basic-memory 0.16.1__py3-none-any.whl → 0.17.4__py3-none-any.whl

basic_memory/__init__.py

@@ -1,7 +1,7 @@
 """basic-memory - Local-first knowledge management combining Zettelkasten with knowledge graphs"""
 
 # Package version - updated by release automation
-__version__ = "0.16.1"
+__version__ = "0.17.4"
 
 # API version for FastAPI - independent of package version
 __api_version__ = "v0"

basic_memory/alembic/env.py

@@ -1,17 +1,32 @@
 """Alembic environment configuration."""
 
+import asyncio
 import os
 from logging.config import fileConfig
 
-from sqlalchemy import engine_from_config
-from sqlalchemy import pool
+# Allow nested event loops (needed for pytest-asyncio and other async contexts)
+# Note: nest_asyncio doesn't work with uvloop, so we handle that case separately
+try:
+    import nest_asyncio
+
+    nest_asyncio.apply()
+except (ImportError, ValueError):
+    # nest_asyncio not available or can't patch this loop type (e.g., uvloop)
+    pass
+
+from sqlalchemy import engine_from_config, pool
+from sqlalchemy.ext.asyncio import AsyncEngine, create_async_engine
 
 from alembic import context
 
 from basic_memory.config import ConfigManager
 
-# set config.env to "test" for pytest to prevent logging to file in utils.setup_logging()
-os.environ["BASIC_MEMORY_ENV"] = "test"
+# Trigger: only set test env when actually running under pytest
+# Why: alembic/env.py is imported during normal operations (MCP server startup, migrations)
+#      but we only want test behavior during actual test runs
+# Outcome: prevents is_test_env from returning True in production, enabling watch service
+if os.getenv("PYTEST_CURRENT_TEST") is not None:
+    os.environ["BASIC_MEMORY_ENV"] = "test"
 
 # Import after setting environment variable  # noqa: E402
 from basic_memory.models import Base  # noqa: E402
@@ -20,12 +35,22 @@ from basic_memory.models import Base  # noqa: E402
 # access to the values within the .ini file in use.
 config = context.config
 
+# Load app config - this will read environment variables (BASIC_MEMORY_DATABASE_BACKEND, etc.)
+# due to Pydantic's env_prefix="BASIC_MEMORY_" setting
 app_config = ConfigManager().config
-# Set the SQLAlchemy URL from our app config
-sqlalchemy_url = f"sqlite:///{app_config.database_path}"
-config.set_main_option("sqlalchemy.url", sqlalchemy_url)
 
-# print(f"Using SQLAlchemy URL: {sqlalchemy_url}")
+# Set the SQLAlchemy URL based on database backend configuration
+# If the URL is already set in config (e.g., from run_migrations), use that
+# Otherwise, get it from app config
+# Note: alembic.ini has a placeholder URL "driver://user:pass@localhost/dbname" that we need to override
+current_url = config.get_main_option("sqlalchemy.url")
+if not current_url or current_url == "driver://user:pass@localhost/dbname":
+    from basic_memory.db import DatabaseType
+
+    sqlalchemy_url = DatabaseType.get_db_url(
+        app_config.database_path, DatabaseType.FILESYSTEM, app_config
+    )
+    config.set_main_option("sqlalchemy.url", sqlalchemy_url)
 
 # Interpret the config file for Python logging.
 if config.config_file_name is not None:
@@ -69,28 +94,89 @@ def run_migrations_offline() -> None:
         context.run_migrations()
 
 
+def do_run_migrations(connection):
+    """Execute migrations with the given connection."""
+    context.configure(
+        connection=connection,
+        target_metadata=target_metadata,
+        include_object=include_object,
+        render_as_batch=True,
+        compare_type=True,
+    )
+    with context.begin_transaction():
+        context.run_migrations()
+
+
+async def run_async_migrations(connectable):
+    """Run migrations asynchronously with AsyncEngine."""
+    async with connectable.connect() as connection:
+        await connection.run_sync(do_run_migrations)
+    await connectable.dispose()
+
+
 def run_migrations_online() -> None:
     """Run migrations in 'online' mode.
 
