Add test data generation tool. (#217)

Added a tool to populate AIPscan with randomly generated example data.

README.md

@@ -155,6 +155,29 @@ Shut down the AIPscan Docker containers and remove the rabbitmq volumes:
 docker-composer down --volumes
 ```
 
+## Tools
+
+The `tools` directory contains scripts that can be run by developers and system
+adminsitrators.
+
+#### Test data generator
+
+The test data generator, `tools/generate-test-data.py`, tool populates
+AIPscan's databse with randomly generated example data.
+
+### Running tools
+
+These should be run using the same system user and virtual environment that
+AIPscan is running under.
+
+Here's how you would run the `generate-test-data.py` tool, for example:
+
+    $ cd <path to AIPscan base directory>
+    $ sudo -u <system user> /bin/bash
+    $ source <path to virtual environment>/bin/activate
+    $ python3 tools/generate-test-data.py
+
+
 # Usage
 
 * Ensure that the Flask Server, RabbitMQ server, and Celery worker queue are up and running.

requirements/test.txt

@@ -1,5 +1,6 @@
 -r base.txt
 
+faker==14.2.1
 flake8==5.0.4
 pytest==5.4.3
 pytest_cov==2.11.1

tools/app/tool.py

@@ -0,0 +1,5 @@
+import sys
+
+sys.path.append("../AIPscan")
+
+config_name = "default"

tools/generate-test-data.py

@@ -0,0 +1,70 @@
+from app import tool
+from faker import Faker
+from flask import Flask
+from helpers import data
+
+from AIPscan import db
+from AIPscan.models import FetchJob
+from config import CONFIGS
+
+app = Flask(__name__)
+app.config.from_object(CONFIGS[tool.config_name])
+
+db.init_app(app)
+
+fake = Faker()
+randint = fake.random.randint
+
+with app.app_context():
+    # Add example storage services
+    ss_to_create = 2
+
+    print(f"Creating pipeline and {ss_to_create} storage services...")
+    pipeline = data.create_pipeline()
+
+    ss_ids = []
+    fetch_jobs = {}
+
+    default_created = False
+    for _ in range(ss_to_create):
+        is_default = len(ss_ids) == 0
+
+        ss = data.create_storage_service(is_default)
+        ss_ids.append(ss.id)
+
+        fetch_job = data.create_fetch_job(ss.id)
+        fetch_jobs[ss.id] = fetch_job.id
+
+    # Populate storage service locations
+    storage_locations_per_ss = 2
+    ss_locations_to_create = ss_to_create * storage_locations_per_ss
+
+    print(
+        f"Creating {ss_locations_to_create} storage service locations (and their AIPs)..."
+    )
+
+    aip_batches_created = 0
+    total_aip_batches = len(ss_ids) * storage_locations_per_ss
+    for ss_id in ss_ids:
+        for _ in range(storage_locations_per_ss):
+            # Add location
+            sl = data.create_location(ss_id)
+
+            # Add AIPs
+            aip_batches_created += 1
+
+            print(f"Creating AIPs ({aip_batches_created}/{total_aip_batches})...")
+
+            aipcount = 0
+            for _ in range(1, randint(100, 300)):
+                aip = data.create_aip(pipeline.id, ss_id, sl.id, fetch_jobs[ss.id])
+                data.create_aip_files(100, 300, aip.id)
+                aipcount += 1
+
+            # Update package/AIP counts in fetch job
+            fetch_job = FetchJob.query.get(fetch_jobs[ss_id])
+            fetch_job.total_packages += aipcount
+            fetch_job.total_aips += aipcount
+            db.session.commit()
+
+    print("Done.")

tools/helpers/data.py

@@ -0,0 +1,116 @@
+from datetime import date
+
+from faker import Faker
+
+from AIPscan import db
+from AIPscan.models import (
+    AIP,
+    FetchJob,
+    File,
+    Pipeline,
+    StorageLocation,
+    StorageService,
+)
+
+fake = Faker()
+randint = fake.random.randint
+
+
+def create_pipeline():
+    pipeline = Pipeline(origin_pipeline=fake.uuid4(), dashboard_url=fake.url())
+
+    db.session.add(pipeline)
+    db.session.commit()
+
+    return pipeline
+
+
+def create_storage_service(default):
+    ss = StorageService(
+        name=fake.text(20)[:-1],
+        url=fake.url(),
+        user_name=fake.profile()["username"],
+        api_key=fake.password(),
+        download_limit=0,
+        download_offset=0,
+        default=default,
+    )
+
+    db.session.add(ss)
+    db.session.commit()
+
+    return ss
