fix(api): make alembic migrations idempotent for fresh databases (#129)

fix(api): make alembic migrations idempotent for fresh databases

api/Dockerfile

@@ -27,4 +27,4 @@ EXPOSE 8000
 HEALTHCHECK --interval=30s --timeout=3s \
     CMD python -c "import urllib.request; urllib.request.urlopen('http://localhost:8000/health')"
 
-CMD ["uvicorn", "cartsnitch_api.main:app", "--host", "0.0.0.0", "--port", "8000"]
+CMD ["sh", "-c", "python -m alembic upgrade head && uvicorn cartsnitch_api.main:app --host 0.0.0.0 --port 8000"]

api/alembic/env.py

@@ -6,7 +6,7 @@ from logging.config import fileConfig
 from sqlalchemy import engine_from_config, pool
 
 from alembic import context
-from cartsnitch_api.models import Base  # noqa: F401 — imports all models for autogenerate
+from cartsnitch_api.models.base import Base  # noqa: F401 — imports all models for autogenerate
 
 config = context.config
 if config.config_file_name is not None:
@@ -47,6 +47,10 @@ def run_migrations_online() -> None:
         context.configure(connection=connection, target_metadata=target_metadata)
         with context.begin_transaction():
             context.run_migrations()
+        # Create any tables defined in models but not yet created by migrations.
+        # This bootstraps fresh databases that have no legacy schema.
+        # checkfirst=True ensures this is a no-op on existing databases.
+        Base.metadata.create_all(bind=connection, checkfirst=True)
 
 
 if context.is_offline_mode():

api/alembic/versions/001_encrypt_session_data.py

@@ -33,6 +33,21 @@ def _is_fernet_token(value: str) -> bool:
 
 
 def upgrade() -> None:
+    conn = op.get_bind()
+    inspector = sa.inspect(conn)
+
+    # Fresh DB — table created by Base.metadata.create_all with correct TEXT type
+    if not inspector.has_table("user_store_accounts"):
+        return
+
+    # Already migrated? Skip if session_data is already TEXT (not JSON)
+    cols = {c["name"]: c for c in inspector.get_columns("user_store_accounts")}
+    if "session_data" not in cols:
+        return
+    col_type = str(cols["session_data"]["type"]).lower()
+    if "text" in col_type and "json" not in col_type:
+        return  # already TEXT — nothing to do
+
     # Change column type from JSON to TEXT to hold Fernet ciphertext
     op.alter_column(
         "user_store_accounts",
@@ -43,7 +58,6 @@ def upgrade() -> None:
         postgresql_using="session_data::text",
     )
 
-    conn = op.get_bind()
     rows = conn.execute(
         text("SELECT id, session_data FROM user_store_accounts WHERE session_data IS NOT NULL")
     ).fetchall()

api/alembic/versions/002_better_auth_tables.py

@@ -21,81 +21,94 @@ depends_on = None
 
 
 def upgrade() -> None:
+    conn = op.get_bind()
+    inspector = sa.inspect(conn)
+
     # --- Extend users table for Better-Auth compatibility ---
-    op.add_column("users", sa.Column("email_verified", sa.Boolean(), nullable=False, server_default="false"))
-    op.add_column("users", sa.Column("image", sa.Text(), nullable=True))
+    # Guard: on a fresh DB Base.metadata.create_all (called in env.py after migrations)
+    # creates the users table with all columns, so migration 002 must not re-run add_column.
+    if inspector.has_table("users"):
+        existing_user_cols = [c["name"] for c in inspector.get_columns("users")]
+        if "email_verified" not in existing_user_cols:
+            op.add_column("users", sa.Column("email_verified", sa.Boolean(), nullable=False, server_default="false"))
+        if "image" not in existing_user_cols:
+            op.add_column("users", sa.Column("image", sa.Text(), nullable=True))
 
