fix for #1073 handling mangled zipcodes and update regress to include these

git-svn-id: http://svn.osgeo.org/postgis/trunk@7517 b70326c6-7e19-0410-871a-916f4a2858ee

diff --git a/extras/tiger_geocoder/tiger_2010/normalize/normalize_address.sql b/extras/tiger_geocoder/tiger_2010/normalize/normalize_address.sql
index c8582e5da360047ae7c0c55d8622f3136df62d16..8b3185563d92542e6a64a10fe4aafdcadb9dee05 100644 (file)
--- a/extras/tiger_geocoder/tiger_2010/normalize/normalize_address.sql
+++ b/extras/tiger_geocoder/tiger_2010/normalize/normalize_address.sql
@@ -89,18 +89,24 @@ BEGIN
 
   -- Assume that the address begins with a digit, and extract it from
   -- the input string.
-  addressString := substring(rawInput from '^([0-9].*?)[ ,/.]');
+  addressString := substring(rawInput from E'^([0-9].*?)[ ,/.]');
 
   IF debug_flag THEN
     raise notice '% addressString: %', clock_timestamp(), addressString;
   END IF;
 
-  -- There are two formats for zip code, the normal 5 digit, and
+  -- There are two formats for zip code, the normal 5 digit , and
   -- the nine digit zip-4.  It may also not exist.
-  zipString := substring(rawInput from ws || '([0-9]{5})$');
+  
+  zipString := substring(rawInput from ws || E'([0-9]{5})$');
   IF zipString IS NULL THEN
-    zipString := substring(rawInput from ws || '([0-9]{5})-[0-9]{4}$');
-    -- Check if all we got was a zipcode, of either form
+    -- Check if the zip is just a partial or a one with -s
+    -- or one that just has more than 5 digits
+    zipString := COALESCE(substring(rawInput from ws || '([0-9]{5})-[0-9]{0,4}$'), 
+                substring(rawInput from ws || '([0-9]{2,5})$'),
+                substring(rawInput from ws || '([0-9]{6,14})$'));
+   
+     -- Check if all we got was a zipcode, of either form
     IF zipString IS NULL THEN
       zipString := substring(rawInput from '^([0-9]{5})$');
       IF zipString IS NULL THEN

diff --git a/extras/tiger_geocoder/tiger_2010/regress/geocode_regress b/extras/tiger_geocoder/tiger_2010/regress/geocode_regress
index 7d34b89d0f14eb31d2a78dd530bcc60b4ad4291a..de46774c8106e01af39019e382b32ea5e9500b83 100644 (file)
--- a/extras/tiger_geocoder/tiger_2010/regress/geocode_regress
+++ b/extras/tiger_geocoder/tiger_2010/regress/geocode_regress
@@ -38,3 +38,16 @@
  Federal Ct, Boston, MA 02110              | POINT(-71.0567205 42.354134)              |     24
  Federal Ln, Dedham, MA 02026              | POINT(-71.183565 42.238295)               |     33
 
+ 212 3rd Ave N, Minneapolis, MN 55401  | POINT(-93.2718114 44.9850188)             |      1
+ 212 3rd Ave S, Minneapolis, MN 55404  | POINT(-93.2635498396584 44.9806201397408) |      3
+ 212 3rd Ave S, Minneapolis, MN 55401  | POINT(-93.2633388442551 44.9808723431463) |      3
+ 212 3rd Ave NE, Minneapolis, MN 55413 | POINT(-93.2592025 44.990786)              |      4
+ 212 3rd Ave S, Minneapolis, MN 55415  | POINT(-93.2641725306122 44.9798755102041) |      4
+ 212 3rd Ave SE, Minneapolis, MN 55414 | POINT(-93.2526539591837 44.9848287755102) |      5
+ 212 3rd Ave N, Biwabik, MN 55708      | POINT(-92.3450754489796 47.5343256938775) |      8
+ 212 3rd Ave N, Lakefield, MN 56150    | POINT(-95.1733443061224 43.6785488163265) |      9
+ 212 3rd Ave N, Bayport, MN 55003      | POINT(-92.7789324000695 45.019825623949)  |      9
+ 212 3rd Ave N, Long Prairie, MN 56347 | POINT(-94.8618907833944 45.977553389256)  |     10
+
+ Plymouth, MN 55340 | POINT(-93.5757500937089 45.076212213583) |    100
+

