Load JPEG files independent of suffix

* cleaned up load()
* added dirty hack to load() (see comments)
* added test for loading tests/images/02.docx (a jpeg file with wrong suffix)
* all tests still working

piexif/_load.py

@@ -1,5 +1,5 @@
 import struct
-
+import sys
 from ._common import *
 from ._exif import *
 
@@ -17,13 +17,22 @@ def load(input_data):
     :return: Exif data({"0th":dict, "Exif":dict, "GPS":dict, "Interop":dict, "1st":dict, "thumbnail":bytes})
     :rtype: dict
     """
-    exif_dict = {"0th":{},
-                 "Exif":{},
-                 "GPS":{},
-                 "Interop":{},
-                 "1st":{},
-                 "thumbnail":None}
-    exifReader = _ExifReader(input_data)
+    exif_dict = {"0th": {},
+                 "Exif": {},
+                 "GPS": {},
+                 "Interop": {},
+                 "1st": {},
+                 "thumbnail": None}
+    # this is just a hack to make it work.
+    # The load function (IMHO) has to be redesigned because it has to deal with
+    # incoming bytes and strings (filenames)
+    if isinstance(input_data, str):
+        with open(input_data, 'rb') as f:
+            data = f.read()
+    else:
+        data = input_data
+
+    exifReader = _ExifReader(data)
     if exifReader.tiftag is None:
         return exif_dict
 
@@ -71,20 +80,6 @@ def __init__(self, data):
             self.tiftag = data
         elif data[0:4] == b"Exif":  # Exif
             self.tiftag = data[6:]
-        elif data[-4:].lower() in (".jpg", "jpeg", ".jpe", ".tif", "tiff"):
-            with open(data, 'rb') as f:
-                data = f.read()
-            if data[0:2] == b"\xff\xd8":  # JPEG
-                segments = split_into_segments(data)
-                app1 = get_exif_seg(segments)
-                if app1:
-                    self.tiftag = app1[10:]
-                else:
-                    self.tiftag = None
-            elif data[0:2] in (b"\x49\x49", b"\x4d4d"):  # TIFF
-                self.tiftag = data
-            else:
-                raise ValueError("Given file is neither JPEG nor TIFF.")
         else:
             raise ValueError("Given file is neither JPEG nor TIFF.")
 

tests/s_test.py

@@ -19,6 +19,7 @@
 INPUT_FILE2 = os.path.join("tests", "images", "02.jpg")
 INPUT_FILE_PEN = os.path.join("tests", "images", "r_pen.jpg")
 NOEXIF_FILE = os.path.join("tests", "images", "noexif.jpg")
+NODOCX_FILE = os.path.join("tests", "images", "02.docx")
 # JPEG without APP0 and APP1 segments
 NOAPP01_FILE = os.path.join("tests", "images", "noapp01.jpg")
 INPUT_FILE_TIF = os.path.join("tests", "images", "01.tif")
@@ -89,14 +90,23 @@ class ExifTests(unittest.TestCase):
 # load ------
     def test_no_exif_load(self):
         exif_dict = piexif.load(NOEXIF_FILE)
-        none_dict = {"0th":{},
-                     "Exif":{},
-                     "GPS":{},
-                     "Interop":{},
-                     "1st":{},
-                     "thumbnail":None}
+        none_dict = {"0th": {},
+                     "Exif": {},
+                     "GPS": {},
+                     "Interop": {},
+                     "1st": {},
+                     "thumbnail": None}
         self.assertEqual(exif_dict, none_dict)
 
+    def test_jpg_in_docx_load(self):
+        """
+        Loads an Image with jpg suffix and a copy with docx suffix and compares
+        exif data.
+        """
+        exif_dict_docx = piexif.load(NODOCX_FILE)
+        exif_dict_jpg = piexif.load(INPUT_FILE2)
+        self.assertEqual(exif_dict_docx, exif_dict_jpg)
+
     def test_load(self):
         files = glob.glob(os.path.join("tests", "images", "r_*.jpg"))
         for input_file in files:
@@ -113,6 +123,7 @@ def test_load_m(self):
         print("********************\n\n" + INPUT_FILE1 + "\n")
         self._compare_piexifDict_PILDict(exif, e)
 
+
     def test_load_tif(self):
         exif = piexif.load(INPUT_FILE_TIF)
         zeroth_ifd = exif["0th"]
@@ -148,7 +159,7 @@ def test_load_fail(self):
         with self.assertRaises(ValueError):
             exif = piexif.load(os.path.join("tests", "images", "notjpeg.jpg"))
 
-        with self.assertRaises(ValueError):
+        with self.assertRaises(FileNotFoundError):
             exif = piexif.load("foo")
 
     def test_load_from_pilImage_property(self):