+
+
+def create_fetch_job(storage_service_id):
+    fetch_job = FetchJob(
+        total_packages=0,
+        total_aips=0,
+        total_deleted_aips=0,
+        download_start=date.today(),
+        download_end=date.today(),
+        download_directory=fake.file_path(),
+        storage_service_id=storage_service_id,
+    )
+    fetch_job.total_dips = 0
+    fetch_job.total_sips = 0
+    fetch_job.total_replicas = 0
+
+    db.session.add(fetch_job)
+    db.session.commit()
+
+    return fetch_job
+
+
+def create_location(storage_service_id):
+    location = StorageLocation(
+        current_location=fake.file_path(),
+        description=fake.text(20)[:-1],
+        storage_service_id=storage_service_id,
+    )
+
+    db.session.add(location)
+    db.session.commit()
+
+    return location
+
+
+def create_aip(pipeline_id, storage_service_id, storage_location_id, fetch_job_id):
+    aip = AIP(
+        uuid=fake.uuid4(),
+        transfer_name=fake.text(20)[:-1],
+        create_date=date.today(),
+        mets_sha256=fake.sha256(),
+        size=randint(10000, 100_000_000),
+        storage_service_id=storage_service_id,
+        storage_location_id=storage_location_id,
+        fetch_job_id=fetch_job_id,
+        origin_pipeline_id=pipeline_id,
+    )
+
+    db.session.add(aip)
+    db.session.commit()
+
+    return aip
+
+
+def create_aip_files(min, max, aip_id):
+    for _ in range(1, randint(min, max)):
+        aipfile = File(
+            aip_id=aip_id,
+            name=fake.text(20)[:-1],
+            filepath=fake.file_path(),
+            uuid=fake.uuid4(),
+            file_type="original",
+            size=randint(1000, 1_000_000),
+            date_created=date.today(),
+            puid=fake.text(20)[:-1],
+            file_format=fake.text(20)[:-1],
+            format_version=fake.text(20)[:-1],
+            checksum_type=fake.text(20)[:-1],
+            checksum_value=fake.text(20)[:-1],
+            premis_object="",
+        )
+
+        db.session.add(aipfile)
+        db.session.commit()

tools/tests/test_data.py

@@ -0,0 +1,88 @@
+import datetime
+
+import pytest
+
+from tools.helpers import data
+
+
+@pytest.fixture
+def mock_db_add(mocker):
+    mocker.patch("AIPscan.db.session.add")
+    mocker.patch("AIPscan.db.session.commit")
+
+
+def test_create_storage_service(mock_db_add):
+    ss = data.create_storage_service(True)
+
+    assert ss.name
+    assert type(ss.name) == str
+
+    assert ss.url
+    assert type(ss.url) == str
+
+    assert ss.user_name
+    assert type(ss.user_name) == str
+
+    assert ss.api_key
+    assert type(ss.api_key) == str
+
+    assert ss.default
+    assert type(ss.default) == bool
+
+    ss = data.create_storage_service(False)
+    assert not ss.default
+
+
+def test_create_fetch_job(mock_db_add):
+    ss = data.create_storage_service(True)
+    ss.id = 1
+
+    fetch_job = data.create_fetch_job(ss.id)
+
+    assert fetch_job.download_start
+    assert type(fetch_job.download_start) == datetime.date
+
+    assert fetch_job.download_end
+    assert type(fetch_job.download_end) == datetime.date
+
+    assert fetch_job.download_directory
+    assert type(fetch_job.download_directory) == str
+
+    assert fetch_job.storage_service_id == ss.id
+
+
+def test_create_location(mock_db_add):
+    location = data.create_location(1)
+
+    assert location.current_location
+    assert type(location.current_location) == str
+
+    assert location.description
+    assert type(location.description) == str
+
+    assert location.storage_service_id == 1
+
+
+def test_create_aip(mock_db_add):
+    aip = data.create_aip(1, 2, 3, 4)
+
+    assert aip.uuid
+    assert type(aip.uuid) == str
+
+    assert aip.transfer_name
+    assert type(aip.transfer_name) == str
+
+    assert aip.create_date
+    assert type(aip.create_date) == datetime.date
+
+    assert aip.mets_sha256
+    assert type(aip.mets_sha256) == str
+
+    assert aip.size
+    assert type(aip.size) == int
+
+    assert aip.origin_pipeline_id == 1
+    assert aip.storage_service_id == 2
+    assert aip.storage_location_id == 3
+    assert aip.fetch_job_id == 4
+    assert aip.origin_pipeline_id == 1