Implement self-contained archives to import and export entire users b…

…etween instances (bookwyrm-social#38)

Co-authored-by: Daniel Burgess <developerdannymate@gmail.com>
Co-authored-by: Hugh Rundle <hugh@hughrundle.net>
Co-authored-by: dannymate <dannymate@noreply.codeberg.org>
Co-authored-by: hughrun <hughrun@noreply.codeberg.org>
Reviewed-on: https://codeberg.org/GuildAlpha/bookwyrm/pulls/38
Co-authored-by: CSDUMMI <csdummi.misquality@simplelogin.co>
Co-committed-by: CSDUMMI <csdummi.misquality@simplelogin.co>

bookwyrm/forms/forms.py

@@ -25,6 +25,10 @@ class ImportForm(forms.Form):
     csv_file = forms.FileField()
 
 
+class ImportUserForm(forms.Form):
+    archive_file = forms.FileField()
+
+
 class ShelfForm(CustomForm):
     class Meta:
         model = models.Shelf

bookwyrm/importers/__init__.py

@@ -1,6 +1,7 @@
 """ import classes """
 
 from .importer import Importer
+from .bookwyrm_import import BookwyrmImporter
 from .calibre_import import CalibreImporter
 from .goodreads_import import GoodreadsImporter
 from .librarything_import import LibrarythingImporter

bookwyrm/importers/bookwyrm_import.py

@@ -0,0 +1,18 @@
+"""Import data from Bookwyrm export files"""
+from bookwyrm import settings
+from bookwyrm.models.bookwyrm_import_job import BookwyrmImportJob
+
+class BookwyrmImporter:
+    """Import a Bookwyrm User export JSON file.
+    This is kind of a combination of an importer and a connector.
+    """
+
+    def process_import(self, user, archive_file, settings):
+        """import user data from a Bookwyrm export file"""
+
+        required = [k for k in settings if settings.get(k) == "on"]
+
+        job = BookwyrmImportJob.objects.create(
+            user=user, archive_file=archive_file, required=required
+        )
+        return job

bookwyrm/migrations/0179_bookwyrmexportjob_bookwyrmimportjob_childjob_parentjob.py

@@ -0,0 +1,71 @@
+# Generated by Django 3.2.19 on 2023-08-31 22:57
+
+from django.conf import settings
+import django.contrib.postgres.fields
+from django.db import migrations, models
+import django.db.models.deletion
+import django.utils.timezone
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('bookwyrm', '0178_auto_20230328_2132'),
+    ]
+
+    operations = [
+        migrations.CreateModel(
+            name='ParentJob',
+            fields=[
+                ('id', models.AutoField(auto_created=True, primary_key=True, serialize=False, verbose_name='ID')),
+                ('task_id', models.UUIDField(blank=True, null=True, unique=True)),
+                ('created_date', models.DateTimeField(default=django.utils.timezone.now)),
+                ('updated_date', models.DateTimeField(default=django.utils.timezone.now)),
+                ('complete', models.BooleanField(default=False)),
+                ('status', models.CharField(choices=[('pending', 'Pending'), ('active', 'Active'), ('complete', 'Complete'), ('stopped', 'Stopped')], default='pending', max_length=50, null=True)),
+                ('user', models.ForeignKey(on_delete=django.db.models.deletion.CASCADE, to=settings.AUTH_USER_MODEL)),
+            ],
+            options={
+                'abstract': False,
+            },
+        ),
+        migrations.CreateModel(
+            name='BookwyrmExportJob',
+            fields=[
+                ('parentjob_ptr', models.OneToOneField(auto_created=True, on_delete=django.db.models.deletion.CASCADE, parent_link=True, primary_key=True, serialize=False, to='bookwyrm.parentjob')),
+                ('export_data', models.FileField(null=True, upload_to='')),
+            ],
+            options={
+                'abstract': False,
+            },
+            bases=('bookwyrm.parentjob',),
+        ),
+        migrations.CreateModel(
+            name='BookwyrmImportJob',
+            fields=[
+                ('parentjob_ptr', models.OneToOneField(auto_created=True, on_delete=django.db.models.deletion.CASCADE, parent_link=True, primary_key=True, serialize=False, to='bookwyrm.parentjob')),
+                ('archive_file', models.FileField(blank=True, null=True, upload_to='')),
+                ('import_data', models.JSONField(null=True)),
+                ('required', django.contrib.postgres.fields.ArrayField(base_field=models.CharField(blank=True, max_length=50), blank=True, size=None)),
+            ],
+            options={
+                'abstract': False,
+            },
+            bases=('bookwyrm.parentjob',),
+        ),
+        migrations.CreateModel(
+            name='ChildJob',
+            fields=[
+                ('id', models.AutoField(auto_created=True, primary_key=True, serialize=False, verbose_name='ID')),
+                ('task_id', models.UUIDField(blank=True, null=True, unique=True)),
+                ('created_date', models.DateTimeField(default=django.utils.timezone.now)),
+                ('updated_date', models.DateTimeField(default=django.utils.timezone.now)),
+                ('complete', models.BooleanField(default=False)),
+                ('status', models.CharField(choices=[('pending', 'Pending'), ('active', 'Active'), ('complete', 'Complete'), ('stopped', 'Stopped')], default='pending', max_length=50, null=True)),
+                ('parent_job', models.ForeignKey(on_delete=django.db.models.deletion.CASCADE, related_name='child_jobs', to='bookwyrm.parentjob')),
+            ],
+            options={
+                'abstract': False,
+            },
+        ),
+    ]

