Merge pull request #291 from Shopify/handle-multi-word-joiner

Handle multiple word joiners when splitting addresses

CHANGELOG.md

@@ -29,6 +29,9 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 - Nil.
 ---
 
+## [1.12.0] - 2024-10-11
+- Update Address splitting methods to split on the first delimiter [#291](https://github.com/Shopify/worldwide/pull/291)
+
 ## [1.11.1] - 2024-10-02
 - Configure regexp timeout in Worldwide#Phone [#290](https://github.com/Shopify/worldwide/pull/290)
 

Gemfile.lock

@@ -13,7 +13,7 @@ GIT
 PATH
   remote: .
   specs:
-    worldwide (1.11.1)
+    worldwide (1.12.0)
       activesupport (>= 7.0)
       i18n
       phonelib (~> 0.8)

lang/typescript/.changeset/address-split-multi-word-joiners.md

@@ -0,0 +1,5 @@
+---
+'@shopify/worldwide': patch
+---
+
+Handle input having multiple word joiners in address-splitting functions

lang/typescript/src/extended-address/splitAddress1.test.ts

@@ -82,6 +82,13 @@ describe('splitAddress1', () => {
     });
   });
 
+  test('splits the address on the first delimiter', () => {
+    expect(splitAddress1('CL', 'Main \u2060123\u2060abc', false)).toEqual({
+      streetName: 'Main',
+      streetNumber: '123\u2060abc',
+    });
+  });
+
   test('returns full address object when separated by delimiter and decorator', () => {
     expect(splitAddress1('BR', 'Main, \u2060123', false)).toEqual({
       streetName: 'Main',

lang/typescript/src/extended-address/splitAddress2.test.ts

@@ -38,6 +38,13 @@ describe('splitAddress2', () => {
     });
   });
 
+  test('splits the address on the first delimiter', () => {
+    expect(splitAddress2('CL', 'dpto 4 \u2060Centro\u2060abc')).toEqual({
+      line2: 'dpto 4',
+      neighborhood: 'Centro\u2060abc',
+    });
+  });
+
   describe('language override', () => {
     test("returns default format if language override doesn't match", () => {
       expect(splitAddress2('TW', '30號\u2060大甲區')).toEqual({

lang/typescript/src/utils/address-fields.ts

@@ -45,10 +45,12 @@ export function splitAddressField(
   fieldDefinition: FieldConcatenationRule[],
   concatenatedAddress: string,
 ): Partial<Address> {
-  const values = concatenatedAddress.split(RESERVED_DELIMITER);
+  const [firstField, ...rest] = concatenatedAddress.split(RESERVED_DELIMITER);
+  const secondField = rest.join(RESERVED_DELIMITER);
+  const values = [firstField, secondField];
 
-  const parsedAddressObject = values.reduce((obj, value, index) => {
-    if (value !== '') {
+  const parsedAddressObject = fieldDefinition.reduce((obj, field, index) => {
+    if (values[index]) {
       // Decorator is included as a suffix in the previous sub-field value,
       // so we need to strip it from the current field by looking ahead at the
       // next field's definition
@@ -57,12 +59,15 @@ export function splitAddressField(
       const fieldValue =
         nextFieldDecorator &&
         nextFieldDecorator.length > 0 &&
-        value.endsWith(nextFieldDecorator)
-          ? value.substring(0, value.length - nextFieldDecorator.length)
-          : value;
+        values[index].endsWith(nextFieldDecorator)
+          ? values[index].substring(
+              0,
+              values[index].length - nextFieldDecorator.length,
+            )
+          : values[index];
       return {
         ...obj,
-        [fieldDefinition[index].key]: fieldValue,
+        [field.key]: fieldValue,
       };
     }
 

lib/worldwide/address.rb

@@ -167,13 +167,15 @@ def concatenate_address2
 
     def split_address1
       additional_fields = region.combined_address_format.dig(script_from_string(address1), "address1") || []
-      split_fields_arr = address1&.split(RESERVED_DELIMITER) || []
+      number_of_fields = additional_fields.size
+      split_fields_arr = address1&.split(RESERVED_DELIMITER, number_of_fields) || []
       split_fields(additional_fields, split_fields_arr)
     end
 
     def split_address2
       additional_fields = region.combined_address_format.dig(script_from_string(address2), "address2") || []
-      split_fields_arr = address2&.split(RESERVED_DELIMITER) || []
+      number_of_fields = additional_fields.size
+      split_fields_arr = address2&.split(RESERVED_DELIMITER, number_of_fields) || []
       split_fields(additional_fields, split_fields_arr)
     end
 

lib/worldwide/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Worldwide
-  VERSION = "1.11.1"
+  VERSION = "1.12.0"
 end

test/worldwide/address_test.rb

@@ -251,6 +251,13 @@ class AddressTest < ActiveSupport::TestCase
       assert_equal expected_hash, address.split_address1
     end
 
+    test "split_address1 splits on the first delimiter" do
+      address = Address.new(address1: "Main Street⁠123⁠ n° 145", country_code: "BR")
+      expected_hash = { "street_name" => "Main Street", "street_number" => "123⁠ n° 145" }
+
+      assert_equal expected_hash, address.split_address1
+    end
+
     test "split_address1 returns only street number if no string is present before the delimiter" do
       cl_address = Address.new(address1: "⁠123", country_code: "CL")
       br_address = Address.new(address1: "⁠123", country_code: "BR")
@@ -349,6 +356,13 @@ class AddressTest < ActiveSupport::TestCase
       assert_equal expected_hash, br_address.split_address2
     end
 
+    test "split_address2 splits on the first delimiter" do
+      address = Address.new(address2: "1樓⁠士林區⁠123", country_code: "TW")
+      expected_hash = { "line2" => "1樓", "neighborhood" => "士林區⁠123" }
+
+      assert_equal expected_hash, address.split_address2
+    end
+
     test "split_address2 returns line2 and neighborhood when both values are present and separated by a delimiter" do
       cl_address = Address.new(address2: "1樓⁠士林區", country_code: "TW")
       expected_hash = { "line2" => "1樓", "neighborhood" => "士林區" }