normalize_address: Internal address prefix sometimes get caught in post dir

[robe]

— this works

SELECT * FROM normalize_address('529 Main Street, Apt 201, Boston, MA 02129');

 address | predirabbrev | streetname | streettypeabbrev | postdirabbrev | internal | location | stateabbrev |  zip  | parsed 
---------+--------------+------------+------------------+---------------+----------+----------+-------------+-------+--------
     529 |              | Main       | St               |               | Apt 201  | Boston   | MA          | 02129 | t

But this has Suite bleed into postdirabbrev because it starts with S

SELECT * FROM normalize_address('529 Main Street, Suite 201, Boston, MA 02129');
 address | predirabbrev | streetname | streettypeabbrev | postdirabbrev | internal  | location | stateabbrev |  zip  | parsed 
---------+--------------+------------+------------------+---------------+-----------+----------+-------------+-------+--------
     529 |              | Main       | St               | S             | Suite 201 | Boston   | MA          | 02129 | t

[robe]

Also # leaves out internal address.

SELECT * FROM normalize_address('529 Main Street, #201, Boston, MA 02129');

 address | predirabbrev | streetname | streettypeabbrev | postdirabbrev | internal | location | stateabbrev |  zip  | parsed 
---------+--------------+------------+------------------+---------------+----------+----------+-------------+-------+--------
     529 |              | Main       | St               |               |          | Boston   | MA          | 02129 | t

[mikepease]

This seems like another example of the internal component confusing the normalizer.

—doesn't work

select abs(ST_X(geomout)::numeric(8,5))

'W, '

ST_Y(geomout)::numeric(8,5)

'N' as lat_lon, *

from geocode('155 Wabasha St S Ste 129, St. Paul, MN 55107')

—works!

select abs(ST_X(geomout)::numeric(8,5))

'W, '

ST_Y(geomout)::numeric(8,5)

'N' as lat_lon, *

from geocode('155 Wabasha St S Suite 129, St. Paul, MN 55107')

[robe]

This seems to work fine now. Not sure what changed. Put in a regression test to verify at r13987 for built-in normalize_address, and r13988 for address_standardizer augmented pagc_normalize_address