X-Git-Url: http://git.freeside.biz/gitweb/?a=blobdiff_plain;f=FS%2FFS%2FMisc%2FGeo.pm;h=e41ba5d766e7a9fd8e3133bec9a1609349bc340a;hb=519f2393ecc06f548c76a677490add1adb1c1edf;hp=6bd817cfccdef98ad2768ef5544b9ed167f57d20;hpb=ecc15d03711690d2b2aeeda2bd8ff1119956c583;p=freeside.git diff --git a/FS/FS/Misc/Geo.pm b/FS/FS/Misc/Geo.pm index 6bd817cfc..e41ba5d76 100644 --- a/FS/FS/Misc/Geo.pm +++ b/FS/FS/Misc/Geo.pm @@ -333,85 +333,22 @@ sub standardize_usps { addr_clean=> 'Y' } } -my %ezlocate_error = ( # USA_Geo_002 documentation - 10 => 'State not found', - 11 => 'City not found', - 12 => 'Invalid street address', - 14 => 'Street name not found', - 15 => 'Address range does not exist', - 16 => 'Ambiguous address', - 17 => 'Intersection not found', #unused? -); - -sub standardize_ezlocate { - my $self = shift; - my $location = shift; - my $class; - #if ( $location->{country} eq 'US' ) { - # $class = 'USA_Geo_004Tool'; - #} - #elsif ( $location->{country} eq 'CA' ) { - # $class = 'CAN_Geo_001Tool'; - #} - #else { # shouldn't be a fatal error, just pass through unverified address - # warn "standardize_teleatlas: address lookup in '".$location->{country}. - # "' not available\n"; - # return $location; - #} - #my $path = $conf->config('teleatlas-path') || ''; - #local @INC = (@INC, $path); - #eval "use $class;"; - #if ( $@ ) { - # die "Loading $class failed:\n$@". - # "\nMake sure the TeleAtlas Perl SDK is installed correctly.\n"; - #} - - $class = 'Geo::EZLocate'; # use our own library - eval "use $class 0.02"; #Geo::EZLocate 0.02 for error handling - die $@ if $@; - - my $userid = $conf->config('ezlocate-userid') - or die "no ezlocate-userid configured\n"; - my $password = $conf->config('ezlocate-password') - or die "no ezlocate-password configured\n"; - - my $tool = $class->new($userid, $password); - my $match = $tool->findAddress( - $location->{address1}, - $location->{city}, - $location->{state}, - $location->{zip}, #12345-6789 format is allowed - ); - warn "ezlocate returned match:\n".Dumper($match) if $DEBUG > 1; - # error handling - B codes indicate success - die $ezlocate_error{$match->{MAT_STAT}}."\n" - unless $match->{MAT_STAT} =~ /^B\d$/; - - my %result = ( - address1 => $match->{MAT_ADDR}, - address2 => $location->{address2}, - city => $match->{MAT_CITY}, - state => $match->{MAT_ST}, - country => $location->{country}, - zip => $match->{MAT_ZIP}, - latitude => $match->{MAT_LAT}, - longitude => $match->{MAT_LON}, - censustract => $match->{FIPS_ST}.$match->{FIPS_CTY}. - sprintf('%07.2f',$match->{CEN_TRCT}), - addr_clean => 'Y', - ); - if ( $match->{STD_ADDR} ) { - # then they have a postal standardized address for us - %result = ( %result, - address1 => $match->{STD_ADDR}, - address2 => $location->{address2}, - city => $match->{STD_CITY}, - state => $match->{STD_ST}, - zip => $match->{STD_ZIP}.'-'.$match->{STD_P4}, - ); - } - - \%result; +sub _tomtom_query { # helper method for the below + my %args = @_; + my $result = Geo::TomTom::Geocoding->query(%args); + die "TomTom geocoding error: ".$result->message."\n" + unless ( $result->is_success ); + my ($match) = $result->locations; + my $type = $match->{type}; + # match levels below "intersection" should not be considered clean + my $clean = ($type eq 'addresspoint' || + $type eq 'poi' || + $type eq 'house' || + $type eq 'intersection' + ) ? 'Y' : ''; + warn "tomtom returned $type match\n" if $DEBUG; + warn Dumper($match) if $DEBUG > 1; + ($match, $clean); } sub standardize_tomtom { @@ -419,8 +356,7 @@ sub standardize_tomtom { # much better, but incompatible with ezlocate my $self = shift; my $location = shift; - my $class = 'Geo::TomTom::Geocoding'; - eval "use $class"; + eval "use Geo::TomTom::Geocoding; use Geo::StreetAddress::US"; die $@ if $@; my $key = $conf->config('tomtom-userid') @@ -428,12 +364,25 @@ sub standardize_tomtom { my $country = code2country($location->{country}); my ($address1, $address2) = ($location->{address1}, $location->{address2}); + my $subloc = ''; + + # trim whitespace + $address1 =~ s/^\s+//; + $address1 =~ s/\s+$//; + $address2 =~ s/^\s+//; + $address2 =~ s/\s+$//; + # try to fix some cases of the address fields being switched if ( $address2 =~ /^\d/ and $address1 !