Skip to content

Commit

Permalink
Continued to refactor the OpenStreetMap parser - still debugging, do …
Browse files Browse the repository at this point in the history 
…not use this version
  • Loading branch information
@nigelhorne
nigelhorne committed Feb 6, 2024
1 parent eab72e2 commit a314eb6
Showing 1 changed file with 27 additions and 24 deletions.
51 changes: 27 additions & 24 deletions createdatabase.PL
View file
Original file line number Diff line number Diff line change
Expand Up @@ -3715,7 +3715,9 @@ sub extract_osm_home
$is_in =~ s/^UK, UK$/GB/;
$is_in =~ s/GB, GB$/GB/;
$is_in =~ s/, UK$/, GB/;
$is_in =~ s/^UK,\s?United Kingdom,\s?/GB, /;
$is_in =~ s/(England|Scotland|Wales), GB/GB/;
$is_in =~ s/,\s?(England|Scotland|Wales)$/, GB/;
$is_in =~ s/Yorkshire, UK/Yorkshire, GB/;
$is_in =~ s/, Europe$//;
$is_in =~ s/;\s?/, /g;
Expand All @@ -3732,8 +3734,8 @@ sub extract_osm_home

my $state;

print __LINE__, "\n";
if(defined($name) && ($name !~ /,/) && (!defined($place)) && ($state = us_state2code(uc($is_in)))) {
print __LINE__, "\n";
# name/place/is_in = Danville/undef/Pennsylvania
if($file =~ /north-america/) {
# WA can be in US or Australia
Expand Down Expand Up @@ -3765,9 +3767,9 @@ print __LINE__, "\n";
@place_fields = ($place);
}

