fix: treat hexadecimal glyph-index names (Skia g%X) as unmapped glyphs

src/parse/pdf_resources/page_font.h

@@ -2612,12 +2612,14 @@ namespace pdflib
     };
 
     // Subset generators (FontForge, fontTools, mPDF, ...) name glyphs by
-    // bare index: /gid00043, /g43, /glyph43, /cid43, /index43. Such a name
+    // bare index: /gid00043, /g43, /glyph43, /cid43, /index43. Skia/PDF
+    // (Chrome print-to-PDF, Notion exports) writes `g%X`, so the index may
+    // carry hexadecimal digits: /g15C, /gA1. Such a name
     // identifies the glyph inside the embedded font program but carries no
     // reading text, so keeping it fabricates plausible-looking garbage
     // ('gid00043gid00049...') that downstream quality gates cannot detect
     // (docling-project/docling-parse#238).
-    std::regex re_gid(R"((gid|glyph|g|cid|index)\d+)", std::regex::icase);
+    std::regex re_gid(R"((gid|glyph|g|cid|index)[0-9A-Fa-f]+)", std::regex::icase);
 
     // A small family of PDF producers replaces every descriptive glyph name
     // with an uppercase `G<decimal Unicode>` name. Do not infer that convention

tests/constants.py

@@ -1,7 +1,7 @@
 from tests.regression_page_selection import REGRESSION_PAGE_SELECTION
 
 HF_DATASET_REPO_ID = "docling-project/regression-dataset-for-docling-parse"
-HF_DATASET_REVISION = "6d3458e2a6a729defd700a7b91e57b6cb2ce0290"
+HF_DATASET_REVISION = "4103ae9b54a52f6890b57d6711614defd03d77b8"
 
 REGRESSION_DIR = "tests/data/regression"
 

tests/test_unit_nul_mapping.py

@@ -51,3 +51,15 @@ def test_nul_mapping_with_unknown_identity_remains_unresolved():
         _extract_character(mapping="0000", glyph_name="gid00043")
         == "GLYPH<name:gid00043>"
     )
+
+
+@pytest.mark.parametrize("glyph_name", ["g15C", "g15c", "gA1", "GID00043"])
+def test_nul_mapping_with_hexadecimal_glyph_index_name_remains_unresolved(
+    glyph_name: str,
+):
+    # Skia/PDF names Type 3 glyphs `g%X` (hexadecimal gid), so a bare index
+    # may carry letters A-F; it still identifies no reading text.
+    assert (
+        _extract_character(mapping="0000", glyph_name=glyph_name)
+        == f"GLYPH<name:{glyph_name}>"
+    )