Add tests for fuzzy matching

OCHA-DAP · Oct 31, 2023 · 8fb6ff5 · 8fb6ff5
1 parent 1c3286d
commit 8fb6ff5
Show file tree

Hide file tree

Showing 4 changed files with 76 additions and 3 deletions.
diff --git a/...dzgev9-a6yu7vtcewhblm56aul1iifrfv5ubvvjvt7b-pub-gid-1040329566-single-true-output-csv.csv b/...dzgev9-a6yu7vtcewhblm56aul1iifrfv5ubvvjvt7b-pub-gid-1040329566-single-true-output-csv.csv
@@ -127,4 +127,4 @@ Association Malienne pour la Solidarité et le Développement,,,,,,,,
 Care Best Initiative,,,,,,,,
 AECID,,,,,,,,
 Norwegian Project Office / Rural Rehabilitation Association for Afghanistan,,,,,,,,
-Social and Humanitarian Assistance Organization,,,,,,,,
+Social and Humanitarian Assistance Organization,,,,,,,,
diff --git a/...org_type_12nkxos71u6lkttpyytfwi2bdq9yen1j7jplhocn96-0-export-format-csv-gid-797752596.csv b/...org_type_12nkxos71u6lkttpyytfwi2bdq9yen1j7jplhocn96-0-export-format-csv-gid-797752596.csv
@@ -13,4 +13,4 @@ HR.info ID,Preferred Term,Scope,Date Added (dd/mm/yyyy),Date updated (dd/mm/yyyy
 444,Private sector,"Includes private sector organizations such as corporate charity arms or foundations, for-profit recruitment firms and consultancies",16/03/2017,,,,
 445,Red Cross / Red Crescent,"Includes IFRC, ICRC and national Red Cross/Crescent societies",16/03/2017,,276,Red Cross/Red Crescent Movement,Exact Match
 446,Religious,Includes faith-based organizations and organizations with affiliations to religious bodies,16/03/2017,,,,
-447,United Nations,"Includes all UN agencies, funds and programmes",16/03/2017,,272,International Organization,Broader
+447,United Nations,"Includes all UN agencies, funds and programmes",16/03/2017,,272,International Organization,Broader
diff --git a/.../sector_1sxsircxhxmzce0pwafcht-whjbdi9uqoefesubilgc4-export-format-csv-gid-2037574879.csv b/.../sector_1sxsircxhxmzce0pwafcht-whjbdi9uqoefesubilgc4-export-format-csv-gid-2037574879.csv
@@ -14,4 +14,4 @@ HRinfo ID,Preferred Term,Preferred Term (fr),ACRONYM,Group Type,Homepage,Date Cr
 2,Early Recovery,Relèvement rapide,ERY,Cluster,http://earlyrecovery.global/,03/16/2017,4/4/2018,Added French name
 5,Emergency Telecommunications,Télécommunications d'urgence,TEL,Cluster,https://www.etcluster.org/,03/16/2017,4/4/2018,Added French name
 5404,Gender Based Violence,Violences basées sur le genre,PRO-GBV,Area of Responsibility (Sub-cluster),http://gbvaor.net/,03/16/2017,4/5/2018,Updated French name based feedback from OCHA-FIS
-5405,"Housing, Land and Property","Logement, terre, et biens",PRO-HLP,Area of Responsibility (Sub-cluster),http://www.globalprotectioncluster.org/en/areas-of-responsibility/housing-land-and-property.html,03/16/2017,4/4/2018,Added French name from 2012 Côte d'Ivoire workshop report
+5405,"Housing, Land and Property","Logement, terre, et biens",PRO-HLP,Area of Responsibility (Sub-cluster),http://www.globalprotectioncluster.org/en/areas-of-responsibility/housing-land-and-property.html,03/16/2017,4/4/2018,Added French name from 2012 Côte d'Ivoire workshop report
diff --git a/tests/test_mappings.py b/tests/test_mappings.py
@@ -0,0 +1,73 @@
+from hapi.pipelines.utilities.mappings import get_code_from_name
+
+
+def test_get_code_from_name():
+    sector_lookup = {
+        "Emergency Shelter and NFI": "SHL",
+        "Camp Coordination / Management": "CCM",
+        "Mine Action": "PRO - MIN",
+        "Food Security": "FSC",
+        "Water Sanitation Hygiene": "WSH",
+        "Logistics": "LOG",
+        "Child Protection": "PRO - CPN",
+        "Protection": "PRO",
+        "Education": "EDU",
+        "Nutrition": "NUT",
+        "Health": "HEA",
+        "Early Recovery": "ERY",
+        "Emergency Telecommunications": "TEL",
+        "Gender Based Violence": "PRO - GBV",
+        "Housing, Land and Property": "PRO - HLP",
+    }
+    sector_map = {
+        "abris": "SHL",
+        "cccm": "CCM",
+        "coordination": "CCM",
+        "education": "EDU",
+        "eha": "WSH",
+        "erl": "ERY",
+        "nutrition": "NUT",
+        "operatioanl presence: water, sanitation & hygiene": "WSH",
+        "operational presence: education in emergencies": "EDU",
+        "operational presence: emergency shelter & non-food items": "SHL",
+        "operational presence: food security & agriculture": "FSC",
+        "operational presence: health": "HEA",
+        "operational presence: nutrition": "NUT",
+        "operational presence: protection": "PRO",
+        "protection": "PRO",
+        "sa": "FSC",
+        "sante": "HEA",
+        "wash": "WSH",
+    }
+
+    org_type_lookup = {
+        "Academic / Research": "431",
+        "Donor": "433",
+        "Embassy": "434",
+        "Government": "435",
+        "International NGO": "437",
+        "International Organization": "438",
+        "Media": "439",
+        "Military": "440",
+        "National NGO": "441",
+        "Other": "443",
+        "Private sector": "444",
+        "Red Cross / Red Crescent": "445",
+        "Religious": "446",
+        "United Nations": "447",
+    }
+    org_type_map = {
+        "agence un": "447",
+        "govt": "435",
+        "ingo": "437",
+        "mouv. cr": "445",
+        "nngo": "441",
+        "ong int": "437",
+        "ong nat": "441",
+        "un agency": "447",
+    }
+    assert get_code_from_name("NATIONAL_NGO", org_type_lookup, org_type_map) == "441"
+    assert get_code_from_name("COOPÉRATION_INTERNATIONALE", org_type_lookup, org_type_map) is None
+    assert get_code_from_name("LOGISTIQUE", sector_lookup, sector_map) == "LOG"
+    assert get_code_from_name("CCCM", sector_lookup, sector_map) == "CCM"
+    assert get_code_from_name("Santé", sector_lookup, sector_map) == "HEA"