Merge pull request #216 from American-Institutes-for-Research/HEA-820/nom-de-mesure-locale-regex-updates

Address remaining missing regex pattern see HEA-820

Author: rhunwicks

pipelines/assets/livelihood_activity.py

@@ -243,7 +243,7 @@ def get_livelihood_activity_regexes() -> list:
         "nbr_pattern": r"(?:n[bo]?r?e?|no)\.?",
         "vendu_pattern": r"(?:quantité )?vendu(?:e|s|ss|es|ses)?",
         "separator_pattern": r" ?[:-]?",
-        "name_of_local_measure_pattern": r"(?:name of (?:meas(?:ure)?\.?)|nom(?: (?:de la mesure(?: locale)?|de mesure locale|du mesure|d'unité|mesure locale|unité de mesure))?)",
+        "name_of_local_measure_pattern": r"(?:name of (?:meas(?:ure)?\.?)|nom(?: (?:de la mesure(?: locale?)?|de mesure locale?|du mesure|d'unité|mesure(?: locale?)?|unité de mesure))?)",
     }
     # Compile the regexes
     compiled_regexes = []

pipelines_tests/test_assets/test_livelihood_activity_regexes.json

@@ -1068,5 +1068,13 @@
     "is_start": true,
     "product_id": "farine de ble",
     "attribute": "name_of_local_measure"
+  },
+  "nom de la mesure local": {
+    "attribute": "name_of_local_measure"
+  },
+  "eau pour le bétail: nom mesure": {
+    "is_start": true,
+    "product_id": "eau pour le bétail",
+    "attribute": "name_of_local_measure"
   }
 }