Use full-document Q&A and refine ask row

Co-authored-by: Codex <codex@openai.com>

app.py

@@ -16,7 +16,6 @@ from services import ingest
 from services.providers import get_default_api_key, instantiate_client, list_providers, validate_api_key
 from services.rag_pipeline import (
     AnalysisResult,
-    answer_query,
     answer_query_from_full_document,
     build_cached_document_artifacts,
     generate_analysis_progress,
@@ -133,6 +132,12 @@ APP_CSS = """
 #analysis-output {
     min-height: 100px;
 }
+#chat-question-row {
+    align-items: flex-start;
+}
+#ask-button button {
+    border-radius: 0.6rem !important;
+}
 #status-output,
 #chat-status {
     min-height: 1.5rem;
@@ -407,7 +412,7 @@ def _format_chat_entry(answer_text: str, citations: list[dict[str, Any]], *, pro
     if provenance == "analysis_based":
         prefix = "_Based on the summary and analysis._\n\n"
     elif provenance == "full_document":
-        prefix = "_Deeper full-document answer._\n\n"
+        prefix = "_Full-document answer._\n\n"
 
     if not citations:
         return prefix + answer_text
@@ -701,17 +706,16 @@ def ask_question(
         return
 
     question_text = question.strip()
-    yield chat_history, session_state, "Searching analysis for an answer...", _deeper_answer_updates(False), ""
+    yield chat_history, session_state, "Reading the full document for an answer...", _deeper_answer_updates(False), ""
 
     try:
         provider_client = instantiate_client(
             record["api_config"]["provider"],
             record["api_config"]["api_key"],
         )
-        yield chat_history, session_state, "Searching analysis for an answer...", _deeper_answer_updates(False), ""
-        answer = answer_query(
+        yield chat_history, session_state, "Reading the full document for an answer...", _deeper_answer_updates(False), ""
+        answer = answer_query_from_full_document(
             provider_client,
-            AnalysisResult.model_validate(record["analysis"]) if record.get("analysis") else None,
             record.get("vector_store"),
             question_text,
             doc_text=record.get("doc_text"),
@@ -731,9 +735,8 @@ def ask_question(
         {"role": "assistant", "content": _format_chat_entry(answer.answer, citations, provenance=answer.provenance)}
     ]
     record["chat_history"] = chat_history
-    record["pending_deeper_question"] = question_text if answer.deeper_answer_available else None
-    deeper_hint = answer.consent_prompt if answer.deeper_answer_available else ""
-    yield chat_history, session_state, "", _deeper_answer_updates(answer.deeper_answer_available), deeper_hint
+    record["pending_deeper_question"] = None
+    yield chat_history, session_state, "", _deeper_answer_updates(False), ""
 
 
 def run_deeper_answer(
@@ -891,15 +894,15 @@ def build_app() -> gr.Blocks:
                 with gr.Group(elem_id="chat-panel"):
                     gr.Markdown("## Ask Further Questions")
                     chatbot = gr.Chatbot(show_label=False)
-                    with gr.Row():
+                    gr.Markdown("Question")
+                    with gr.Row(elem_id="chat-question-row"):
                         with gr.Column(scale=6, min_width=0):
-                            gr.Markdown("Question")
                             question_input = gr.Textbox(
                                 label="Question",
                                 placeholder="What would you like to know?",
                                 show_label=False,
                             )
-                        ask_button = gr.Button("Ask", scale=1, variant="primary")
+                        ask_button = gr.Button("Ask", scale=1, variant="primary", elem_id="ask-button")
                     chat_status = gr.Markdown(elem_id="chat-status")
                     deeper_answer_button = gr.Button("Run deeper full-document answer", visible=False)
                     deeper_answer_hint = gr.Markdown("")

tests/test_app.py

@@ -1,4 +1,5 @@
-from app import _format_chat_entry, _stream_chat_entry
+from app import _empty_session, _format_chat_entry, _session_record, ask_question, _stream_chat_entry
+from services.rag_pipeline import AnswerResult, Citation
 
 
 def test_format_chat_entry_wraps_supporting_snippets_in_details() -> None:
@@ -24,3 +25,41 @@ def test_stream_chat_entry_appends_collapsible_snippets_after_text_stream() -> N
 
     assert frames[-2][-1]["content"] == "_Based on the summary and analysis._\n\nAnswer text"
     assert "<details><summary>Supporting snippet (1)</summary>" in frames[-1][-1]["content"]
+
+
+def test_ask_question_uses_full_document_answers(monkeypatch) -> None:
+    session_state = _empty_session()
+    record = _session_record(session_state)
+    record.update(
+        {
+            "api_config": {"provider": "qwen", "api_key": "hf_test_token"},
+            "analysis": {},
+            "vector_store": object(),
+            "doc_text": "Full bill text",
+            "chat_history": [],
+            "pending_deeper_question": "old question",
+        }
+    )
+
+    monkeypatch.setattr("app.instantiate_client", lambda provider, api_key: object())
+
+    calls: list[tuple[object | None, str | None]] = []
+
+    def fake_answer_query_from_full_document(provider_client, vector_store, question, *, doc_text=None):
+        calls.append((vector_store, doc_text))
+        return AnswerResult(
+            answer="Full-document answer",
+            citations=[Citation(ref_id=1, snippet="Clause text")],
+            provenance="full_document",
+        )
+
+    monkeypatch.setattr("app.answer_query_from_full_document", fake_answer_query_from_full_document)
+
+    frames = list(ask_question("What does the bill require?", session_state, []))
+
+    assert calls == [(record["vector_store"], "Full bill text")]
+    assert "Reading the full document for an answer..." == frames[0][2]
+    assert "<details><summary>Supporting snippet (1)</summary>" in frames[-1][0][-1]["content"]
+    assert frames[-1][3]["visible"] is False
+    assert frames[-1][4] == ""
+    assert record["pending_deeper_question"] is None