Merge pull request #1714 from mvanhorn/fix-nonembedded-cid-fonts-nondict-resource

fix: guard find_nonembedded_cid_fonts against non-dictionary resources

src/ocrmypdf/pdfa.py

@@ -12,7 +12,7 @@ from importlib.resources import files as package_files
 from pathlib import Path
 
 import pikepdf
-from pikepdf import Array, Dictionary, Name, Pdf, Stream
+from pikepdf import Array, Dictionary, Name, Object, Pdf, Stream
 
 log = logging.getLogger(__name__)
 
@@ -137,11 +137,13 @@ def file_claims_pdfa(filename: Path):
     return pdfa_dict
 
 
-def _cid_font_is_embedded(type0_font: Dictionary) -> bool:
+def _cid_font_is_embedded(type0_font: Object) -> bool:
     """Return True if a Type0 font's CID descendant carries embedded glyphs."""
     for descendant in type0_font.get(Name.DescendantFonts, []):
         descriptor = descendant.get(Name.FontDescriptor, None)
-        if descriptor is not None and any(
+        # A malformed PDF may store a non-dictionary here; `key in descriptor`
+        # raises on those, so require a real dictionary before probing it.
+        if isinstance(descriptor, Dictionary) and any(
             key in descriptor for key in (Name.FontFile, Name.FontFile2, Name.FontFile3)
         ):
             return True
@@ -174,9 +176,13 @@ def find_nonembedded_cid_fonts(pdf: Pdf) -> set[str]:
     def scan_resources(resources, depth: int = 0) -> None:
         if resources is None or depth > 10:
             return
+        # A well-formed PDF stores dictionaries under /Font and /XObject, but a
+        # malformed one (common in OCR workloads) may store an array, a name, or
+        # another non-dictionary object. Only such dictionaries have .values(),
+        # so guard with isinstance rather than let the scan crash (issue #1713).
         fonts = resources.get(Name.Font, None)
-        if fonts is not None:
-            for font in fonts.values():
+        if isinstance(fonts, Dictionary):
+            for font in fonts.as_dict().values():
                 try:
                     if font.get(Name.Subtype) != Name.Type0:
                         continue
@@ -186,8 +192,8 @@ def find_nonembedded_cid_fonts(pdf: Pdf) -> set[str]:
                 except (AttributeError, TypeError, KeyError):
                     continue
         xobjects = resources.get(Name.XObject, None)
-        if xobjects is not None:
-            for xobj in xobjects.values():
+        if isinstance(xobjects, Dictionary):
+            for xobj in xobjects.as_dict().values():
                 if xobj.get(Name.Subtype) == Name.Form and Name.Resources in xobj:
                     scan_resources(xobj[Name.Resources], depth + 1)
 

tests/test_pdfa.py

@@ -94,6 +94,52 @@ class TestFindNonembeddedCidFonts:
         with pikepdf.open(path) as pdf:
             assert find_nonembedded_cid_fonts(pdf) == {'ZZZ+Hidden'}
 
+    def test_non_dictionary_font_and_xobject_resources_are_ignored(self, tmp_path):
+        # A malformed PDF may carry a /Font or /XObject resource that is not a
+        # dictionary (an array, a name, an empty value). Scanning must skip it
+        # rather than raise when iterating its values (regression test for the
+        # crash reported in issue #1713).
+        path = tmp_path / 'malformed_resources.pdf'
+        with pikepdf.new() as pdf:
+            page = pdf.add_blank_page()
+            page.Resources = pikepdf.Dictionary(
+                Font=pikepdf.Array([]),
+                XObject=pikepdf.Array([]),
+            )
+            pdf.save(path)
+        with pikepdf.open(path) as pdf:
+            assert find_nonembedded_cid_fonts(pdf) == set()
+
+    def test_non_dictionary_font_descriptor_is_reported(self, tmp_path):
+        # A Type0 font whose descendant carries a non-dictionary /FontDescriptor
+        # has no embedded glyph data, so it must be reported -- not crash. This
+        # is the same malformed-resource bug class as #1713, one level deeper:
+        # `key in descriptor` raises ValueError on a non-dictionary.
+        path = tmp_path / 'bad_descriptor.pdf'
+        with pikepdf.new() as pdf:
+            page = pdf.add_blank_page()
+            cidfont = pdf.make_indirect(
+                pikepdf.Dictionary(
+                    Type=Name.Font,
+                    Subtype=Name.CIDFontType2,
+                    BaseFont=Name('/BOGUS+CID'),
+                    FontDescriptor=Name.NotADictionary,
+                )
+            )
+            type0 = pdf.make_indirect(
+                pikepdf.Dictionary(
+                    Type=Name.Font,
+                    Subtype=Name.Type0,
+                    BaseFont=Name('/BOGUS+CID'),
+                    Encoding=Name.Identity_H,
+                    DescendantFonts=pikepdf.Array([cidfont]),
+                )
+            )
+            page.Resources = pikepdf.Dictionary(Font=pikepdf.Dictionary(F0=type0))
+            pdf.save(path)
+        with pikepdf.open(path) as pdf:
+            assert find_nonembedded_cid_fonts(pdf) == {'BOGUS+CID'}
+
 
 @pytest.fixture
 def nonembedded_cid_pdf(tmp_path):