allowing multiple summaries per bill text

tests/test_bill_text_summary_model.py

@@ -0,0 +1,36 @@
+from pipelines.orm.data_science_dev.congress import BillText, BillTextSummary
+
+
+def test_default_summary_prefers_primary_summary() -> None:
+    primary_summary = BillTextSummary(id=1, bill_text_id=10, summary="primary")
+    latest_summary = BillTextSummary(id=2, bill_text_id=10, summary="latest")
+    bill_text = BillText(
+        id=10,
+        bill_id=5,
+        version_code="ih",
+        summaries=[latest_summary],
+        primary_summary=primary_summary,
+    )
+
+    assert bill_text.default_summary() is primary_summary
+
+
+def test_default_summary_falls_back_to_latest_summary() -> None:
+    latest_summary = BillTextSummary(id=2, bill_text_id=10, summary="latest")
+    older_summary = BillTextSummary(id=1, bill_text_id=10, summary="older")
+    bill_text = BillText(
+        id=10,
+        bill_id=5,
+        version_code="ih",
+        summaries=[latest_summary, older_summary],
+    )
+
+    assert bill_text.latest_summary() is latest_summary
+    assert bill_text.default_summary() is latest_summary
+
+
+def test_default_summary_is_none_without_summaries() -> None:
+    bill_text = BillText(id=10, bill_id=5, version_code="ih")
+
+    assert bill_text.latest_summary() is None
+    assert bill_text.default_summary() is None

tests/test_extract_bill_topics.py

@@ -0,0 +1,71 @@
+from sqlalchemy.dialects import postgresql
+
+from pipelines.jobs.extract_bill_topics import (
+    _select_bill_text_for_topic_extraction,
+    create_select_bills_for_topic_extraction,
+)
+from pipelines.orm.data_science_dev.congress import Bill, BillText, BillTextSummary
+
+
+def _compile_sql(statement: object) -> str:
+    return str(
+        statement.compile(
+            dialect=postgresql.dialect(),
+            compile_kwargs={"literal_binds": True},
+        )
+    )
+
+
+def test_select_bill_text_for_topic_extraction_uses_primary_summary() -> None:
+    primary_summary = BillTextSummary(id=1, bill_text_id=10, summary="primary")
+    newest_summary = BillTextSummary(id=2, bill_text_id=10, summary="newest")
+    bill_text = BillText(
+        id=10,
+        bill_id=5,
+        version_code="ih",
+        summaries=[newest_summary],
+        primary_summary=primary_summary,
+    )
+    bill = Bill(
+        id=5,
+        congress=119,
+        bill_type="hr",
+        number=1,
+        bill_texts=[bill_text],
+    )
+
+    selected = _select_bill_text_for_topic_extraction(bill)
+
+    assert selected is bill_text
+    assert selected.default_summary() is primary_summary
+
+
+def test_select_bill_text_for_topic_extraction_uses_latest_summary_without_primary() -> None:
+    newest_summary = BillTextSummary(id=2, bill_text_id=10, summary="newest")
+    older_summary = BillTextSummary(id=1, bill_text_id=10, summary="older")
+    bill_text = BillText(
+        id=10,
+        bill_id=5,
+        version_code="ih",
+        summaries=[newest_summary, older_summary],
+    )
+    bill = Bill(
+        id=5,
+        congress=119,
+        bill_type="hr",
+        number=1,
+        bill_texts=[bill_text],
+    )
+
+    selected = _select_bill_text_for_topic_extraction(bill)
+
+    assert selected is bill_text
+    assert selected.default_summary() is newest_summary
+
+
+def test_create_select_bills_for_topic_extraction_uses_summary_exists_subquery() -> None:
+    sql = _compile_sql(create_select_bills_for_topic_extraction())
+
+    assert "bill_text_summary" in sql
+    assert "EXISTS" in sql
+    assert "bill_text.summary" not in sql

tests/test_summarize_bills.py

@@ -0,0 +1,58 @@
+from sqlalchemy.dialects import postgresql
+
+from pipelines.jobs.summarize_bills import (
+    create_select_bill_texts_for_summarization,
+    store_bill_summary_result,
+)
+from pipelines.orm.data_science_dev.congress import BillText, BillTextSummary
+
+
+class FakeSession:
+    def __init__(self) -> None:
+        self.added: list[object] = []
+
+    def add(self, value: object) -> None:
+        self.added.append(value)
+
+
+def _compile_sql(statement: object) -> str:
+    return str(
+        statement.compile(
+            dialect=postgresql.dialect(),
+            compile_kwargs={"literal_binds": True},
+        )
+    )
+
+
+def test_store_bill_summary_result_creates_summary_row() -> None:
+    session = FakeSession()
+    bill_text = BillText(id=10, bill_id=5, version_code="ih")
+
+    summary_row = store_bill_summary_result(
+        session=session,
+        bill_text=bill_text,
+        summary="A summary",
+        model="gpt-5.4-mini",
+    )
+
+    assert session.added == [summary_row]
+    assert isinstance(summary_row, BillTextSummary)
+    assert summary_row.bill_text is bill_text
+    assert summary_row.summary == "A summary"
+    assert summary_row.summarization_model == "gpt-5.4-mini"
+    assert summary_row.summarization_system_prompt_version == "v1.2"
+    assert summary_row.summarization_user_prompt_version == "v1"
+
+
+def test_create_select_bill_texts_for_summarization_excludes_existing_summaries() -> None:
+    sql = _compile_sql(create_select_bill_texts_for_summarization(force=False))
+
+    assert "bill_text_summary" in sql
+    assert "NOT (EXISTS" in sql or "NOT EXISTS" in sql
+    assert "bill_text.summary" not in sql
+
+
+def test_create_select_bill_texts_for_summarization_force_skips_summary_filter() -> None:
+    sql = _compile_sql(create_select_bill_texts_for_summarization(force=True))
+
+    assert "bill_text_summary" not in sql