bookwyrm/migrations/0182_merge_20230905_2240.py

@@ -0,0 +1,14 @@
+# Generated by Django 3.2.19 on 2023-09-05 22:40
+
+from django.db import migrations
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('bookwyrm', '0179_bookwyrmexportjob_bookwyrmimportjob_childjob_parentjob'),
+        ('bookwyrm', '0181_merge_20230806_2302'),
+    ]
+
+    operations = [
+    ]

bookwyrm/models/__init__.py

@@ -26,6 +26,7 @@
 from .group import Group, GroupMember, GroupMemberInvitation
 
 from .import_job import ImportJob, ImportItem
+from .bookwyrm_import_job import BookwyrmImportJob
 
 from .site import SiteSettings, Theme, SiteInvite
 from .site import PasswordReset, InviteRequest

bookwyrm/models/bookwyrm_export_job.py

@@ -0,0 +1,222 @@
+import logging
+
+from django.db.models import FileField
+from django.db.models import Q
+from django.core.serializers.json import DjangoJSONEncoder
+from django.core.files.base import ContentFile
+
+from bookwyrm import models
+from bookwyrm.settings import DOMAIN
+from bookwyrm.tasks import app, IMPORTS
+from bookwyrm.models.job import ParentJob, ParentTask, SubTask, create_child_job
+from uuid import uuid4
+from bookwyrm.utils.tar import BookwyrmTarFile
+
+logger = logging.getLogger(__name__)
+
+class BookwyrmExportJob(ParentJob):
+    """entry for a specific request to export a bookwyrm user"""
+    export_data = FileField(null=True)
+
+    def start_job(self):
+        """Start the job"""
+        start_export_task.delay(job_id=self.id, no_children=True)
+
+        return self
+
+@app.task(queue=IMPORTS, base=ParentTask)
+def start_export_task(**kwargs):
+    """trigger the child tasks for each row"""
+    job = BookwyrmExportJob.objects.get(id=kwargs["job_id"])
+
+    # don't start the job if it was stopped from the UI
+    if job.complete:
+        return
+
+    # This is where ChildJobs get made
+    job.export_data = ContentFile(b"", str(uuid4()))
+
+    json_data = json_export(job.user)
+    tar_export(json_data, job.user, job.export_data)
+
+    job.save(update_fields=["export_data"])
+
+def tar_export(json_data: str, user, f):
+    f.open("wb")
+    with BookwyrmTarFile.open(mode="w:gz", fileobj=f) as tar:
+        tar.write_bytes(json_data.encode("utf-8"))
+
+        # Add avatar image if present
+        if getattr(user, "avatar", False):
+            tar.add_image(user.avatar, filename="avatar")
+
+        editions, books = get_books_for_user(user)
+        for book in editions:
+            tar.add_image(book.cover)
+
+    f.close()
+
+def json_export(user):
+    """Generate an export for a user"""
+    # user
+    exported_user = {}
+    vals = [
+        "username",
+        "name",
+        "summary",
+        "manually_approves_followers",
+        "hide_follows",
+        "show_goal",
+        "show_suggested_users",
+        "discoverable",
+        "preferred_timezone",
+        "default_post_privacy",
+    ]
+    for k in vals:
+        exported_user[k] = getattr(user, k)
+
+    if getattr(user, "avatar", False):
+        exported_user["avatar"] = f'https://{DOMAIN}{getattr(user, "avatar").url}'
+
+    # reading goals
+    reading_goals = models.AnnualGoal.objects.filter(user=user).distinct()
+    goals_list = []
+    try:
+        for goal in reading_goals:
+            goals_list.append(
+                {"goal": goal.goal, "year": goal.year, "privacy": goal.privacy}
+            )
+    except Exception:
+        pass
+
+    try:
+        readthroughs = models.ReadThrough.objects.filter(
+            user=user
+        ).distinct().values()
+        readthroughs = list(readthroughs)
+    except Exception as e:
+        readthroughs = []
+
+    # books
+    editions, books = get_books_for_user(user)
+    final_books = []
+
+    for book in books.values():
+        edition = editions.filter(id=book["id"])
