setting up ORM

pipelines/orm/data_science_dev/posts/tables.py

@@ -1,13 +1,36 @@
-"""Posts parent table with PostgreSQL weekly range partitioning on date column."""
+"""Posts parent table and PostTopic table for the data_science_dev database."""
 
 from __future__ import annotations
 
-from pipelines.orm.data_science_dev.base import DataScienceDevBase
+from pipelines.orm.data_science_dev.base import (
+    DataScienceDevBase,
+    DataScienceDevTableBase,
+)
 from pipelines.orm.data_science_dev.posts.columns import PostsColumns
 
 
+from sqlalchemy import BigInteger, Index, SmallInteger
+from sqlalchemy.orm import Mapped, mapped_column
+
+
 class Posts(PostsColumns, DataScienceDevBase):
     """Parent partitioned table for posts, partitioned by week on `date`."""
 
     __tablename__ = "posts"
     __table_args__ = ({"postgresql_partition_by": "RANGE (date)"},)
+
+
+class PostTopic(DataScienceDevTableBase):
+    """Stores BERTopic topic assignments for posts.
+
+    post_id references main.posts but without a FK constraint
+    since posts is a partitioned table.
+    """
+
+    __tablename__ = "post_topic"
+    __table_args__ = (Index("ix_post_topic_post_id", "post_id"),)
+
+    post_id: Mapped[int] = mapped_column(BigInteger)
+    topic_id: Mapped[int] = mapped_column(SmallInteger)
+    topic_label: Mapped[str | None]
+    model_version: Mapped[str | None]