Merge pull request #14 from googleads/new_address_fields

feat(Formatter): Add formatting methods for new GA4 address info fields

Author: lindsey-volta

google/ads/datamanager_util/format.py

@@ -184,6 +184,55 @@ def format_region_code(self, region_code: str) -> str:
             raise ValueError("Region code must be two characters")
         return region_code
 
+    def _format_location_string(self, value: str, label: str) -> str:
+        """Returns the normalized and formatted location string.
+
+        Args:
+          value: the string to format.
+          label: the label used for error messages.
+        Raises:
+          ValueError: If the provided value is invalid.
+        """
+        if value is None:
+            raise ValueError(f"{label} is None")
+        value = value.strip().lower()
+        # Removes all punctuation and special characters, leaving only
+        # alphanumeric characters and whitespace.
+        value = re.sub(r"[^\w\s]|_", "", value)
+        if not value:
+            raise ValueError(f"{label} is blank or empty")
+        return value
+
+    def format_address_line(self, address_line: str) -> str:
+        """Returns the normalized and formatted address line as a string.
+
+        Args:
+          address_line: the address line.
+        Raises:
+          ValueError: If the provided address line is invalid.
+        """
+        return self._format_location_string(address_line, "Address line")
+
+    def format_city(self, city: str) -> str:
+        """Returns the normalized and formatted city as a string.
+
+        Args:
+          city: the city.
+        Raises:
+          ValueError: If the provided city is invalid.
+        """
+        return self._format_location_string(city, "City")
+
+    def format_administrative_area(self, administrative_area: str) -> str:
+        """Returns the normalized and formatted administrative area as a string.
+
+        Args:
+          administrative_area: the administrative area.
+        Raises:
+          ValueError: If the provided administrative area is invalid.
+        """
+        return self._format_location_string(administrative_area, "Administrative area")
+
     def hash_string(self, s: str) -> bytes:
         """Returns bytes containing the hash of the string.
 
@@ -195,8 +244,7 @@ def hash_string(self, s: str) -> bytes:
         """
         if s is None:
             raise ValueError("String is None")
-        s = "".join(s.split())
-        if len(s) == 0:
+        if len(s.strip()) == 0:
             raise ValueError("String is blank or empty")
         return hashlib.sha256(s.encode()).digest()
 
@@ -350,6 +398,55 @@ def process_postal_code(self, postal_code: str) -> str:
         """
         return self.format_postal_code(postal_code)
 
+    def process_address_line(
+        self,
+        address_line: str,
+        encoding: Encoding,
+        encrypter: Optional[Encrypter] = None,
+    ) -> str:
+        """Formats, hashes, and encodes an address line.
+
+        Args:
+            address_line: The address line to process.
+            encoding: The encoding to use.
+            encrypter: An optional Encrypter to use for encryption.
+
+        Returns:
+            The processed address line.
+        """
+        formatted_address_line = self.format_address_line(address_line)
+        if encrypter:
+            return self._hash_encode_and_encrypt(
+                formatted_address_line, encoding, encrypter
+            )
+        return self._hash_and_encode(formatted_address_line, encoding)
+
+    def process_city(self, city: str) -> str:
+        """Processes a city.
+
+        This is a convenience method that simply calls format_city.
+
+        Args:
+            city: The city to process.
+
+        Returns:
+            The processed city.
+        """
+        return self.format_city(city)
+
+    def process_administrative_area(self, administrative_area: str) -> str:
+        """Processes an administrative area.
+
+        This is a convenience method that simply calls format_administrative_area.
+
+        Args:
+            administrative_area: The administrative area to process.
+
+        Returns:
+            The processed administrative area.
+        """
+        return self.format_administrative_area(administrative_area)
+
     def _hash_and_encode(
         self, normalized_string: str, encoding: Encoding
     ) -> str:

pyproject.toml

@@ -18,7 +18,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "google-ads-datamanager-util"
-version = "0.3.0"
+version = "0.4.0"
 description = "Utilities for the Data Manager API"
 readme = "./README.rst"
 requires-python = ">=3.9, <3.14"

