moviewyrm/bookwyrm/tests/test_goodreads_import.py

253 lines
11 KiB
Python
Raw Normal View History

2021-01-02 17:42:50 +00:00
''' testing import '''
2021-01-02 19:29:50 +00:00
from collections import namedtuple
2021-01-13 21:54:01 +00:00
import csv
2021-01-02 17:42:50 +00:00
import pathlib
2021-01-02 19:29:50 +00:00
from unittest.mock import patch
2021-01-02 17:42:50 +00:00
from django.test import TestCase
import responses
2021-02-20 16:02:36 +00:00
from bookwyrm import models, importer
from bookwyrm.goodreads_import import GoodreadsImporter
from bookwyrm import importer
2021-01-02 19:29:50 +00:00
from bookwyrm.settings import DOMAIN
2021-01-02 17:42:50 +00:00
class GoodreadsImport(TestCase):
''' importing from goodreads csv '''
def setUp(self):
2021-02-20 16:02:36 +00:00
self.importer = GoodreadsImporter()
2021-01-02 17:42:50 +00:00
''' use a test csv '''
datafile = pathlib.Path(__file__).parent.joinpath(
'data/goodreads.csv')
2021-02-20 16:02:36 +00:00
self.csv = open(datafile, 'r', encoding=self.importer.encoding)
2021-01-02 17:42:50 +00:00
self.user = models.User.objects.create_user(
'mouse', 'mouse@mouse.mouse', 'password', local=True)
2021-01-02 19:29:50 +00:00
models.Connector.objects.create(
identifier=DOMAIN,
name='Local',
local=True,
connector_file='self_connector',
base_url='https://%s' % DOMAIN,
books_url='https://%s/book' % DOMAIN,
covers_url='https://%s/images/covers' % DOMAIN,
search_url='https://%s/search?q=' % DOMAIN,
priority=1,
)
2021-01-13 22:02:01 +00:00
work = models.Work.objects.create(title='Test Work')
self.book = models.Edition.objects.create(
title='Example Edition',
remote_id='https://example.com/book/1',
parent_work=work
)
2021-01-02 19:29:50 +00:00
2021-01-02 17:42:50 +00:00
def test_create_job(self):
''' creates the import job entry and checks csv '''
2021-02-20 16:02:36 +00:00
import_job = self.importer.create_job(
2021-01-02 19:29:50 +00:00
self.user, self.csv, False, 'public')
2021-01-02 17:42:50 +00:00
self.assertEqual(import_job.user, self.user)
self.assertEqual(import_job.include_reviews, False)
self.assertEqual(import_job.privacy, 'public')
import_items = models.ImportItem.objects.filter(job=import_job).all()
2021-01-02 19:29:50 +00:00
self.assertEqual(len(import_items), 3)
2021-01-02 17:42:50 +00:00
self.assertEqual(import_items[0].index, 0)
self.assertEqual(import_items[0].data['Book Id'], '42036538')
self.assertEqual(import_items[1].index, 1)
self.assertEqual(import_items[1].data['Book Id'], '52691223')
self.assertEqual(import_items[2].index, 2)
self.assertEqual(import_items[2].data['Book Id'], '28694510')
2021-01-02 19:29:50 +00:00
def test_create_retry_job(self):
''' trying again with items that didn't import '''
2021-02-20 16:02:36 +00:00
import_job = self.importer.create_job(
2021-01-02 19:29:50 +00:00
self.user, self.csv, False, 'unlisted')
import_items = models.ImportItem.objects.filter(
job=import_job
).all()[:2]
2021-02-20 16:02:36 +00:00
retry = self.importer.create_retry_job(
2021-01-02 19:29:50 +00:00
self.user, import_job, import_items)
self.assertNotEqual(import_job, retry)
self.assertEqual(retry.user, self.user)
self.assertEqual(retry.include_reviews, False)
self.assertEqual(retry.privacy, 'unlisted')
retry_items = models.ImportItem.objects.filter(job=retry).all()
self.assertEqual(len(retry_items), 2)
self.assertEqual(retry_items[0].index, 0)
self.assertEqual(retry_items[0].data['Book Id'], '42036538')
self.assertEqual(retry_items[1].index, 1)
self.assertEqual(retry_items[1].data['Book Id'], '52691223')
def test_start_import(self):
''' begin loading books '''
2021-02-20 16:02:36 +00:00
import_job = self.importer.create_job(
2021-01-02 19:29:50 +00:00
self.user, self.csv, False, 'unlisted')
MockTask = namedtuple('Task', ('id'))
mock_task = MockTask(7)
2021-02-20 16:02:36 +00:00
with patch('bookwyrm.importer.import_data.delay') as start:
2021-01-02 19:29:50 +00:00
start.return_value = mock_task
2021-02-20 16:02:36 +00:00