-    In this scenario we need to create an Engine
-    and associate a connection with the context.
+    Supports both sync engines (SQLite) and async engines (PostgreSQL with asyncpg).
     """
-    connectable = engine_from_config(
-        config.get_section(config.config_ini_section, {}),
-        prefix="sqlalchemy.",
-        poolclass=pool.NullPool,
-    )
-
-    with connectable.connect() as connection:
-        context.configure(
-            connection=connection,
-            target_metadata=target_metadata,
-            include_object=include_object,
-            render_as_batch=True,
-        )
-
-        with context.begin_transaction():
-            context.run_migrations()
+    # Check if a connection/engine was provided (e.g., from run_migrations)
+    connectable = context.config.attributes.get("connection", None)
+
+    if connectable is None:
+        # No connection provided, create engine from config
+        url = context.config.get_main_option("sqlalchemy.url")
+
+        # Check if it's an async URL (sqlite+aiosqlite or postgresql+asyncpg)
+        if url and ("+asyncpg" in url or "+aiosqlite" in url):
+            # Create async engine for asyncpg or aiosqlite
+            connectable = create_async_engine(
+                url,
+                poolclass=pool.NullPool,
+                future=True,
+            )
+        else:
+            # Create sync engine for regular sqlite or postgresql
+            connectable = engine_from_config(
+                context.config.get_section(context.config.config_ini_section, {}),
+                prefix="sqlalchemy.",
+                poolclass=pool.NullPool,
+            )
+
+    # Handle async engines (PostgreSQL with asyncpg)
+    if isinstance(connectable, AsyncEngine):
+        # Try to run async migrations
+        # nest_asyncio allows asyncio.run() from within event loops, but doesn't work with uvloop
+        try:
+            asyncio.run(run_async_migrations(connectable))
+        except RuntimeError as e:
+            if "cannot be called from a running event loop" in str(e):
+                # We're in a running event loop (likely uvloop) - need to use a different approach
+                # Create a new thread to run the async migrations
+                import concurrent.futures
+
+                def run_in_thread():
+                    """Run async migrations in a new event loop in a separate thread."""
+                    new_loop = asyncio.new_event_loop()
+                    asyncio.set_event_loop(new_loop)
+                    try:
+                        new_loop.run_until_complete(run_async_migrations(connectable))
+                    finally:
+                        new_loop.close()
+
+                with concurrent.futures.ThreadPoolExecutor() as executor:
+                    future = executor.submit(run_in_thread)
+                    future.result()  # Wait for completion and re-raise any exceptions
+            else:
+                raise
+    else:
+        # Handle sync engines (SQLite) or sync connections
+        if hasattr(connectable, "connect"):
+            # It's an engine, get a connection
+            with connectable.connect() as connection:
+                do_run_migrations(connection)
+        else:
+            # It's already a connection
+            do_run_migrations(connectable)
 
 
 if context.is_offline_mode():

basic_memory/alembic/versions/314f1ea54dc4_add_postgres_full_text_search_support_.py

