instance allow list

CONTRIBUTING.md

@@ -5,7 +5,7 @@ When it matures enough, PyFedi will aim to work in a way consistent with the [Co
 Please discuss your ideas in an issue at https://codeberg.org/rimu/pyfedi/issues before 
 starting any large pieces of work to ensure alignment with the roadmap, architecture and processes.
 
-The general style and philosphy behind the way things have been constructed is well described by 
+The general style and philosophy behind the way things have been constructed is well described by 
 [The Grug Brained Developer](https://grugbrain.dev/). If that page resonates with you then you'll
 probably enjoy your time here! The codebase needs to be simple enough that new developers of all
 skill levels can easily understand what's going on and onboard quickly without a lot of upfront 

app/activitypub/util.py

@@ -5,7 +5,7 @@ from typing import Union, Tuple
 from flask import current_app
 from sqlalchemy import text
 from app import db, cache
-from app.models import User, Post, Community, BannedInstances, File, PostReply
+from app.models import User, Post, Community, BannedInstances, File, PostReply, AllowedInstances
 import time
 import base64
 import requests
@@ -14,7 +14,7 @@ from cryptography.hazmat.primitives.asymmetric import padding
 from app.constants import *
 from urllib.parse import urlparse
 
-from app.utils import get_request, allowlist_html, html_to_markdown
+from app.utils import get_request, allowlist_html, html_to_markdown, get_setting
 
 
 def public_key():
@@ -184,6 +184,15 @@ def instance_blocked(host: str) -> bool:
     return instance is not None
 
 
+@cache.memoize(150)
+def instance_allowed(host: str) -> bool:
+    host = host.lower()
+    if 'https://' in host or 'http://' in host:
+        host = urlparse(host).hostname
+    instance = AllowedInstances.query.filter_by(domain=host.strip()).first()
+    return instance is not None
+
+
 def find_actor_or_create(actor: str) -> Union[User, Community, None]:
     user = None
     # actor parameter must be formatted as https://server/u/actor or https://server/c/actor
@@ -197,8 +206,12 @@ def find_actor_or_create(actor: str) -> Union[User, Community, None]:
             return None
     elif actor.startswith('https://'):
         server, address = extract_domain_and_actor(actor)
-        if instance_blocked(server):
-            return None
+        if get_setting('use_allowlist', False):
+            if not instance_allowed(server):
+                return None
+        else:
+            if instance_blocked(server):
+                return None
         user = User.query.filter_by(
             ap_profile_id=actor).first()  # finds users formatted like https://kbin.social/u/tables
         if user.banned:

app/models.py

@@ -451,6 +451,12 @@ class BannedInstances(db.Model):
     created_at = db.Column(db.DateTime, default=datetime.utcnow)
 
 
+class AllowedInstances(db.Model):
+    id = db.Column(db.Integer, primary_key=True)
+    domain = db.Column(db.String(256), index=True)
+    created_at = db.Column(db.DateTime, default=datetime.utcnow)
+
+
 class Instance(db.Model):
     id = db.Column(db.Integer, primary_key=True)
     domain = db.Column(db.String(256), index=True)

env.sample

@@ -0,0 +1,10 @@
+SECRET_KEY=
+SERVER_NAME='127.0.0.1:5000'
+DATABASE_URL=postgresql+psycopg2://pyfedi:pyfedi@127.0.0.1/pyfedi
+MAIL_SERVER=''
+RECAPTCHA3_PUBLIC_KEY=''
+RECAPTCHA3_PRIVATE_KEY=''
+MODE='development'
+FULL_AP_CONTEXT=True
+CACHE_TYPE='FileSystemCache'
+CACHE_DIR='/dev/shm/pyfedi'

migrations/versions/6a9bec0c492e_allowed_instances.py

@@ -0,0 +1,39 @@
+"""allowed instances
+
+Revision ID: 6a9bec0c492e
+Revises: c88bbba381b5
+Create Date: 2023-11-03 20:23:43.536572
+
+"""
+from alembic import op
+import sqlalchemy as sa
+
+
+# revision identifiers, used by Alembic.
+revision = '6a9bec0c492e'
+down_revision = 'c88bbba381b5'
+branch_labels = None
+depends_on = None
+
+
+def upgrade():
+    # ### commands auto generated by Alembic - please adjust! ###
+    op.create_table('allowed_instances',
+    sa.Column('id', sa.Integer(), nullable=False),
+    sa.Column('domain', sa.String(length=256), nullable=True),
+    sa.Column('created_at', sa.DateTime(), nullable=True),
+    sa.PrimaryKeyConstraint('id')
+    )
+    with op.batch_alter_table('allowed_instances', schema=None) as batch_op:
+        batch_op.create_index(batch_op.f('ix_allowed_instances_domain'), ['domain'], unique=False)
+
+    # ### end Alembic commands ###
+
+
+def downgrade():
+    # ### commands auto generated by Alembic - please adjust! ###
+    with op.batch_alter_table('allowed_instances', schema=None) as batch_op:
+        batch_op.drop_index(batch_op.f('ix_allowed_instances_domain'))
+
+    op.drop_table('allowed_instances')
+    # ### end Alembic commands ###