feat: get analysis datasets

common/constants/constants.py

@@ -0,0 +1,8 @@
+"""Contains constants"""
+# Units in bytes
+KB = 1024
+MB = 1024 * KB
+GB = 1024 * MB
+TB = 1024 * GB
+
+DATASET_MAX_SIZE = 10 * MB

common/helpers/get_mime_type_from_extension.py

@@ -0,0 +1,7 @@
+"""Contains a method for getting a mimetype from a file extension"""
+import mimetypes
+
+def get_mimetype_from_extension(filename: str) -> str:
+    """Retrieve a mimetype from a given filename"""
+    mimetype, _ = mimetypes.guess_type(filename)
+    return mimetype

file_management/contract/dto/dataset_to.py

@@ -12,8 +12,11 @@
 class DatasetTO(BaseTO):
     """Contains the fields for a dataset"""
     id: str | None
-    uploadedBy: UserTO | None
-    createdOn: datetime | None
+    uploadedBy: str | None
+    sizeBytes: int | None
+    createdAt: datetime | None
+    updatedAt: datetime | None
+    mimeType: str | None
     url: str | None
     filename: str | None
 
@@ -24,9 +27,12 @@ def from_model(cls, instance: Dataset) -> 'DatasetTO | None':
         if not instance:
             return None
         return cls(
-            uploadedBy=instance.uploaded_by,
-            createdOn=instance.created_on,
+            uploadedBy=instance.uploaded_by.id,
+            createdAt=instance.created_at,
             id=instance.id,
+            sizeBytes=instance.size_bytes,
+            mimeType=instance.mime_type,
+            updatedAt=instance.updated_at,
             filename=instance.filename,
             url=instance.url
         )

file_management/contract/io/create_presigned_url_upload_file_in.py

@@ -6,3 +6,4 @@ class CreatePresignedUrlUploadFileIn(serializers.Serializer):
     """Request input for an analysis creation"""
     filename = serializers.CharField()
     analysis_id = serializers.IntegerField()
+    size_bytes = serializers.IntegerField()

file_management/contract/repository/file_management_repository.py

@@ -1,6 +1,7 @@
 """This module contains the File Management repository"""
 
 from abc import abstractmethod
+from typing import List
 
 from file_management.contract.dto.dataset_to import DatasetTO
 from file_management.contract.dto.s3_presigned_url_to import S3PresignedUrlTO
@@ -11,7 +12,7 @@ class FileManagementRepository:
 
     @abstractmethod
     def create_presigned_url_upload_file(
-        self, filename: str, user_id: str
+        self, filename: str, user_id: str, size_bytes: int
     ) -> tuple[S3PresignedUrlTO, DatasetTO]:
         """
         Create a presigned URL to allow the frontend to upload a file
@@ -38,3 +39,12 @@ def get_dataset_by_id(self, dataset_id) -> DatasetTO:
         """
         Retrieve a dataset by id
         """
+
+    @abstractmethod
+    def get_analysis_datasets(self, analysis_id: int) -> List[DatasetTO]:
+        """Retrieve the datasets of the analysis
+
+        Keyword arguments:
+        analysis_id -- the id of the analysis
+        Return: A list of DatasetTO
+        """

file_management/interfaces/controllers/create_presigned_url_upload_file_controller.py

@@ -23,5 +23,6 @@ def create_presigned_url_upload_file_controller(request):
         raise BadRequestException("All fields are required")
     filename = data.validated_data["filename"]
     analysis_id = data.validated_data["analysis_id"]
-    url = service.create_presigned_url_upload_file(request.user, filename, analysis_id)
+    size_bytes = data.validated_data["size_bytes"]
+    url = service.create_presigned_url_upload_file(request.user, filename, analysis_id, size_bytes)
     return api_response_success(data=url)

file_management/interfaces/controllers/get_analysis_datasets_controller.py

@@ -0,0 +1,16 @@
+"""Contains the controller for getting the analysis datasets"""
+from rest_framework.decorators import api_view
+
+from common.exceptions.exceptions import BadRequestException
+from common.helpers.api_responses import api_response_success
+from file_management.service.impl.file_management_service_impl import FileManagementServiceImpl
+
+@api_view(["GET"])
+def get_analysis_datasets_controller(request):
+    """Retrieve the analysis datasets"""
+    analysis_id = request.query_params.get("analysis_id", None)
+    if not analysis_id:
+        raise BadRequestException("The analysis id is required")
+    service = FileManagementServiceImpl()
+    datasets = service.get_analysis_datasets(request.user, analysis_id)
+    return api_response_success(data=datasets)