@@ -0,0 +1,131 @@
+"""Add Postgres full-text search support with tsvector and GIN indexes
+
+Revision ID: 314f1ea54dc4
+Revises: e7e1f4367280
+Create Date: 2025-11-15 18:05:01.025405
+
+"""
+
+from typing import Sequence, Union
+
+from alembic import op
+import sqlalchemy as sa
+
+
+# revision identifiers, used by Alembic.
+revision: str = "314f1ea54dc4"
+down_revision: Union[str, None] = "e7e1f4367280"
+branch_labels: Union[str, Sequence[str], None] = None
+depends_on: Union[str, Sequence[str], None] = None
+
+
+def upgrade() -> None:
+    """Add PostgreSQL full-text search support.
+
+    This migration:
+    1. Creates search_index table for Postgres (SQLite uses FTS5 virtual table)
+    2. Adds generated tsvector column for full-text search
+    3. Creates GIN index on the tsvector column for fast text queries
+    4. Creates GIN index on metadata JSONB column for fast containment queries
+
+    Note: These changes only apply to Postgres. SQLite continues to use FTS5 virtual tables.
+    """
+    # Check if we're using Postgres
+    connection = op.get_bind()
+    if connection.dialect.name == "postgresql":
+        # Create search_index table for Postgres
+        # For SQLite, this is a FTS5 virtual table created elsewhere
+        from sqlalchemy.dialects.postgresql import JSONB
+
+        op.create_table(
+            "search_index",
+            sa.Column("id", sa.Integer(), nullable=False),  # Entity IDs are integers
+            sa.Column("project_id", sa.Integer(), nullable=False),  # Multi-tenant isolation
+            sa.Column("title", sa.Text(), nullable=True),
+            sa.Column("content_stems", sa.Text(), nullable=True),
+            sa.Column("content_snippet", sa.Text(), nullable=True),
+            sa.Column("permalink", sa.String(), nullable=True),  # Nullable for non-markdown files
+            sa.Column("file_path", sa.String(), nullable=True),
+            sa.Column("type", sa.String(), nullable=True),
+            sa.Column("from_id", sa.Integer(), nullable=True),  # Relation IDs are integers
+            sa.Column("to_id", sa.Integer(), nullable=True),  # Relation IDs are integers
+            sa.Column("relation_type", sa.String(), nullable=True),
+            sa.Column("entity_id", sa.Integer(), nullable=True),  # Entity IDs are integers
+            sa.Column("category", sa.String(), nullable=True),
+            sa.Column("metadata", JSONB(), nullable=True),  # Use JSONB for Postgres
+            sa.Column("created_at", sa.DateTime(timezone=True), nullable=True),
+            sa.Column("updated_at", sa.DateTime(timezone=True), nullable=True),
+            sa.PrimaryKeyConstraint(
+                "id", "type", "project_id"
+            ),  # Composite key: id can repeat across types
+            sa.ForeignKeyConstraint(
+                ["project_id"],
+                ["project.id"],
+                name="fk_search_index_project_id",
+                ondelete="CASCADE",
+            ),
+            if_not_exists=True,
+        )
+
+        # Create index on project_id for efficient multi-tenant queries
+        op.create_index(
+            "ix_search_index_project_id",
+            "search_index",
+            ["project_id"],
+            unique=False,
+        )
+
+        # Create unique partial index on permalink for markdown files
+        # Non-markdown files don't have permalinks, so we use a partial index
+        op.execute("""
+            CREATE UNIQUE INDEX uix_search_index_permalink_project
+            ON search_index (permalink, project_id)
+            WHERE permalink IS NOT NULL
+        """)
+
+        # Add tsvector column as a GENERATED ALWAYS column
+        # This automatically updates when title or content_stems change
+        op.execute("""
+            ALTER TABLE search_index
+            ADD COLUMN textsearchable_index_col tsvector
+            GENERATED ALWAYS AS (
+                to_tsvector('english',
+                    coalesce(title, '') || ' ' ||
+                    coalesce(content_stems, '')
+                )
+            ) STORED
+        """)
+
+        # Create GIN index on tsvector column for fast full-text search
+        op.create_index(
+            "idx_search_index_fts",
+            "search_index",
+            ["textsearchable_index_col"],
+            unique=False,
+            postgresql_using="gin",
+        )
+
+        # Create GIN index on metadata JSONB for fast containment queries
+        # Using jsonb_path_ops for smaller index size and better performance
+        op.execute("""
+            CREATE INDEX idx_search_index_metadata_gin
+            ON search_index
+            USING GIN (metadata jsonb_path_ops)
+        """)
+
+
+def downgrade() -> None:
+    """Remove PostgreSQL full-text search support."""
+    connection = op.get_bind()
+    if connection.dialect.name == "postgresql":
+        # Drop indexes first
+        op.execute("DROP INDEX IF EXISTS idx_search_index_metadata_gin")
+        op.drop_index("idx_search_index_fts", table_name="search_index")
+        op.execute("DROP INDEX IF EXISTS uix_search_index_permalink_project")
+        op.drop_index("ix_search_index_project_id", table_name="search_index")
+
+        # Drop the generated column
+        op.execute("ALTER TABLE search_index DROP COLUMN IF EXISTS textsearchable_index_col")
+
+        # Drop the search_index table
+        op.drop_table("search_index")

basic_memory/alembic/versions/5fe1ab1ccebe_add_projects_table.py

@@ -21,6 +21,12 @@ depends_on: Union[str, Sequence[str], None] = None
 
 def upgrade() -> None:
     # ### commands auto generated by Alembic - please adjust! ###
+
+    # SQLite FTS5 virtual table handling is SQLite-specific
+    # For Postgres, search_index is a regular table managed by ORM
+    connection = op.get_bind()
+    is_sqlite = connection.dialect.name == "sqlite"
+
     op.create_table(
         "project",
         sa.Column("id", sa.Integer(), nullable=False),
@@ -55,7 +61,9 @@ def upgrade() -> None:
         batch_op.add_column(sa.Column("project_id", sa.Integer(), nullable=False))
         batch_op.drop_index(
             "uix_entity_permalink",
-            sqlite_where=sa.text("content_type = 'text/markdown' AND permalink IS NOT NULL"),
+            sqlite_where=sa.text("content_type = 'text/markdown' AND permalink IS NOT NULL")
+            if is_sqlite
+            else None,
         )
         batch_op.drop_index("ix_entity_file_path")
         batch_op.create_index(batch_op.f("ix_entity_file_path"), ["file_path"], unique=False)
@@ -67,12 +75,16 @@ def upgrade() -> None:
             "uix_entity_permalink_project",
             ["permalink", "project_id"],
             unique=True,
-            sqlite_where=sa.text("content_type = 'text/markdown' AND permalink IS NOT NULL"),
+            sqlite_where=sa.text("content_type = 'text/markdown' AND permalink IS NOT NULL")
+            if is_sqlite
+            else None,
         )
         batch_op.create_foreign_key("fk_entity_project_id", "project", ["project_id"], ["id"])
 
