From 4071f240e7c85813fa76ec17bf174117f28df2f5 Mon Sep 17 00:00:00 2001
From: Ainur T <1525421+nurikk@users.noreply.github.com>
Date: Fri, 18 Sep 2026 16:05:55 +0000
Subject: [PATCH] fix: fall back from null cmap mappings

Signed-off-by: Ainur T <1525421+nurikk@users.noreply.github.com>
---
 src/parse/pdf_resources/page_font.h |  1 +
 tests/test_unit_nul_mapping.py      | 53 +++++++++++++++++++++++++++++
 2 files changed, 54 insertions(+)
 create mode 100644 tests/test_unit_nul_mapping.py

diff --git a/src/parse/pdf_resources/page_font.h b/src/parse/pdf_resources/page_font.h
index 2d28feff..090d32ef 100644
--- a/src/parse/pdf_resources/page_font.h
+++ b/src/parse/pdf_resources/page_font.h
@@ -2566,6 +2566,7 @@ namespace pdflib
       return cmap_initialized
 	and cmap_numb_to_char.count(numb)==1
 	and cmap_numb_to_char.at(numb).size()>0
+	and cmap_numb_to_char.at(numb)!=std::string(1, '\0')
 	and cmap_numb_to_char.at(numb)!=replacement_char;
     };
 
diff --git a/tests/test_unit_nul_mapping.py b/tests/test_unit_nul_mapping.py
new file mode 100644
index 00000000..fd95bf0b
--- /dev/null
+++ b/tests/test_unit_nul_mapping.py
@@ -0,0 +1,53 @@
+from io import BytesIO
+
+import pytest
+
+from docling_parse.pdf_parser import DecodeConfig, DoclingPdfParser
+from tests.pdf_builder import content_stream, simple_page_pdf
+
+
+def _extract_character(mapping: str, glyph_name: str) -> str:
+    cmap = (
+        "/CIDInit /ProcSet findresource begin\n"
+        "12 dict begin\nbegincmap\n"
+        "/CIDSystemInfo << /Registry (Adobe) /Ordering (UCS) /Supplement 0 >> def\n"
+        "/CMapName /Test-UCS def\n/CMapType 2 def\n"
+        "1 begincodespacerange\n<00> <FF>\nendcodespacerange\n"
+        f"1 beginbfrange\n<41> <41> <{mapping}>\nendbfrange\n"
+        "endcmap\nCMapName currentdict /CMap defineresource pop\nend\nend"
+    )
+    pdf = simple_page_pdf(
+        "BT /F1 24 Tf 20 100 Td (A) Tj ET",
+        resources="/Font << /F1 5 0 R >>",
+        extra_objects=[
+            "<< /Type /Font /Subtype /Type1 /BaseFont /XXXXXX+FakeSubset "
+            "/FirstChar 65 /LastChar 65 /Widths [600] "
+            f"/Encoding << /Differences [65 /{glyph_name}] >> "
+            "/ToUnicode 6 0 R >>",
+            content_stream(cmap),
+        ],
+    )
+    parser = DoclingPdfParser(loglevel="fatal")
+    doc = parser.load(
+        BytesIO(pdf),
+        decode_config=DecodeConfig(keep_glyphs=True),
+    )
+    page = doc.get_page(1)
+    assert len(page.char_cells) == 1
+    return page.char_cells[0].text
+
+
+def test_authoritative_minus_overrides_conflicting_identity_before_normalization():
+    assert _extract_character(mapping="2212", glyph_name="plus") == "-"
+
+
+@pytest.mark.parametrize(("glyph_name", "expected"), [("minus", "-"), ("plus", "+")])
+def test_nul_mapping_recovers_from_known_glyph_identity(glyph_name: str, expected: str):
+    assert _extract_character(mapping="0000", glyph_name=glyph_name) == expected
+
+
+def test_nul_mapping_with_unknown_identity_remains_unresolved():
+    assert (
+        _extract_character(mapping="0000", glyph_name="gid00043")
+        == "GLYPH<name:gid00043>"
+    )