Two new utilities that, together, can generate a zips.txt file customized for your...

[migration-tools.git] / enrich_zips

diff --git a/enrich_zips b/enrich_zips
new file mode 100755 (executable)
index 0000000..28587df
--- /dev/null
+++ b/enrich_zips
@@ -0,0 +1,72 @@
+#!/usr/bin/perl -w
+
+use strict;
+use Getopt::Long;
+
+my $VERSION = '1.01';
+
+=pod
+
+NAME
+
+       enrich_zips - Utility to add county information to city/state/ZIP data (and optionally use zips.txt format)
+
+USAGE
+
+       enrich_zips --db US.txt < citystatezip.tsv
+       enrich_zips --makezips --db US.txt < citystatezip.tsv > zips.txt
+
+NOTES
+
+       Geonames database can be downloaded from http://download.geonames.org/export/zip/US.zip
+
+=cut
+
+my ($db, $makezips, %zips, $warn);
+my $result = GetOptions ("db=s"     => \$db,
+                         "makezips" => \$makezips,
+                         "warn"     => \$warn);
+die
+       "Please specify the location of the Geonames database with --db US.txt\n" .
+       "HINT: You can download it at http://download.geonames.org/export/zip/US.zip" .
+  "\n\nProgram halted"
+unless defined($db);
+
+open DB, $db or die "Couldn't open Geonames database $db: $!\n";
+
+# Slurp in the Geonames database
+while (<DB>) {
+       chomp;
+       my @f = split(/\t/);
+       @{$zips{$f[1]}} = @f[4,2,5];
+       # @{$zips{"33166"}} == ("FL", "Miami Springs", "Miami-Dade")
+}
+
+while (<>) {
+
+       chomp;
+       (my $city, my $state, my $zip) = split(/\t/) or next;
+       my $county = "";
+       my ($dbcity, $dbstate);
+
+       if (defined $zips{$zip}) {
+               ($dbstate, $dbcity, $county) = @{$zips{$zip}};
+       } 
+
+       if ($warn) {
+               if (!defined $zips{$zip}) {
+                       print STDERR "No county data found for ZIP code $zip ($city, $state)\n\n";
+                       next;
+               }
+               if ($city ne $dbcity || $state ne $dbstate) {
+                       print STDERR "Patron data input says $zip is $city, $state\n".
+                                    "Geonames database says $zip is $dbcity, $dbstate\n\n";
+               }
+       }
+
+       if ($makezips) {
+               print "|" . join("|", ($state, $city, $zip, "1", "", $county)) . "||\n";
+       } else {
+               print join("\t", ($city, $state, $zip, $county)) . "\n";
+       }
+}