     # --- Create sessions table ---
-    op.create_table(
-        "sessions",
-        sa.Column("id", sa.Text(), nullable=False),
-        sa.Column("token", sa.Text(), nullable=False),
-        sa.Column("user_id", sa.Text(), nullable=False),
-        sa.Column("expires_at", sa.DateTime(timezone=True), nullable=False),
-        sa.Column("ip_address", sa.Text(), nullable=True),
-        sa.Column("user_agent", sa.Text(), nullable=True),
-        sa.Column("created_at", sa.DateTime(timezone=True), server_default=sa.func.now(), nullable=False),
-        sa.Column("updated_at", sa.DateTime(timezone=True), server_default=sa.func.now(), nullable=False),
-        sa.PrimaryKeyConstraint("id"),
-    )
-    op.create_index("ix_sessions_token", "sessions", ["token"], unique=True)
-    op.create_index("ix_sessions_user_id", "sessions", ["user_id"])
+    if not inspector.has_table("sessions"):
+        op.create_table(
+            "sessions",
+            sa.Column("id", sa.Text(), nullable=False),
+            sa.Column("token", sa.Text(), nullable=False),
+            sa.Column("user_id", sa.Text(), nullable=False),
+            sa.Column("expires_at", sa.DateTime(timezone=True), nullable=False),
+            sa.Column("ip_address", sa.Text(), nullable=True),
+            sa.Column("user_agent", sa.Text(), nullable=True),
+            sa.Column("created_at", sa.DateTime(timezone=True), server_default=sa.func.now(), nullable=False),
+            sa.Column("updated_at", sa.DateTime(timezone=True), server_default=sa.func.now(), nullable=False),
+            sa.PrimaryKeyConstraint("id"),
+        )
+        op.create_index("ix_sessions_token", "sessions", ["token"], unique=True)
+        op.create_index("ix_sessions_user_id", "sessions", ["user_id"])
 
     # --- Create accounts table ---
-    op.create_table(
-        "accounts",
-        sa.Column("id", sa.Text(), nullable=False),
-        sa.Column("user_id", sa.Text(), nullable=False),
-        sa.Column("account_id", sa.Text(), nullable=False),
-        sa.Column("provider_id", sa.Text(), nullable=False),
-        sa.Column("access_token", sa.Text(), nullable=True),
-        sa.Column("refresh_token", sa.Text(), nullable=True),
-        sa.Column("access_token_expires_at", sa.DateTime(timezone=True), nullable=True),
-        sa.Column("refresh_token_expires_at", sa.DateTime(timezone=True), nullable=True),
-        sa.Column("scope", sa.Text(), nullable=True),
-        sa.Column("id_token", sa.Text(), nullable=True),
-        sa.Column("password", sa.Text(), nullable=True),
-        sa.Column("created_at", sa.DateTime(timezone=True), server_default=sa.func.now(), nullable=False),
-        sa.Column("updated_at", sa.DateTime(timezone=True), server_default=sa.func.now(), nullable=False),
-        sa.PrimaryKeyConstraint("id"),
-    )
-    op.create_index("ix_accounts_user_id", "accounts", ["user_id"])
+    if not inspector.has_table("accounts"):
+        op.create_table(
+            "accounts",
+            sa.Column("id", sa.Text(), nullable=False),
+            sa.Column("user_id", sa.Text(), nullable=False),
+            sa.Column("account_id", sa.Text(), nullable=False),
+            sa.Column("provider_id", sa.Text(), nullable=False),
+            sa.Column("access_token", sa.Text(), nullable=True),
+            sa.Column("refresh_token", sa.Text(), nullable=True),
+            sa.Column("access_token_expires_at", sa.DateTime(timezone=True), nullable=True),
+            sa.Column("refresh_token_expires_at", sa.DateTime(timezone=True), nullable=True),
+            sa.Column("scope", sa.Text(), nullable=True),
+            sa.Column("id_token", sa.Text(), nullable=True),
+            sa.Column("password", sa.Text(), nullable=True),
+            sa.Column("created_at", sa.DateTime(timezone=True), server_default=sa.func.now(), nullable=False),
+            sa.Column("updated_at", sa.DateTime(timezone=True), server_default=sa.func.now(), nullable=False),
+            sa.PrimaryKeyConstraint("id"),
+        )
+        op.create_index("ix_accounts_user_id", "accounts", ["user_id"])
 
     # --- Create verifications table ---
-    op.create_table(
-        "verifications",
-        sa.Column("id", sa.Text(), nullable=False),
-        sa.Column("identifier", sa.Text(), nullable=False),
-        sa.Column("value", sa.Text(), nullable=False),
-        sa.Column("expires_at", sa.DateTime(timezone=True), nullable=False),
-        sa.Column("created_at", sa.DateTime(timezone=True), server_default=sa.func.now(), nullable=False),
-        sa.Column("updated_at", sa.DateTime(timezone=True), server_default=sa.func.now(), nullable=False),
-        sa.PrimaryKeyConstraint("id"),
-    )
+    if not inspector.has_table("verifications"):
+        op.create_table(
+            "verifications",
+            sa.Column("id", sa.Text(), nullable=False),
+            sa.Column("identifier", sa.Text(), nullable=False),
+            sa.Column("value", sa.Text(), nullable=False),
+            sa.Column("expires_at", sa.DateTime(timezone=True), nullable=False),
+            sa.Column("created_at", sa.DateTime(timezone=True), server_default=sa.func.now(), nullable=False),
+            sa.Column("updated_at", sa.DateTime(timezone=True), server_default=sa.func.now(), nullable=False),
+            sa.PrimaryKeyConstraint("id"),
+        )
 
