Merge pull request #290 from SELab-2/seeder-improvements

Seeder improvements

backend/.gitignore

@@ -1,6 +1,6 @@
 .venv
 .idea
 staticfiles
-db.sqlite3
+db.sqlite3*
 __pycache__
 *.mo

backend/api/management/commands/seed_db_new.py

@@ -0,0 +1,141 @@
+from random import choices, randint
+from django.core.management.base import BaseCommand
+from authentication.models import User
+from api.seeders.faker import faker
+from api.seeders.seeder import seed_students, seed_assistants, seed_teachers, seed_courses, seed_projects, seed_groups, \
+    seed_submissions
+from api.models.course import Course
+from api.models.group import Group
+from api.models.project import Project
+from api.models.teacher import Teacher
+from api.models.student import Student
+from api.models.assistant import Assistant
+from api.models.submission import Submission
+
+import time
+
+
+def _seed_users(num: int = 500, student_prob: int = 70, assistant_prob: int = 20, teacher_prob: int = 10):
+    fake = faker()
+    User.objects.all().delete()
+
+    users: list[User] = User.objects.bulk_create(
+        [fake.fake_user(fake, id) for id in range(num)]
+    )
+
+    roles = choices(['student'] * student_prob + ['assistant'] * assistant_prob + ['teacher'] * teacher_prob, k=num)
+
+    for user, role in zip(users, roles):
+        if role == 'assistant':
+            Assistant.create(user)
+        elif role == 'teacher':
+            Teacher.create(user)
+        elif role == 'student':
+            Student.create(user, student_id=user.id)
+
+        user.faculties.add(
+            *[fake.real_faculty() for _ in range(randint(1, 2))]
+        )
+
+
+def _seed_courses(num: int = 200):
+    fake = faker()
+    Course.objects.all().delete()
+
+    courses: list[Course] = Course.objects.bulk_create(
+        [fake.fake_course(fake) for _ in range(num)]
+    )
+
+    for course in courses:
+        course.students.add(
+            *[fake.real_student() for _ in range(randint(10, 100))]
+        )
+        course.teachers.add(
+            *[fake.real_teacher() for _ in range(randint(1, 3))]
+        )
+        course.assistants.add(
+            *[fake.real_assistant() for _ in range(randint(2, 5))]
+        )
+
+
+def _seed_projects(num: int = 1_000):
+    fake = faker()
+    Project.objects.all().delete()
+
+    Project.objects.bulk_create(
+        [fake.fake_project(fake) for _ in range(num)]
+    )
+
+
+def _seed_groups(num: int = 3_000):
+    fake = faker()
+    Group.objects.all().delete()
+
+    groups = Group.objects.bulk_create(
+        [fake.fake_group(fake) for _ in range(num)]
+    )
+
+    for group in groups:
+        group.students.add(
+            *[fake.real_student() for _ in range(randint(1, group.project.group_size))]
+        )
+
+
+def format_time(execution_time):
+    if execution_time < 1:
+        return f"{execution_time * 1000:.2f} milliseconds"
+    elif execution_time < 60:
+        return f"{execution_time:.2f} seconds"
+    elif execution_time < 3600:
+        return f"{execution_time / 60:.2f} minutes"
+    else:
+        return f"{execution_time / 3600:.2f} hours"
+
+
+class Command(BaseCommand):
+    help = 'seed the db with data'
+
+    amount_of_students = 50_000
+    amount_of_assistants = 5_000
+    amount_of_teachers = 1_500
+    amount_of_courses = 1_500
+    amount_of_projects = 3_000
+    amount_of_groups = 3_000
+    amount_of_submissions = 3_000
+
+    def handle(self, *args, **options):
+        # Reset DB
+        User.objects.all().delete()
+        Student.objects.all().delete()
+        Assistant.objects.all().delete()
+        Teacher.objects.all().delete()
+        Course.objects.all().delete()
+        Project.objects.all().delete()
+        Submission.objects.all().delete()
+
+        # Seed students
+        fake = faker()
+        start_time = time.time()
+        seed_students(fake, self.amount_of_students, 0)
+
+        # Seed assistants
+        seed_assistants(fake, self.amount_of_assistants, self.amount_of_students)
+
+        # Seed teachers
+        seed_teachers(fake, self.amount_of_teachers, self.amount_of_students + self.amount_of_assistants)
+
+        # Seed courses
+        seed_courses(faker(), self.amount_of_courses)
+
+        # Seed projects
+        seed_projects(faker(), self.amount_of_projects)
+
+        # Seed groups
+        seed_groups(faker(), self.amount_of_groups)
+
+        # Seed submissions
+        seed_submissions(faker(), self.amount_of_submissions)
+
+        end_time = time.time()
+        execution_time = end_time - start_time
+        self.stdout.write(self.style.SUCCESS(f"Successfully seeded db in {format_time(execution_time)}!"))

