Hi Kam,

 

 

If the string always consists of: organization name (one or more any characters) <space> street name (one or more any characters, except space) <space> 'Street' <space> shire name (one or more any characters, except space) <space> lot number (one or more any characters)   StringSearcher with the following expression would extract the elements: ^(.+)\\s(\^\\s]+\\sStreet)\\s(\^\\s]+)\\s(.+)$

 

Takashi

---

Hi Takashi,

 

 

I guess that is one way of doing it, but I would need to repeat this process for Road, Drive, Way, Crescent etc etc... And then it will not be very accurate as Nottingham has quite a few double worded street names, i.e. Castle Bridge Road

 

 

Is there any way of doing it outside of RegEX too? I.e look into a DBF/CSV file that contains the street name in the string and then uses that street name to be the Regex calculation

 

 

Example:

 

Dataset 1 containing the full string

 

Dataset 2 containing street names

 

 

Both datasets enter one transformer

 

Transformer looks at first feature and checks to see if there is a street name within the string

 

IF So then the string searcher value = StreetName.*

 

 

Which would then strip out the rest of the string from the street name start... and do this for every feature it finds with a street name from the dataset 2 containing street names

 

 

 

I hope this makes sense.

 

 

---

Hi,

 

 

I would first try to make a match / relation between the organisations and their respective street names. I would try to do this in the database and not using FME, using something like (untested):

 

 

select *

 

from organisation, address

 

where address.street like '%' || organisation.name || '%'

 

 

You could do this with a SQLCreator, for instance. The result should be something one row for each organisation with the matching street name, like:

 

 

org_name

 

street_name

 

 

It would then be a simple matter of using a StringSearcher with a regexp that returns the part of org_name that preceeds street_name.

 

 

David

---

if that pattern. like mr.Takashi suggests is consistent, i would strip every line from right to left.

 

Using a regexp for the postalcode and zip leaves city, street and org.

 

Stripping city usig a city dataset leaves street and org.

 

Then stripping streets using your adres dataset wold leave Org.

 

At least in your example string.

 

 

 

etc.

---