streamline migration from 0.8.x #397

boofilsic/settings.py

@@ -192,6 +192,10 @@ REDIRECT_URIS = env(
 
 # Timeout of requests to Mastodon, in seconds
 MASTODON_TIMEOUT = env("NEODB_LOGIN_MASTODON_TIMEOUT", default=10)  # type: ignore
+TAKAHE_REMOTE_TIMEOUT = MASTODON_TIMEOUT
+
+NEODB_USER_AGENT = f"NeoDB/{NEODB_VERSION} (+{SITE_INFO.get('site_url', 'undefined')})"
+TAKAHE_USER_AGENT = NEODB_USER_AGENT
 
 # Scope when creating Mastodon apps
 # Alternatively, use "read write follow" to avoid re-authorize when migrating to a future version with more features

catalog/sites/discogs.py

@@ -135,7 +135,7 @@ class DiscogsMaster(AbstractSite):
 def get_discogs_data(data_type: str, discogs_id):
     if data_type not in ("releases", "masters"):
         raise ValueError("data_type can only be in ('releases' or masters')")
-    user_agent_string = "Neodb/0.1"
+    user_agent_string = settings.NEODB_USER_AGENT
     user_token = settings.DISCOGS_API_KEY
     headers = {
         "User-Agent": user_agent_string,

catalog/sites/fedi.py

@@ -32,7 +32,10 @@ class FediverseInstance(AbstractSite):
         "Performance": Performance,
         "PerformanceProduction": PerformanceProduction,
     }
-    request_header = {"User-Agent": "NeoDB/0.5", "Accept": "application/activity+json"}
+    request_header = {
+        "User-Agent": settings.NEODB_USER_AGENT,
+        "Accept": "application/activity+json",
+    }
 
     @classmethod
     def id_to_url(cls, id_value):

catalog/sites/rss.py

@@ -41,7 +41,7 @@ class RSS(AbstractSite):
             feed = pickle.load(open(_local_response_path + get_mock_file(url), "rb"))
         else:
             req = urllib.request.Request(url)
-            req.add_header("User-Agent", "NeoDB/0.5")
+            req.add_header("User-Agent", settings.NEODB_USER_AGENT)
             try:
                 feed = podcastparser.parse(url, urllib.request.urlopen(req, timeout=3))
             except:

common/management/commands/cron.py

@@ -41,10 +41,10 @@ class Command(BaseCommand):
 
     def handle(self, *args, **options):
         if options["cancel"]:
-            JobManager.cancel()
+            JobManager.cancel_all()
         if options["schedule"]:
-            JobManager.cancel()  # cancel previously scheduled jobs if any
-            JobManager.schedule()
+            JobManager.cancel_all()  # cancel previously scheduled jobs if any
+            JobManager.schedule_all()
         if options["runonce"]:
             for job_id in options["runonce"]:
                 run = JobManager.run(job_id)

common/management/commands/setup.py

@@ -18,6 +18,7 @@ class Command(BaseCommand):
             domain=domain,
             local=True,
             service_domain=service_domain,
+            state="updated",
             notes="NeoDB",
             nodeinfo={},
         )

common/models.py

@@ -53,12 +53,12 @@ class JobManager:
         return target
 
     @classmethod
-    def schedule(cls):
+    def schedule_all(cls):
         for j in cls.registry:
             j.schedule()
 
     @classmethod
-    def cancel(cls):
+    def cancel_all(cls):
         for j in cls.registry:
             j.cancel()
 
@@ -77,11 +77,11 @@ class JobManager:
         return registry.get_job_ids()
 
     @classmethod
-    def schedule_all(cls):
+    def reschedule_all(cls):
         # TODO rewrite lazy import in a better way
         from catalog.jobs import DiscoverGenerator, PodcastUpdater
         from mastodon.jobs import MastodonSiteCheck
         from users.jobs import MastodonUserSync
 
-        cls.cancel()
-        cls.schedule()
+        cls.cancel_all()
+        cls.schedule_all()

common/setup.py

@@ -20,12 +20,15 @@ class Setup:
     """
 
     def create_site(self, domain, service_domain):
-        TakaheDomain.objects.create(
+        TakaheDomain.objects.update_or_create(
             domain=domain,
-            local=True,
-            service_domain=service_domain,
-            notes="NeoDB",
-            nodeinfo={},
+            defaults={
+                "local": True,
+                "service_domain": service_domain,
+                "notes": "NeoDB",
+                "nodeinfo": {},
+                "state": "updated",
+            },
         )
         TakaheConfig.objects.update_or_create(
             key="public_timeline",
@@ -156,8 +159,9 @@ class Setup:
         # Register cron jobs if not yet
         if settings.DISABLE_CRON:
             logger.info("Cron jobs are disabled.")
+            JobManager.cancel_all()
         else:
-            JobManager.schedule_all()
+            JobManager.reschedule_all()
 
         logger.info("Finished post-migration setup.")
 

