JSON cache per python instance - working

1 files changed, 21 insertions, 4 deletions

diff --git a/core/models/wallets.py b/core/models/wallets.py
index 1cc42e8..b357344 100644
--- a/core/models/wallets.py
+++ b/core/models/wallets.py
@@ -79,6 +79,19 @@ class Wallet(models.Model):
     but we may change this if we need to do a Django query on e.g. personame
     
     ManyToMany field uses modern Django: a hidden Class, unlike CaveAndEntrances which is explict and visible.
+    
+    We parse all the JSON on initial reset/import but we only keep data on the Wallet objects that we need for
+    indexing: people, caves and year. So that we can quickly make reports on e.g. all wallets for a particular cave.
+    All the other fields in the JSON are parsed and loaded from file only dynamiclally, when a report is beging generated,
+    but since this only happens for a subset of wallets (e.g. for a specific year) the speed penalty is fine. Indeed it
+    might be faster overall as db operations in django are a bit slow. 
+    This other data is the stuff which generates the tick-lists.
+    
+    A trick to minimize the number of times we hit the file to load JSON data is to use a field on every wallet object
+    called 'JSONdata', but this is not part of the schema and there is no corresponding field in the db. The property
+    'JSONdata' only lives for as long as the ephemeral python wallet object. We use this to cache the JSON data so that
+    queries of several different things, e.g. 'name', or 'survexnotrequired', do not repeatedly re-read the JSON 
+    which has not changed.
     """
 
     fpath = models.CharField(max_length=200)
@@ -86,6 +99,8 @@ class Wallet(models.Model):
     walletdate = models.DateField(blank=True, null=True)
     walletyear = models.DateField(blank=True, null=True)
     caves =  models.ManyToManyField("Cave", related_name="wallets")
+    # not yet 
+    #persons = models.ManyToManyField("Person", related_name="wallets")
 
     class Meta:
         ordering = ("walletname",)
@@ -99,9 +114,11 @@ class Wallet(models.Model):
         Do it every time it is queried, to be sure the result is fresh
         
         import DataIssue locally to prevent import cycle problem"""
-        # jsonfile = Path(self.fpath, 'contents.json')
-
-        # Get from git repo instead
+        
+        if hasattr(self, "JSONdata"):
+            print(f'seen it already {self}')
+            return self.JSONdata
+  
         # :drawings: walletjson/2022/2022#01/contents.json
         # fpath = /mnt/d/EXPO/expofiles/surveyscans/1999/1999#02
         fp = Path(self.fpath)
@@ -145,7 +162,7 @@ class Wallet(models.Model):
                             from troggle.core.models.troggle import DataIssue
                             DataIssue.objects.update_or_create(parser="wallets", message=message, url=wurl)
                             
- 
+        self.JSONdata = waldata
         return waldata
 
     def check_survexlist(self):