i.e. For Australian postcodes, we look at the suburb and postcode combination. Matching these two will give us the state as the particular suburb and postcode combination can only exist in one state.

However when looking at the data, I believe for New Zealand postcodes we can’t do this. To the point that a combination of Suburb and Postcode is not unique to a particular town.

Am I correct in my understanding of this?

Does anyone have any suggestions regarding how we could handle New Zealand postcodes in a similar way to Australia?

For instance from post code directory pdfs simply select all text and dump to plain text file, then parse.

Its very easy as the “junk” text such as “please visit our website..” is repeated in large blocks throughout the document and can be easily stripped out in a matter of seconds.

That will leave u with some nice blocks of text starting with “Lobby Name Box/Bag Range Postcode”, “Postcode Listings – Urban Areas ” and “To check that an address is serviced by Rural Delivery” that you can use to change parsers for a specific post code type.

If you look at the plain text dump you’ll see that RD parsing is a piece of cake as are boxs and bags (lobby/start/end/post code). Its a very basic 3 columns format. Very simple.

Urban addresses look complicated but they are not if u consider that all New Zealand address fall into about 12 categories (streets/roads/lanes/etc). So when parsing the text for urban addresses just check and compare against these. If the word your on is one of these categories then you know uve just hit then end of a street, from there u should be expecting a suburb/town, so if u hit a numeric digit instead of a letter then you know you got a street range, then after the suburb/town the next series of digits is the post code.

So the logic to parse the documents from NZ Post is very very simple, although given the copyright probably illegal. I estimate maybe 2-3 hours of coding and you’d have the whole database in 100% accurate form straight from the horses mouth so to speak.

Thx.

Will be very useful.

Roslyn are suburbs in Palmerston North & Dunedin.

EXAMPLE:

SELECT *
FROM `suburbs`
WHERE `postcode` = 4414

Returns(3 rows I’ve removed 2 for brevity):

907 Roslyn Palmerston North 4414

So the suburb id is 907

BUT…

if we do:

SELECT *
FROM `streets`
WHERE `postcode` = 4414

and we find Tremaine Ave in the result set we see:

39456 Tremaine Avenue 647-1057 4414 0 906 175.602 -40.343 1
Edit Delete 39457 Tremaine Avenue 666-1050 4414 2 906 175.602 -40.343 1

we see the suburb id is 906 ( tremaine ave, 4412 is in Palmerston North)

…

SELECT *
FROM `suburbs`
WHERE `suburbid` =906

returns:

906 Roslyn Dunedin 9010

Regards Brett

How do I use your software? Run it against a csv file or ???

cheers

- Rachel

Many thanks for putting in the work to provide this - I started scraping PDFs and it’s nasty work.

I’ve found a problem with the suburb table though. If you look at the ‘Vernon Terrace’ street field, join the suburb table onto the postcode column and join another suburb table onto the suburb column. Basically the suburb column and postcode column contradict eachother: suburb reads Hillsborough in Auckland and postcode reads Christchurch.

I hope this is of help to you, and again many thanks for doing this!