Merge pull request #8560 from Knio/tom/bufferedio

Fix file position desync when calling into libtiff

Author: radarhere

Tests/test_file_libtiff.py

@@ -1098,6 +1098,25 @@ def test_exif_transpose(self) -> None:
 
                     assert_image_similar(base_im, im, 0.7)
 
+    @pytest.mark.parametrize(
+        "test_file",
+        [
+            "Tests/images/old-style-jpeg-compression-no-samplesperpixel.tif",
+            "Tests/images/old-style-jpeg-compression.tif",
+        ],
+    )
+    def test_buffering(self, test_file: str) -> None:
+        # load exif first
+        with Image.open(open(test_file, "rb", buffering=1048576)) as im:
+            exif = dict(im.getexif())
+
+        # load image before exif
+        with Image.open(open(test_file, "rb", buffering=1048576)) as im2:
+            im2.load()
+            exif_after_load = dict(im2.getexif())
+
+        assert exif == exif_after_load
+
     @pytest.mark.valgrind_known_error(reason="Backtrace in Python Core")
     def test_sampleformat_not_corrupted(self) -> None:
         # Assert that a TIFF image with SampleFormat=UINT tag is not corrupted

src/PIL/TiffImagePlugin.py

@@ -1216,10 +1216,6 @@ def seek(self, frame: int) -> None:
     def _seek(self, frame: int) -> None:
         self.fp = self._fp
 
-        # reset buffered io handle in case fp
-        # was passed to libtiff, invalidating the buffer
-        self.fp.tell()
-
         while len(self._frame_pos) <= frame:
             if not self.__next:
                 msg = "no more images in TIFF file"
@@ -1303,10 +1299,6 @@ def load_end(self) -> None:
         if not self.is_animated:
             self._close_exclusive_fp_after_loading = True
 
-            # reset buffered io handle in case fp
-            # was passed to libtiff, invalidating the buffer
-            self.fp.tell()
-
             # load IFD data from fp before it is closed
             exif = self.getexif()
             for key in TiffTags.TAGS_V2_GROUPS:
@@ -1381,8 +1373,17 @@ def _load_libtiff(self) -> Image.core.PixelAccess | None:
             logger.debug("have fileno, calling fileno version of the decoder.")
             if not close_self_fp:
                 self.fp.seek(0)
+            # Save and restore the file position, because libtiff will move it
+            # outside of the Python runtime, and that will confuse
+            # io.BufferedReader and possible others.
+            # NOTE: This must use os.lseek(), and not fp.tell()/fp.seek(),
+            # because the buffer read head already may not equal the actual
+            # file position, and fp.seek() may just adjust it's internal
+            # pointer and not actually seek the OS file handle.
+            pos = os.lseek(fp, 0, os.SEEK_CUR)
             # 4 bytes, otherwise the trace might error out
             n, err = decoder.decode(b"fpfp")
+            os.lseek(fp, pos, os.SEEK_SET)
         else:
             # we have something else.
             logger.debug("don't have fileno or getvalue. just reading")