tests/formatter_test.py

@@ -326,3 +326,114 @@ def test_process_postal_code_valid_inputs(self):
         self.assertEqual(
             "1229-076", self.formatter.process_postal_code(" 1229-076  ")
         )
+
+    def test_format_address_line_valid_inputs(self):
+        self.assertEqual(
+            "1800 amphibious blvd",
+            self.formatter.format_address_line(" 1800 Amphibious Blvd.  "),
+        )
+
+    def test_format_address_line_invalid_inputs(self):
+        with self.assertRaises(ValueError):
+            self.formatter.format_address_line(None)
+        with self.assertRaises(ValueError):
+            self.formatter.format_address_line("")
+        with self.assertRaises(ValueError):
+            self.formatter.format_address_line("   ")
+
+    def test_format_city_valid_inputs(self):
+        self.assertEqual(
+            "mountain view",
+            self.formatter.format_city(" Mountain View  "),
+        )
+        self.assertEqual(
+            "mountain view",
+            self.formatter.format_city("Mountain View,"),
+        )
+
+    def test_format_city_invalid_inputs(self):
+        with self.assertRaises(ValueError):
+            self.formatter.format_city(None)
+        with self.assertRaises(ValueError):
+            self.formatter.format_city("")
+        with self.assertRaises(ValueError):
+            self.formatter.format_city("   ")
+
+    def test_format_administrative_area_valid_inputs(self):
+        self.assertEqual(
+            "ca",
+            self.formatter.format_administrative_area(" CA  "),
+        )
+        self.assertEqual(
+            "california",
+            self.formatter.format_administrative_area(" California  "),
+        )
+        self.assertEqual(
+            "ca",
+            self.formatter.format_administrative_area("C.A."),
+        )
+        self.assertEqual(
+            "ca",
+            self.formatter.format_administrative_area("CA."),
+        )
+
+    def test_format_administrative_area_invalid_inputs(self):
+        with self.assertRaises(ValueError):
+            self.formatter.format_administrative_area(None)
+        with self.assertRaises(ValueError):
+            self.formatter.format_administrative_area("")
+        with self.assertRaises(ValueError):
+            self.formatter.format_administrative_area("   ")
+
+    def test_process_address_line_valid_inputs_hex_encoding(self):
+        encoded_hash = (
+            "ff75e73a0e768cc1fa28a64faebbceccb562d7c05f2ffcdd8d100abad73e4579"
+        )
+        self.assertEqual(
+            encoded_hash,
+            self.formatter.process_address_line(
+                " 1800 Amphibious Blvd.  ", Encoding.HEX
+            ),
+        )
+
+    def test_process_address_line_valid_inputs_base64_encoding(self):
+        encoded_hash = "/3XnOg52jMH6KKZPrrvOzLVi18BfL/zdjRAKutc+RXk="
+        self.assertEqual(
+            encoded_hash,
+            self.formatter.process_address_line(
+                " 1800 Amphibious Blvd.  ", Encoding.BASE64
+            ),
+        )
+
+    def test_process_city_valid_inputs(self):
+        self.assertEqual(
+            "mountain view",
+            self.formatter.process_city(" Mountain View  "),
+        )
+
+    def test_process_administrative_area_valid_inputs(self):
+        self.assertEqual(
+            "ca",
+            self.formatter.process_administrative_area(" CA  "),
+        )
+
+    def test_format_example_address(self):
+        # 1800 Amphibious Blvd.
+        # Mountain View, CA 94045
+        self.assertEqual(
+            "1800 amphibious blvd",
+            self.formatter.format_address_line("1800 Amphibious Blvd."),
+        )
+        self.assertEqual(
+            "mountain view",
+            self.formatter.format_city("Mountain View"),
+        )
+        self.assertEqual(
+            "ca",
+            self.formatter.format_administrative_area("CA."),
+        )
+        self.assertEqual(
+            "94045",
+            self.formatter.format_postal_code("94045"),
+        )
+