compose.yml

@@ -39,6 +39,8 @@ x-shared:
       NEODB_SEARCH_URL: ${NEODB_SEARCH_URL:-typesense://user:eggplant@typesense:8108/catalog}
       NEODB_EMAIL_URL:
       NEODB_EMAIL_FROM: no-reply@${NEODB_SITE_DOMAIN}
+      NEODB_FANOUT_LIMIT_DAYS:
+      TAKAHE_FANOUT_LIMIT_DAYS:
       NEODB_DOWNLOADER_PROXY_LIST:
       NEODB_DOWNLOADER_BACKUP_PROXY:
       NEODB_DOWNLOADER_SAVE_DIR:
@@ -53,8 +55,8 @@ x-shared:
       TAKAHE_MEDIA_BACKEND: local://www/media/
       TAKAHE_MEDIA_ROOT: /www/media
       TAKAHE_USE_PROXY_HEADERS: true
-      TAKAHE_STATOR_CONCURRENCY: 4
-      TAKAHE_STATOR_CONCURRENCY_PER_MODEL: 2
+      TAKAHE_STATOR_CONCURRENCY: ${TAKAHE_STATOR_CONCURRENCY:-4}
+      TAKAHE_STATOR_CONCURRENCY_PER_MODEL: ${TAKAHE_STATOR_CONCURRENCY_PER_MODEL:-2}
       TAKAHE_DEBUG: ${NEODB_DEBUG:-True}
       TAKAHE_VENV: /takahe-venv
       SPOTIFY_API_KEY:

journal/migrations/0020_shelflogentry_unique_shelf_log_entry.py

@@ -2,6 +2,14 @@
 
 from django.db import migrations, models
 
+_sql = """DELETE
+FROM journal_shelflogentry a USING journal_shelflogentry b
+WHERE a.ctid < b.ctid
+    AND a.item_id=b.item_id
+    AND a.owner_id=b.owner_id
+    AND a.timestamp=b.timestamp
+    AND a.shelf_type=b.shelf_type"""
+
 
 class Migration(migrations.Migration):
 
@@ -10,6 +18,12 @@ class Migration(migrations.Migration):
     ]
 
     operations = [
+        migrations.RunSQL("SET CONSTRAINTS ALL IMMEDIATE;"),
+        migrations.RunSQL(
+            sql=_sql,
+            reverse_sql=migrations.RunSQL.noop,
+        ),
+        migrations.RunSQL("SET CONSTRAINTS ALL DEFERRED;"),
         migrations.AddConstraint(
             model_name="shelflogentry",
             constraint=models.UniqueConstraint(

mastodon/api.py

@@ -53,7 +53,7 @@ API_CREATE_APP = "/api/v1/apps"
 # GET
 API_SEARCH = "/api/v2/search"
 
-USER_AGENT = f"NeoDB/{settings.NEODB_VERSION} (+{settings.SITE_INFO.get('site_url', 'undefined')})"
+USER_AGENT = settings.NEODB_USER_AGENT
 
 get = functools.partial(requests.get, timeout=settings.MASTODON_TIMEOUT)
 put = functools.partial(requests.put, timeout=settings.MASTODON_TIMEOUT)

