Bugfix/global filtering

CHANGELOG.md

@@ -4,6 +4,12 @@ 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/).
 
+## [0.10.13] - 2024-10-16
+
+### Added
+
+- Added global food security data with additional p-code mappings
+
 ## [0.10.12] - 2024-10-16
 
 ### Fixed

requirements.txt

@@ -60,7 +60,7 @@ hdx-python-api==6.3.4
     # via
     #   hapi-pipelines (pyproject.toml)
     #   hdx-python-scraper
-hdx-python-country==3.8.1
+hdx-python-country==3.8.2
     # via
     #   hapi-pipelines (pyproject.toml)
     #   hdx-python-api
@@ -228,7 +228,7 @@ ruamel-yaml==0.18.6
     # via hdx-python-utilities
 ruamel-yaml-clib==0.2.8
     # via ruamel-yaml
-setuptools==75.1.0
+setuptools==75.2.0
     # via ckanapi
 shellingham==1.5.4
     # via typer

src/hapi/pipelines/app/pipelines.py

@@ -246,6 +246,7 @@ def output_food_security(self):
                 admins=self.admins,
                 adminone=self.adminone,
                 admintwo=self.admintwo,
+                countryiso3s=self.countries,
                 configuration=self.configuration,
             )
             food_security.populate()
@@ -311,6 +312,7 @@ def output_funding(self):
             funding = Funding(
                 session=self.session,
                 metadata=self.metadata,
+                countryiso3s=self.countries,
                 locations=self.locations,
                 configuration=self.configuration,
             )

src/hapi/pipelines/configs/core.yaml

@@ -44,6 +44,7 @@ admin1:
     "CMR|Far-North": "CM004"
     "CMR|West": "CM008"
     "CPV|Santo Antao": "CV06"
+    "DJI|Djibouti Ville": "DJ04"
     "ETH|B. Gumuz": "ET06"
     "HTI|Nord-Ouest": "HT09"
     "HTI|Sud": "HT07"
@@ -90,6 +91,8 @@ admin2:
     "AF08|Onaba": "AF0805"
     "AF14|Khost": "AF1401"
     "AF23|Chaghcharan": "AF2301"
+    "AO09|Município dos Gambos (ex-chiange)": "AO09073"
+    "AO16|Mocamedes": "AO16139"
     "CD43|Ville de Gbadolite": "CD4301"
     "CD83|Territoire de LODJA": "CD8303"
     "CO08|Distrito Especial, Industrial Y Portuario De Barranquilla": "CO08001"
@@ -132,6 +135,7 @@ admin2:
     "NG027|Munya": "NG027018"
     "PH08|Samar": "PH08060"
     "PK2|Mekran": "PK211"
+    "PK5|Dera Ismail Khan": "PK509"
     "PS01|Ramallah and Albireh": "PS0130"
     "RUS|Sankt-peterburg": "RU004002"
     "SD01|Sharq El Nile": "SD01004"
@@ -144,6 +148,7 @@ admin2:
     "TD11|La kabbia": "TD1102"
     "TD17|Iriba": "TD1703"
     "TD22|Tibest-Ouest": "TD2202"
+    "TG05|Kpendial- Ouest": "TG0513"
     "TZ12|Mbozi": "TZ2606"  # TODO: this unit is not getting picked up because of the mismatched admin1
     "UA12|Apostolivskyi": "UA1206"
     "UA12|Marhanetska": "UA1208"
@@ -271,6 +276,10 @@ admin2:
     - "maguindanao"
     - "meru south"
     - "metropolitan manila"
+    - "GHA|ashanti"
+    - "GHA|bono"
+    - "GHA|bono east"
+    - "GHA|central"
 
 orphan_admin2s:
   SS0001: "SSD"

src/hapi/pipelines/configs/food_security.yaml

@@ -25,6 +25,7 @@ food_security:
   # This is where all "Areas" are non-admin units and so there is only admin 1
   # data available in "Level 1"
   adm1_only:
+    - "DJI"
     - "HTI"
     - "MMR"
     - "SOM"
@@ -34,6 +35,16 @@ food_security:
   adm2_only:
     - "PSE"
 