+        book["edition"] = edition.values()[0]
+        # authors
+        book["authors"] = list(edition.first().authors.all().values())
+        # readthroughs
+        book_readthroughs = models.ReadThrough.objects.filter(
+            user=user, book=book["id"]
+        ).distinct().values()
+        book["readthroughs"] = list(book_readthroughs)
+        # shelves
+        shelf_books = models.ShelfBook.objects.filter(
+            user=user, book=book['id']
+        ).distinct()
+        shelves_from_books = models.Shelf.objects.filter(
+            shelfbook__in=shelf_books, user=user
+        )
+
+        book["shelves"] = list(shelves_from_books.values())
+        book["shelf_books"] = {}
+
+        for shelf in shelves_from_books:
+            shelf_contents = models.ShelfBook.objects.filter(
+                user=user, shelf=shelf
+            ).distinct()
+
+            book["shelf_books"][shelf.identifier] = list(shelf_contents.values())
+
+        # book lists
+        book_lists = models.List.objects.filter(
+            books__in=[book['id']], user=user
+        ).distinct()
+        book["lists"] = list(book_lists.values())
+        book["list_items"] = {}
+        for blist in book_lists:
+            list_items = models.ListItem.objects.filter(book_list=blist).distinct()
+            book["list_items"][blist.name] = list(list_items.values())
+
+        # reviews
+        reviews = models.Review.objects.filter(
+            user=user, book=book["id"]
+        ).distinct()
+
+        book["reviews"] = list(reviews.values())
+
+        # comments
+        comments = models.Comment.objects.filter(
+            user=user, book=book["id"]
+        ).distinct()
+
+        book["comments"] = list(comments.values())
+        logger.error("FINAL COMMENTS")
+        logger.error(book["comments"])
+
+        # quotes
+        quotes = models.Quotation.objects.filter(
+            user=user, book=book["id"]
+        ).distinct()
+        # quote_statuses = models.Status.objects.filter(
+        #     id__in=quotes, user=kwargs["user"]
+        # ).distinct()
+
+        book["quotes"] = list(quotes.values())
+
+        logger.error("FINAL QUOTES")
+        logger.error(book["quotes"])
+
+        # append everything
+        final_books.append(book)
+
+    # saved book lists
+    saved_lists = models.List.objects.filter(
+        id__in=user.saved_lists.all()
+    ).distinct()
+    saved_lists = [l.remote_id for l in saved_lists]
+
+    # follows
+    follows = models.UserFollows.objects.filter(
+        user_subject=user
+    ).distinct()
+    following = models.User.objects.filter(
+        userfollows_user_object__in=follows
+    ).distinct()
+    follows = [f.remote_id for f in following]
+
+    # blocks
+    blocks = models.UserBlocks.objects.filter(user_subject=user).distinct()
+    blocking = models.User.objects.filter(
+        userblocks_user_object__in=blocks
+    ).distinct()
+
+    blocks = [b.remote_id for b in blocking]
+
+    data = {
+        "user": exported_user,
+        "goals": goals_list,
+        "books": final_books,
+        "saved_lists": saved_lists,
+        "follows": follows,
+        "blocked_users": blocks,
+    }
+
+    return DjangoJSONEncoder().encode(data)
+
+def get_books_for_user(user):
+    """Get all the books and editions related to a user
+    :returns: tuple of editions, books
+    """
+    all_books = models.Edition.viewer_aware_objects(user)
+    editions = all_books.filter(
+        Q(shelves__user=user)
+        | Q(readthrough__user=user)
+        | Q(review__user=user)
+        | Q(list__user=user)
+        | Q(comment__user=user)
+        | Q(quotation__user=user)
+    ).distinct()
+    books = models.Book.objects.filter(id__in=editions).distinct()
+    return editions, books