diff --git a/extras/tiger_geocoder/tiger_2010/regress/geocode_regress.sql b/extras/tiger_geocoder/tiger_2010/regress/geocode_regress.sql
index c4622cf3e30fefa672548a17168a5aa636a2bc8a..0f17d22466860b366fcabd11fa6f18cc69180476 100644 (file)
--- a/extras/tiger_geocoder/tiger_2010/regress/geocode_regress.sql
+++ b/extras/tiger_geocoder/tiger_2010/regress/geocode_regress.sql
@@ -25,4 +25,8 @@ SELECT pprint_addy(addy) As address, ST_AsText(geomout) As pt, rating FROM geoco
 --Test misspellings and missing zip --
 SELECT pprint_addy(addy) As address, ST_AsText(geomout) As pt, rating FROM geocode('101 Fedaral Street, Boston, MA',1);
 SELECT pprint_addy(addy) As address, ST_AsText(geomout) As pt, rating FROM geocode('101 Fedaral Street, Boston, MA',50);
+
+--Geocoding mangled zipcodes
+SELECT pprint_addy(addy) As address, ST_AsText(geomout) As pt, rating FROM geocode('212 3rd Ave N, MINNEAPOLIS, MN 553404');
+SELECT pprint_addy(addy) As address, ST_AsText(geomout) As pt, rating FROM geocode('212 3rd Ave N, MINNEAPOLIS, MN 55340-');
 \timing
\ No newline at end of file

diff --git a/extras/tiger_geocoder/tiger_2010/regress/normalize_address_regress b/extras/tiger_geocoder/tiger_2010/regress/normalize_address_regress
index 488bbe220a69ee3dd5d4bd027c3d708962457814..9b406a8c388f2f64cff4a403d243e297c753ba9d 100644 (file)
--- a/extras/tiger_geocoder/tiger_2010/regress/normalize_address_regress
+++ b/extras/tiger_geocoder/tiger_2010/regress/normalize_address_regress
@@ -2,7 +2,13 @@
 
      529 |              | Main       | St               |               |          | Boston   | MA          | 02129 | t
 
-     529 |              | Main       | St               |               |          | Boston   | MA          | 02129 | t
+     529 |              | Main       | St               |               |          | Boston,MA |             | 02129 | t
+
+     529 |              | Main       | St               |               |          | Boston   | MA          |     | t
+
+     529 |              | Main       | St               |               |          | Boston   | MA          | 021 | t
+
+     212 |              | 3rd        | Ave              | N             |          | MINNEAPOLIS | MN          | 553404 | t
 
-     529 |              | Main       | St               |               |          | Boston MA |             |     | t
+     212 |              | 3rd        | Ave              | N             |          | MINNEAPOLIS | MN          | 55401 | t
 

diff --git a/extras/tiger_geocoder/tiger_2010/regress/normalize_address_regress.sql b/extras/tiger_geocoder/tiger_2010/regress/normalize_address_regress.sql
index ff8a1412dddc026a41eff592e5f26f0cb6dd30af..8381090c2e29b143a5f08342c591e911f90fcd33 100644 (file)
--- a/extras/tiger_geocoder/tiger_2010/regress/normalize_address_regress.sql
+++ b/extras/tiger_geocoder/tiger_2010/regress/normalize_address_regress.sql
@@ -3,7 +3,13 @@ SELECT * FROM normalize_address('3937 43RD AVE S, MINNEAPOLIS, MN 55406');
 -- comma in wrong spot
 SELECT * FROM normalize_address('529 Main Street, Boston MA, 02129');
 -- comma in right spot
-SELECT * FROM normalize_address('529 Main Street, Boston MA, 02129');
+SELECT * FROM normalize_address('529 Main Street, Boston,MA 02129');
+-- partial address
+SELECT * FROM normalize_address('529 Main Street, Boston, MA');
 -- partial address
-SELECT * FROM normalize_address('529 Main Street, Boston MA');
+SELECT * FROM normalize_address('529 Main Street, Boston, MA 021');
+
+-- Mangled zipcodes
+SELECT * FROM normalize_address('212 3rd Ave N, MINNEAPOLIS, MN 553404');
+SELECT * FROM normalize_address('212 3rd Ave N, MINNEAPOLIS, MN 55401-');
 \timing
\ No newline at end of file