+  # This is where there is only admin 2 data available in "Level 1" whether
+  # "Area" is blank or not
+  adm2_in_level1:
+    - "LBN"
+
+  # This is where "Level 1" is non-admin units and so there is only admin 1
+  # data available in "Area"
+  adm1_in_area:
+    - "KEN"
+
   # The errors below were picked up from the logging which outputs all the
   # fuzzy matches
   adm1_errors:

src/hapi/pipelines/database/food_security.py

@@ -2,7 +2,7 @@
 
 from dataclasses import dataclass
 from logging import getLogger
-from typing import Dict, Optional, Set
+from typing import Dict, List, Optional, Set
 
 from hapi_schema.db_food_security import DBFoodSecurity
 from hdx.api.configuration import Configuration
@@ -38,13 +38,15 @@ def __init__(
         admins: admins.Admins,
         adminone: AdminLevel,
         admintwo: AdminLevel,
+        countryiso3s: List[str],
         configuration: Configuration,
     ):
         super().__init__(session)
         self._metadata = metadata
         self._admins = admins
         self._adminone = adminone
         self._admintwo = admintwo
+        self._countryiso3s = countryiso3s
         self._configuration = configuration
         self._country_status = {}
 
@@ -212,7 +214,23 @@ def process_subnational(
             admin_level == "admintwo"
             and countryiso3 in food_sec_config["adm1_only"]
         ):
-            return None
+            self._country_status[countryiso3] = (
+                "Level 1: Admin 1, Area: ignored"
+            )
+            adminoneinfo = self.get_adminoneinfo(
+                food_sec_config["adm_ignore_patterns"],
+                warnings,
+                dataset_name,
+                countryiso3,
+                row["Level 1"],
+            )
+            return self.get_adminone_admin2_ref(
+                food_sec_config,
+                warnings,
+                errors,
+                dataset_name,
+                adminoneinfo,
+            )
         # The YAML configuration "adm2_only" specifies locations where
         # "Level 1" is not populated and "Area" is admin 2. (These are
         # exceptions since "Level 1" would normally be populated if "Area" is
@@ -234,6 +252,43 @@ def process_subnational(
                 adminoneinfo,
             )
 
+        if countryiso3 in food_sec_config["adm2_in_level1"]:
+            row["Area"] = row["Level 1"]
+            row["Level 1"] = None
+            adminoneinfo = AdminInfo(countryiso3, "NOT GIVEN", "", None, False)
+            self._country_status[countryiso3] = (
+                "Level 1: Admin 2, Area: ignored"
+            )
+            return self.get_admintwo_admin2_ref(
+                food_sec_config,
+                warnings,
+                errors,
+                dataset_name,
+                row,
+                adminoneinfo,
+            )
+
+        if countryiso3 in food_sec_config["adm1_in_area"]:
+            if admin_level == "adminone":
+                return None
+            self._country_status[countryiso3] = (
+                "Level 1: ignored, Area: Admin 1"
+            )
+            adminoneinfo = self.get_adminoneinfo(
+                food_sec_config["adm_ignore_patterns"],
+                warnings,
+                dataset_name,
+                countryiso3,
+                row["Area"],
+            )
+            return self.get_adminone_admin2_ref(
+                food_sec_config,
+                warnings,
+                errors,
+                dataset_name,
+                adminoneinfo,
+            )
+
         adminone_name = row["Level 1"]
 
         if not adminone_name:
@@ -334,7 +389,7 @@ def populate(self) -> None:
                 if "#" in row["Date of analysis"]:  # ignore HXL row
                     continue
                 countryiso3 = row["Country"]
-                if countryiso3 not in self._configuration["HAPI_countries"]:
+                if countryiso3 not in self._countryiso3s:
                     continue
                 provider_admin1_name = get_provider_name(row, "Level 1")
                 provider_admin2_name = get_provider_name(row, "Area")

src/hapi/pipelines/database/funding.py

@@ -1,6 +1,7 @@
 """Functions specific to the funding theme."""
 
 from logging import getLogger
+from typing import List
 
 from hapi_schema.db_funding import DBFunding
 from hdx.api.configuration import Configuration
@@ -21,11 +22,13 @@ def __init__(
         self,
         session: Session,
         metadata: Metadata,
+        countryiso3s: List[str],
         locations: locations,
         configuration: Configuration,
     ):
         super().__init__(session)
         self._metadata = metadata
+        self._countryiso3s = countryiso3s
         self._locations = locations
         self._configuration = configuration
 
@@ -45,6 +48,8 @@ def populate(self) -> None:
             if dataset["archived"]:
                 continue
             admin_code = dataset.get_location_iso3s()[0]
+            if admin_code not in self._countryiso3s:
+                continue
             resource = [
                 r
                 for r in dataset.get_resources()