Cache genome build regardless how accessed

environment.yml

@@ -74,7 +74,7 @@ dependencies:
       - django-postgres-extra==2.0.5
       - django_recaptcha==3.0.0
       - django-registration-redux==2.9
-      - django_termsandconditions==2.0.9
+      - django_termsandconditions==2.0.12
       - django-reversion-compare==0.15.0
       - django_threadlocals==0.10
       - djangorestframework==3.14.0

library/django_utils/django_object_managers.py

@@ -0,0 +1,60 @@
+from dataclasses import dataclass
+from typing import Dict, Any, Optional
+
+from django.db.models import Manager, QuerySet, Q
+from frozendict import frozendict
+
+
+@dataclass(frozen=True)
+class CachedQuery:
+    model: Any
+    args: Any
+    kwargs: Any
+
+
+_CACHED_QUERIES: Dict[CachedQuery, Any] = {}
+# TODO, could make this a LRU cache so it doesn't grow infinitely for larger models
+
+
+class QuerySetCaching(QuerySet):
+
+    def __init__(self, model=None, query=None, using=None, hints=None):
+        super().__init__(model=model, query=query, using=using, hints=hints)
+
+    def get(self, *args, **kwargs):
+        cq: Optional[CachedQuery] = None
+        kwargs_fd = None
+        if kwargs:
+            # turn into a frozen dict so we can hash it
+            kwargs_fd = frozendict(kwargs)
+        try:
+            cq = CachedQuery(model=self.model, args=args, kwargs=kwargs_fd)
+        except TypeError:
+            # unhashable argument
+            pass
+
+        if cq:
+            if existing := _CACHED_QUERIES.get(cq):
+                return existing
+            else:
+                result = super().get(*args, **kwargs)
+                _CACHED_QUERIES[cq] = result
+                return result
+        else:
+            return super().get(*args, **kwargs)
+
+
+class CachingObjectManager(Manager):
+    """
+    Best used as both the default objects and as Meta._base_manager_name = 'objects'
+    Will cache results for queries - so only do this on models that are read from but never written to.
+
+    We could invalidate the cache when save() is called, but that only works if the save is done in the only instance
+    that has already cached values.
+
+    Note that currently it only caches the results of get(...)
+    """
+
+    def __init__(self):
+        super().__init__()
+        self._queryset_class = QuerySetCaching

requirements.txt

@@ -30,7 +30,7 @@ django_model_utils==4.2.0
 django-postgres-extra==2.0.5
 django_recaptcha==3.0.0
 django-registration-redux==2.9
-django_termsandconditions==2.0.9
+django_termsandconditions==2.0.12
 django-reversion-compare==0.15.0
 djangorestframework==3.14.0
 easy_thumbnails==2.8.1

snpdb/models/models_genome.py

@@ -13,6 +13,7 @@
 from genes.models_enums import AnnotationConsortium
 from library.cache import timed_cache
 from library.django_utils import SortMetaOrderingMixin
+from library.django_utils.django_object_managers import CachingObjectManager
 from library.genomics.fasta_wrapper import FastaFileWrapper
 from library.utils import invert_dict
 from snpdb.genome.fasta_index import load_genome_fasta_index
@@ -28,6 +29,8 @@ class GenomeBuild(models.Model, SortMetaOrderingMixin):
 
         Build & Contig are populated via migration snpdb 0006 """
 
+    objects = CachingObjectManager()
+
     name = models.TextField(primary_key=True)
     accession = models.TextField(null=True)
     alias = models.TextField(null=True, unique=True)
@@ -36,22 +39,20 @@ class GenomeBuild(models.Model, SortMetaOrderingMixin):
 
     class Meta:
         ordering = ["name"]
+        base_manager_name = 'objects'
 
     def is_version(self, version: int) -> bool:
         return str(version) in self.name
 
     @classmethod
-    @timed_cache(ttl=60)
     def grch37(cls) -> 'GenomeBuild':
         return cls.objects.get(pk='GRCh37')
 
     @classmethod
-    @timed_cache(ttl=60)
     def grch38(cls) -> 'GenomeBuild':
         return cls.objects.get(pk='GRCh38')
 
     @classmethod
-    @timed_cache(ttl=60)
     def legacy_build(cls) -> 'GenomeBuild':
         """ Use this for hacks - makes it easy to find / fix later """
         return cls.objects.get(pk='GRCh37')