Chore: Refactor Django orcabus_id (#784)

lib/workload/stateless/stacks/metadata-manager/app/fields.py

@@ -1,48 +1,55 @@
-import hashlib
-
+import ulid
+from django.core.validators import RegexValidator
 from django.db import models
 
+ULID_REGEX_STR = r"[0123456789ABCDEFGHJKMNPQRSTVWXYZ]{26}"
+ulid_validator = RegexValidator(regex=ULID_REGEX_STR,
+                                message='ULID is expected to be 26 characters long',
+                                code='invalid_orcabus_id')
+
+
+def get_ulid() -> str:
+    return ulid.new().str
 
-class HashField(models.CharField):
-    description = (
-        "HashField is related to some base fields (other columns) in a model and"
-        "stores its hashed value for better indexing performance."
-    )
 
-    def __init__(self, base_fields, *args, **kwargs):
-        """
-        :param base_fields: name of fields storing the value to be hashed
-        """
-        self.base_fields = base_fields
-        kwargs["max_length"] = 64
-        super(HashField, self).__init__(*args, **kwargs)
+class UlidField(models.CharField):
+    description = "An OrcaBus internal ID (ULID)"
+
+    def __init__(self, *args, **kwargs):
+        kwargs['max_length'] = 26  # ULID length
+        kwargs['validators'] = [ulid_validator]
+        kwargs['default'] = get_ulid
+        super().__init__(*args, **kwargs)
 
     def deconstruct(self):
         name, path, args, kwargs = super().deconstruct()
         del kwargs["max_length"]
-        if self.base_fields is not None:
-            kwargs["base_fields"] = self.base_fields
+        del kwargs['validators']
+        del kwargs['default']
         return name, path, args, kwargs
 
-    def pre_save(self, instance, add):
-        self.calculate_hash(instance)
-        return super(HashField, self).pre_save(instance, add)
 
-    def calculate_hash(self, instance):
-        sha256 = hashlib.sha256()
-        for field in self.base_fields:
-            value = getattr(instance, field)
-            sha256.update(value.encode("utf-8"))
-        setattr(instance, self.attname, sha256.hexdigest())
+class OrcaBusIdField(UlidField):
+    description = "An OrcaBus internal ID (based on ULID)"
+
+    def __init__(self, prefix='', *args, **kwargs):
+        self.prefix = prefix
+        super().__init__(*args, **kwargs)
 
+    @property
+    def non_db_attrs(self):
+        return super().non_db_attrs + ("prefix",)
 
-class HashFieldHelper(object):
-    def __init__(self):
-        self.__sha256 = hashlib.sha256()
+    def from_db_value(self, value, expression, connection):
+        if value and self.prefix != '':
+            return f"{self.prefix}.{value}"
+        else:
+            return value
 
-    def add(self, value):
-        self.__sha256.update(value.encode("utf-8"))
-        return self
+    def to_python(self, value):
+        # This will be called when the function
+        return self.get_prep_value(value)
 
-    def calculate_hash(self):
-        return self.__sha256.hexdigest()
+    def get_prep_value(self, value):
+        # We just want the last 26 characters which is the ULID (ignoring any prefix) when dealing with the database
+        return value[-26:]

lib/workload/stateless/stacks/metadata-manager/app/migrations/0003_alter_contact_orcabus_id_and_more.py

