use nuts code to detect country and subdivision

CHANGELOG.md

@@ -5,6 +5,10 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/)
 
 ## Unreleased
 
+### Added
+
+- Detection for DE and FR subdivisions using NUTS codes
+
 ### Dependencies
 
 - remove flake8

geoconvert/convert.py

@@ -2,10 +2,13 @@
 import re
 
 from .data import (
+    ALL_NUTS_CODES,
     BR_POSTCODES_RANGE,
     CA_POSTCODE_FIRST_LETTER_TO_PROVINCE_CODE,
     DE_HAUPTSTADT,
     DE_POSTCODE_RANGE,
+    NUTS_CODES_BY_COUNTRY,
+    all_nuts_regex,
     br_postcode_regex,
     br_state_code_regex,
     br_state_name_regex,
@@ -27,6 +30,7 @@
     fr_regions,
     language_to_capital_names,
     language_to_country_names,
+    nuts_regexes_by_country,
     us_postcode_regex,
     us_state_code_regex,
     us_state_name_regex,
@@ -150,6 +154,10 @@ def de_address_to_land_code(text):
     code_match = re.search(de_land_code_regex, text)
     if code_match:
         return code_match.group("code").upper()
+    # Look for NUTS code in the plain text
+    nuts_match = re.search(nuts_regexes_by_country["DE"], text)
+    if nuts_match:
+        return NUTS_CODES_BY_COUNTRY["DE"].get(nuts_match.group().upper())
 
 
 def de_postcode_to_land_code(text):
@@ -242,6 +250,10 @@ def fr_address_to_dept_code(text):
     code = fr_postcode_to_dept_code(text)
     if code is not None:
         return code
+    # Look for NUTS code in the plain text
+    nuts_match = re.search(nuts_regexes_by_country["FR"], text)
+    if nuts_match:
+        return NUTS_CODES_BY_COUNTRY["FR"].get(nuts_match.group().upper())
     # Look for the dept name in plain text
     return fr_dept_name_to_dept_code(text)
 
@@ -658,9 +670,16 @@ def _guess_country_and_subdivision_codes(text, lang=None):
     """
     Just guess the subdivision when no country is explicitly given.
     """
+    # Look for NUTS code in the plain text
+    nuts_match = re.search(all_nuts_regex, text)
+    if nuts_match:
+        nuts_code = nuts_match.group().upper()
+        return nuts_code[:2], ALL_NUTS_CODES.get(nuts_code)
+
     country_code, subdivision_code = _guess_country_then_subdivision_codes(text, lang)
     if country_code is not None:
         return country_code, subdivision_code
+
     return _guess_subdivision_then_country_codes(text)
 
 

geoconvert/data/__init__.py

@@ -39,6 +39,12 @@
     de_landers,
     de_postcode_regex,
 )
+from .subdivisions.nuts import (
+    ALL_NUTS_CODES,
+    NUTS_CODES_BY_COUNTRY,
+    all_nuts_regex,
+    nuts_regexes_by_country,
+)
 from .subdivisions.united_states import (
     us_postcode_regex,
     us_state_code_regex,