backend/api/seeders/faker.py

@@ -0,0 +1,24 @@
+from faker import Faker
+from api.seeders.providers.real_provider import real_group_provider, real_course_provider, real_faculty_provider, \
+    real_file_extension_provider, real_structure_check_provider, real_project_provider, real_student_provider, \
+    real_teacher_provider, real_assistant_provider, real_submission_provider
+from api.seeders.providers.fake_provider import ModelProvider
+
+
+def faker():
+    fake = Faker()
+
+    fake.add_provider(real_faculty_provider())
+    fake.add_provider(real_student_provider())
+    fake.add_provider(real_assistant_provider())
+    fake.add_provider(real_teacher_provider())
+    fake.add_provider(real_course_provider())
+    fake.add_provider(real_project_provider())
+    fake.add_provider(real_group_provider())
+    fake.add_provider(real_submission_provider())
+    fake.add_provider(real_file_extension_provider())
+    fake.add_provider(real_structure_check_provider())
+
+    fake.add_provider(ModelProvider)
+
+    return fake

backend/api/seeders/providers/fake_provider.py

@@ -0,0 +1,86 @@
+from faker.providers import BaseProvider
+from django.utils import timezone
+from api.models.course import Course
+from api.models.project import Project
+from api.models.group import Group
+from authentication.models import User
+
+
+class ModelProvider(BaseProvider):
+    min_salt = 1
+    max_salt = 100_000
+
+    def fake_username(self, first_name: str, last_name: str) -> str:
+        """Fake username for users"""
+        rand = self.random_int(min=self.min_salt, max=self.max_salt)
+        return f"{first_name.lower()}{last_name.lower()}{rand}"[:12]
+
+    def fake_user(self, fake, id: int, staff_probability: float = 0.001) -> User:
+        """Create a fake user"""
+        return User(
+            id=id,
+            first_name=fake.first_name(),
+            last_name=fake.last_name(),
+            username=fake.unique.user_name(),
+            email=fake.unique.email(),
+            create_time=timezone.now(),
+            last_enrolled=timezone.now().year,
+            is_staff=fake.boolean(chance_of_getting_true=staff_probability)
+        )
+
+    def fake_course(self, fake, min_year=2022, max_year=2025) -> Course:
+        """Create a fake course"""
+        course = Course(
+            name=fake.catch_phrase(),
+            academic_startyear=fake.random_int(min=min_year, max=max_year),
+            faculty=fake.real_faculty(),
+            description=fake.paragraph()
+        )
+
+        return course
+
+    def fake_project(self,
+                     fake,
+                     min_start_date_dev=-100,
+                     max_start_date_dev=100,
+                     min_deadline_dev=1,
+                     max_deadline_dev=100,
+                     visible_prob=80,
+                     archived_prob=10,
+                     score_visible_prob=30,
+                     locked_groups_prob=30,
+                     min_max_score=1,
+                     max_max_score=100,
+                     min_group_size=1,
+                     max_group_size=15):
+        """Create a fake project"""
+        start_date = timezone.now() + timezone.timedelta(
+            days=fake.random_int(min=min_start_date_dev, max=max_start_date_dev)
+        )
+
+        course = fake.real_course()
+
+        return Project(
+            name=fake.catch_phrase(),
+            description=fake.paragraph(),
+            visible=fake.boolean(chance_of_getting_true=visible_prob),
+            archived=fake.boolean(chance_of_getting_true=archived_prob),
+            score_visible=fake.boolean(chance_of_getting_true=score_visible_prob),
+            locked_groups=fake.boolean(chance_of_getting_true=locked_groups_prob),
+            deadline=start_date + timezone.timedelta(days=fake.random_int(min=min_deadline_dev, max=max_deadline_dev)),
+            course=course,
+            start_date=start_date,
+            max_score=fake.random_int(min=min_max_score, max=max_max_score),
+            group_size=fake.random_int(min=min_group_size, max=max_group_size)
+        )
+
+    def fake_group(self, fake, min_score: int = 0):
+        """Create a fake group"""
+        project: Project = fake.real_project()
+
+        group: Group = Group(
+            project=fake.real_project(),
+            score=fake.random_int(min=min_score, max=project.max_score)
+        )
+
+        return group

backend/api/seeders/providers/real_provider.py

@@ -0,0 +1,80 @@
+from faker.providers import DynamicProvider
+from authentication.models import Faculty
+from api.models.student import Student
+from api.models.assistant import Assistant
+from api.models.teacher import Teacher
+from api.models.course import Course
+from api.models.group import Group
+from api.models.project import Project
+from api.models.submission import Submission
+from api.models.checks import FileExtension, StructureCheck
+
+
+def real_faculty_provider():
+    return DynamicProvider(
+        provider_name="real_faculty",
+        elements=Faculty.objects.all(),
+    )
+
+
+def real_student_provider():
+    return DynamicProvider(
+        provider_name="real_student",
+        elements=Student.objects.all(),
+    )
+
+
+def real_assistant_provider():
+    return DynamicProvider(
+        provider_name="real_assistant",
+        elements=Assistant.objects.all(),
+    )
+
+
+def real_teacher_provider():
+    DynamicProvider(
+        provider_name="real_teacher",
+        elements=Teacher.objects.all(),
+    )
+
+
+def real_course_provider():
+    DynamicProvider(
+        provider_name="real_course",
+        elements=Course.objects.all(),
+    )
+
+
+def real_project_provider():
+    DynamicProvider(
+        provider_name="real_project",
+        elements=Project.objects.all(),
+    )
+
+
+def real_group_provider():
+    DynamicProvider(
+        provider_name="real_group",
+        elements=Group.objects.all(),
+    )
+
+
+def real_submission_provider():
+    DynamicProvider(
+        provider_name="real_submission",
+        elements=Submission.objects.all(),
+    )
+
+
+def real_file_extension_provider():
+    DynamicProvider(
+        provider_name="real_file_extension",
+        elements=FileExtension.objects.all(),
+    )
+
+
+def real_structure_check_provider():
+    DynamicProvider(
+        provider_name="real_structure_check",
+        elements=StructureCheck.objects.all(),
+    )