Fixing multiple caves with same kataser no

1 files changed, 31 insertions, 5 deletions

diff --git a/core/models_caves.py b/core/models_caves.py
index 2304109..c5d3945 100644
--- a/core/models_caves.py
+++ b/core/models_caves.py
@@ -5,6 +5,7 @@ import logging
 import re
 import json
 from subprocess import call
+from collections import defaultdict
 
 from urllib.parse import urljoin
 
@@ -21,6 +22,7 @@ from django.template import Context, loader
 
 from troggle.core.models import TroggleModel, Person, Expedition
 from troggle.core.models_survex import SurvexStation
+from troggle.core.models import DataIssue
 
 class Area(TroggleModel):
     short_name = models.CharField(max_length=100)
@@ -35,7 +37,7 @@ class Area(TroggleModel):
             return str(self.short_name)
 
     def kat_area(self):
-        if self.short_name in ["1623", "1626"]:
+        if self.short_name in ["1623", "1626", "1624", "1627"]:
             return self.short_name
         elif self.parent:
             return self.parent.kat_area()
@@ -515,30 +517,47 @@ def get_scan_path(instance, filename):
     return os.path.join('./',year,year+r'#'+number,str(instance.contents)+str(instance.number_in_wallet)+r'.jpg')
 
 Gcavelookup = None
+Gcave_count = None
 def GetCaveLookup():
     """lookup function modelled on GetPersonExpeditionNameLookup
     repeated assignment each call, needs refactoring
+    
+    Does NOT detect duplicates! Needs fixing.
+    Needs to be a proper funciton that raises an exception if there is a duplicate.
+    OR we could set it to return None if there are duplictes, and require the caller to 
+    fall back on doing the actual database query it wants rathe rthna using  this cache shortcut
     """
     global Gcavelookup
     if Gcavelookup:
         return Gcavelookup
-    Gcavelookup = {"NONEPLACEHOLDER":None}
+    Gcavelookup = {"NONEPLACEHOLDER": None}
+    global Gcave_count
+    Gcave_count  = defaultdict(int) # sets default value to int(0)
+    
     for cave in Cave.objects.all():
-        Gcavelookup[cave.official_name.lower()] = cave
+        key = cave.official_name.lower()
+        Gcavelookup[key] = cave
+        Gcave_count[key] += 1
         if cave.kataster_number:
-            Gcavelookup[cave.kataster_number] = cave
+            Gcavelookup[cave.kataster_number] = cave # DUPLICATE as we have 1623-55 and 1626-55
+            Gcave_count[cave.kataster_number]  += 1
         if cave.unofficial_number:
             Gcavelookup[cave.unofficial_number.lower()] = cave
+            Gcave_count[cave.unofficial_number.lower()] += 1
         if cave.filename:
             # this is the slug - usually..
             Gcavelookup[cave.filename.replace(".html","").lower()] = cave
+            Gcave_count[cave.filename.replace(".html","").lower()] += 1
         if cave.slug():
             slug = cave.slug()
             Gcavelookup[slug.lower()] = cave
+            Gcave_count[slug.lower()] += 1
     # These are exact matches! edit to check for prefix only!
     # mostly taken from expoweb/noinfo/cave-number-index 
     # and Becka's email of 25 may 2020 on new kataster numbers
     # this should be re-done as a JSON file upload
+    
+    # These might alse create more duplicate entries, so re-write it to check
     Gcavelookup["1987-02"] = Gcavelookup["267"]
     Gcavelookup["1990-01"] = Gcavelookup["171"]
     Gcavelookup["1990-02"] = Gcavelookup["172"]
@@ -624,7 +643,7 @@ def GetCaveLookup():
     Gcavelookup["2016-jb-01"] = Gcavelookup["289"]
     Gcavelookup["2017-pw-01"] = Gcavelookup["277"]
     Gcavelookup["2018-dm-07"] = Gcavelookup["359"]
-    Gcavelookup["2017_cucc_24"] = Gcavelookup["291"]
+    Gcavelookup["2017_cucc_24"] = Gcavelookup["291"] # note _ not - here
     Gcavelookup["2017_cucc_23"] = Gcavelookup["295"]
     Gcavelookup["2017_cucc_28"] = Gcavelookup["290"]
     Gcavelookup["bs17"] = Gcavelookup["283"]
@@ -679,5 +698,12 @@ def GetCaveLookup():
             addmore[id] = Gcavelookup[id].unofficial_number.lower()
     with open("cave-lookup.json", 'w') as f:
         json.dump(addmore, f)
+        
+    for c in Gcave_count:
+        if Gcave_count[c] > 1:
+            message = " ** Duplicate cave id: {}:{}:{}".format(Gcave_count[c], Gcavelookup[c], c)
+            #print(message)
+            #DataIssue.objects.create(parser='caves', message=message)
+            # logdataissues[Gcavelookup[c]]=message # pending troggle-wide issues logging system
 
     return Gcavelookup