Merge pull request #1967 from anthrotype/pickle

sfnt: add __getstate__ and __setstate__ to SFNTReader to make it pickelable

Lib/fontTools/ttLib/sfnt.py

@@ -12,7 +12,8 @@ classes, since whenever to number of tables changes or whenever
 a table's length chages you need to rewrite the whole file anyway.
 """
 
-from fontTools.misc.py23 import *
+from io import BytesIO
+from fontTools.misc.py23 import Tag
 from fontTools.misc import sstruct
 from fontTools.ttLib import TTLibError
 import struct
@@ -122,29 +123,29 @@ class SFNTReader(object):
 	def close(self):
 		self.file.close()
 
-	def __deepcopy__(self, memo):
-		"""Overrides the default deepcopy of SFNTReader object, to make it work
-		in the case when TTFont is loaded with lazy=True, and thus reader holds a
-		reference to a file object which is not pickleable.
-		We work around it by manually copying the data into a in-memory stream.
-		"""
-		from copy import deepcopy
+	# We define custom __getstate__ and __setstate__ to make SFNTReader pickle-able
+	# and deepcopy-able. When a TTFont is loaded as lazy=True, SFNTReader holds a
+	# reference to an external file object which is not pickleable. So in __getstate__
+	# we store the file name and current position, and in __setstate__ we reopen the
+	# same named file after unpickling.
 
-		cls = self.__class__
-		obj = cls.__new__(cls)
-		for k, v in self.__dict__.items():
-			if k == "file":
-				pos = v.tell()
-				v.seek(0)
-				buf = BytesIO(v.read())
-				v.seek(pos)
-				buf.seek(pos)
-				if hasattr(v, "name"):
-					buf.name = v.name
-				obj.file = buf
-			else:
-				obj.__dict__[k] = deepcopy(v, memo)
-		return obj
+	def __getstate__(self):
+		if isinstance(self.file, BytesIO):
+			# BytesIO is already pickleable, return the state unmodified
+			return self.__dict__
+
+		# remove unpickleable file attribute, and only store its name and pos
+		state = self.__dict__.copy()
+		del state["file"]
+		state["_filename"] = self.file.name
+		state["_filepos"] = self.file.tell()
+		return state
+
+	def __setstate__(self, state):
+		if "file" not in state:
+			self.file = open(state.pop("_filename"), "rb")
+			self.file.seek(state.pop("_filepos"))
+		self.__dict__.update(state)
 
 
 # default compression level for WOFF 1.0 tables and metadata

Tests/ttLib/sfnt_test.py

@@ -1,7 +1,59 @@
-from fontTools.misc.py23 import *
-from fontTools.ttLib.sfnt import calcChecksum
+import io
+import copy
+import pickle
+from fontTools.ttLib.sfnt import calcChecksum, SFNTReader
+import pytest
 
 
 def test_calcChecksum():
     assert calcChecksum(b"abcd") == 1633837924
     assert calcChecksum(b"abcdxyz") == 3655064932
+
+
+EMPTY_SFNT = b"\x00\x01\x00\x00" + b"\x00" * 8
+
+
+def pickle_unpickle(obj):
+    return pickle.loads(pickle.dumps(obj))
+
+
+class SFNTReaderTest:
+    @pytest.mark.parametrize("deepcopy", [copy.deepcopy, pickle_unpickle])
+    def test_pickle_protocol_FileIO(self, deepcopy, tmp_path):
+        fontfile = tmp_path / "test.ttf"
+        fontfile.write_bytes(EMPTY_SFNT)
+        reader = SFNTReader(fontfile.open("rb"))
+
+        reader2 = deepcopy(reader)
+
+        assert reader2 is not reader
+        assert reader2.file is not reader.file
+
+        assert isinstance(reader2.file, io.BufferedReader)
+        assert isinstance(reader2.file.raw, io.FileIO)
+        assert reader2.file.name == reader.file.name
+        assert reader2.file.tell() == reader.file.tell()
+
+        for k, v in reader.__dict__.items():
+            if k == "file":
+                continue
+            assert getattr(reader2, k) == v
+
+    @pytest.mark.parametrize("deepcopy", [copy.deepcopy, pickle_unpickle])
+    def test_pickle_protocol_BytesIO(self, deepcopy, tmp_path):
+        buf = io.BytesIO(EMPTY_SFNT)
+        reader = SFNTReader(buf)
+
+        reader2 = deepcopy(reader)
+
+        assert reader2 is not reader
+        assert reader2.file is not reader.file
+
+        assert isinstance(reader2.file, io.BytesIO)
+        assert reader2.file.tell() == reader.file.tell()
+        assert reader2.file.getvalue() == reader.file.getvalue()
+
+        for k, v in reader.__dict__.items():
+            if k == "file":
+                continue
+            assert getattr(reader2, k) == v

Tests/varLib/varLib_test.py

@@ -22,6 +22,9 @@ def reload_font(font):
     """(De)serialize to get final binary layout."""
     buf = BytesIO()
     font.save(buf)
+    # Close the font to release filesystem resources so that on Windows the tearDown
+    # method can successfully remove the temporary directory created during setUp.
+    font.close()
     buf.seek(0)
     return TTFont(buf)