     # drop the search index table. it will be recreated
-    op.drop_table("search_index")
+    # Only drop for SQLite - Postgres uses regular table managed by ORM
+    if is_sqlite:
+        op.drop_table("search_index")
 
     # ### end Alembic commands ###
 

basic_memory/alembic/versions/647e7a75e2cd_project_constraint_fix.py

@@ -25,43 +25,51 @@ def upgrade() -> None:
     The UNIQUE constraint prevents multiple projects from having is_default=FALSE,
     which breaks project creation when the service sets is_default=False.
 
-    Since SQLite doesn't support dropping specific constraints easily, we'll
-    recreate the table without the problematic constraint.
+    SQLite: Recreate the table without the constraint (no ALTER TABLE support)
+    Postgres: Use ALTER TABLE to drop the constraint directly
     """
-    # For SQLite, we need to recreate the table without the UNIQUE constraint
-    # Create a new table without the UNIQUE constraint on is_default
-    op.create_table(
-        "project_new",
-        sa.Column("id", sa.Integer(), nullable=False),
-        sa.Column("name", sa.String(), nullable=False),
-        sa.Column("description", sa.Text(), nullable=True),
-        sa.Column("permalink", sa.String(), nullable=False),
-        sa.Column("path", sa.String(), nullable=False),
-        sa.Column("is_active", sa.Boolean(), nullable=False),
-        sa.Column("is_default", sa.Boolean(), nullable=True),  # No UNIQUE constraint!
-        sa.Column("created_at", sa.DateTime(), nullable=False),
-        sa.Column("updated_at", sa.DateTime(), nullable=False),
-        sa.PrimaryKeyConstraint("id"),
-        sa.UniqueConstraint("name"),
-        sa.UniqueConstraint("permalink"),
-    )
-
-    # Copy data from old table to new table
-    op.execute("INSERT INTO project_new SELECT * FROM project")
-
-    # Drop the old table
-    op.drop_table("project")
-
-    # Rename the new table
-    op.rename_table("project_new", "project")
-
-    # Recreate the indexes
-    with op.batch_alter_table("project", schema=None) as batch_op:
-        batch_op.create_index("ix_project_created_at", ["created_at"], unique=False)
-        batch_op.create_index("ix_project_name", ["name"], unique=True)
-        batch_op.create_index("ix_project_path", ["path"], unique=False)
-        batch_op.create_index("ix_project_permalink", ["permalink"], unique=True)
-        batch_op.create_index("ix_project_updated_at", ["updated_at"], unique=False)
+    connection = op.get_bind()
+    is_sqlite = connection.dialect.name == "sqlite"
+
+    if is_sqlite:
+        # For SQLite, we need to recreate the table without the UNIQUE constraint
+        # Create a new table without the UNIQUE constraint on is_default
+        op.create_table(
+            "project_new",
+            sa.Column("id", sa.Integer(), nullable=False),
+            sa.Column("name", sa.String(), nullable=False),
+            sa.Column("description", sa.Text(), nullable=True),
+            sa.Column("permalink", sa.String(), nullable=False),
+            sa.Column("path", sa.String(), nullable=False),
+            sa.Column("is_active", sa.Boolean(), nullable=False),
+            sa.Column("is_default", sa.Boolean(), nullable=True),  # No UNIQUE constraint!
+            sa.Column("created_at", sa.DateTime(), nullable=False),
+            sa.Column("updated_at", sa.DateTime(), nullable=False),
+            sa.PrimaryKeyConstraint("id"),
+            sa.UniqueConstraint("name"),
+            sa.UniqueConstraint("permalink"),
+        )
+
+        # Copy data from old table to new table
+        op.execute("INSERT INTO project_new SELECT * FROM project")
+
+        # Drop the old table
+        op.drop_table("project")
+
+        # Rename the new table
+        op.rename_table("project_new", "project")
+
+        # Recreate the indexes
+        with op.batch_alter_table("project", schema=None) as batch_op:
+            batch_op.create_index("ix_project_created_at", ["created_at"], unique=False)
+            batch_op.create_index("ix_project_name", ["name"], unique=True)
+            batch_op.create_index("ix_project_path", ["path"], unique=False)
+            batch_op.create_index("ix_project_permalink", ["permalink"], unique=True)
+            batch_op.create_index("ix_project_updated_at", ["updated_at"], unique=False)
+    else:
+        # For Postgres, we can simply drop the constraint
+        with op.batch_alter_table("project", schema=None) as batch_op:
+            batch_op.drop_constraint("project_is_default_key", type_="unique")
 
