ENH: Add possibility to get names of fonts

See #153

Co-authored-by: tiarno <[email protected]>

PyPDF2/_page.py

@@ -39,6 +39,7 @@
  Iterator,
  List,
  Optional,
+ Set,
  Tuple,
  Union,
  cast,
@@ -1338,6 +1339,35 @@ def extractText(
  deprecate_with_replacement("extractText", "extract_text")
  return self.extract_text(Tj_sep=Tj_sep, TJ_sep=TJ_sep)
 
+ def _get_fonts(self) -> Tuple[Set[str], Set[str]]:
+ """
+ Get the names of embedded fonts and unembedded fonts.
+
+ :return: (Set of embedded fonts, set of unembedded fonts)
+ """
+ obj = self.get_object()
+ assert obj is not None
+
+ fonts: Set[str] = set()
+ embedded: Set[str] = set()
+ if isinstance(obj, ArrayObject):
+ embedded_fonts: Set[str] = set()
+ for i in obj:
+ if hasattr(i, "keys"):
+ f, e = _get_fonts_walk(i, fonts, embedded_fonts)
+ if f is not None:
+ fonts = fonts.union(f)
+ if e is not None:
+ embedded = embedded.union(e)
+ elif isinstance(obj, DictionaryObject):
+ f, e = _get_fonts_walk(obj["/Resources"], fonts, embedded)
+ if f is not None:
+ fonts = fonts.union(f)
+ if e is not None:
+ embedded = embedded.union(e)
+ unembedded = fonts - embedded
+ return embedded, unembedded
+
  mediabox = _create_rectangle_accessor(PG.MEDIABOX, ())
  """
  A :class:`RectangleObject<PyPDF2.generic.RectangleObject>`, expressed in default user space units,
@@ -1486,3 +1516,29 @@ def __getitem__(self, index: int) -> PageObject:
  def __iter__(self) -> Iterator[PageObject]:
  for i in range(len(self)):
  yield self[i]
+
+
+def _get_fonts_walk(
+ obj: DictionaryObject, fnt: Set[str], emb: Set[str]
+) -> Tuple[Optional[Set[str]], Optional[Set[str]]]:
+ """
+ If there is a key called 'BaseFont', that is a font that is used in the document.
+ If there is a key called 'FontName' and another key in the same dictionary object
+ that is called 'FontFilex' (where x is null, 2, or 3), then that fontname is
+ embedded.
+
+ We create and add to two sets, fnt = fonts used and emb = fonts embedded.
+ """
+ if not hasattr(obj, "keys"):
+ return None, None
+ fontkeys = ("/FontFile", "/FontFile2", "/FontFile3")
+ if "/BaseFont" in obj:
+ fnt.add(obj["/BaseFont"])
+ if "/FontName" in obj:
+ if [x for x in fontkeys if x in obj]: # test to see if there is FontFile
+ emb.add(obj["/FontName"])
+
+ for key in obj.keys():
+ _get_fonts_walk(obj[key], fnt, emb)
+
+ return fnt, emb # return the sets for each page

tests/test_page.py

@@ -266,3 +266,58 @@ def test_extract_text_operator_t_star(): # L1266, L1267
  reader = PdfReader(BytesIO(get_pdf_from_url(url, name=name)))
  for page in reader.pages:
  page.extract_text()
+
+
+@pytest.mark.parametrize(
+ ("pdf_path", "password", "embedded", "unembedded"),
+ [
+ (
+ os.path.join(RESOURCE_ROOT, "crazyones.pdf"),
+ None,
+ {
+ "/HHXGQB+SFTI1440",
+ "/TITXYI+SFRM0900",
+ "/YISQAD+SFTI1200",
+ },
+ set(),
+ ),
+ (
+ os.path.join(RESOURCE_ROOT, "attachment.pdf"),
+ None,
+ {
+ "/HHXGQB+SFTI1440",
+ "/TITXYI+SFRM0900",
+ "/YISQAD+SFTI1200",
+ },
+ set(),
+ ),
+ (
+ os.path.join(RESOURCE_ROOT, "libreoffice-writer-password.pdf"),
+ "openpassword",
+ {"/BAAAAA+DejaVuSans"},
+ set(),
+ ),
+ (
+ os.path.join(RESOURCE_ROOT, "imagemagick-images.pdf"),
+ None,
+ set(),
+ {"/Helvetica"},
+ ),
+ (os.path.join(RESOURCE_ROOT, "imagemagick-lzw.pdf"), None, set(), set()),
+ (
+ os.path.join(RESOURCE_ROOT, "reportlab-inline-image.pdf"),
+ None,
+ set(),
+ {"/Helvetica"},
+ ),
+ ],
+)
+def test_get_fonts(pdf_path, password, embedded, unembedded):
+ reader = PdfReader(pdf_path, password=password)
+ a = set()
+ b = set()
+ for page in reader.pages:
+ a_tmp, b_tmp = page._get_fonts()
+ a = a.union(a_tmp)
+ b = b.union(b_tmp)
+ assert (a, b) == (embedded, unembedded)