print __LINE__, "\n";
if((scalar(@is_in_fields) > 1) && ($is_in_fields[1] eq 'US') && ($state = $us->{'state2code'}{uc($is_in_fields[0])})) {
if($place_fields[1] eq $is_in_fields[0]) {
print __LINE__, "\n";
# name/place/is_in = Bemidji/Bemidji, Minnesota/Minnesota, USA
return {
CITY => $name,
Expand All @@ -3788,9 +3790,9 @@ print __LINE__, "\n";
COUNTRY => $country
}
}
print __LINE__, "\n";
my @name_fields = split(/[,;]\s?/, $name) if(defined($name));
if((scalar(@is_in_fields) == 5) && ($is_in_fields[4] eq 'US') && ($state = us_state2code($is_in_fields[1]))) {
print __LINE__, "\n";
$row = {
STATE => $state,
COUNTRY => 'US'
Expand All @@ -3809,8 +3811,8 @@ print __LINE__, "\n";
}
return $row
}
print __LINE__, "\n";
if((scalar(@is_in_fields) == 5) && ($is_in_fields[4] eq 'US') && ($state = us_state2code($is_in_fields[3]))) {
print __LINE__, "\n";
# name/place/is_in = Forest Park/Forest Park (Columbus, Ohio)/Nortland, Columbus, Franklin, Ohio, USA
$row = {
CITY => $is_in_fields[1],
Expand All @@ -3820,8 +3822,8 @@ print __LINE__, "\n";
$row->{'NAME'} = $name if(defined($name));
return $row
}
print __LINE__, "\n";
if((scalar(@is_in_fields) == 2) && (scalar(@place_fields) == 2) && ($state = us_state2code($is_in_fields[0])) && ($state eq $is_in_fields[1]) && ($place_fields[1] eq $is_in_fields[0])) {
print __LINE__, "\n";
# name/place/is_in = Mifflinville/Mifflinville, Pennsylvania/Pennsylvania,PA
my $country = 'US' if($file =~ /north-america/); # WA can be in US or Australia
die if(!defined($country));
Expand All @@ -3831,8 +3833,8 @@ print __LINE__, "\n";
COUNTRY => $country
}
}
print __LINE__, "\n";
if((scalar(@is_in_fields) == 2) && defined($name) && (!defined($place)) && us_state2code($is_in_fields[0]) && (us_state2code($is_in_fields[0]) eq $is_in_fields[1])) {
print __LINE__, "\n";
# name/place/is_in = Catawissa/undef/Pennsylvania,PA
my $country = 'US' if($file =~ /north-america/); # WA can be in US or Australia
die if(!defined($country));
Expand All @@ -3842,8 +3844,8 @@ print __LINE__, "\n";
COUNTRY => $country
}
}
print __LINE__, "\n";
if((scalar(@is_in_fields) == 2) && defined($name) && (scalar(@place_fields) == 3) && ($state = us_state2code($is_in_fields[1]))) {
print __LINE__, "\n";
# name/place/is_in = Enterprise/Enterprise, Lake County, California/Lake, California
my $country = 'US' if($file =~ /north-america/); # WA can be in US or Australia
die if(!defined($country));
Expand All @@ -3857,8 +3859,8 @@ print __LINE__, "\n";
}
return $row;
}
print __LINE__, "\n";
if((scalar(@place_fields) == 2) && defined($name) && ($name eq $place_fields[0]) && ($us->{'code2state'}{$is_in})) {
print __LINE__, "\n";
# name/place/is_in = Epping/Epping, New Hampshire/NH
my $country = 'US' if($file =~ /north-america/); # WA can be in US or Australia
die if(!defined($country));
Expand All @@ -3868,8 +3870,8 @@ print __LINE__, "\n";
COUNTRY => $country
}
}
print __LINE__, "\n";
if((scalar(@place_fields) == 2) && (scalar(@is_in_fields) == 2) && ($state = us_state2code($is_in_fields[1]))) {
print __LINE__, "\n";
# name/place/is_in = West Athens/West Athens, California/Los Angeles, California
my $country = 'US' if($file =~ /north-america/); # WA can be in US or Australia
die if(!defined($country));
Expand All @@ -3887,7 +3889,6 @@ print __LINE__, "\n";
}
return $row
}
print __LINE__, "\n";
if(defined($name) && (scalar(@is_in_fields) == 2) && (!defined($place)) && ($state = us_state2code($is_in_fields[1]))) {
print __LINE__, "\n";
my $country;
Expand Down Expand Up @@ -3919,8 +3920,8 @@ print __LINE__, "\n";
}
return $row
}
print __LINE__, "\n";
if((scalar(@place_fields) == 3) && defined($name) && ($state = us_state2code($is_in))) {
print __LINE__, "\n";
# name/place/is_in = Waterville/Cummings Township, Lycoming County, Pennsylvania#Waterville/Pennsylvania
my $country = 'US' if($file =~ /north-america/); # WA can be in US or Australia
die if(!defined($country));
Expand All @@ -3934,8 +3935,8 @@ print __LINE__, "\n";
}
return $row
}
print __LINE__, "\n";
if((scalar(@is_in_fields) == 4) && (scalar(@place_fields) == 2) && ($is_in_fields[3] eq 'US') && $us->{'code2state'}{$is_in_fields[2]}) {
print __LINE__, "\n";
# name/place/is_in = Archbold/Archbold, Ohio/Fulton County, Ohio, OH, USA
$row = {
CITY => $place_fields[0],
Expand All @@ -3947,8 +3948,8 @@ print __LINE__, "\n";
}
return $row
}
print __LINE__, "\n";
if(defined($name) && ($name !~ /,/) && (!defined($place)) && ($is_in eq 'Washington DC')) {
print __LINE__, "\n";
