Fix PC encoding for Brazil

pgeocode.py

@@ -10,6 +10,7 @@
 import numpy as np
 import pandas as pd
 
+
 __version__ = "0.2.1"
 
 STORAGE_DIR = os.environ.get(
@@ -250,8 +251,26 @@ def query_postal_code(self, codes):
         if isinstance(codes, str):
             codes = [codes]
             single_entry = True
+            check_zip_code = _CheckPostalCode(codes)
+            try:
+                if not check_zip_code(codes):
+                    warnings.warn(
+                        f"{codes} is a invalid postal code for {self.country}"
+                    )
+            except NotImplementedError:
+                pass
         else:
             single_entry = False
+            for code in codes:
+                check_zip_code = _CheckPostalCode(code)
+                try:
+                    if not check_zip_code(code):
+                        warnings.warn(
+                            f"{code} is a invalid"
+                            + f"postal code for {self.country}"
+                        )
+                except NotImplementedError:
+                    pass
 
         if not isinstance(codes, pd.DataFrame):
             codes = pd.DataFrame(codes, columns=["postal_code"])
@@ -385,3 +404,35 @@ def haversine_distance(x, y):
 
     c = 2 * np.arcsin(np.sqrt(a))
     return EARTH_RADIUS * c
+
+
+class _CheckPostalCode:
+    def __init__(self, country):
+        self.country = country
+        self.implemented = ["BR"]
+
+    def __call__(self, postal_code):
+        if self.country not in self.implemented:
+            raise NotImplementedError
+        else:
+            return self._check_is_valid_zip_code(postal_code)
+
+    def _is_valid_zip_brazil(self, zip_code):
+        # in the Brazil de zip_codes valids in Geoname are terminated
+        # with 000 e have 8 digits
+        if isinstance(zip_code, list):
+            zip_code = "".join(zip_code)
+        if isinstance(zip_code, str) and zip_code.find("-") != -1:
+            zip_code = "".join(zip_code.split("-"))
+
+        if not isinstance(zip_code, str):
+            zip_code = str(zip_code)
+
+        return all([len(zip_code) == 8, zip_code[5:] == "000"])
+
+    def _check_is_valid_zip_code(self, zip_code):
+        # each country have a deferent postal code format
+        if isinstance(zip_code, list):
+            zip_code = "".join(zip_code)
+
+        return self._is_valid_zip_brazil(zip_code)

test_pgeocode.py

@@ -11,7 +11,12 @@
 from numpy.testing import assert_allclose, assert_array_equal
 
 import pgeocode
-from pgeocode import GeoDistance, Nominatim, haversine_distance
+from pgeocode import (
+    GeoDistance,
+    Nominatim,
+    haversine_distance,
+    _CheckPostalCode,
+)
 
 
 @pytest.fixture
@@ -179,3 +184,21 @@ def test_haversine_distance():
     d_pred = haversine_distance(x, y)
     # same distance +/- 3 km
     assert_allclose(d_ref, d_pred, atol=3)
+
+
+class TestCheckPostalCode:
+    """ Test for class CheckPostalCode"""
+
+    a = _CheckPostalCode("BR")
+
+    def test_is_valid_zip_brazil_false(self, zip_code="69000-010"):
+        assert self.a._is_valid_zip_brazil(zip_code) is False
+
+    def test_check_is_valid_zip_code_false(self, zip_code="69000-010"):
+        assert self.a._check_is_valid_zip_code(zip_code) is False
+
+    def test_is_valid_zip_brazil_true(self, zip_code="69000-000"):
+        assert self.a._is_valid_zip_brazil(zip_code) is True
+
+    def test_check_is_valid_zip_code_true(self, zip_code="69000-000"):
+        assert self.a._check_is_valid_zip_code(zip_code) is True