add pagc regression testing. Will upload the regress output once I look into the...

git-svn-id: http://svn.osgeo.org/postgis/trunk@11254 b70326c6-7e19-0410-871a-916f4a2858ee

diff --git a/extras/tiger_geocoder/tiger_2011/regress/pagc_normalize_address_regress.sql b/extras/tiger_geocoder/tiger_2011/regress/pagc_normalize_address_regress.sql
new file mode 100644 (file)
index 0000000..68a228b
--- /dev/null
+++ b/extras/tiger_geocoder/tiger_2011/regress/pagc_normalize_address_regress.sql
@@ -0,0 +1,97 @@
+--$Id: normalize_address_regress.sql 10309 2012-09-20 11:54:44Z robe $\r
+SELECT set_geocode_setting('use_pagc_address_parser', 'true');\r
+\timing\r
+SELECT '#887' As ticket, * FROM normalize_address('2450 N COLORADO ST, PHILADELPHIA, PA, 19132');\r
+SELECT '#1051a' As ticket, * FROM normalize_address('212 3rd Ave N Suite 560, Minneapolis, MN 55401');\r
+SELECT '#1051b' As ticket, * FROM normalize_address('3937 43RD AVE S, MINNEAPOLIS, MN 55406');\r
+SELECT '#1051c' As ticket, * FROM normalize_address('212 N 3rd Ave, Minneapolis, MN 55401');\r
+-- City missing ,  -- NOTE this one won't normalize right if you don't have MN data loaded\r
+SELECT '#1051d' As ticket, * FROM normalize_address('212 3rd Ave N Minneapolis, MN 55401'); \r
+-- comma in wrong spot\r
+SELECT * FROM normalize_address('529 Main Street, Boston MA, 02129');\r
+-- comma in right spot\r
+SELECT * FROM normalize_address('529 Main Street, Boston,MA 02129');\r
+-- partial address\r
+SELECT * FROM normalize_address('529 Main Street, Boston, MA');\r
+-- Full address with suite using ,\r
+SELECT * FROM normalize_address('529 Main Street, Apt 201, Boston, MA 02129');\r
+-- Full address with apart using space\r
+SELECT * FROM normalize_address('529 Main Street Apt 201, Boston, MA 02129');\r
+-- Partial address with apartment\r
+SELECT * FROM normalize_address('529 Main Street, Apt 201, Boston, MA');\r
+\r
+--This one fails so lead out for now\r
+SELECT '#1108a' As ticket, * FROM normalize_address('529 Main Street, Suite 201, Boston, MA 02129');\r
+\r
+-- Partial and Mangled zipcodes\r
+SELECT '#1073a' As ticket, * FROM normalize_address('212 3rd Ave N, MINNEAPOLIS, MN 553404');\r
+SELECT '#1073b' As ticket, * FROM normalize_address('212 3rd Ave N, MINNEAPOLIS, MN 55401-');\r
+SELECT '#1073c' As ticket, * FROM normalize_address('529 Main Street, Boston, MA 021');\r
+\r
+-- comma in wrong position\r
+SELECT '#1086a' As ticket, * FROM normalize_address('949 N 3rd St, New Hyde Park, NY, 11040');\r
+\r
+-- comma in right position --\r
+SELECT '#1086b' As ticket, * FROM normalize_address('949 N 3rd St, New Hyde Park, NY 11040');\r
+\r
+-- country roads and highways with spaces in street type\r
+SELECT '#1076a' As ticket, * FROM normalize_address('16725 Co Rd 24, Plymouth, MN 55447'); \r
+SELECT '#1076b' As ticket, * FROM normalize_address('16725 County Road 24, Plymouth, MN 55447'); \r
+SELECT '#1076c' As ticket, * FROM normalize_address('13800 County Hwy 9, Andover, MN 55304');\r
+SELECT '#1076d' As ticket, * FROM normalize_address('13800 9, Andover, MN 55304');\r
+-- this one is a regular street that happens to have a street type as the name\r
+SELECT '#1076e' As ticket, * FROM normalize_address('14 Forest Road, Acton, MA');\r
+\r
+-- A country road with a letter name and direction \r
+-- NOTE this doesn't completely normalize right since the direction W is being cut off --\r
+SELECT '#1076f' As ticket, * FROM normalize_address('1940 County Road C W, Roseville, MN 55113'); \r
+\r
+-- Route with a name that sounds like a direction --\r
+SELECT '#1076g' As ticket, * FROM normalize_address('3900 Route 6, Eastham, Massachusetts 02642');\r
+\r