takahe/management/commands/backfill_takahe.py

@@ -0,0 +1,88 @@
+from django.conf import settings
+from django.contrib.contenttypes.models import ContentType
+from django.core.management.base import BaseCommand
+from django.db.models import Count, F
+from loguru import logger
+from tqdm import tqdm
+
+from catalog.common import *
+from catalog.common.models import *
+from catalog.models import *
+from journal.models import *
+from takahe.utils import *
+from users.models import APIdentity
+from users.models import User as NeoUser
+
+
+def content_type_id(cls):
+    return ContentType.objects.get(app_label="journal", model=cls.__name__.lower()).pk
+
+
+class Command(BaseCommand):
+    def add_arguments(self, parser):
+        parser.add_argument(
+            "--verbose",
+            action="store_true",
+        )
+        parser.add_argument(
+            "--post",
+            action="store_true",
+        )
+        parser.add_argument(
+            "--like",
+            action="store_true",
+        )
+        parser.add_argument(
+            "--post-new",
+            action="store_true",
+        )
+        parser.add_argument("--start", default=0, action="store")
+        parser.add_argument("--count", default=0, action="store")
+
+    def process_post(self):
+        logger.info(f"Processing posts...")
+        qs = Piece.objects.filter(
+            polymorphic_ctype__in=[
+                content_type_id(ShelfMember),
+                content_type_id(Comment),
+                content_type_id(Review),
+            ]
+        ).order_by("id")
+        if self.starting_id:
+            qs = qs.filter(id__gte=self.starting_id)
+        tracker = tqdm(qs.iterator(), total=self.count_est or qs.count())
+        for p in tracker:
+            tracker.set_postfix_str(f"{p.id}")
+            if p.__class__ == ShelfMember:
+                mark = Mark(p.owner, p.item)
+                Takahe.post_mark(mark, self.post_new)
+            elif p.__class__ == Comment:
+                if p.item.__class__ in [PodcastEpisode, TVEpisode]:
+                    Takahe.post_comment(p, self.post_new)
+            elif p.__class__ == Review:
+                Takahe.post_review(p, self.post_new)
+
+    def process_like(self):
+        logger.info(f"Processing likes...")
+        qs = Like.objects.order_by("id")
+        tracker = tqdm(qs)
+        for like in tracker:
+            post_id = like.target.latest_post_id
+            if post_id:
+                Takahe.like_post(post_id, like.owner.pk)
+            else:
+                logger.warning(f"Post not found for like {like.id}")
+
+    def handle(self, *args, **options):
+        self.verbose = options["verbose"]
+        self.post_new = options["post_new"]
+        self.starting_id = int(options["start"])
+        self.count_est = int(options["count"])
+
+        if options["post"]:
+            self.process_post()
+
+        if options["like"]:
+            self.process_like()
+
+        self.stdout.write(self.style.SUCCESS(f"Done."))

takahe/models.py

@@ -526,7 +526,7 @@ class Identity(models.Model):
         based on probing host-meta.
         """
         with httpx.Client(
-            timeout=settings.SETUP.REMOTE_TIMEOUT,
+            timeout=settings.TAKAHE_REMOTE_TIMEOUT,
             headers={"User-Agent": settings.TAKAHE_USER_AGENT},
         ) as client:
             try:
@@ -565,7 +565,7 @@ class Identity(models.Model):
 
         # Go make a Webfinger request
         with httpx.Client(
-            timeout=settings.SETUP.REMOTE_TIMEOUT,
+            timeout=settings.TAKAHE_REMOTE_TIMEOUT,
             headers={"User-Agent": settings.TAKAHE_USER_AGENT},
         ) as client:
             try:
@@ -1066,6 +1066,7 @@ class Post(models.Model):
         attachments: list | None = None,
         attachment_attributes: list | None = None,
         type_data: dict | None = None,
+        edited: datetime.datetime | None = None,
     ):
         with transaction.atomic():
             # Strip all HTML and apply linebreaks filter
@@ -1099,6 +1100,12 @@ class Post(models.Model):
             self.state_changed = timezone.now()
             self.state_next_attempt = None
             self.state_locked_until = None
+            if edited and edited < timezone.now():
+                self.published = edited
+                if timezone.now() - edited > datetime.timedelta(
+                    days=settings.FANOUT_LIMIT_DAYS
+                ):
+                    self.state = "edited_fanned_out"  # add post quietly if it's old
             self.save()
 
     @classmethod
@@ -1109,13 +1116,13 @@ class Post(models.Model):
             handle = handle.lower()
             if "@" in handle:
                 username, domain = handle.split("@", 1)
+                local = False
             else:
                 username = handle
                 domain = author.domain_id
+                local = author.local
             identity = Identity.by_username_and_domain(
-                username=username,
-                domain=domain,
-                fetch=True,
+                username=username, domain=domain, fetch=True, local=local
             )
             if identity is not None:
                 mentions.add(identity)