 
 def downgrade() -> None:

basic_memory/alembic/versions/6830751f5fb6_merge_multiple_heads.py

@@ -0,0 +1,24 @@
+"""Merge multiple heads
+
+Revision ID: 6830751f5fb6
+Revises: a2b3c4d5e6f7, g9a0b3c4d5e6
+Create Date: 2025-12-29 12:46:46.476268
+
+"""
+from typing import Sequence, Union
+
+
+
+# revision identifiers, used by Alembic.
+revision: str = '6830751f5fb6'
+down_revision: Union[str, Sequence[str], None] = ('a2b3c4d5e6f7', 'g9a0b3c4d5e6')
+branch_labels: Union[str, Sequence[str], None] = None
+depends_on: Union[str, Sequence[str], None] = None
+
+
+def upgrade() -> None:
+    pass
+
+
+def downgrade() -> None:
+    pass

basic_memory/alembic/versions/a2b3c4d5e6f7_add_search_index_entity_cascade.py

@@ -0,0 +1,56 @@
+"""Add cascade delete FK from search_index to entity
+
+Revision ID: a2b3c4d5e6f7
+Revises: f8a9b2c3d4e5
+Create Date: 2025-12-02 07:00:00.000000
+
+"""
+
+from typing import Sequence, Union
+
+from alembic import op
+
+
+# revision identifiers, used by Alembic.
+revision: str = "a2b3c4d5e6f7"
+down_revision: Union[str, None] = "f8a9b2c3d4e5"
+branch_labels: Union[str, Sequence[str], None] = None
+depends_on: Union[str, Sequence[str], None] = None
+
+
+def upgrade() -> None:
+    """Add FK with CASCADE delete from search_index.entity_id to entity.id.
+
+    This migration is Postgres-only because:
+    - SQLite uses FTS5 virtual tables which don't support foreign keys
+    - The FK enables automatic cleanup of search_index entries when entities are deleted
+    """
+    connection = op.get_bind()
+    dialect = connection.dialect.name
+
+    if dialect == "postgresql":
+        # First, clean up any orphaned search_index entries where entity no longer exists
+        op.execute("""
+            DELETE FROM search_index
+            WHERE entity_id IS NOT NULL
+            AND entity_id NOT IN (SELECT id FROM entity)
+        """)
+
+        # Add FK with CASCADE - nullable FK allows search_index entries without entity_id
+        op.create_foreign_key(
+            "fk_search_index_entity_id",
+            "search_index",
+            "entity",
+            ["entity_id"],
+            ["id"],
+            ondelete="CASCADE",
+        )
+
+
+def downgrade() -> None:
+    """Remove the FK constraint."""
+    connection = op.get_bind()
+    dialect = connection.dialect.name
+
+    if dialect == "postgresql":
+        op.drop_constraint("fk_search_index_entity_id", "search_index", type_="foreignkey")

basic_memory/alembic/versions/cc7172b46608_update_search_index_schema.py

@@ -21,6 +21,12 @@ depends_on: Union[str, Sequence[str], None] = None
 def upgrade() -> None:
     """Upgrade database schema to use new search index with content_stems and content_snippet."""
 
+    # This migration is SQLite-specific (FTS5 virtual tables)
+    # For Postgres, the search_index table is created via ORM models
+    connection = op.get_bind()
+    if connection.dialect.name != "sqlite":
+        return
+
     # First, drop the existing search_index table
     op.execute("DROP TABLE IF EXISTS search_index")
 
@@ -59,6 +65,13 @@ def upgrade() -> None:
 
 def downgrade() -> None:
     """Downgrade database schema to use old search index."""
+
+    # This migration is SQLite-specific (FTS5 virtual tables)
+    # For Postgres, the search_index table is managed via ORM models
+    connection = op.get_bind()
+    if connection.dialect.name != "sqlite":
+        return
+
     # Drop the updated search_index table
     op.execute("DROP TABLE IF EXISTS search_index")