# perl script to parse comma delimited output of wab.exe
$liner = "----------------------------------------------------------------------";

while (<>) {
  if ($. == 1) {  # first line tells us the fields:
    @fields = split /\,/, $_;
    chop $fields[$#fields]; # remove '\n'
    #print "Number of fields: $#fields.\n";
  } else {
    @record = split /,/;
    if ($#record < $#fields) { # if their are less commas separated fields than expected,
                               # first guess that the line got split over more than one line
      $a = $#record; # remember which field was split over two lines
      $_ .= <>;
      @record = split /,/;
      if ($record[$a] !~ s/^"(.*)"$/$1/s) {
        print "ERROR line $., concat 2 lines failed. (Didn't find start & end quotes). $_";
      }
      if ($#record != $#fields) {
        print "ERROR line $., concat 2 lines failed. (Expected $#fields, found $#record). $_";
      } else { &store_record(); }
    } elsif ($#record > $#fields) { # extra comma?
      print "ERROR line $., comma in field? (Expected $#fields, found $#record). $_";
    } else { &store_record(); }
  }
}

#print "-----------\n"; foreach (@fields) {print "<$_>\n";} print "-----------\n";

# create a sorted list of records
foreach (0..$#{$db{$fields[0]}}) {
  $a = sprintf "%-30s %-30s %-30s %d",
       $db{'Last Name'}[$_], $db{'First Name'}[$_], $db{'Middle Name'}[$_], $_;
  $a =~ tr/A-Z/a-z/; # change to lowercase for sorting list
  $srt{$a} = $_;
}
#foreach (sort keys %srt) { print "key: <$_> value: <$srt{$_}>\n"; } # debug sort hash

$el_cnt = 0;
foreach (sort keys %srt) {
  $i = $srt{$_};
  &get_record($i);
  &validate(0); # arg = 0 means skip get_record
  if ($el eq 'el') {
    $el_cnt++;
    &print_adr();
  } elsif ($el ne '') {
    print "ERROR: unexpected contents of Office Location field: '$el'.\n";
  }
}

print "$liner\n";
$a = keys %add_err;
print  "Total: $#{$db{$fields[0]}} Easter Letter: $el_cnt ";
printf "Errors: $a  E.L. deliverable: %d\n", $el_cnt - $a;

# print out errors found during address validation
print "Capitalization errors:\n";
foreach (@caps_err) {print "  $_\n";}
print "Zip code errors:\n";
foreach (@zip_err) {print "  $_\n";}
$a = keys %add_err;
print "Incomplete addresses on Easter Letter list: ($a)\n";
#foreach (sort keys %add_err) {printf "  %-40s: $add_err{$_}\n", $_;}
foreach (sort keys %srt) {
  $i = $srt{$_};
  &get_record($i);
  $who = sprintf "%13s, %s", $last_name, $first_name;
  if (exists $add_err{$who}) {
    printf "  %-43s: $add_err{$who}\n", $who;
  }
}

sub print_adr {
  my($mn) = '';
  if ($middle_name ne '') {$mn = " $middle_name";}
  print "$first_name$mn $last_name\n";
  print "$address\n";
  print "$city, $state  $zip\n";
  print "  $country\n";
  print "\n";
}

sub validate { # validate address, using various tests
  if ($_[0]) {&get_record();}
  my($who) = sprintf "%13s, %s", $last_name, $first_name;

  # check for correct Zip codes for common cities
  if ($city =~ m/Loveland/i && $zip !~ /^8053[78]/) {
    push @zip_err,  "line $line $who '$city $zip'";
  }
  if ($city =~ m/Berthoud/i && $zip !~ /^80513/) {
    push @zip_err,  "line $line $who '$city $zip'";
  }
  if ($city =~ m/Boulder/i && $zip !~ /^8030[123]/) {
    push @zip_err,  "line $line $who '$city $zip'";
  }

  # capitailzations
  if ($city ne '' && $city !~ m/^[A-Z]/) { push @caps_err, "line $line $who '$city'"; }

  # incomplete address for Easter Letter records
  if ($el eq 'el') {
    # city must begin with a capital letter (flushes out blank city fields)
    if ($city !~ /^[A-Z]/) { $add_err{$who} .= "city? "; }

    # address must be filled in
    if ($address =~ /^\s*$/) { $add_err{$who} .= "street? "; }

    # US addresses
    if ($country eq '' || $country =~ m/^USA$/ || country =~ /^U\.S\.A\.$/) {
      if (not ($zip =~ /^\d\d\d\d\d$/ || $zip =~ m/^\d\d\d\d\d\-\d\d\d\d$/)) { $add_err{$who} .= "zip? "; }
      if ($state !~ /^[A-Z][A-Z]$/) { $add_err{$who} .= "state? "; }
    # Canadian addresses
    } elsif ($country =~ /^Canada$/i) {
      if ($zip !~ /^[A-Z]\d[A-Z]\s*\d[A-Z]\d$/) { $add_err{$who} .= "zip? "; }
      if ($state !~ /^[A-Z][A-Z]$/) { $add_err{$who} .= "state? "; }
    # all other addresses
    } else {
    }
  }
}

sub get_record {
  my($i) = $_[0];

  $first_name   = $db{"First Name"}[$i];
  $middle_name  = $db{"Middle Name"}[$i];
  $last_name    = $db{"Last Name"}[$i];
  $address      = $db{"Home Street"}[$i];
  $city         = $db{"Home City"}[$i];
  $state        = $db{"Home State"}[$i];
  $zip          = $db{"Home Postal Code"}[$i];
  $country      = $db{"Home Country"}[$i];
  $el           = $db{"Office Location"}[$i];
  $line         = $db{"line"}[$i];
} # get_record --------------------------------------------------------------

# ---------------------------------------------------------------------------
sub store_record {
  chop $record[$#record]; # remove '\n'
  foreach (0..$#record) {
    push @{ $db{$fields[$_]} }, $record[$_];
    push @{ $db{line} }, $.;
  }
} # store_record ------------------------------------------------------------