+-- Street that has same name as type  --\r
+SELECT '#1076h' As ticket, * FROM normalize_address('4533 PARK AVE S, MINNEAPOLIS, MN 55407');\r
+-- same street with alternate county name\r
+SELECT '#1076i' As ticket, * FROM normalize_address('4533 County Road 33, MINNEAPOLIS, MN 55407'); \r
+\r
+-- Same case of street type that has name as a type --\r
+-- this matches -\r
+SELECT '#1109a' As ticket, * from normalize_address('4373 LAKE DRIVE, ROBBINSDALE, MN 55422');\r
+\r
+-- this failed --\r
+SELECT '#1109b' As ticket, * from normalize_address('4373 LAKE DR, ROBBINSDALE, MN 55422');\r
+\r
+-- another type (Is) that is part of street name but a compound street name\r
+SELECT '#1074a' As ticket, * FROM normalize_address('3420 RHODE ISLAND AVE S, ST. LOUIS PARK, MN 55426');\r
+\r
+-- another type that is part of street name --\r
+SELECT '#1074b' As ticket, * FROM normalize_address('26 Court Street, Boston,MA 02109');\r
+\r
+-- service roads and interstates\r
+SELECT '#1112a' As ticket, * FROM normalize_address('8401 W 35W Service Dr NE, Blaine, MN 55449');\r
+SELECT '#1112b' As ticket, * FROM normalize_address('8401 35W, Blaine, MN 55449');\r
+SELECT '#1112c' As ticket, * FROM normalize_address('8401 35W West, Blaine, MN 55449');\r
+SELECT '#1112d' As ticket, * FROM normalize_address('8401 West 35W, Blaine, MN 55449');\r
+SELECT '#1112e' As ticket, * FROM normalize_address('8401 W 35W, Blaine, MN 55449');\r
+\r
+-- Testing pretty print of highway addresses\r
+-- These tests excerpted from Brian Hamlin's CASS failures\r
+-- in #1077\r
+SELECT '#1125a' As ticket, pprint_addy(normalize_address('19596 COUNTY ROAD 480, COLCORD, OK 74338'));\r
+SELECT '#1125b' As ticket, pprint_addy(addy), addy.* FROM normalize_address('4345 353 Rte, SALAMANCA, NY 14779') AS addy;\r
+SELECT '#1125c' As ticket, pprint_addy(addy), addy.* FROM normalize_address('19799 STATE ROUTE O, COSBY, MO 64436') AS addy;\r
+\r
+-- some more to test interstate permutations\r
+SELECT '#1125d' As ticket, pprint_addy(addy), addy.* FROM normalize_address('Interstate 90,Boston, MA') As addy;\r
+-- this one is wrong (because the lack of space trips it up) but will fix later\r
+SELECT '#1125e' As ticket, pprint_addy(addy), addy.* FROM normalize_address('I-90,Boston, MA') As addy;\r
+SELECT '#1125f' As ticket, pprint_addy(addy), addy.* FROM normalize_address('I 90,Boston, MA') As addy;\r
+\r
+-- location with prefixes getting caught in post prefix\r
+SELECT '#1310a' As ticket, pprint_addy(addy), addy.* FROM normalize_address('1110 W CAPITOL AVE, WEST SACRAMENTO, CA') As addy;\r
+\r
+-- #1614 County Rd\r
+SELECT '#1614a' As ticket, pprint_addy(addy), addy.* FROM normalize_address('8435 COUNTY RD 20 SE, ROCHESTER, MN 55904') As addy;\r
+SELECT '#1614b' As ticket, pprint_addy(addy), addy.* FROM normalize_address('3208 U.S. 52, Rochester, MN 55901') As addy;\r
+\timing\r
+SELECT set_geocode_setting('use_pagc_address_parser', 'false');\r

diff --git a/extras/tiger_geocoder/tiger_2011/regress/regress.sql b/extras/tiger_geocoder/tiger_2011/regress/regress.sql
index c97cf7ae12ce10af37de4607cdc9ebb1b17d4fe8..90df009bfbcf0c3d503801cfb0790619631443a9 100644 (file)
--- a/extras/tiger_geocoder/tiger_2011/regress/regress.sql
+++ b/extras/tiger_geocoder/tiger_2011/regress/regress.sql
@@ -7,6 +7,8 @@ SELECT set_geocode_setting('debug_geocode_intersection', 'false');
 SELECT set_geocode_setting('reverse_geocode_numbered_roads', '1'); -- prefer numbered highway name
 \o normalize_address_regress.out
 \i normalize_address_regress.sql
+\o pagc_normalize_address_regress.out
+\i pagc_normalize_address_regress.sql
 \o geocode_regress.out
 \i geocode_regress.sql
 \o reverse_geocode_regress.out