New format: wip spi input converter

diff --git a/README.format b/README.format
index e08627cb0d3390524f2ec4e1d3c2fc2e0c51e0c8..5ec79035cb7e09eb84aa04dbc86b76c459c2a7a8 100644 (file)
--- a/README.format
+++ b/README.format
@@ -2,10 +2,10 @@ We define a common election and ballot input format.
 
 Format consists of lines:
 
-  [| OPTION]
+  [| OPTION...]
   CAND = DESCRIPTION ... [| OPTION...]
-  VOTERNAME : CAND, (CAND=CAND=...) ... [| OPTION...]
-
+  VOTERNAME : CAND CAND=CAND=... ... [| OPTION...]
+  .
 
 We define normalised and non-normalised inputs.  Normalised is for
 input by compute programs (which do not need to check conformance with
@@ -15,27 +15,26 @@ messing about by numans.
 In normalised files lines must occor in the order above (and
 whitespace is normalised, although readers should avoid relying on
 this other than that they may assume lack of leading and trailing
-whitespace on each line).
+whitespace on each line).  Normalised files always contain a
+(possibly-empty) `| OPTION' section on each line.
+
+Normalised files alwyas contain a `.' line at the end.  In
+non-normalised files anything after the `.' line is ignored, but it
+may be missing.
 
 Non-normalised files can also contain empty lines and
   # COMMENT...
 
 In non-normalised files any CAND may be omitted in which case CAND =
 CAND is assumed.  In non-normalised files an empty description is
-replaced with a copy of CAND.
+replaced with a copy of CAND.  In non-normalised files there may be
+multiple lines for any one CAND (provided there is only one non-empty
+DESCRIPTION).
 
-CAND and VOTERNAME may contain any characters except control
-characters and 7-bit ASCII punctuation, but including ASCII
-  ampersand &
-  apostrophe '
-  hyphen -
-  full stop .
-  and underscore (_).  Processors are not required to
-generate unambiguous output when CAND or VOTERNAME contains characters
-other than 7-bit alphanumerics plus _.
+CAND and VOTERNAME may contain 7-bit alphanumerics and _..
 
-DESCRIPTION may contain any characters except | (and cannot contain
-leading or trailing whitespace).
+DESCRIPTION may contain any characters except | and newline (and
+cannot contain leading or trailing whitespace).
 
 VOTERNAMEs need not be distinct and may be empty.
 
@@ -64,4 +63,5 @@ Currently defined OPTNAMEs (default values shown)
 
  Voter      Weight=1.0
 
-OPTIONs wihin a particular scope may occur in any order.
+OPTIONs wihin a particular scope may occur in any order.  Whether they
+may be repeated is up to the option.

diff --git a/normalise b/normalise
index ca5ecf7975742a29bb521de392b4295f68dabb3d..6bf8ea664d7b54fce5b65f3c2faea20fe90a51d7 100755 (executable)
--- a/normalise
+++ b/normalise
@@ -2,14 +2,11 @@
 
 use strict;
 
-our @options, @candiates, @voters;
+our @options, %candiates, @ballots;
 
-our %seen_cand, %need_cand;
+my $candvoter_re = '\w+';
 
-my $candvoter_re = '[^\000-\037!"#$%()*+,/0-\136`-\177]+';
-
-sub normalise_opts ($) {
-    my ($os) = @_;
+sub normalise_opts_list ($) {
     my @o;
     foreach my $o (split /\s+/, $os) {
        if ($o =~ m/^\w+$/) {
@@ -21,7 +18,13 @@ sub normalise_opts ($) {
            badinput "bad option \`$o'";
        }
     }
-    return @o ? " | @o" : "";
+    return @o;
+}
+
+sub normalise_opts ($) {
+    my ($os) = @_;
+    my @o = normalise_opts_list $os;
+    return " | @o";
 }
 
 while (<>) {
@@ -29,12 +32,49 @@ while (<>) {
     next if m/^\#/;
     s/^\s+//;
     s/\s+$//;
-    if (m/^\|\s*(\w+(?:\=\S+)?)$/) {
-       push @options, "| $1";
+    if (m/^\|/) {
+       push @options, normalise_opts_list $';
     } elsif (m/^($candvoter_re?)\s*=\s*([^|]+?)\s*|(.*)?$/) {
        my ($cand,$desc,$opts) = ($1,$2,$3);
+       push @{ $candidates{$cand}{Opts} }, normalise_opts $opts;
+       if (length $desc) {
+           badinput "multiple descriptions for $cand" if
+               defined $candidates{$cand}{Desc};
+           $candidates{$cand}{Desc} = $desc;
+       }
        $desc=$cand unless length $desc;
-       $opts = normalise_opts $opts;
-       push @candidates, "$cand = $desc".$opts;
-    } elsif (m/^($candvoter_re?)?\s*\:/) {
-       
+       push @candidates, "$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 my $p =~ m/\w+/g;
+           } else {
+               badinput "bad vote preference \`$p'";
+           }
+       }
+       push @ballots, "$voter : @p".normalise_opts $opts;
+    } elsif (m/^\.$/) {
+    } else {
+       badinput "unknown line format \`$_'";
+    }
+}
+
+print "| @options\n" or die $!;
+
+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 $!;

diff --git a/spi2loose b/spi2loose
new file mode 100755 (executable)
index 0000000..9965502
--- /dev/null
+++ b/spi2loose
@@ -0,0 +1,8 @@
+#!/usr/bin/perl -w
+use strict;
+while (<>) {
+      m/^(\w+) (\w+)$/ or die "$_ ?";
+      print "$1 : ".(join " ", split //, $2)."\n" or die $!;
+}
+      
+