topic Re: A better way to parse an address? in Python Questions

A better way to parse an address?

JoeBorgione — Fri, 10 Dec 2021 20:43:02 GMT

I looked at the standardize address tool to see if it would do what I need and it works for most of my address data, but there are a number that don't fit the perfect model.( See Error Trapping )

At any rate, holding the rank of Hack Specialist .1 in the Python Legion, I'm running a series of scripts to parse out address components from a single string address in the form of

1234 S Main ST or 1234 E Olive Branch Dr or 1234 S 300 E.

The house number, pre-dir and suf-type/suf-dir aren't too bad, but teasing the street itself out is a little more challenging as it may be multiple words. What I've come up with is a series of splits and joins that get the job done, but there has got to be a better way. Any pointers are appreciated.

Here is what I do:

def myStreetName(inString):
  a = inString.split(' ')
  b = a.pop()                    #takes off suf
  c = ' '.join(a)                  # put it back together
  d = c.split(' ')                 #split it back out again
  e = d.pop(0)                 #takes off housenum
  f = ' '.join(d)                  #put it back together again
  g = f.split(' ')                 # split again
  h = g.pop(0)                 # get rid of pre-dir
  street = ' '.join(g)          #leaves just the street
  return street                 #home free

myStreetName(!fullAddress!)

Re: A better way to parse an address?

BruceHarold — Wed, 12 Aug 2015 16:04:12 GMT

Hi Joe

You have outed us on Standardize Addresses having issues at 10.x.

The easiest thing to do is download the 9.x styles and use Standardize Addresses with an old style.

http://www.arcgis.com/home/item.html?id=d36e3c27f12342d3b54f697048c71658

For interest, attached is a script I did for a tool for internal use - to find the parts of an address. It is far from bulletproof.

Another, canonical, way to standardize addresses (worldwide) is to leverage the World Geocode Service, for example:

http://geocode.arcgis.com/arcgis/rest/services/World/GeocodeServer/find?text=380%20new%20york%20st%20redlands&outFields=…

You grab the returned JSON and take it from there.

The only catch is if you're storing the results you must use the forStorage parameter and use a token.

Regards

Re: A better way to parse an address?

DarrenWiens2 — Fri, 10 Dec 2021 20:43:05 GMT

Your split string ('a') already contains a usable list of all you need.

If your addresses always take the form:

- first element = house number

- second element = pre-dir

- last element = suf-dir

- street name = everything between second and last element

... you can parse it apart like this:

>>> inString = '1234 E Olive Branch Dr'
... splitString = inString.split(' ')
... a = splitString[0]
... b = splitString[1]
... c = ' '.join(splitString[2:-1]) # everything between 2nd and last element, joined back together with a space
... d = splitString[-1]
... print (a,b,c,d)
...
('1234', 'E', 'Olive Branch', 'Dr')

Re: A better way to parse an address?

DanPatterson_Retired — Fri, 10 Dec 2021 20:43:08 GMT

>>> inString = "1234 E Olive Branch Dr"
>>> a = (inString.strip()).split(" ")
>>> addr = a[0]
>>> NSEW = [i for i in a if len(i)==1][0]
>>> pre = "{} {}".format(addr,NSEW)
>>> street = inString.replace(pre,"").strip()
>>> street

eg.. 'Main ST' 'Olive Branch Dr' '300 E' (not sure of the last on...just in case there was an E street N

We really should quit playing around....

Re: A better way to parse an address?

JoeBorgione — Wed, 12 Aug 2015 16:48:59 GMT

Bruce: Been using 9.3 locators, but now I'm being 'forced' into the more modern world....

Darren- that's what I'm talking about! Thanks!

Dan- not sure I follow: MAIN, OLIVE BRANCH, and 300 are all the valid StreetNames for the examples.

Re: A better way to parse an address?

BruceHarold — Wed, 12 Aug 2015 16:52:02 GMT

Standardize Addresses doesn't use a locator, it uses a style, so for the US you should be safe to install the old stuff and carry on.

Re: A better way to parse an address?

DanPatterson_Retired — Wed, 12 Aug 2015 16:53:54 GMT

Ahhh I had assumed 'main st' etc would be valid names..oh well another slice

Re: A better way to parse an address?

MichaelHilstrom — Thu, 17 Dec 2015 17:33:27 GMT

hello,

I am looking for a way to batch address parsing. Is there a python script?

Mike H

Re: A better way to parse an address?

JoeBorgione — Thu, 17 Dec 2015 17:52:12 GMT

I just use the python code provided above and run it a few times to calculate the values of each field. I'm sure some young, ambitious, aspiring GIS analyst could work that into a tool that is just a double click away from greatness. But since I'm just a cranky, gray bearded, old school GIS analyst, I do it my way...

Re: A better way to parse an address?

MichaelHilstrom — Thu, 17 Dec 2015 18:00:20 GMT

Joe,

Where is the python script? I cannot see it…

PS:I am new to the discussion groups.

I was able to download a python script last year for exporting attribute tables to excel spread sheets.

Thanks,

Michael Hilstrom, R.G.

Environmental Consultant

ADOA - Risk Management Division l State of Arizona

100 North 15th Avenue, Suite 301, Phoenix, AZ 85007

p: 602.542.2863 | michael.hilstrom@azdoa.gov<mailto:michael.hilstrom@azdoa.gov>

https://staterisk.az.gov<https://staterisk.az.gov/>

How am I doing? Please take a moment to answer a few questions.

https://www.surveymonkey.com/r/VOCRiskLossPrev

Re: A better way to parse an address?

JoeBorgione — Fri, 10 Dec 2021 20:43:10 GMT

Right in Darrens' post above. This is how I run it in the attribute calculator:

def myParser(inString):
  splitString = inString.split(' ')       ###splits the address into a list
  a = splitString[0]                         ### list item 0 is the house number          
  b = splitString[1]                         ### list item 1 is the prefix                
  c = ' '.join(splitString[2:-1])         ###  this rerurns the street name
  d = splitString[-1]                       ### this returns the suf dir or street type            
  return a                                     ### depending on what you want to return,
  #return b                                   ### un-comment and re-comment out the
  #return c                                   ### appropriate value(s)
  #return d‍‍‍‍‍‍‍‍‍‍

It'll look something like this where I am calculating the value of HouseNumber using the field PT_ADD as the input:

Re: A better way to parse an address?

MichaelHilstrom — Thu, 17 Dec 2015 18:21:56 GMT

Ok, Thanks I will try it in the field calculator…I am a bit inexperienced…I used the field calc. before, but that was over 10 years ago..

I really do appreciate the help

Thanks, Mike

Re: A better way to parse an address?

JoshuaBixby — Thu, 17 Dec 2015 19:39:45 GMT

Like all other tools, the best address parser is the one that someone else already wrote. Lately, I have been using usaddress and have been quiet happy with it. Granted, it doesn't validate addresses, but it also doesn't carry all the bulky overhead of other geocoding services when all I want to do is parse addresses.

Re: A better way to parse an address?

TedKowal — Mon, 19 Jun 2017 16:01:29 GMT

There is a lot of good stuff listed above, In the course of my work I occasionally have to parse addresses and I get data address dumps from our tax division which are horribly bad! So I have been slowly working on a parser to accommodate this data.... for what it is worth .... (I am still learning python)