fix: keep known good locodes

Signed-off-by: Evan Prodromou <[email protected]>
Open-Earth-Foundation · Mar 21, 2024 · 5e9af68 · 5e9af68
1 parent 1136129
commit 5e9af68
Show file tree

Hide file tree

Showing 9 changed files with 12 additions and 5 deletions.
diff --git a/harmonize/data/processed/UNLOCODE_2023-2/Actor.csv b/harmonize/data/processed/UNLOCODE_2023-2/Actor.csv
@@ -55352,6 +55352,7 @@ GB BEO,city,Burrelton,GB-PKN,UNLOCODE:2023-2
 GB B39,city,Burridge,GB-HAM,UNLOCODE:2023-2
 GB BWR,city,Burrow,GB-SOM,UNLOCODE:2023-2
 GB BBI,city,Burrowbridge,GB-SOM,UNLOCODE:2023-2
+GB BPT,city,Burry Port,GB-CMN,UNLOCODE:2023-2
 GB BGH,city,Burscough,GB-LAN,UNLOCODE:2023-2
 GB BUE,city,Bursledon,GB-HAM,UNLOCODE:2023-2
 GB BEM,city,Burslem,GB-STS,UNLOCODE:2023-2

diff --git a/harmonize/data/processed/UNLOCODE_2023-2/Actor.delete.csv b/harmonize/data/processed/UNLOCODE_2023-2/Actor.delete.csv
@@ -253,7 +253,6 @@ FR ORY
 FR LYS
 GA MBY
 GA OGU
-GB BPT
 GB EMA
 GB ELL
 GB LGW

diff --git a/harmonize/data/processed/UNLOCODE_2023-2/ActorIdentifier.csv b/harmonize/data/processed/UNLOCODE_2023-2/ActorIdentifier.csv
@@ -55352,6 +55352,7 @@ GB BEO,GB BEO,UNLOCODE,UNLOCODE:2023-2
 GB B39,GB B39,UNLOCODE,UNLOCODE:2023-2
 GB BWR,GB BWR,UNLOCODE,UNLOCODE:2023-2
 GB BBI,GB BBI,UNLOCODE,UNLOCODE:2023-2
+GB BPT,GB BPT,UNLOCODE,UNLOCODE:2023-2
 GB BGH,GB BGH,UNLOCODE,UNLOCODE:2023-2
 GB BUE,GB BUE,UNLOCODE,UNLOCODE:2023-2
 GB BEM,GB BEM,UNLOCODE,UNLOCODE:2023-2

diff --git a/harmonize/data/processed/UNLOCODE_2023-2/ActorIdentifier.delete.csv b/harmonize/data/processed/UNLOCODE_2023-2/ActorIdentifier.delete.csv
@@ -253,7 +253,6 @@ FR ORY
 FR LYS
 GA MBY
 GA OGU
-GB BPT
 GB EMA
 GB ELL
 GB LGW

diff --git a/harmonize/data/processed/UNLOCODE_2023-2/ActorName.csv b/harmonize/data/processed/UNLOCODE_2023-2/ActorName.csv
@@ -64766,6 +64766,7 @@ GB BEO,Burrelton,und,0,UNLOCODE:2023-2
 GB B39,Burridge,und,0,UNLOCODE:2023-2
 GB BWR,Burrow,und,0,UNLOCODE:2023-2
 GB BBI,Burrowbridge,und,0,UNLOCODE:2023-2
+GB BPT,Burry Port,und,0,UNLOCODE:2023-2
 GB BGH,Burscough,und,0,UNLOCODE:2023-2
 GB BUE,Bursledon,und,0,UNLOCODE:2023-2
 GB BEM,Burslem,und,0,UNLOCODE:2023-2

diff --git a/harmonize/data/processed/UNLOCODE_2023-2/ActorName.delete.csv b/harmonize/data/processed/UNLOCODE_2023-2/ActorName.delete.csv
@@ -253,7 +253,6 @@ FR ORY
 FR LYS
 GA MBY
 GA OGU
-GB BPT
 GB EMA
 GB ELL
 GB LGW

diff --git a/harmonize/data/processed/UNLOCODE_2023-2/Territory.csv b/harmonize/data/processed/UNLOCODE_2023-2/Territory.csv
@@ -49363,6 +49363,7 @@ GB BVO,604833,-10167,UNLOCODE:2023-2
 GB BUF,588500,-29167,UNLOCODE:2023-2
 GB B39,508833,-12500,UNLOCODE:2023-2
 GB BBI,510667,-29167,UNLOCODE:2023-2
+GB BPT,516833,-42500,UNLOCODE:2023-2
 GB BGH,535833,-28333,UNLOCODE:2023-2
 GB UTO,524000,11500,UNLOCODE:2023-2
 GB KBK,537333,-1333,UNLOCODE:2023-2

diff --git a/harmonize/data/processed/UNLOCODE_2023-2/Territory.delete.csv b/harmonize/data/processed/UNLOCODE_2023-2/Territory.delete.csv
@@ -253,7 +253,6 @@ FR ORY
 FR LYS
 GA MBY
 GA OGU
-GB BPT
 GB EMA
 GB ELL
 GB LGW

diff --git a/harmonize/scripts/process_unlocode_2023.py b/harmonize/scripts/process_unlocode_2023.py
@@ -73,6 +73,13 @@
     "type"
 ]
 
+# Known good UNLOCODEs that we want to keep
+
+KNOWN_GOOD = [
+    'GB BPT',
+    'GB ELL'
+    'AU PBN'
+]
 
 def write_csv(name, rows):
     with open(f'{OUTPUT_DIR}/{name}.csv', mode='w') as csvfile:
@@ -138,7 +145,7 @@ def handle_input_row(row, subdivs):
         '|.+Airport'
     )
 
-    if re.search(regex, row["Name"]) or row["ISO 3166-1"].strip() == "XZ":
+    if (re.search(regex, row["Name"]) or row["ISO 3166-1"].strip() == "XZ") and not actor_id in KNOWN_GOOD:
         # We want to delete bad existing rows
 
         write_output_row("Actor.delete", ["actor_id"], {