Commods (upload): report hex ord of nonprinting characters when rejecting

[ypp-sc-tools.db-test.git] / yarrg / Commods.pm

diff --git a/yarrg/Commods.pm b/yarrg/Commods.pm
index c6ed2536399a8e713efce419bf04e3d052678b68..48caf79e9ba3f5ca1f0d2f158c4d8f36d18f242d 100644 (file)
--- a/yarrg/Commods.pm
+++ b/yarrg/Commods.pm
@@ -40,7 +40,7 @@ BEGIN {
                      %oceans %commods %clients %commodclasses
                      %vessels %shotname2damage
                      &parse_pctb_commodmap %pctb_commodmap @pctb_commodmap
-                     &get_our_version &check_tsv_line
+                     &get_our_version &check_tsv_line &errsan
                      &pipethrough_prep &pipethrough_run
                      &pipethrough_run_along &pipethrough_run_finish
                      &pipethrough_run_gzip &http_useragent &version_core
@@ -102,7 +102,7 @@ my %colours; # eg $colours{'c'}{'black'}= $sources
 my (@rawcm, @nocm); # eg $rawcm[0]='fine rum'; $rawcm[1]='fine %c cloth'
 
 my %colour_ordvals; # $colour_ordvals{'c'}{'green'}= '30';
-my %commodclasses; # $commodclasses{'dye'}= '3';
+our %commodclasses; # $commodclasses{'dye'}= '3';
 
 # IMPORTANT
 #  when extending the format of source-info in a non-backward
@@ -137,7 +137,7 @@ sub parse_info1 ($$$) {
            @ctx= (sub { push @nocm, lc $_; });
        } elsif (m/^commodclasses$/) {
            @ctx= (sub {
-               die unless m/^\*([-a-z]+)$/;
+               die unless m/^\*([_a-z]+)$/;
                $commodclasses{$1}= scalar keys %commodclasses;
            });
        } elsif (m/^ocean (\w+)$/) {
@@ -207,23 +207,27 @@ sub parse_info1 ($$$) {
            $commods{$ucname}{Srcs} .= $ss;
            my $c= $commods{$ucname};
            $c->{Volume}= 1000;
+           $c->{Flags}= '';
            my ($ordval, $ordclassval);
            foreach my $prop (defined $props ? split /\s+/, $props : ()) {
                if ($prop =~ m/^([1-9]\d*)(k?)g$/) {
                    $c->{Mass}= $1 * ($2 ? 1000 : 1);
                } elsif ($prop =~ m/^([1-9]\d*)l$/) {
                    $c->{Volume}= $1 * 1000;
-               } elsif ($prop =~ m/^\*([-a-z]+)$/) {
+               } elsif ($prop =~ m/^\*([_a-z]+)$/) {
                    $c->{Class}= $1;
                    die "$1" unless exists $commodclasses{$1};
                    $ordclassval= 1e7 + $commodclasses{$1} * 1e7;
                } elsif ($prop =~ m/^\@(\d+\+?)$/) {
                    $ordval= $1;
                    $ordval =~ s/^(\d+)\+$/ $1 + $lnoix * 10 /e;
+               } elsif ($prop =~ m/^\!([a-z]+)$/) {
+                   $c->{Flags} .= $1;
                } else {
                    die "unknown property $prop for $ucname";
                }
            }
+           $c->{ClassOrdval}= $ordclassval;
            if (defined $ordbase && defined $ordval && defined $ordclassval) {
                my $ordvalout= $ordbase + $ordval + $ordclassval;
                $c->{Ordval}= $ordvalout;
@@ -406,28 +410,39 @@ sub cgipostform ($$$) {
 
 our %check_tsv_done;
 
+sub errsan ($) {
+    my ($value) = @_;
+    $value =~ s/[^-+\'. A-Za-z0-9]/ sprintf "\\x%02x",ord $& /ge;
+    return "\"$value\"";
+}
+
 sub check_tsv_line ($$) {
     my ($l, $bad_data_callback) = @_;
     my $bad_data= sub { &$bad_data_callback("bad data: line $.: $_[0]"); };
     
     chomp($l) or &$bad_data('missing end-of-line');
 
-    $l !~ m/\P{IsPrint}/ or &$bad_data('nonprinting char(s)');
+    $l !~ m/\P{IsPrint}/ or &$bad_data('nonprinting char(s) '.sprintf "%#x", ord $&);
     my @v= split /\t/, $l, -1;
     @v==6 or &$bad_data('wrong number of fields');
+    $v[1] =~ s/^\s+//; $v[1] =~ s/\s+$//; # ooo don't check :-(
     my ($commod,$stall) = @v;
 
     !keys %commods or
        defined $commods{$commod} or
-       &$bad_data("unknown commodity \`$commod'");
+       &$bad_data("unknown commodity ".errsan($commod));
     
-    $stall =~ m/^\p{IsUpper}|^[0-9]/ or &$bad_data("stall not capitalised");
-    !exists $check_tsv_done{$commod,$stall} or &$bad_data("repeated data");
+    $stall =~ m/^\p{IsAlnum}/ or
+       &$bad_data("stall does not start with alphanumeric ".errsan($stall));
+    !exists $check_tsv_done{$commod,$stall} or
+       &$bad_data("repeated data ".errsan($commod).",".errsan($stall));
     $check_tsv_done{$commod,$stall}= 1;
     foreach my $i (2..5) {
        my $f= $v[$i];
-       $f =~ m/^(|0|[1-9][0-9]{0,5}|\>1000)$/ or &$bad_data("bad field $i");
-       ($i % 2) or ($f !~ m/\>/) or &$bad_data("> in field $i price");
+       $f =~ m/^(|0|[1-9][0-9]{0,5}|\>1000)$/ or
+           &$bad_data("bad field $i ".errsan($f));
+       ($i % 2) or ($f !~ m/\>/) or
+           &$bad_data("> in field $i price ".errsan($f));
     }
 
     foreach my $i (2,4) {