[postgis-users] pagc normalize address min

[maplabs at light42.com]

Hi All -
 
  I hacked up a copy of the recent normalize address, and kept some notes on what I had to do to make a minimal version run in one database, pl-pgsql only. Pg 9.2, postgis 2.1 trunk

  without a lot of other observations, I can say that the example in the docs is really really "gordian"

http://postgis.net/docs/Pagc_Normalize_Address.html  (exmaple 2)

  on my setup with 4000 records that runs in 47s (!)

the following version runs in 7 seconds and does a similar thing, with the current funciton I have here

with g as  ( SELECT address, normalize_address(address||', '||city||' CA'||' '||zip) As sa           FROM tmd_mls_250ct13_typed ) SELECT   address,    (g.sa).address,    (g.sa).preDirAbbrev,    (g.sa).streetName,    (g.sa).streetTypeAbbrev,    (g.sa).postDirAbbrev,    (g.sa).location,    (g.sa).stateAbbrev,    (g.sa).zip,    (g.sa).parsed FROM g;

I want say both, that I know this is a hard problem and lots of sticky pats with pagc,
and, well, the future is yet unfolding.. 

all for now from Berkeley, Califonia

--
Brian M Hamlin
OSGeo California Chapter
blog.light42.com