Avoid disambiguation from street names and city names in fr_dept_name…

…_to_dept_code
jurismarches · Jan 8, 2025 · 25712e5 · 25712e5
1 parent aab94a4
commit 25712e5
Show file tree

Hide file tree

Showing 3 changed files with 42 additions and 6 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -3,6 +3,10 @@ All notable changes to this project will be documented in this file.
 
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/)
 
+### Fixed
+
+- Avoid disambiguation from street names and city names in fr_dept_name_to_dept_code
+
 ## [5.3.1] - 2024-11-06
 
 ### Added

diff --git a/geoconvert/convert.py b/geoconvert/convert.py
@@ -296,11 +296,38 @@ def fr_postcode_to_dept_code(text):
 # Keep backward compatibility
 address_to_zipcode = fr_postcode_to_dept_code
 
+# Avoid "rue de Paris" situations
+fr_street_names_re = "|".join(
+    (
+        "boulevard",
+        "avenue",
+        "chemin",
+        "rue",
+        "route",
+        "impasse",
+        "place",
+        "passage",
+        "ruelle",
+        "quai",
+        "all.e",
+    )
+)
+fr_street_name_cleaning_re = re.compile(
+    rf"({fr_street_names_re})[^\d\(,\n-]{{,20}}", flags=re.I
+)
+# Avoid "Ville-sur-Loire" situations
+fr_town_name_cleaning_re = re.compile(r"\w+.\b(sous|sur)\b.(\w+)", flags=re.I)
+
 
 def fr_dept_name_to_dept_code(text):
     """
     Return the departement number from the departement name
     """
+    # Avoid "rue de Paris" situations
+    text = fr_street_name_cleaning_re.sub("", text)
+    # Avoid "Ville-sur-Loire" situations
+    text = fr_town_name_cleaning_re.sub("", text)
+
     # There is no space in french dept names, but hyphens instead.
     text = safe_string(text).replace(" ", "-")
 

diff --git a/tests/test_subdivisions/test_france.py b/tests/test_subdivisions/test_france.py
@@ -228,12 +228,17 @@ def test_fr_postcode_to_dept_code(self, input_data, expected):
             ("Loire Atlanti)que", "44"),
             ("Yonne", "89"),
             ("Saint Pierre et Miquelon", "975"),
-            ("Tout savoir sur Saint Barthélemy", "977"),
-            ("Tout savoir sur saint-barthelemy", "977"),
-            ("Tout savoir sur saint Barthélémy", "977"),
-            # There may be some mistakes, so be careful what is passed
-            ("Rue de la Réunion, 75000 Paris", "974"),
-            ("Rue de l'Orne, 44800 Saint-Herblain", "61"),
+            ("Tout savoir à propos de Saint Barthélemy", "977"),
+            ("Tout savoir à propos de saint-barthelemy", "977"),
+            ("Tout savoir à propos de saint Barthélémy", "977"),
+            # Avoid disambiguations
+            # due to street names
+            ("Rue de la Réunion 61000 Alencon", None),  # "réunion" could mean 974
+            ("Rue de Paris, Nantes", None),  # "paris" could mean 75
+            ("Rue de l'Orne, 44800 Saint-Herblain", None),  # "Orne" could be 61
+            # due to city names
+            ("Sully sur Loire (Loiret)", "45"),  # "Loire" could mean 42
+            ("Gournay-sous-Marne (Seine saint Denis)", "93"),  # "Marne" could be 51
         ],
     )
     def test_fr_dept_name_dept_code(self, input_data, expected):