     # --- Migrate existing password hashes to accounts table ---
-    # For each user with a hashed_password, create a 'credential' account row
-    conn = op.get_bind()
-    users = conn.execute(
-        text("SELECT id, hashed_password FROM users WHERE hashed_password IS NOT NULL")
-    ).fetchall()
+    # Only run on existing (non-fresh) DBs that already have users table with data
+    if inspector.has_table("users"):
+        users = conn.execute(
+            text("SELECT id, hashed_password FROM users WHERE hashed_password IS NOT NULL")
+        ).fetchall()
 
-    for user_id, hashed_password in users:
-        user_id_str = str(user_id)
-        conn.execute(
-            text(
-                "INSERT INTO accounts (id, user_id, account_id, provider_id, password, created_at, updated_at) "
-                "VALUES (gen_random_uuid()::text, :user_id, :account_id, 'credential', :password, now(), now())"
-            ),
-            {"user_id": user_id_str, "account_id": user_id_str, "password": hashed_password},
-        )
+        for user_id, hashed_password in users:
+            user_id_str = str(user_id)
+            conn.execute(
+                text(
+                    "INSERT INTO accounts (id, user_id, account_id, provider_id, password, created_at, updated_at) "
+                    "VALUES (gen_random_uuid()::text, :user_id, :account_id, 'credential', :password, now(), now())"
+                ),
+                {"user_id": user_id_str, "account_id": user_id_str, "password": hashed_password},
+            )
 
 
 def downgrade() -> None:
-    op.drop_table("verifications")
-    op.drop_table("accounts")
-    op.drop_index("ix_sessions_user_id", table_name="sessions")
-    op.drop_index("ix_sessions_token", table_name="sessions")
-    op.drop_table("sessions")
-    op.drop_column("users", "image")
-    op.drop_column("users", "email_verified")
+    op.execute(text("DROP INDEX IF EXISTS ix_accounts_user_id"))
+    op.execute(text("DROP TABLE IF EXISTS verifications"))
+    op.execute(text("DROP TABLE IF EXISTS accounts"))
+    op.execute(text("DROP INDEX IF EXISTS ix_sessions_user_id"))
+    op.execute(text("DROP INDEX IF EXISTS ix_sessions_token"))
+    op.execute(text("DROP TABLE IF EXISTS sessions"))
+    op.execute(text("ALTER TABLE users DROP COLUMN IF EXISTS image"))
+    op.execute(text("ALTER TABLE users DROP COLUMN IF EXISTS email_verified"))

api/alembic/versions/003_make_users_hashed_password_nullable.py

@@ -19,8 +19,25 @@ depends_on = None
 
 
 def upgrade() -> None:
-    op.alter_column("users", "hashed_password", existing_type=sa.String(255), nullable=True)
+    conn = op.get_bind()
+    inspector = sa.inspect(conn)
+
+    # Fresh DB — nothing to alter
+    if not inspector.has_table("users"):
+        return
+
+    cols = {c["name"]: c for c in inspector.get_columns("users")}
+    if "hashed_password" in cols and not cols["hashed_password"]["nullable"]:
+        op.alter_column("users", "hashed_password", existing_type=sa.String(255), nullable=True)
 
 
 def downgrade() -> None:
-    op.alter_column("users", "hashed_password", existing_type=sa.String(255), nullable=False)
+    conn = op.get_bind()
+    inspector = sa.inspect(conn)
+
+    if not inspector.has_table("users"):
+        return
+
+    cols = {c["name"]: c for c in inspector.get_columns("users")}
+    if "hashed_password" in cols and cols["hashed_password"]["nullable"]:
+        op.alter_column("users", "hashed_password", existing_type=sa.String(255), nullable=False)

