Looking for UDF in mapping for splitting street

[stefan_grube]

Hi experts,

I need to split a string into street, house number and supplement, for example:

"Hauptstr. 15 b" -> Street = Hauptstr, House number = 15, supplement = b

Does anyone have implemented this already and likes to share the Java code for UDF?

Thanks in advance

Stefan

[former_member184720 · Accepted solution]

var1 is your inpuy - Type String

var2 is the position - Type Int (1 for Street, 2- House Number, 3 - suppliment)

String[] splited = var1.split("\\s+");

if(splited.length > var2)

{

return splited[var2];

}

else

{

return "";

}

[stefan_grube · Accepted solution]

Thank you Hareesh, I think this helps me a lot.

Addresses not always have supplement, but sometimes the street comes with two parts:

Schlierseer Str. 3

Esterfelder Stiege 60 a

Nikolausstr. 15

Forsthausstr. 1-3 b

How can I identify, which part includes the number?

[former_member184720 · Accepted solution]

>>Addresses not always have supplement

This should be fine as long as there are no additional spaces in the input

>>>But sometimes the street comes with two parts; How can I identify, which part includes the number?

Hmm.. that's difficult to handle without knowing the pattern

At least does the street name ends with  "."  ?? like below

Schlierseer Str. 3

Nikolausstr. 15

Forsthausstr. 1-3 b

[stefan_grube · Accepted solution]

I have only some test data, so I cannot be sure, that real data come always like this.

I think I will check for numbers, as street name and supplement should not include any number.

[former_member184720 · Accepted solution]

If that's true then below should work..

String[] splited = var1.split("\\s+");

ArrayList<String> list = new ArrayList<String>();

list.add("");

int flag = 0;

for (int i = 0; i < splited.length; i++) {

if ((!splited[i].matches(".*\\d+.*"))) {

if (flag == 0) {

list.set(0, list.get(0) + " " + splited[i]);

}

else {

list.add(splited[i]);

}

}

else {

flag = 1;

list.add(splited[i]);

}

}

if (list.size() > var2) {

return list.get(var2);

}

else {

return "";

}

[engswee · Accepted solution]

Hi Stefan

Here is another option you can try out. It uses the capturing groups feature of regex.

I'm not sure what your source structure is, but the UDF below assumes the possibility that you might have multiple records having the street field. So it is written for execution type = "All values of a context", with single input and 3 output fields. Since it's all values of a context, it does not handle context change or suppress values.

You can change the delimiter into something that you won't be expecting in an address field, I just tried with the equals sign here, but it can be swapped with pipe (|), exclamation, question mark or some other character.

public void splitAddress(String[] address, ResultList street, ResultList house, ResultList supp, Container container)throws StreamTransformationException {
  String delimiter = "=";

  for(int i = 0; i < address.length; i++) {
   String extract = null;
   // Extract using regex capturing groups 
   if(address[i].matches(".*\\d$")) {
    extract = address[i].replaceAll("(\\D+)\\W+(\\d+[-]*\\d*)", "$1" + delimiter + "$2");
   } else {
    extract = address[i].replaceAll("(\\D+)\\W+(\\d+[-]*\\d*)\\W+(\\D+)", "$1" + delimiter + "$2" + delimiter + "$3");
   }
   // Split extracted content to output fields
   String[] lines = extract.split(delimiter);
   if(lines.length > 0) {
    street.addValue(lines[0]);
    house.addValue(lines[1]);
    if(lines.length > 2) {
     supp.addValue(lines[2]);
    } else {
     supp.addValue(""); // Optional, can be removed if null context is preferred
    }
   }
  }
 }

I've tested this on Eclipse on the 5 values you provided and it manages to split it for all 5 cases. It also handles the case where the number field has a dash in the middle.

Rgds

Eng Swee

[stefan_grube · Accepted solution]

This works perfect. Thank you a lot.