ENH: Extract the /FontFile and store it in the new FontDescriptor class

pypdf/_font.py

@@ -2,11 +2,12 @@
 from dataclasses import dataclass, field
 from typing import Any, Optional, Union, cast
 
-from pypdf.generic import ArrayObject, DictionaryObject, IndirectObject
+from pypdf.generic import ArrayObject, DictionaryObject, IndirectObject, StreamObject
 
 from ._cmap import get_encoding
 from ._codecs.adobe_glyphs import adobe_glyphs
 from ._utils import logger_warning
+from .errors import PdfReadError
 
 
 @dataclass(frozen=True)
@@ -32,6 +33,7 @@ class FontDescriptor:
     bbox: tuple[float, float, float, float] = field(default_factory=lambda: (-100.0, -200.0, 1000.0, 900.0))
 
     character_widths: dict[str, int] = field(default_factory=lambda: {"default": 500})
+    font_file: Union[StreamObject, None] = None
 
     @staticmethod
     def _parse_font_descriptor(font_kwargs: dict[str, Any], font_descriptor_obj: DictionaryObject) -> dict[str, Any]:
@@ -59,6 +61,18 @@ def _parse_font_descriptor(font_kwargs: dict[str, Any], font_descriptor_obj: Dic
             bbox_tuple = tuple(map(float, font_kwargs["bbox"]))
             assert len(bbox_tuple) == 4, bbox_tuple
             font_kwargs["bbox"] = bbox_tuple
+        # Find the binary stream for this font if there is one
+        for source_key in ["/FontFile", "/FontFile2", "/FontFile3"]:
+            if source_key in font_descriptor_dict:
+                if "font_file" in font_kwargs:
+                    raise PdfReadError(f"More than one /FontFile found in {font_descriptor_obj}")
+
+                try:
+                    font_file = font_descriptor_dict[source_key].get_object()
+                    font_kwargs["font_file"] = font_file
+                except PdfReadError as e:
+                    logger_warning(f"Failed to get '{source_key}' in {font_descriptor_dict}: {e}", __name__)
+
         return font_kwargs
 
     @staticmethod

tests/test_cmap.py

@@ -8,7 +8,7 @@
 from pypdf._cmap import get_encoding, parse_bfchar
 from pypdf._codecs import charset_encoding
 from pypdf._font import Font
-from pypdf.generic import ArrayObject, DictionaryObject, IndirectObject, NameObject, NullObject
+from pypdf.generic import ArrayObject, DictionaryObject, EncodedStreamObject, IndirectObject, NameObject, NullObject
 
 from . import get_data_from_url
 
@@ -139,6 +139,8 @@ def test_iss1533():
     reader.pages[0].extract_text()  # no error
     font = Font.from_font_resource(reader.pages[0]["/Resources"]["/Font"]["/F"])
     assert font.character_map["\x01"] == "Ü"
+    assert type(font.font_descriptor.font_file) is EncodedStreamObject
+    assert font.font_descriptor.font_file["/Subtype"] == "/CIDFontType0C"
 
 
 @pytest.mark.enable_socket

tests/test_font.py

@@ -1,7 +1,16 @@
 """Test font-related functionality."""
+from pathlib import Path
 
-from pypdf._font import FontDescriptor
-from pypdf.generic import DictionaryObject, NameObject
+import pytest
+
+from pypdf import PdfReader
+from pypdf._font import Font, FontDescriptor
+from pypdf.errors import PdfReadError
+from pypdf.generic import DictionaryObject, EncodedStreamObject, NameObject
+
+TESTS_ROOT = Path(__file__).parent.resolve()
+PROJECT_ROOT = TESTS_ROOT.parent
+RESOURCE_ROOT = PROJECT_ROOT / "resources"
 
 
 def test_font_descriptor():
@@ -28,3 +37,29 @@ def test_font_descriptor():
     assert my_font.italic_angle == 0
     assert my_font.flags == 33
     assert my_font.bbox == (-113.0, -250.0, 749.0, 801.0)
+
+
+def test_font_file():
+    reader = PdfReader(RESOURCE_ROOT / "multilang.pdf")
+
+    # /FontFile
+    font = Font.from_font_resource(reader.pages[0]["/Resources"]["/Font"]["/F2"])
+    assert type(font.font_descriptor.font_file) is EncodedStreamObject
+    assert len(font.font_descriptor.font_file.get_data()) == 5116
+
+    # /FontFile2
+    font_resource = reader.pages[0]["/Resources"]["/Font"]["/F1"]
+    font = Font.from_font_resource(font_resource)
+    assert type(font.font_descriptor.font_file) is EncodedStreamObject
+    assert len(font.font_descriptor.font_file.get_data()) == 28464
+
+    with pytest.raises(PdfReadError) as exception:
+        font_resource[NameObject("/FontDescriptor")][NameObject("/FontFile")] = NameObject("xyz")
+        font = Font.from_font_resource(font_resource)
+    assert "More than one /FontFile" in exception.value.args[0]
+
+    # /FontFile3
+    reader = PdfReader(RESOURCE_ROOT / "attachment.pdf")
+    font = Font.from_font_resource(reader.pages[0]["/Resources"]["/Font"]["/F1"])
+    assert type(font.font_descriptor.font_file) is EncodedStreamObject
+    assert len(font.font_descriptor.font_file.get_data()) == 2168

tests/test_text_extraction.py

@@ -143,6 +143,7 @@ def test_font_class_to_dict():
             "x_height": 500.0,
             "italic_angle": 0.0,
             "flags": 32,
+            "font_file": None,
             "bbox": (
                 -100.0,
                 -200.0,