Fixes for TIFF parsing issues

(python-openxml#187, python-openxml#184, python-openxml#1451)

Author: FireMasterK

src/docx/image/tiff.py

@@ -236,6 +236,9 @@ def _parse_value(cls, stream_rdr, offset, value_count, value_offset):
         The length of the string, including a terminating '\x00' (NUL) character, is in
         `value_count`.
         """
+        if value_count <= 4:
+            return stream_rdr.read_str(value_count - 1, offset, 8)
+
         return stream_rdr.read_str(value_count - 1, value_offset)
 
 

tests/image/test_jpeg.py

@@ -1,10 +1,12 @@
 """Unit test suite for docx.image.jpeg module"""
 
 import io
+import os
 
 import pytest
 
 from docx.image.constants import JPEG_MARKER_CODE, MIME_TYPE
+from docx.image.image import Image
 from docx.image.helpers import BIG_ENDIAN, StreamReader
 from docx.image.jpeg import (
     Exif,
@@ -21,6 +23,7 @@
 )
 from docx.image.tiff import Tiff
 
+from ..unitutil.file import test_file_dir
 from ..unitutil.mock import (
     ANY,
     call,
@@ -31,6 +34,21 @@
 )
 
 
+class DescribeJpeg_integration:
+    """Integration tests using real image files."""
+
+    def it_can_parse_a_jpeg_with_inline_ascii_exif_values(self):
+        """Regression test for #184/#187 — iOS JPEGs whose EXIF Software field
+        is short enough (<=4 bytes) to be stored inline in the Value/Offset
+        slot rather than at a separate file offset.
+
+        Prior to the fix this raised UnexpectedEndOfFileError.
+        """
+        jpeg_path = os.path.join(test_file_dir, "ios-exif-inline-ascii.jpg")
+        image = Image.from_file(jpeg_path)
+        assert image.content_type == MIME_TYPE.JPEG
+
+
 class DescribeJpeg:
     def it_knows_its_content_type(self):
         jpeg = Jpeg(None, None, None, None)

tests/image/test_tiff.py

@@ -417,6 +417,18 @@ def it_can_parse_an_ascii_string_IFD_entry(self):
         val = _AsciiIfdEntry._parse_value(stream_rdr, None, 7, 0)
         assert val == "foobar"
 
+    @pytest.mark.parametrize("byte_order", [BIG_ENDIAN, LITTLE_ENDIAN])
+    def it_can_parse_an_inline_ascii_string_ifd_entry(self, byte_order):
+        bytes_ = {
+            BIG_ENDIAN: b"\x00\x01\x00\x02\x00\x00\x00\x048.3\x00",
+            LITTLE_ENDIAN: b"\x01\x00\x02\x00\x04\x00\x00\x008.3\x00",
+        }[byte_order]
+        stream_rdr = StreamReader(io.BytesIO(bytes_), byte_order)
+
+        ifd_entry = _AsciiIfdEntry.from_stream(stream_rdr, 0)
+
+        assert (ifd_entry.tag, ifd_entry.value) == (1, "8.3")
+
 
 class Describe_ShortIfdEntry:
     def it_can_parse_a_short_int_IFD_entry(self):