self.importer.start_import(import_job)
2021-01-02 19:29:50 +00:00
import_job.refresh_from_db()
self.assertEqual(import_job.task_id, '7')
2021-01-02 17:42:50 +00:00
@responses.activate
def test_import_data(self):
''' resolve entry '''
2021-02-20 16:02:36 +00:00
import_job = self.importer.create_job(
2021-01-02 19:29:50 +00:00
self.user, self.csv, False, 'unlisted')
book = models.Edition.objects.create(title='Test Book')
with patch(
'bookwyrm.models.import_job.ImportItem.get_book_from_isbn'
) as resolve:
resolve.return_value = book
2021-02-20 16:02:36 +00:00
with patch('bookwyrm.importer.handle_imported_book'):
importer.import_data(self.importer.service, import_job.id)
2021-01-02 19:29:50 +00:00
import_item = models.ImportItem.objects.get(job=import_job, index=0)
self.assertEqual(import_item.book.id, book.id)
2021-01-13 21:54:01 +00:00
def test_handle_imported_book(self):
''' goodreads import added a book, this adds related connections '''
2021-01-13 22:02:01 +00:00
shelf = self.user.shelf_set.filter(identifier='read').first()
2021-01-13 21:54:01 +00:00
self.assertIsNone(shelf.books.first())
2021-01-13 22:02:01 +00:00
import_job = models.ImportJob.objects.create(user=self.user)
2021-01-13 21:54:01 +00:00
datafile = pathlib.Path(__file__).parent.joinpath('data/goodreads.csv')
csv_file = open(datafile, 'r')
for index, entry in enumerate(list(csv.DictReader(csv_file))):
2021-02-20 16:02:36 +00:00
entry = self.importer.parse_fields(entry)
2021-01-13 21:54:01 +00:00
import_item = models.ImportItem.objects.create(
job_id=import_job.id, index=index, data=entry, book=self.book)
break
2021-02-07 19:29:24 +00:00
with patch('bookwyrm.models.activitypub_mixin.broadcast_task.delay'):
2021-02-20 16:02:36 +00:00
importer.handle_imported_book(
self.importer.service, self.user, import_item, False, 'public')
2021-01-13 21:54:01 +00:00
shelf.refresh_from_db()
self.assertEqual(shelf.books.first(), self.book)
2021-01-13 22:02:01 +00:00
readthrough = models.ReadThrough.objects.get(user=self.user)
2021-01-13 21:54:01 +00:00
self.assertEqual(readthrough.book, self.book)
# I can't remember how to create dates and I don't want to look it up.
self.assertEqual(readthrough.start_date.year, 2020)
self.assertEqual(readthrough.start_date.month, 10)
self.assertEqual(readthrough.start_date.day, 21)
self.assertEqual(readthrough.finish_date.year, 2020)
self.assertEqual(readthrough.finish_date.month, 10)
self.assertEqual(readthrough.finish_date.day, 25)
def test_handle_imported_book_already_shelved(self):
''' goodreads import added a book, this adds related connections '''
2021-02-07 19:29:24 +00:00
with patch('bookwyrm.models.activitypub_mixin.broadcast_task.delay'):
shelf = self.user.shelf_set.filter(identifier='to-read').first()
models.ShelfBook.objects.create(
shelf=shelf, user=self.user, book=self.book)
2021-01-13 21:54:01 +00:00
2021-01-13 22:02:01 +00:00
import_job = models.ImportJob.objects.create(user=self.user)
2021-01-13 21:54:01 +00:00
datafile = pathlib.Path(__file__).parent.joinpath('data/goodreads.csv')
csv_file = open(datafile, 'r')
for index, entry in enumerate(list(csv.DictReader(csv_file))):
2021-02-20 16:02:36 +00:00
entry = self.importer.parse_fields(entry)
2021-01-13 21:54:01 +00:00
import_item = models.ImportItem.objects.create(
job_id=import_job.id, index=index, data=entry, book=self.book)
break
2021-02-07 19:29:24 +00:00
with patch('bookwyrm.models.activitypub_mixin.broadcast_task.delay'):
2021-02-20 16:02:36 +00:00
importer.handle_imported_book(
self.importer.service, self.user, import_item, False, 'public')
2021-01-13 21:54:01 +00:00
shelf.refresh_from_db()
self.assertEqual(shelf.books.first(), self.book)
self.assertIsNone(
2021-01-13 22:02:01 +00:00
self.user.shelf_set.get(identifier='read').books.first())
readthrough = models.ReadThrough.objects.get(user=self.user)
2021-01-13 21:54:01 +00:00
self.assertEqual(readthrough.book, self.book)
self.assertEqual(readthrough.start_date.year, 2020)
