From 150b9152824c995cc732ed08213241ca2b4c1e8d Mon Sep 17 00:00:00 2001
From: kcar <kcar@kevlarai.com>
Date: Mon, 8 Jun 2026 18:44:30 +0100
Subject: [PATCH] [verified] fix exam auto-map duplicate continued parts

(cherry picked from commit 31c51cb7aa33d7f2e1102cea4ffabfefee259faa)
---
 routers/exam/templates.py    |  6 ++++++
 tests/test_exam_templates.py | 21 +++++++++++++++++++++
 2 files changed, 27 insertions(+)

diff --git a/routers/exam/templates.py b/routers/exam/templates.py
index 48ad039..736b35f 100644
--- a/routers/exam/templates.py
+++ b/routers/exam/templates.py
@@ -507,6 +507,12 @@ def _map_first_pass_to_rows(template_id: str, first_pass: Dict[str, Any], pdf_by
                 questions.append({"id": parent_id, "template_id": template_id, "label": parent_label, "order": len(q_ids) - 1, "max_marks": 0, "is_container": True, "source": "ai", "confirmed": False, "confidence": 0.7, "derivation": "docling-inferred-main-question"})
             pid = _ai_id(template_id, "part", label)
             first_part_by_page.setdefault(page_index, pid)
+            # B1 live-route papers can carry continuation bands for the same part label
+            # on later pages. The UUID is intentionally stable per template+part label,
+            # so only insert the first question row; later continuations still map
+            # response/context regions through first_part_by_page.
+            if any(q["id"] == pid for q in questions):
+                continue
             bounds = None
             y1, y2 = band.get("y_start"), band.get("y_end")
             if margins["left"] is not None and margins["right"] is not None and y1 is not None and y2 is not None:
diff --git a/tests/test_exam_templates.py b/tests/test_exam_templates.py
index 6ac1676..307e9ef 100644
--- a/tests/test_exam_templates.py
+++ b/tests/test_exam_templates.py
@@ -595,6 +595,27 @@ def test_box_to_canvas_uses_cropbox_as_page_origin():
     assert templates_mod._box_to_canvas(box, 1, pages) == {"x": 0.0, "y": 25.0, "w": 80.0, "h": 40.0}
 
 
+def test_auto_map_deduplicates_continued_part_labels(monkeypatch):
+    monkeypatch.setattr(templates_mod, "_pdf_page_geometry", lambda _pdf: [
+        {"media_x0": 0.0, "crop_x0": 0.0, "crop_y0": 0.0, "page_pt_w": 600.0, "page_pt_h": 800.0, "rendered_w": 600.0, "rendered_h": 800.0, "page_top": 0.0},
+        {"media_x0": 0.0, "crop_x0": 0.0, "crop_y0": 0.0, "page_pt_w": 600.0, "page_pt_h": 800.0, "rendered_w": 600.0, "rendered_h": 800.0, "page_top": 800.0},
+    ])
+    first_pass = _first_pass_template()
+    first_pass["meta"]["n_pages"] = 2
+    first_pass["pages"]["2"] = {
+        "role": "question", "role_source": "auto", "margins_enabled": True,
+        "main_bands": [],
+        "part_bands": [{"label": "01.1", "question": "01", "y_start": 760, "y_end": 600, "label_box": {"l": 50, "t": 760, "r": 90, "b": 740, "coord_origin": "BOTTOMLEFT"}, "source": "auto", "confirmed": False}],
+        "furniture": [], "figures": [], "tables": [],
+    }
+
+    rows = templates_mod._map_first_pass_to_rows("t1", first_pass, b"%PDF", [])
+
+    question_ids = [q["id"] for q in rows["questions"]]
+    assert len(question_ids) == len(set(question_ids))
+    assert [q["label"] for q in rows["questions"]].count("01.1") == 1
+
+
 def test_response_region_types_are_mapped_to_response_form_enum(monkeypatch):
     monkeypatch.setattr(templates_mod, "_pdf_page_geometry", lambda _pdf: [{"media_x0": 0.0, "crop_x0": 0.0, "crop_y0": 0.0, "page_pt_w": 600.0, "page_pt_h": 800.0, "rendered_w": 600.0, "rendered_h": 800.0, "page_top": 0.0}])
     first_pass = _first_pass_template()