api/alembic/versions/004_fix_user_id_text.py

@@ -25,7 +25,21 @@ depends_on = None
 
 
 def upgrade() -> None:
-    # Step 1: Drop existing FK constraints
+    conn = op.get_bind()
+    inspector = sa.inspect(conn)
+
+    # Fresh DB — no tables yet, nothing to convert
+    if not inspector.has_table("users"):
+        return
+
+    # Check if already TEXT (Base.metadata.create_all uses TEXT for fresh DB)
+    users_cols = {c["name"]: c for c in inspector.get_columns("users")}
+    if "id" in users_cols:
+        id_type = str(users_cols["id"]["type"]).lower()
+        if "text" in id_type and "uuid" not in id_type:
+            return  # already TEXT — nothing to do
+
+    # Step 1: Drop existing FK constraints (ignore if they don't exist)
     op.execute(text("ALTER TABLE user_store_accounts DROP CONSTRAINT IF EXISTS user_store_accounts_user_id_fkey"))
     op.execute(text("ALTER TABLE purchases DROP CONSTRAINT IF EXISTS purchases_user_id_fkey"))
 

api/alembic/versions/005_add_email_inbound_token.py

@@ -18,6 +18,15 @@ depends_on = None
 
 
 def upgrade() -> None:
+    conn = op.get_bind()
+    inspector = sa.inspect(conn)
+    # Guard: on a fresh DB Base.metadata.create_all creates users table with the column already present
+    if not inspector.has_table("users"):
+        return
+    existing_cols = [c["name"] for c in inspector.get_columns("users")]
+    if "email_inbound_token" in existing_cols:
+        return
+
     # Add column nullable first so existing rows can be backfilled
     op.add_column(
         "users",
@@ -25,11 +34,10 @@ def upgrade() -> None:
     )
 
     # Backfill existing users with unique tokens
-    connection = op.get_bind()
-    result = connection.execute(sa.text("SELECT id FROM users WHERE email_inbound_token IS NULL"))
+    result = conn.execute(sa.text("SELECT id FROM users WHERE email_inbound_token IS NULL"))
     for (user_id,) in result:
         token = secrets.token_urlsafe(16)
-        connection.execute(
+        conn.execute(
             sa.text("UPDATE users SET email_inbound_token = :token WHERE id = :id"),
             {"token": token, "id": user_id},
         )

api/alembic/versions/006_email_inbound_token_server_default.py

@@ -15,6 +15,16 @@ depends_on = None
 
 
 def upgrade() -> None:
+    conn = op.get_bind()
+    inspector = sa.inspect(conn)
+    # Guard: on a fresh DB Base.metadata.create_all already sets the server_default
+    if not inspector.has_table("users"):
+        return
+    cols = {c["name"]: c for c in inspector.get_columns("users")}
+    if "email_inbound_token" not in cols:
+        return
+    if cols["email_inbound_token"].get("default") is not None:
+        return
     op.alter_column(
         "users",
         "email_inbound_token",

api/src/cartsnitch_api/models/user.py

@@ -4,7 +4,8 @@ import secrets
 from datetime import datetime
 from typing import TYPE_CHECKING
 
-from sqlalchemy import DateTime, ForeignKey, String, Text, UniqueConstraint
+import sqlalchemy as sa
+from sqlalchemy import Boolean, DateTime, ForeignKey, String, Text, UniqueConstraint
 from sqlalchemy.orm import Mapped, mapped_column, relationship
 
 from cartsnitch_api.constants import AccountStatus
@@ -23,13 +24,20 @@ class User(TimestampMixin, Base):
 
     id: Mapped[str] = mapped_column(Text, primary_key=True)
     email: Mapped[str] = mapped_column(String(255), nullable=False, unique=True)
-    hashed_password: Mapped[str] = mapped_column(String(255), nullable=False)
+    hashed_password: Mapped[str | None] = mapped_column(String(255), nullable=True)
     display_name: Mapped[str | None] = mapped_column(String(100))
+    email_verified: Mapped[bool] = mapped_column(
+        Boolean, nullable=False, server_default="false"
+    )
+    image: Mapped[str | None] = mapped_column(Text, nullable=True)
     email_inbound_token: Mapped[str] = mapped_column(
         String(22),
         nullable=False,
         unique=True,
         default=lambda: secrets.token_urlsafe(16),
+        server_default=sa.text(
+            "replace(replace(trim(trailing '=' from encode(gen_random_bytes(16), 'base64')), '+', '-'), '/', '_')"
+        ),
     )
 
     # Relationships