takahe/utils.py

@@ -366,7 +366,11 @@ class Takahe:
             raise ValueError(f"Cannot find post to reply: {reply_to_pk}")
         if post:
             post.edit_local(
-                pre_conetent, content, visibility=visibility, type_data=data
+                pre_conetent,
+                content,
+                visibility=visibility,
+                type_data=data,
+                edited=post_time,
             )
         else:
             post = Post.create_local(
@@ -435,7 +439,7 @@ class Takahe:
             comment.visibility, user.preference.mastodon_publish_public
         )
         existing_post = None if share_as_new_post else comment.latest_post
-        post = Takahe.post(  # TODO post as Article?
+        post = Takahe.post(
             comment.owner.pk,
             pre_conetent,
             content,

users/management/commands/generate_usernames.py

@@ -0,0 +1,77 @@
+from datetime import timedelta
+
+from django.apps import apps
+from django.core.management.base import BaseCommand
+from django.utils import timezone
+from tqdm import tqdm
+
+from users.models.user import _RESERVED_USERNAMES
+
+User = apps.get_model("users", "User")
+_RESERVED_USERNAMES = [
+    "connect",
+    "oauth2_login",
+    "__",
+    "admin",
+    "api",
+    "me",
+]
+
+
+class Command(BaseCommand):
+    help = "Generate unique username"
+
+    def process_users(self, users):
+        count = 0
+        for user in users:
+            if not user.is_active:
+                un = f"-{user.pk}-"
+            else:
+                un = user.mastodon_username
+            if not un:
+                un = f"_{user.pk}"
+            if un.lower() in _RESERVED_USERNAMES:
+                un = f"__{un}"
+            if User.objects.filter(username__iexact=un).exists():  # type: ignore
+                un = f"{un}_{user.pk}"
+            print(f"{user} -> {un}")
+            user.username = un
+            user.save(update_fields=["username"])
+            count += 1
+        print(f"{count} users updated")
+
+    def handle(self, *args, **options):
+        print("Processing active users")
+        # recent logged in users
+        proactive_users = User.objects.filter(  # type: ignore
+            username__isnull=True,
+            is_active=True,
+            last_login__gt=timezone.now() - timedelta(days=30),
+        ).order_by("date_joined")
+        # users with mastodon token still active
+        active_users = (
+            User.objects.filter(  # type: ignore
+                username__isnull=True,
+                is_active=True,
+            )
+            .exclude(mastodon_token="")
+            .order_by("date_joined")
+        )
+        # users with mastodon handler still reachable
+        reachable_users = User.objects.filter(  # type: ignore
+            username__isnull=True,
+            is_active=True,
+            mastodon_last_reachable__gt=timezone.now() - timedelta(days=7),
+        ).order_by("date_joined")
+        # all other users
+        users = User.objects.filter(  # type: ignore
+            username__isnull=True,
+        ).order_by("date_joined")
+        print(f"{proactive_users.count()} proactive users")
+        self.process_users(proactive_users)
+        print(f"{active_users.count()} active users")
+        self.process_users(active_users)
+        print(f"{reachable_users.count()} reachable users")
+        self.process_users(reachable_users)
+        print(f"{users.count()} other users")
+        self.process_users(users)

users/migrations/0013_init_identity.py

@@ -1,7 +1,10 @@
 # Generated by Django 4.2.4 on 2023-08-09 16:54
 
+from datetime import timedelta
+
 from django.conf import settings
 from django.db import migrations, models, transaction
+from django.utils import timezone
 from loguru import logger
 from tqdm import tqdm
 
@@ -22,19 +25,17 @@ def init_domain(apps, schema_editor):
             "No users found, skip domain migration (if you are running initial migration for new site, pls ignore this)"
         )
         return
-    d = TakaheDomain.objects.filter(domain=domain).first()
-    if not d:
-        logger.info(f"Creating takahe domain {domain}")
-        TakaheDomain.objects.create(
-            domain=domain,
-            local=True,
-            service_domain=service_domain,
-            notes="NeoDB",
-            nodeinfo={},
-        )
-    else:
-        logger.info(f"Takahe domain {domain} already exists")
-
+    logger.info(f"Creating takahe domain {domain}")
+    TakaheDomain.objects.update_or_create(
+        domain=domain,
+        defaults={
+            "local": True,
+            "service_domain": service_domain,
+            "notes": "NeoDB",
+            "nodeinfo": {},
+            "state": "updated",
+        },
+    )
     TakaheConfig.objects.update_or_create(
         key="public_timeline",
         user=None,
@@ -85,7 +86,9 @@ def init_identity(apps, schema_editor):
             local=True,
             username=username,
             domain_name=domain,
-            deleted=None if user.is_active else user.updated,
+            deleted=None
+            if user.is_active
+            else user.mastodon_last_reachable + timedelta(days=90),
         )
         takahe_user = TakaheUser.objects.create(
             pk=user.pk, email=handler, admin=user.is_superuser
@@ -93,12 +96,14 @@ def init_identity(apps, schema_editor):
         takahe_identity = TakaheIdentity.objects.create(
             pk=user.pk,
             actor_uri=f"https://{service_domain or domain}/@{username}@{domain}/",
-            profile_uri=user.url,
+            profile_uri=f"https://{service_domain or domain}/users/{username}/",
             username=username,
             domain=tdomain,
             name=username,
             local=True,
-            discoverable=not user.preference.no_anonymous_view,
+            discoverable=True,
+            state="updated",
+            state_next_attempt=timezone.now() + timedelta(days=365 * 99),
         )
         takahe_identity.generate_keypair()
         takahe_user.identities.add(takahe_identity)