# name/place/is_in = Bellevue/undef/Washington DC
# is_in will have been split into 2 by the space
my $country = 'US' if($file =~ /north-america/); # WA can be in US or Australia
Expand All @@ -3959,8 +3960,8 @@ print __LINE__, "\n";
COUNTRY => $country
}
}
print __LINE__, "\n";
if(defined($name) && ($name !~ /,/) && defined($place) && ($is_in eq 'Washington DC')) {
print __LINE__, "\n";
# name/place/is_in = Congress Heights/Congress Heights/Washington DC
# is_in will have been split into 2 by the space
my $country = 'US' if($file =~ /north-america/); # WA can be in US or Australia
Expand All @@ -3971,10 +3972,10 @@ print __LINE__, "\n";
COUNTRY => $country
}
}
print __LINE__, "\n";
if(defined($name) && ($name !~ /,/) && defined($place) && ($place =~ /^\Q$name\E\s.+Washington.*/) && ($is_in eq 'Washington DC')) {
# name/place/is_in = Petworth/Petworth (Washington, D.C.)/Washington DC
# is_in will have been split into 2 by the space
print __LINE__, "\n";
my $country = 'US' if($file =~ /north-america/); # WA can be in US or Australia
die if(!defined($country));
return {
Expand All @@ -3984,15 +3985,16 @@ print __LINE__, "\n";
}
}
if(defined($name) && defined($place) && ($place eq 'county') && (scalar(@is_in_fields) == 4) && ($is_in_fields[3] eq 'US') && (us_state2code($is_in_fields[0]) eq $is_in_fields[2])) {
print __LINE__, "\n";
# name/place/is_in = Bartholomew/county/Indiana,Ind.,IN,USA
return {
COUNTY => $name,
STATE => $is_in_fields[2],
COUNTRY => 'US'
}
}
print __LINE__, "\n";
if(defined($name) && (scalar(@place_fields) == 2) && (scalar(@is_in_fields) == 1) && ($state = us_state2code($place_fields[1])) && ($is_in =~ /(.+)\sCounty$/)) {
print __LINE__, "\n";
# name/place/is_in = Hard Rock/Hardrock, Arizona/Navajo County
$row = {
CITY => $name,
Expand All @@ -4005,16 +4007,16 @@ print __LINE__, "\n";
$row->{'COUNTRY'} = $country;
return $row
}
print __LINE__, "\n";
if(defined($name) && defined($place) && ($is_in eq 'US') && ($state = us_state2code($name))) {
print __LINE__, "\n";
# name/place/is_in = Massachusetts/Massachusetts/USA
return {
STATE => $state,
COUNTRY => 'US'
}
}
print __LINE__, "\n";
if(defined($name) && defined($place) && (scalar(@is_in_fields) == 2) && ($state = us_state2code($is_in_fields[1]))) {
print __LINE__, "\n";
my $country = 'US' if($file =~ /north-america/); # WA can be in US or Australia
die if(!defined($country));
$row = {
Expand All @@ -4032,8 +4034,8 @@ print __LINE__, "\n";
}
return $row
}
print __LINE__, "\n";
if(defined($name) && (scalar(@is_in_fields) == 3) && ($is_in_fields[2] eq 'US') && ($state = us_state2code($is_in_fields[1]))) {
print __LINE__, "\n";
if(scalar(@place_fields) == 1) {
# name/place/is_in = Bird Springs Overlook/Monument Valley/Navajo County; Arizona; United States of America
$row = {
Expand All @@ -4055,7 +4057,6 @@ print __LINE__, "\n";
COUNTRY => 'US'
}
}
print __LINE__, "\n";
if(defined($name) && (scalar(@is_in_fields) == 4)) {
print __LINE__, "\n";
if($file =~ /north-america/) {
Expand Down Expand Up @@ -4092,8 +4093,8 @@ print __LINE__, "\n";
}
}
}
print __LINE__, "\n";
if(defined($name) && (scalar(@is_in_fields) == 6) && ($state = us_state2code($is_in_fields[2])) && ($is_in_fields[5] eq 'US')) {
print __LINE__, "\n";
# name/place/is_in = Paloma del Sol/undef/Temecula;Riverside;California;Calif;CA;USA
return {
NAME => $name,
Expand Down Expand Up @@ -4172,6 +4173,7 @@ print __LINE__, "\n";
}
} else {
# name/place/is_in = Boyle/undef/Alberta,Alta.,AB,Canada
print __LINE__, "\n";
return {
'CITY' => $name,
'STATE' => $state,
Expand Down Expand Up @@ -4305,8 +4307,8 @@ print __LINE__, "\n";
print __LINE__, "\n";
}
}
print __LINE__, "\n";
if((scalar(@is_in_fields) == 7) && ($state = ca_province2code($is_in_fields[5]))) {
print __LINE__, "\n";
# name/place/is_in = Gilwood Road/400 Lake Shore Drive/High Prairie,Alberta,Canada;Big Lakes, M.D. of,Alberta,Canada
$row = {
CITY => $is_in_fields[0],
Expand Down Expand Up @@ -4521,10 +4523,11 @@ print __LINE__, "\n";
COUNTRY => 'GB'
}
}
if(defined($name) && (scalar(@place_fields) == 2) && (scalar(@is_in_fields) == 2) && ($is_in_fields[1] eq 'GB')) {
if((scalar(@name_fields) == 1) && (scalar(@is_in_fields) == 2) && ($is_in_fields[1] eq 'GB')) {
# name/place/is_in = Mere/Mere, Wiltshire/Wiltshire, GB
# name/place/is_in = Shedfield/undef/Hampshire, GB
return {
CITY => $name_fields[0],
CITY => $name,
STATE => $is_in_fields[0],
COUNTRY => 'GB'
}
Expand Down

0 comments on commit a314eb6

Please sign in to comment.