@@ -0,0 +1,74 @@
+# Generated by Django 5.1.4 on 2024-12-17 01:44
+
+import app.fields
+from django.db import migrations
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('app', '0002_remove_historicalcontact_history_user_and_more'),
+    ]
+
+    operations = [
+        migrations.AlterField(
+            model_name='contact',
+            name='orcabus_id',
+            field=app.fields.OrcaBusIdField(primary_key=True, serialize=False),
+        ),
+        migrations.AlterField(
+            model_name='historicalcontact',
+            name='orcabus_id',
+            field=app.fields.OrcaBusIdField(db_index=True),
+        ),
+        migrations.AlterField(
+            model_name='historicalindividual',
+            name='orcabus_id',
+            field=app.fields.OrcaBusIdField(db_index=True),
+        ),
+        migrations.AlterField(
+            model_name='historicallibrary',
+            name='orcabus_id',
+            field=app.fields.OrcaBusIdField(db_index=True),
+        ),
+        migrations.AlterField(
+            model_name='historicalproject',
+            name='orcabus_id',
+            field=app.fields.OrcaBusIdField(db_index=True),
+        ),
+        migrations.AlterField(
+            model_name='historicalsample',
+            name='orcabus_id',
+            field=app.fields.OrcaBusIdField(db_index=True),
+        ),
+        migrations.AlterField(
+            model_name='historicalsubject',
+            name='orcabus_id',
+            field=app.fields.OrcaBusIdField(db_index=True),
+        ),
+        migrations.AlterField(
+            model_name='individual',
+            name='orcabus_id',
+            field=app.fields.OrcaBusIdField(primary_key=True, serialize=False),
+        ),
+        migrations.AlterField(
+            model_name='library',
+            name='orcabus_id',
+            field=app.fields.OrcaBusIdField(primary_key=True, serialize=False),
+        ),
+        migrations.AlterField(
+            model_name='project',
+            name='orcabus_id',
+            field=app.fields.OrcaBusIdField(primary_key=True, serialize=False),
+        ),
+        migrations.AlterField(
+            model_name='sample',
+            name='orcabus_id',
+            field=app.fields.OrcaBusIdField(primary_key=True, serialize=False),
+        ),
+        migrations.AlterField(
+            model_name='subject',
+            name='orcabus_id',
+            field=app.fields.OrcaBusIdField(primary_key=True, serialize=False),
+        ),
+    ]

lib/workload/stateless/stacks/metadata-manager/app/models/base.py

@@ -23,6 +23,7 @@
 from simple_history.models import HistoricalRecords
 
 from rest_framework.settings import api_settings
+
 from app.pagination import PaginationConstant
 
 logger = logging.getLogger(__name__)
@@ -42,7 +43,9 @@ def reduce_multi_values_qor(key: str, values: List[str]):
         ):
             values = [values]
         return reduce(
-            operator.or_, (Q(**{"%s__iexact" % key: value})
+            # Apparently the `get_prep_value` from the custom fields.py is not called prior hitting the Db but,
+            # the regular `__exact` still execute that function.
+            operator.or_, (Q(**{"%s__exact" % key: value})
                            for value in values)
         )
 
@@ -102,7 +105,6 @@ def update_or_create_if_needed(self, search_key: dict, data: dict, user_id: str
         """
         is_created = False
         is_updated = False
-
         try:
             obj = self.get(**search_key)
             for key, value in data.items():
@@ -125,27 +127,16 @@ class BaseModel(models.Model):
     class Meta:
         abstract = True
 
-    orcabus_id = models.CharField(
-        primary_key=True,
-        unique=True,
-        editable=False,
-        blank=False,
-        null=False,
-        validators=[
-            RegexValidator(
-                regex=r'[\w]{26}$',
-                message='ULID is expected to be 26 characters long',
-                code='invalid_orcabus_id'
-            )]
-
-    )
-
     def save(self, *args, **kwargs):
-        if not self.orcabus_id:
-            self.orcabus_id = ulid.new().str
+        # To make django validate the constraint before saving it
         self.full_clean()
 
-        return super(BaseModel, self).save(*args, **kwargs)
+        super(BaseModel, self).save(*args, **kwargs)
+
+        # Reload the object from the database to ensure custom fields like OrcaBusIdField
+        # invoke the `from_db_value` method (which provides the annotation) after saving.
+        self.refresh_from_db()
+
 
     @classmethod
     def get_fields(cls):

lib/workload/stateless/stacks/metadata-manager/app/models/contact.py

@@ -1,16 +1,16 @@
 from django.db import models
 
+from app.fields import OrcaBusIdField
 from app.models.base import BaseModel, BaseManager, BaseHistoricalRecords
 
-
 class ContactManager(BaseManager):
     pass
 
 
 class Contact(BaseModel):
-    orcabus_id_prefix = 'ctc.'
     objects = ContactManager()
 
+    orcabus_id = OrcaBusIdField(primary_key=True, prefix='ctc')
     contact_id = models.CharField(
         unique=True,
         blank=True,

lib/workload/stateless/stacks/metadata-manager/app/models/individual.py

@@ -1,5 +1,6 @@
 from django.db import models
 
+from app.fields import OrcaBusIdField
 from app.models.base import BaseModel, BaseManager, BaseHistoricalRecords
 
