Yet another Map-maker post

Discussions around using and interfacing with the Church MLS program.
unixguy
New Member
Posts: 45
Joined: Sun Mar 04, 2007 7:43 pm

#11

Post by unixguy »

Just thought I'd post and say "Thanks!"-- this tool was quite nice for me. I discovered that more addresses matched if I removed the address2 field from consideration due to values like "XYZ Nursing home", or "used to be no contact".

One thing that is proving to be frustrating (not with the tool, but the 'situation') is that apparently google sometimes recognizes "1234 Somestreet Way SE", and other times it must be in "1234 SE Somestreet Way" (moving the "SE"). I don't know if it's actually a county problem or if it's the vendor that Google gets their data from, but it's creating an issue for me, especially since I can't just change all the entries via a regexp, because that seems to break other addresses.

(Our county uses my first example as the standard, and a neighboring county uses the second example as their standard.)

I'm debating modifying your code for my specific case, so that if a lookup fails the first time, it will go through some address munging process to try and catch these odd ball cases. I thought it might make more sense to try and generalize the address munging so that different people could easily add their own functions that scratch their particular itch.

The general process would then be something like:

// how to get the address of a js function?
address_mangling_functions[0]="move_direction_in_address()" ;
address_mangling_functions[1]="replace_court_with_place()";
address_mangling_functions[2]="replace_st_with_court()";

process_without_modifying_addresses();
counter=0;
while (addresses_left_to_process() && length(address_mangling_functions[])>0) {
process_addresses( $address_list, address_mangling_functions[counter]);
counter++;
}

(hope my pidgin p-code that derives from perl, javascript, and c++ is understandable)

The second thing I've found is that a couple homes have street names that are too new for google, so am considering options there-- the best I can come up with is a new custom field to hardcode the lat/long values that I want used. (The geocode fields are already in use with the names that we've assigned each of our geocode districts.)

In this case the javascript would need to know which field that was and when it should use that instead of the values it gets from google. (I've got two difference failure cases-- some that google can't find at all, and some that show up about 40 miles away from the actual location.)
fraserredmond
Member
Posts: 56
Joined: Fri Feb 16, 2007 6:44 pm
Location: New Zealand

#12

Post by fraserredmond »

unixguy wrote:...I thought it might make more sense to try and generalize the address munging so that different people could easily add their own functions that scratch their particular itch.
unixguy wrote:...a new custom field to hardcode the lat/long values that I want used. (The geocode fields are already in use with the names that we've assigned each of our geocode districts.)

In this case the javascript would need to know which field that was and when it should use that instead of the values it gets from google. (I've got two difference failure cases-- some that google can't find at all, and some that show up about 40 miles away from the actual location.)
Great suggestions Unixguy, I'll put them both in, should be done either tomorrow or next week. The python script from dostler that mine was based on had some address-munging to do with flat-numbers, that I left out cause it didn't apply to me, but I guess out in the real world its the type of thing that will often be needed.
Locked

Return to “MLS Support, Help, and Feedback”