self.assertEqual(readthrough.start_date.month, 10)
self.assertEqual(readthrough.start_date.day, 21)
self.assertEqual(readthrough.finish_date.year, 2020)
self.assertEqual(readthrough.finish_date.month, 10)
self.assertEqual(readthrough.finish_date.day, 25)
def test_handle_import_twice(self):
''' re-importing books '''
2021-01-13 22:02:01 +00:00
shelf = self.user.shelf_set.filter(identifier='read').first()
import_job = models.ImportJob.objects.create(user=self.user)
2021-01-13 21:54:01 +00:00
datafile = pathlib.Path(__file__).parent.joinpath('data/goodreads.csv')
csv_file = open(datafile, 'r')
for index, entry in enumerate(list(csv.DictReader(csv_file))):
2021-02-20 16:02:36 +00:00
entry = self.importer.parse_fields(entry)
2021-01-13 21:54:01 +00:00
import_item = models.ImportItem.objects.create(
job_id=import_job.id, index=index, data=entry, book=self.book)
break
2021-02-07 19:29:24 +00:00
with patch('bookwyrm.models.activitypub_mixin.broadcast_task.delay'):
2021-02-20 16:02:36 +00:00
importer.handle_imported_book(
self.importer.service, self.user, import_item, False, 'public')
importer.handle_imported_book(
self.importer.service, self.user, import_item, False, 'public')
2021-01-13 21:54:01 +00:00
shelf.refresh_from_db()
self.assertEqual(shelf.books.first(), self.book)
2021-01-13 22:02:01 +00:00
readthrough = models.ReadThrough.objects.get(user=self.user)
2021-01-13 21:54:01 +00:00
self.assertEqual(readthrough.book, self.book)
# I can't remember how to create dates and I don't want to look it up.
self.assertEqual(readthrough.start_date.year, 2020)
self.assertEqual(readthrough.start_date.month, 10)
self.assertEqual(readthrough.start_date.day, 21)
self.assertEqual(readthrough.finish_date.year, 2020)
self.assertEqual(readthrough.finish_date.month, 10)
self.assertEqual(readthrough.finish_date.day, 25)
def test_handle_imported_book_review(self):
''' goodreads review import '''
2021-01-13 22:02:01 +00:00
import_job = models.ImportJob.objects.create(user=self.user)
2021-01-13 21:54:01 +00:00
datafile = pathlib.Path(__file__).parent.joinpath('data/goodreads.csv')
csv_file = open(datafile, 'r')
entry = list(csv.DictReader(csv_file))[2]
2021-02-20 16:02:36 +00:00
entry = self.importer.parse_fields(entry)
2021-01-13 21:54:01 +00:00
import_item = models.ImportItem.objects.create(
job_id=import_job.id, index=0, data=entry, book=self.book)
2021-02-07 19:29:24 +00:00
with patch('bookwyrm.models.activitypub_mixin.broadcast_task.delay'):
2021-02-20 16:02:36 +00:00
importer.handle_imported_book(
self.importer.service, self.user, import_item, True, 'unlisted')
2021-01-13 22:02:01 +00:00
review = models.Review.objects.get(book=self.book, user=self.user)
2021-01-13 21:54:01 +00:00
self.assertEqual(review.content, 'mixed feelings')
self.assertEqual(review.rating, 2)
self.assertEqual(review.published_date.year, 2019)
self.assertEqual(review.published_date.month, 7)
self.assertEqual(review.published_date.day, 8)
self.assertEqual(review.privacy, 'unlisted')
def test_handle_imported_book_reviews_disabled(self):
''' goodreads review import '''
2021-01-13 22:02:01 +00:00
import_job = models.ImportJob.objects.create(user=self.user)
2021-01-13 21:54:01 +00:00
datafile = pathlib.Path(__file__).parent.joinpath('data/goodreads.csv')
csv_file = open(datafile, 'r')
entry = list(csv.DictReader(csv_file))[2]
2021-02-20 16:02:36 +00:00
entry = self.importer.parse_fields(entry)
2021-01-13 21:54:01 +00:00
import_item = models.ImportItem.objects.create(
job_id=import_job.id, index=0, data=entry, book=self.book)
2021-02-07 19:29:24 +00:00
with patch('bookwyrm.models.activitypub_mixin.broadcast_task.delay'):
2021-02-20 16:02:36 +00:00
importer.handle_imported_book(
self.importer.service, self.user, import_item, False, 'unlisted')
2021-01-13 21:54:01 +00:00
self.assertFalse(models.Review.objects.filter(
2021-01-13 22:02:01 +00:00
book=self.book, user=self.user
2021-01-13 21:54:01 +00:00
).exists())