 
@@ -8,9 +9,9 @@ class IndividualManager(BaseManager):
 
 
 class Individual(BaseModel):
-    orcabus_id_prefix = 'idv.'
     objects = IndividualManager()
 
+    orcabus_id = OrcaBusIdField(primary_key=True, prefix='idv')
     individual_id = models.CharField(
         unique=True,
         blank=True,

lib/workload/stateless/stacks/metadata-manager/app/models/library.py

@@ -2,6 +2,7 @@
 
 from django.db import models
 
+from app.fields import OrcaBusIdField
 from app.models.base import BaseManager, BaseModel, BaseHistoricalRecords
 from app.models.subject import Subject
 from app.models.sample import Sample
@@ -64,9 +65,9 @@ class LibraryProjectLink(models.Model):
 
 
 class Library(BaseModel):
-    orcabus_id_prefix = 'lib.'
     objects = LibraryManager()
 
+    orcabus_id = OrcaBusIdField(primary_key=True, prefix='lib')
     library_id = models.CharField(
         unique=True,
         blank=True,

lib/workload/stateless/stacks/metadata-manager/app/models/project.py

@@ -1,5 +1,6 @@
 from django.db import models
 
+from app.fields import OrcaBusIdField
 from app.models.contact import Contact
 from app.models.base import BaseModel, BaseManager, BaseHistoricalRecords
 
@@ -19,9 +20,9 @@ class ProjectContactLink(models.Model):
 
 
 class Project(BaseModel):
-    orcabus_id_prefix = 'prj.'
     objects = ProjectManager()
 
+    orcabus_id = OrcaBusIdField(primary_key=True, prefix='prj')
     project_id = models.CharField(
         unique=True,
         blank=True,

lib/workload/stateless/stacks/metadata-manager/app/models/sample.py

@@ -1,6 +1,7 @@
 import ulid
 from django.db import models
 
+from app.fields import OrcaBusIdField
 from app.models.base import BaseModel, BaseManager, BaseHistoricalRecords
 
 
@@ -31,9 +32,9 @@ class SampleManager(BaseManager):
 
 
 class Sample(BaseModel):
-    orcabus_id_prefix = 'smp.'
     objects = SampleManager()
 
+    orcabus_id = OrcaBusIdField(primary_key=True, prefix='smp')
     sample_id = models.CharField(
         unique=True,
         blank=True,

lib/workload/stateless/stacks/metadata-manager/app/models/subject.py

@@ -1,5 +1,6 @@
 from django.db import models
 
+from app.fields import OrcaBusIdField
 from app.models.base import BaseModel, BaseManager, BaseHistoricalRecords
 
 
@@ -17,8 +18,9 @@ class SubjectIndividualLink(models.Model):
 
 
 class Subject(BaseModel):
-    orcabus_id_prefix = 'sbj.'
     objects = SubjectManager()
+
+    orcabus_id = OrcaBusIdField(primary_key=True, prefix='sbj')
     subject_id = models.CharField(
         unique=True,
         blank=True,

lib/workload/stateless/stacks/metadata-manager/app/serializers/contact.py

@@ -1,22 +1,14 @@
-from abc import ABC
-
-from rest_framework import serializers
-
+from rest_framework.serializers import ModelSerializer
 from app.models import Contact
-from .base import SerializersBase
-
-
-class ContactBaseSerializer(SerializersBase):
-    prefix = Contact.orcabus_id_prefix
 
 
-class ContactSerializer(ContactBaseSerializer):
+class ContactSerializer(ModelSerializer):
     class Meta:
         model = Contact
         fields = "__all__"
 
 
-class ContactDetailSerializer(ContactBaseSerializer):
+class ContactDetailSerializer(ModelSerializer):
     from .project import ProjectSerializer
 
     project_set = ProjectSerializer(many=True, read_only=True)
@@ -26,7 +18,7 @@ class Meta:
         fields = "__all__"
 
 
-class ContactHistorySerializer(ContactBaseSerializer):
+class ContactHistorySerializer(ModelSerializer):
     class Meta:
         model = Contact.history.model
         fields = "__all__"