Add --full-citations to ask and analyze

CHANGELOG.md

@@ -2,6 +2,11 @@
 
 ## [Unreleased]
 
+### Added
+
+- `--full-citations` on `haiku-rag ask` and `haiku-rag analyze` renders citation
+  text untruncated. `format_citations_rich` takes a `full` argument.
+
 ### Fixed
 
 - Migration to 0.38.0 no longer fails with `UnicodeDecodeError` on a

docs/cli.md

@@ -172,10 +172,16 @@ haiku-rag ask "Does this photo satisfy the spec in the design document?" --image
 
 `ask` runs the [RAG capability](capabilities/rag.md) and always renders citations under the answer. When available, citations use the document title, otherwise they fall back to the URI.
 
+Citation text is truncated to a 300-character preview. To read the whole passage the model saw:
+```bash
+haiku-rag ask "What are the main findings?" --full-citations
+```
+
 Flags:
 
 - `--filter` / `-f`: Restrict searches to documents matching the filter (see [Filtering Search Results](python.md#filtering-search-results))
 - `--image`: Path to an image attached to the question (repeatable). Retrieval stays text-based; the model must have `vision: true` configured.
+- `--full-citations`: Show the full text of each citation instead of a truncated preview
 
 ## Analyze
 
@@ -195,6 +201,7 @@ Flags:
 
 - `--filter` / `-f`: SQL WHERE clause to restrict document access
 - `--image`: Path to an image attached to the question (repeatable). Requires `vision: true` on the analysis model.
+- `--full-citations`: Show the full text of each citation instead of a truncated preview
 
 See [Analysis capability](capabilities/analysis.md) for details and configuration.
 

haiku_rag_slim/haiku/rag/app.py

@@ -638,6 +638,7 @@ class HaikuRAGApp:
         question: str,
         filter: str | None = None,
         images: list[Path] | None = None,
+        full_citations: bool = False,
     ):
         """Ask a question using the RAG system.
 
@@ -645,6 +646,7 @@ class HaikuRAGApp:
             question: The question to ask
             filter: SQL WHERE clause to filter documents
             images: Paths of images to attach to the question
+            full_citations: Render citation text without truncating it
         """
         async with HaikuRAG._covering(
             self.scope, self.config, read_only=True
@@ -660,7 +662,7 @@ class HaikuRAGApp:
             self.console.print("[bold green]Answer:[/bold green]")
             self.console.print(Markdown(answer))
             for renderable in await format_citations_rich(
-                citations, client=self.client
+                citations, client=self.client, full=full_citations
             ):
                 self.console.print(renderable)
 
@@ -669,6 +671,7 @@ class HaikuRAGApp:
         question: str,
         filter: str | None = None,
         images: list[Path] | None = None,
+        full_citations: bool = False,
     ):
         """Answer a question using the analysis capability.
 
@@ -676,6 +679,7 @@ class HaikuRAGApp:
             question: The question to answer
             filter: SQL WHERE clause to filter documents
             images: Paths of images to attach to the question
+            full_citations: Render citation text without truncating it
         """
         async with HaikuRAG._covering(
             self.scope, self.config, read_only=True
@@ -696,7 +700,7 @@ class HaikuRAGApp:
             self.console.print("[bold green]Answer:[/bold green]")
             self.console.print(Markdown(result.answer))
             for renderable in await format_citations_rich(
-                result.citations, client=self.client
+                result.citations, client=self.client, full=full_citations
             ):
                 self.console.print(renderable)
 

haiku_rag_slim/haiku/rag/cli.py

@@ -413,6 +413,11 @@ def ask(
         "--image",
         help="Path to an image to attach to the question (repeatable; requires a vision-capable model)",
     ),
+    full_citations: bool = typer.Option(
+        False,
+        "--full-citations",
+        help="Show the full text of each citation instead of a truncated preview",
+    ),
 ):
     app = create_app(db, covers_set=True)
     asyncio.run(
@@ -420,6 +425,7 @@ def ask(
             question=question,
             filter=filter,
             images=image,
+            full_citations=full_citations,
         )
     )
 
@@ -445,6 +451,11 @@ def analyze(
         "--image",
         help="Path to an image to attach to the question (repeatable; requires a vision-capable model)",
     ),
+    full_citations: bool = typer.Option(
+        False,
+        "--full-citations",
+        help="Show the full text of each citation instead of a truncated preview",
+    ),
 ):
     app = create_app(db, covers_set=True)
     asyncio.run(
@@ -452,6 +463,7 @@ def analyze(
             question=question,
             filter=filter,
             images=image,
+            full_citations=full_citations,
         )
     )
 

haiku_rag_slim/haiku/rag/utils.py

@@ -430,13 +430,15 @@ def truncated(text: str, limit: int) -> str:
 async def format_citations_rich(
     citations: "list[Citation]",
     client: "HaikuRAG | None" = None,
+    full: bool = False,
 ) -> "list[RenderableType]":
     """Format citations as Rich renderables for terminal display.
 
     Each citation becomes a Panel with a compact header (``[N] Title (URI) —
     locator``, with the database name before the locator when ``client`` covers
-    several), a body holding any referenced figures followed by a truncated text
+    several), a body holding any referenced figures followed by a text preview
-    preview, and a dimmed footer that exposes the document and chunk IDs.
+    truncated at ``CITATION_PREVIEW_CHARS``, and a dimmed footer that exposes the
+    document and chunk IDs. ``full`` renders the content untruncated.
 
