
-our @options, @candiates, @voters;
+our @options;
+our %candidates; # $candidates{CAND}{Desc}, {Opts}[]
+our %candref; # $candref{CAND} => msg of why needed
+our @ballots;
 
-our %seen_cand, %need_cand;
+my $candvoter_re = '\w+';
+my $cands_re = '\w+(?,\w+)*';
+my $opt_re = '\w+(?:=\S*)?';
 
-my $candvoter_re = '[^\000-\037!"#$%()*+,/0-\136`-\177]+';
+sub badinput ($) {
+    die "bad input: $_[0]";
+}
 
-sub normalise_opts ($) {
-    my ($os) = @_;
+sub normalise_opts_list ($$) {
+    # $ctx is one of Election Candidate Ballot
+    my ($os,$ctx) = @_;
+    $os //= '';
     my @o;
     foreach my $o (split /\s+/, $os) {
 	if ($o =~ m/^\w+$/) {
 	    push @o, $&;
+	} elsif ($o =~ s/^_[Tt]ie\=//) {
+	    $o =~ m/^($cands_re)([<>])($cands_re)$/
+		or badinput "bad value \`$_' for tie option";
+	    my ($l,$op,$r) = ($1,$2,$3);
+	    ($l,$op,$r) = ($r,'>',$l) if $op eq '<';
+	    $candref{$_} = "tie break spec" foreach $o =~ m/\w+/g;
+	    $l = join ',', sort split /\,/, $l;
+	    $r = join ',', sort split /\,/, $r;
+	    $l =~ m/\b$_\b/ and badinput "reflexive tie"
+		foreach split /\,/, $r;
+	    push @o, "$l$op$r";
 	} elsif ($o =~ m/^\w+\=\S+$/) {
-	    push @o, $&;S
-	} elseif ($o !~ m/\S/) {
+	    push @o, $&;
+	} elsif ($o !~ m/\S/) {
 	} else {
 	    badinput "bad option \`$o'";
 	}
     }
-    return @o ? " | @o" : "";
+    return @o;
+}
+
+sub normalise_opts ($$) {
+    my ($os,$ctx) = @_;
+    my @o = normalise_opts_list $os, $ctx;
+    return " | @o";
+}
+
+sub setcanddesc ($$) {
+    my ($cand,$desc) = @_;
+
+    if (length $desc) {
+	badinput "multiple descriptions for $cand" if
+	    defined $candidates{$cand}{Desc};
+	$candidates{$cand}{Desc} = $desc;
+    }
+}
+
+while (@ARGV) {
+    $_ = shift @ARGV;
+    if (m/^--$/) {
+	last;
+    } elsif (m/^(\w+)=([^|]+)$/) {
+	setcanddesc $1, $2;
+    } elsif (m/^\+($opt_re)$/) {
+	push @options, $1;
+    } elsif (m/^\+(\w+)\+($opt_re)$/) {
+	push @{ $candidates{$1}{Opts} }, $2;
+    } elsif (m/^-/) {
+	die "unknown normalise option \`$_'\n";
+    } else {
+	# oh!
+	unshift @ARGV, $_;
+	last;
+    }
 }
 
 while (<>) {
@@ -29,12 +94,52 @@ while (<>) {
     next if m/^\#/;
     s/^\s+//;
     s/\s+$//;
-    if (m/^\|\s*(\w+(?:\=\S+)?)$/) {
-	push @options, "| $1";
-    } elsif (m/^($candvoter_re?)\s*=\s*([^|]+?)\s*|(.*)?$/) {
+    if (m/^\|/) {
+	push @options, normalise_opts_list $', 'Election';
+    } elsif (m/^($candvoter_re?)\s*=\s*([^|]+?)\s*\|(.*)?$/o) {
+	use Data::Dumper;
+print STDERR Dumper($1,$2,$3);
 	my ($cand,$desc,$opts) = ($1,$2,$3);
-	$desc=$cand unless length $desc;
-	$opts = normalise_opts $opts;
-	push @candidates, "$cand = $desc".$opts;
-    } elsif (m/^($candvoter_re?)?\s*\:/) {
-	
+	push @{ $candidates{$cand}{Opts} }, normalise_opts $opts, 'Candidate';
+	setcanddesc $cand, $desc;
+    } elsif (m/^($candvoter_re?)?\s*\:([^|]*)(?:\|(.*))?$/) {
+	my ($voter,$opts) = ($1,$3);
+	my @p;
+	foreach my $p (split /\s+/, $2) {
+	    if ($p =~ m/^\w+(?:\=\w+)*$/) {
+		push @p, $&;
+		$candidates{$_} //= { } foreach $p =~ m/\w+/g;
+	    } elsif ($p eq '') {
+		# empty entry can only happen if voter casts no prefs at all
+	    } else {
+		badinput "bad vote preference \`$p'";
+	    }
+	}
+	push @ballots, "$voter : @p".normalise_opts $opts, 'Ballot';
+    } elsif (m/^\.$/) {
+    } else {
+	badinput "unknown line format \`$_'";
+    }
+}
+
+print "| @options\n" or die $!;
+
+foreach my $cand (sort keys %candref) {
+    badinput "missing candidate $cand (ref. by $candref{$cand}"
+	unless defined $candidates{$cand};
+}
+
+foreach my $cand (sort keys %candidates) {
+    my $c = $candidates{$cand};
+    $c->{Desc} //= $cand;
+    $c->{Opts} //= [ ];
+    my $opts = $c->{Opts};
+    print "$cand = $c->{Desc} | @$opts\n" or die $!;
+}
+
+sub vsortkey { $_[0] =~ m/:/; return "$' : $`"; }
+
+print $_,"\n" or die $! foreach
+    sort { vsortkey($a) cmp vsortkey($b) } @ballots;
+
+print ".\n" or die $!;