fix: merge mixed-font symbols into line cells

src/parse/page_item_sanitators/cells.h

@@ -184,11 +184,14 @@ namespace pdflib
 
     LOG_S(INFO) << "# char-cells: " << line_cells.size();
 
-    sanitize_bbox(line_cells,
-		  config.horizontal_cell_tolerance,
-		  config.enforce_same_font,
-		  config.line_space_width_factor_for_merge,
-		  config.line_space_width_factor_for_merge_with_space);
+    // Line cells can legitimately mix fonts (e.g., fallback symbol glyphs such as arrows).
+    // Also, PDF content-stream order is not guaranteed to match visual reading order.
+    // Use the order-independent merge path for line construction and do not require font equality.
+    contract_cells_into_lines_v2(line_cells,
+				 config.horizontal_cell_tolerance,
+				 false,
+				 config.line_space_width_factor_for_merge,
+				 config.line_space_width_factor_for_merge_with_space);
 
     LOG_S(INFO) << "# line-cells: " << line_cells.size();
 
@@ -580,6 +583,16 @@ namespace pdflib
 		    erased_cell = true;
 
 		    LOG_S(INFO) << " -> merging cell-" << i << " with " << j << " '" << cells[j].text << "'"<< ": " << cells[i].text;
+		  }
+		else if(cells[j].is_adjacent_to(cells[i], delta_0))
+		  {
+		    cells[j].merge_with(cells[i], delta_1);
+
+		    cells[i].active = false;
+		    erased_cell = true;
+
+		    LOG_S(INFO) << " -> merging reverse cell-" << j << " with " << i << " '" << cells[i].text << "'"<< ": " << cells[j].text;
+		    break;
 		  }		
 	      }
 	  }

tests/test_parse.py

@@ -918,3 +918,28 @@ def test_annotations_match_groundtruth():
         verify_annotations_recursive(true_annotations, pred_dict)
 
         pdf_doc.unload()
+
+
+def test_mixed_font_arrows_are_merged_into_line_cells():
+    """Regression: mixed-font arrow glyphs must stay in the surrounding line text."""
+    parser = DoclingPdfParser(loglevel="fatal")
+    pdf_doc = parser.load(
+        path_or_stream="tests/data/targeted/mixed_font_arrows_01.pdf",
+        lazy=False,
+    )
+
+    expected_line_texts = [
+        "Arrow Ordering Parser Fixture",
+        "Alpha → Beta → Gamma",
+        "One → Two → Three",
+        "North → South",
+    ]
+
+    page = pdf_doc.get_page(1)
+    line_texts = [normalize_text(cell.text) for cell in page.textline_cells]
+
+    assert (
+        line_texts == expected_line_texts
+    ), f"unexpected line cells: {line_texts}"
+
+    pdf_doc.unload()