users/migrations/0014_preference_mastodon_skip_relationship_and_more.py

@@ -6,9 +6,13 @@ from tqdm import tqdm
 
 
 def migrate_relationships(apps, schema_editor):
-    User = apps.get_model("users", "User")
-    APIdentity = apps.get_model("users", "APIdentity")
-    logger.info(f"Migrate user relationship")
+    # User = apps.get_model("users", "User")
+    # APIdentity = apps.get_model("users", "APIdentity")
+    from takahe.models import Block as TakaheBlock
+    from takahe.models import Follow as TakaheFollow
+    from users.models import APIdentity, User
+
+    logger.info(f"Migrate local user relationship")
     for user in tqdm(User.objects.filter(is_active=True)):
         identity = APIdentity.objects.get(user=user)
         for target in user.local_following.all():
@@ -20,11 +24,12 @@ def migrate_relationships(apps, schema_editor):
         for target in user.local_muting.all():
             target_identity = APIdentity.objects.get(user=target)
             identity.mute(target_identity)
-        user.sync_relationship()
+    logger.info(f"Migrate sync user relationship")
     for user in tqdm(User.objects.filter(is_active=True)):
-        identity = APIdentity.objects.get(user=user)
-        for target_identity in identity.follow_requesting_identities:
-            target_identity.accept_follow_request(identity)
+        user.sync_relationship()
+    logger.info(f"Update relationship states")
+    TakaheBlock.objects.all().update(state="sent")
+    TakaheFollow.objects.all().update(state="accepted")
 
 
 class Migration(migrations.Migration):

users/models/user.py

@@ -236,26 +236,23 @@ class User(AbstractUser):
     def sync_relationship(self):
         from .apidentity import APIdentity
 
-        for target in self.mastodon_followers:
-            t = target.split("@")
-            target_identity = APIdentity.objects.filter(
-                user__mastodon_username=t[0], user__mastodon_site=t[1]
-            ).first()
-            if target_identity and not self.identity.is_following(target_identity):
+        def get_identities(accts: list):
+            q = Q(pk__in=[])
+            for acct in accts or []:
+                t = acct.split("@") if acct else []
+                if len(t) == 2:
+                    q = q | Q(mastodon_username=t[0], mastodon_site=t[1])
+            users = User.objects.filter(is_active=True).filter(q)
+            return APIdentity.objects.filter(user__in=users)
+
+        for target_identity in get_identities(self.mastodon_following):
+            if not self.identity.is_following(target_identity):
                 self.identity.follow(target_identity)
-        for target in self.mastodon_blocks:
-            t = target.split("@")
-            target_identity = APIdentity.objects.filter(
-                user__mastodon_username=t[0], user__mastodon_site=t[1]
-            ).first()
-            if target_identity and not self.identity.is_blocking(target_identity):
+        for target_identity in get_identities(self.mastodon_blocks):
+            if not self.identity.is_blocking(target_identity):
                 self.identity.block(target_identity)
-        for target in self.mastodon_mutes:
-            t = target.split("@")
-            target_identity = APIdentity.objects.filter(
-                user__mastodon_username=t[0], user__mastodon_site=t[1]
-            ).first()
-            if target_identity and not self.identity.is_muting(target_identity):
+        for target_identity in get_identities(self.mastodon_mutes):
+            if not self.identity.is_muting(target_identity):
                 self.identity.mute(target_identity)
 
     def sync_identity(self):