file_management/migrations/0002_alter_dataset_options_and_more.py

@@ -0,0 +1,39 @@
+# Generated by Django 5.1.4 on 2024-12-30 15:59
+
+import django.utils.timezone
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('file_management', '0001_initial'),
+    ]
+
+    operations = [
+        migrations.AlterModelOptions(
+            name='dataset',
+            options={'ordering': ['created_at']},
+        ),
+        migrations.RenameField(
+            model_name='dataset',
+            old_name='created_on',
+            new_name='created_at',
+        ),
+        migrations.AddField(
+            model_name='dataset',
+            name='mime_type',
+            field=models.CharField(default='text/csv'),
+        ),
+        migrations.AddField(
+            model_name='dataset',
+            name='size_bytes',
+            field=models.IntegerField(default=0),
+            preserve_default=False,
+        ),
+        migrations.AddField(
+            model_name='dataset',
+            name='updated_at',
+            field=models.DateTimeField(auto_now=True),
+        ),
+    ]

file_management/models/dataset.py

@@ -7,10 +7,13 @@ class Dataset(models.Model):
     id = models.UUIDField(primary_key=True, default=uuid.uuid4)
     filename = models.CharField(max_length=255)
     url = models.URLField(null=True)
-    created_on = models.DateTimeField(auto_now_add=True)
+    created_at = models.DateTimeField(auto_now_add=True)
+    updated_at = models.DateTimeField(auto_now=True)
     uploaded_by = models.ForeignKey('user_management.User', on_delete=models.CASCADE)
+    mime_type = models.CharField(null=False, default="text/csv")
+    size_bytes = models.IntegerField(default=0)
 
     class Meta:
         """Table's metadata"""
         db_table = 'dataset'
-        ordering = ['created_on']
+        ordering = ['created_at']

file_management/repository/file_management_repository_impl.py

@@ -1,6 +1,7 @@
 """This module contains the analysis repository"""
 
 import logging
+from typing import List
 import urllib
 from datetime import timedelta
 from os import getenv
@@ -9,6 +10,7 @@
 from django.db import transaction
 
 from analysis.models.analysis import Analysis
+from common.helpers.get_mime_type_from_extension import get_mimetype_from_extension
 from file_management.contract.dto.dataset_to import DatasetTO
 from file_management.contract.dto.s3_presigned_url_to import S3PresignedUrlTO
 from file_management.contract.repository.file_management_repository import (
@@ -23,17 +25,20 @@
 class FileManagementRepositoryImpl(FileManagementRepository):
     """Analysis repository"""
 
-    def create_presigned_url_upload_file(self, filename: str, user_id: str):
+    def create_presigned_url_upload_file(self, filename: str, user_id: str, size_bytes: int):
         s3_client = boto3.client("s3")
         expires_in = timedelta(hours=1).seconds
         user = User.objects.filter(id=user_id).first()
 
         try:
             with transaction.atomic():
                 new_dataset = Dataset.objects.create(
-                    filename=filename, uploaded_by=user
+                    filename=filename,
+                    uploaded_by=user,
+                    size_bytes=size_bytes,
+                    mime_type=get_mimetype_from_extension(filename)
                 )
-                object_name = f"datasets/{str(new_dataset.id)}"
+                object_name = f"datasets/{filename}"
                 response = s3_client.generate_presigned_post(
                     bucket_name,
                     object_name,
@@ -70,3 +75,8 @@ def attach_file_to_analysis(self, dataset_id: str, analysis_id: int) -> DatasetT
     def get_dataset_by_id(self, dataset_id: str) -> DatasetTO:
         dataset = Dataset.objects.filter(id=dataset_id).first()
         return DatasetTO.from_model(dataset)
+
+    def get_analysis_datasets(self, analysis_id: int) -> List[DatasetTO]:
+        analysis = Analysis.objects.filter(id=analysis_id).first()
+        datasets = analysis.datasets.all()
+        return DatasetTO.from_models(datasets)

file_management/service/file_management_service.py

@@ -1,8 +1,10 @@
 """Contains the abstract class of file management service"""
 
 from abc import abstractmethod
+from typing import List
 
 from common.service.base_service import BaseService
+from file_management.contract.dto.dataset_to import DatasetTO
 from file_management.contract.dto.s3_presigned_url_to import S3PresignedUrlTO
 