~ /^\d/ ) { $address2 = $address1; $address1 = $location->{address2}; } - my $result = $class->query( + # parse sublocation part (unit/suite/apartment...) and clean up + # non-sublocation address2 + ($subloc, $address2) = + subloc_address2($address1, $address2, $location->{country}); + # ask TomTom to standardize address1: + my %args = ( key => $key, T => $address1, L => $location->{city}, @@ -441,40 +390,48 @@ sub standardize_tomtom { PC => $location->{zip}, CC => country2code($country, LOCALE_CODE_ALPHA_3), ); - unless ( $result->is_success ) { - die "TomTom geocoding error: ".$result->message."\n"; + + my ($match, $clean) = _tomtom_query(%args); + + if (!$match or !$clean) { + # Then try cleaning up the input; TomTom is picky about junk in the + # address. Any of these can still be a clean match. + my $h = Geo::StreetAddress::US->parse_location($address1); + # First conservatively: + if ( $h->{sec_unit_type} ) { + my $strip = '\s+' . $h->{sec_unit_type}; + $strip .= '\s*' . $h->{sec_unit_num} if $h->{sec_unit_num}; + $strip .= '$'; + $args{T} =~ s/$strip//; + ($match, $clean) = _tomtom_query(%args); + } + if ( !$match or !$clean ) { + # Then more aggressively: + $args{T} = uc( join(' ', @$h{'number', 'street', 'type'}) ); + ($match, $clean) = _tomtom_query(%args); + } } - my ($match) = $result->locations; - if (!$match) { - die "Location not found.\n"; + + if ( !$match or !$clean ) { # partial matches are not useful + die "Address not found\n"; } - my $type = $match->{type}; - warn "tomtom returned $type match\n" if $DEBUG; - warn Dumper($match) if $DEBUG > 1; my $tract = ''; if ( defined $match->{censusTract} ) { $tract = $match->{censusStateCode}. $match->{censusFipsCountyCode}. join('.', $match->{censusTract} =~ /(....)(..)/); } - # match levels below "intersection" should not be considered clean - my $clean = ($type eq 'addresspoint' || - $type eq 'poi' || - $type eq 'house' || - $type eq 'intersection' - ) ? 'Y' : ''; - - $address2 = normalize_address2($address2, $location->{country}); - $address1 = ''; $address1 = $match->{houseNumber} . ' ' if length($match->{houseNumber}); $address1 .= $match->{street} if $match->{street}; + $address1 .= ' '.$subloc if $subloc; + $address1 = uc($address1); # USPS standards return +{ address1 => $address1, address2 => $address2, - city => $match->{city}, - state => $location->{state}, # this will never change - country => $location->{country}, # ditto + city => uc($match->{city}), + state => uc($location->{state}), + country => uc($location->{country}), zip => ($match->{standardPostalCode} || $match->{postcode}), latitude => $match->{latitude}, longitude => $match->{longitude}, @@ -483,15 +440,16 @@ sub standardize_tomtom { }; } -=iten normalize_address2 STRING, COUNTRY +=iten subloc_address2 ADDRESS1, ADDRESS2, COUNTRY -Given an 'address2' STRING, normalize it for COUNTRY postal standards. -Currently only works for US and CA. +Given 'address1' and 'address2' strings, extract the sublocation part +(from either one) and return it. If the sublocation was found in ADDRESS1, +also return ADDRESS2 (cleaned up for postal standards) as it's assumed to +contain something relevant. =cut -# XXX really ought to be a separate module -my %address2_forms = ( +my %subloc_forms = ( # Postal Addressing Standards, Appendix C # (plus correction of "hanger" to "hangar") US => {qw( @@ -532,26 +490,76 @@ my %address2_forms = ( )}, ); -sub normalize_address2 { +sub subloc_address2 { # Some things seen in the address2 field: # Whitespace # The complete address (with address1 containing part of the company name, # or an ATTN or DBA line, or P.O. Box, or department name, or building/suite # number, etc.) - my ($addr2, $country) = @_; - $addr2 = uc($addr2); - if ( exists($address2_forms{$country}) ) { - my $dict = $address2_forms{$country}; - # protect this - $addr2 =~ s/#\s*(\d)/NUMBER$1/; # /g? - my @words; - # remove all punctuation and spaces - foreach my $w (split(/\W+/, $addr2)) { - if ( exists($dict->{$w}) ) { - push @words, $dict->{$w}; - } else { - push @words, $w; - } + + # try to parse sublocation parts from address1; if they are present we'll + # append them back to address1 after standardizing + my $subloc = ''; + my ($addr1, $addr2, $country) = map uc, @_; + my $dict = $subloc_forms{$country} or return('', $addr2); + + my $found_in = 0; # which address is the sublocation + my $h; + foreach my $string ( + # patterns to try to parse + $addr1, + "$addr1 Nullcity, CA" + ) { + $h = Geo::StreetAddress::US->parse_location($addr1); + last if exists($h->{sec_unit_type}); + } + if (exists($h->{sec_unit_type})) { + $found_in = 1 + } else { + foreach my $string ( + # more patterns + $addr2, + "$addr1, $addr2", + "$addr1, $addr2 Nullcity, CA" + ) { + $h = Geo::StreetAddress::US->parse_location("$addr1, $addr2"); + last if exists($h->{sec_unit_type}); + } + if (exists($h->{sec_unit_type})) { + $found_in = 2; + } + } + if ( $found_in ) { + $subloc = $h->{sec_unit_type}; + # special case: do not combine P.O. box sublocs with address1 + if ( $h->{sec_unit_type} =~ /^P *O *BOX/i ) { + if ( $found_in == 2 ) { + $addr2 = "PO BOX ".$h->{sec_unit_num}; + } # else it's in addr1, and leave it alone + return ('', $addr2); + } elsif ( exists($dict->{$subloc}) ) { + # substitute the official abbreviation + $subloc = $dict->{$subloc}; + } + $subloc .= ' ' . $h->{sec_unit_num} if length($h->{sec_unit_num}); + } # otherwise $subloc = '' + + if ( $found_in == 2 ) { + # address2 should be fully combined into address1 + return ($subloc, ''); + } + # else address2 is not the canonical sublocation, but do our best to + # clean it up + # + # protect this + $addr2 =~ s/#\s*(\d)/NUMBER$1/; # /g? + my @words; + # remove all punctuation and spaces + foreach my $w (split(/\W+/, $addr2)) { + if ( exists($dict->{$w}) ) { + push @words, $dict->{$w}; + } else { + push @words, $w; } my $result = join(' ', @words); # correct spacing of pound sign + number @@ -559,13 +567,64 @@ sub normalize_address2 { warn "normalizing '$addr2' to '$result'\n" if $DEBUG > 1; $addr2 = $result; } - $addr2; + $addr2 = '' if $addr2 eq $subloc; # if it was entered redundantly + ($subloc, $addr2); } +sub standardize_melissa { + my $class = shift; + my $location = shift; + + local $@; + eval "use Geo::Melissa::WebSmart"; + die $@ if $@; + + my $id = $conf->config('melissa-userid') + or die "no melissa-userid configured\n"; + my $geocode = $conf->exists('melissa-enable_geocoding') ? 1 : 0; + + my $request = { + id => $id, + a1 => $location->{address1}, + a2 => $location->{address2}, + city => $location->{city}, + state => $location->{state}, + ctry => $location->{country}, + zip => $location->{zip}, + geocode => $geocode, + }; + my $result = Geo::Melissa::WebSmart->query($request); + if ( $result->code =~ /AS01/ ) { # always present on success + my $addr = $result->address; + warn Dumper $addr if $DEBUG > 1; + my $out = { + address1 => $addr->{Address1}, + address2 => $addr->{Address2}, + city => $addr->{City}->{Name}, + state => $addr->{State}->{Abbreviation}, + country => $addr->{Country}->{Abbreviation}, + zip => $addr->{Zip}, + latitude => $addr->{Latitude}, + longitude => $addr->{Longitude}, + addr_clean => 'Y', + }; + if ( $addr->{Census}->{Tract} ) { + my $censustract = $addr->{County}->{Fips} . $addr->{Census}->{Tract}; + # insert decimal point two digits from the end + $censustract =~ s/(\d\d)$/\.$1/; + $out->{censustract} = $censustract; + $out->{censusyear} = $conf->config('census_year'); + } + # we could do a lot more nuanced reporting of the warning/status codes, + # but the UI doesn't support that yet. + return $out; + } else { + die $result->status_message; + } +} =back =cut - 1;