     When ``client`` is supplied, picture bytes for ``picture_refs`` are fetched and
     rendered inline via ``textual_image``. Without a client, picture refs appear as
@@ -481,7 +483,9 @@ async def format_citations_rich(
                 else Text(f"[Figure: {ref}]", style="italic dim")
             )
 
-        body.append(Text(truncated(c.content, CITATION_PREVIEW_CHARS)))
+        body.append(
+            Text(c.content if full else truncated(c.content, CITATION_PREVIEW_CHARS))
+        )
 
         footer = Text()
         footer.append("doc: ", style="dim")

tests/test_app.py

@@ -307,7 +307,7 @@ async def test_visualize_reports_no_grounding(app, client):
 async def test_ask_prints_question_answer_and_citations(app, client, monkeypatch):
     client.ask.return_value = ("The answer.", [])
 
-    async def no_citations(citations, client=None):
+    async def no_citations(citations, client=None, full=False):
         return ["citation block"]
 
     monkeypatch.setattr("haiku.rag.app.format_citations_rich", no_citations)
@@ -326,7 +326,7 @@ async def test_ask_attaches_image_bytes(app, client, monkeypatch, tmp_path):
     image.write_bytes(b"img")
     client.ask.return_value = ("answer", [])
 
-    async def no_citations(citations, client=None):
+    async def no_citations(citations, client=None, full=False):
         return []
 
     monkeypatch.setattr("haiku.rag.app.format_citations_rich", no_citations)
@@ -336,13 +336,36 @@ async def test_ask_attaches_image_bytes(app, client, monkeypatch, tmp_path):
     assert client.ask.await_args.kwargs["images"] == [b"img"]
 
 
+@pytest.mark.parametrize("verb", ["ask", "analyze"])
+async def test_full_citations_reaches_the_formatter(app, client, monkeypatch, verb):
+    """The flag has to survive the app layer, or the CLI switch does nothing."""
+    client.ask.return_value = ("answer", [])
+    result = AsyncMock()
+    result.answer = "answer"
+    result.citations = []
+    client.analyze.return_value = result
+
+    seen = []
+
+    async def record(citations, client=None, full=False):
+        seen.append(full)
+        return []
+
+    monkeypatch.setattr("haiku.rag.app.format_citations_rich", record)
+
+    await getattr(app, verb)("why?", full_citations=True)
+    await getattr(app, verb)("why?")
+
+    assert seen == [True, False]
+
+
 async def test_analyze_prints_the_answer(app, client, monkeypatch):
     result = AsyncMock()
     result.answer = "computed answer"
     result.citations = []
     client.analyze.return_value = result
 
-    async def no_citations(citations, client=None):
+    async def no_citations(citations, client=None, full=False):
         return []
 
     monkeypatch.setattr("haiku.rag.app.format_citations_rich", no_citations)
@@ -964,7 +987,7 @@ async def test_analyze_prints_citation_renderables(app, client, monkeypatch):
     result.citations = ["c1"]
     client.analyze.return_value = result
 
-    async def one_citation(citations, client=None):
+    async def one_citation(citations, client=None, full=False):
         return ["citation renderable"]
 
     monkeypatch.setattr("haiku.rag.app.format_citations_rich", one_citation)

tests/test_cli.py

@@ -951,10 +951,18 @@ def test_question_commands_dispatch(app_stub, command, method):
 
     assert result.exit_code == 0, result.output
     getattr(app_stub, method).assert_called_once_with(
-        question="why?", filter=None, images=None
+        question="why?", filter=None, images=None, full_citations=False
     )
 
 
+@pytest.mark.parametrize("command, method", [("ask", "ask"), ("analyze", "analyze")])
+def test_full_citations_flag_dispatches(app_stub, command, method):
+    result = runner.invoke(cli, [command, "why?", "--full-citations"] + DB_ARGS)
+
+    assert result.exit_code == 0, result.output
+    assert getattr(app_stub, method).call_args.kwargs["full_citations"] is True
+
+
 @pytest.mark.parametrize(
     "flag, mode_name",
     [

tests/test_utils.py

@@ -907,6 +907,23 @@ async def test_format_citations_rich_truncates_long_content():
     assert "A" * (CITATION_PREVIEW_CHARS + 1) not in output
 
 
+async def test_format_citations_rich_full_keeps_the_whole_content():
+    from haiku.rag.store.models.citation import Citation
+    from haiku.rag.utils import CITATION_PREVIEW_CHARS, format_citations_rich
+
+    content = "A" * (CITATION_PREVIEW_CHARS + 200)
+    citation = Citation(
+        document_id="doc1",
+        chunk_id="chunk1",
+        document_uri="test://doc",
+        content=content,
+    )
+    output = _render_rich(await format_citations_rich([citation], full=True))
+    assert "…" not in output
+    # Rich wraps the body across panel lines, so count the content instead.
+    assert output.count("A") == len(content)
+
+
 async def test_format_citations_rich_picture_marker_without_client():
     from haiku.rag.store.models.citation import Citation
     from haiku.rag.utils import format_citations_rich