From d8d766b56e91dc4125e70e24c1faf0937709a98e Mon Sep 17 00:00:00 2001
From: Research Assistant <research@example.com>
Date: Wed, 10 Jun 2026 01:54:43 +0800
Subject: [PATCH] fix: keep OCR tail support out of body flow

---
 paperforge/worker/ocr_blocks.py   |  4 ++++
 paperforge/worker/ocr_document.py | 20 ++++++++++++++++++++
 2 files changed, 24 insertions(+)

diff --git a/paperforge/worker/ocr_blocks.py b/paperforge/worker/ocr_blocks.py
index d95d1b57..75b63db4 100644
--- a/paperforge/worker/ocr_blocks.py
+++ b/paperforge/worker/ocr_blocks.py
@@ -204,6 +204,10 @@ def build_structured_blocks(
             for zone in refreshed_reference_zones
         ] or None
 
+    from paperforge.worker.ocr_document import _exclude_tail_nonref_from_body_flow
+
+    _exclude_tail_nonref_from_body_flow(rows)
+
     # Sync render_default/index_default after role normalizations
     for row in rows:
         role = row.get("role", "")
diff --git a/paperforge/worker/ocr_document.py b/paperforge/worker/ocr_document.py
index 085d0d9d..0e9f173a 100644
--- a/paperforge/worker/ocr_document.py
+++ b/paperforge/worker/ocr_document.py
@@ -1840,6 +1840,25 @@ def _exclude_frontmatter_side_from_body_flow(blocks: list[dict]) -> None:
         block.setdefault("evidence", []).append("frontmatter_side_zone excluded from body flow")
 
 
+def _exclude_tail_nonref_from_body_flow(blocks: list[dict]) -> None:
+    for block in blocks:
+        if block.get("role") != "body_paragraph":
+            continue
+        if block.get("zone") != "tail_nonref_hold_zone":
+            continue
+        old_role = block.get("role")
+        block["role"] = "backmatter_body"
+        if old_role != block["role"]:
+            record_decision(
+                block,
+                stage="tail_nonref_exclusion",
+                old_role=old_role,
+                new_role=block["role"],
+                reason="tail non-reference body block excluded from body flow",
+            )
+        block.setdefault("evidence", []).append("tail_nonref_hold_zone excluded from body flow")
+
+
 def _get_column(block: dict, page_width: float = 1200) -> int:
     bbox = block.get("bbox") or block.get("block_bbox")
     if bbox and len(bbox) >= 4:
@@ -3274,6 +3293,7 @@ def normalize_document_structure(blocks: list[dict]) -> tuple[DocumentStructure,
         },
     )
     _exclude_frontmatter_side_from_body_flow(blocks)
+    _exclude_tail_nonref_from_body_flow(blocks)
 
     from paperforge.worker.ocr_roles import resolve_final_role