 
@@ -16,3 +18,9 @@ def create_presigned_url_upload_file(self, user, filename: str) -> S3PresignedUr
     @abstractmethod
     def create_presigned_url_download_file(self, user, dataset_id: str) -> S3PresignedUrlTO:
         """Generate a presigned URL for downloading files"""
+
+    @abstractmethod
+    def get_analysis_datasets(self, user, analysis_id: int) -> List[DatasetTO]:
+        """
+        Retrieve all the datasets from the given analysis if the user has the required permissions
+        """

file_management/service/impl/file_management_service_impl.py

@@ -1,9 +1,8 @@
 """Contains the implementation of AnalysisService"""
 
-import urllib
 from analysis.service.impl.analysis_service_impl import AnalysisServiceImpl
-from common.exceptions.exceptions import BadRequestException, NotFoundException
-from file_management.contract.dto.s3_presigned_url_to import S3PresignedUrlTO
+from common.constants.constants import DATASET_MAX_SIZE, MB
+from common.exceptions.exceptions import BadRequestException, NotFoundException, ForbiddenException
 from file_management.repository.file_management_repository_impl import (
     FileManagementRepositoryImpl,
 )
@@ -14,7 +13,9 @@
 from file_management.use_cases.create_presigned_url_upload_file_uc import (
     CreatePresignedUrlUploadFileUC,
 )
+from file_management.use_cases.get_analysis_datasets_uc import GetAnalysisDatasetsUC
 from user_management.repository.role_repository_impl import RoleRepositoryImpl
+from user_management.service.impl.users_service_impl import UsersServiceImpl
 from user_management.usecases.attach_file_to_analysis_uc import AttachFileToAnalysisUC
 from user_management.usecases.get_user_role_in_analysis_uc import (
     GetUserRoleInAnalysisUC,
@@ -30,23 +31,28 @@ def __init__(self):
         )
         self.get_user_role_in_analysis_uc = GetUserRoleInAnalysisUC.get_instance()
         self.attach_file_to_analysis_uc = AttachFileToAnalysisUC.get_instance()
+        self.get_analysis_datasets_uc = GetAnalysisDatasetsUC.get_instance()
         self.create_presigned_url_download_file_uc = (
             CreatePresignedUrlDownloadFileUC.get_instance()
         )
         self.repository = FileManagementRepositoryImpl()
         self.role_repository = RoleRepositoryImpl()
         self.analysis_service = AnalysisServiceImpl()
+        self.user_service = UsersServiceImpl()
 
     def create_presigned_url_upload_file(
-        self, user, filename: str, analysis_id: int
+        self, user, filename: str, analysis_id: int, size_bytes: int
     ) -> str:
         # TODO: validate if the user is in ['FACILITATOR', 'DATA MANAGER']
+        if size_bytes > DATASET_MAX_SIZE:
+            raise BadRequestException(f"The file must be smaller than {DATASET_MAX_SIZE / MB}MB")
+
         self.analysis_service.get_analysis_by_id(
             analysis_id
         )  # Raise 404 if analysis doesn't exist
 
         presigned_url, dataset = self.create_presigned_url_upload_file_uc.exec(
-            self.repository, filename, user.id
+            self.repository, filename, user.id, size_bytes
         )
 
         self.attach_file_to_analysis_uc.exec(self.repository, dataset.id, analysis_id)
@@ -69,3 +75,12 @@ def create_presigned_url_download_file(
         if not response:
             raise BadRequestException()
         return response
+
+    def get_analysis_datasets(self, user, analysis_id: int):
+        self.analysis_service.get_analysis_by_id(analysis_id) # Raises 404 if not found
+        if not self.user_service.is_user_in_analysis(user.id, analysis_id):
+            raise ForbiddenException("You can't see that analysis")
+        datasets = self.get_analysis_datasets_uc.exec(
+            self.repository, analysis_id
+        )
+        return [dataset.to_dict() for dataset in datasets]