[RBT-273] Faulty exif fixes and code cleanup

piexif/_exif.py

@@ -1,3 +1,6 @@
+from struct import calcsize
+
+
 class TYPES:
     Byte = 1
     Ascii = 2
@@ -13,6 +16,29 @@ class TYPES:
     DFloat = 12
 
 
+TYPE_FORMAT = {
+    TYPES.Byte: 'B',
+    TYPES.Ascii: None,
+    TYPES.Short: 'H',
+    TYPES.Long: 'L',
+    TYPES.Rational: 'LL',
+    TYPES.SByte: 'b',
+    TYPES.Undefined: None,
+    TYPES.SShort: 'h',
+    TYPES.SLong: 'l',
+    TYPES.SRational: 'll',
+    TYPES.Float: 'f',
+    TYPES.DFloat: 'd',
+}
+
+
+TYPE_LENGTH = {
+    t: calcsize('=' + f)
+    for t, f in TYPE_FORMAT.items()
+    if f
+}
+
+
 SIMPLE_NUMERICS = [
     TYPES.Byte,
     TYPES.Short,

piexif/_load.py

@@ -103,35 +103,74 @@ def __init__(self, data):
                 else:
                     raise InvalidImageDataError("Given file is neither JPEG nor TIFF.")
 
+    def _unpack_from(self, format, pointer):
+        return unpack_from(self.endian_mark + format, self.tiftag, pointer)
+
+    def _read_tag(self, pointer):
+        tag, value_type, value_num = self._unpack_from("HHL", pointer)
+        # Treat unknown types as `Undefined`
+        value_length = TYPE_LENGTH.get(value_type, 1) * value_num
+        if value_length > 4:
+            data_pointer = self._unpack_from("L", pointer + 8)[0]
+        else:
+            data_pointer = pointer + 8
+
+        format = TYPE_FORMAT.get(value_type, None)
+
+        if format is None:
+            raw_value = self.tiftag[data_pointer:data_pointer+value_length]
+            # Ascii, Undefined and unknown types
+            if value_type == TYPES.Ascii:
+                # Crop ending zero
+                raw_value = raw_value.split(b'\0')[0]
+            values = (raw_value, )
+        else:
+            # Unpacked types
+            values = self._unpack_from(format * value_num, data_pointer)
+            # Collate rationals
+            if len(format) > 1:
+                stride = len(format)
+                values = tuple(
+                    values[i*stride:(i+1)*stride] for i in range(value_num)
+                )
+        return tag, value_type, tuple(values)
+
     def get_ifd_dict(self, pointer, ifd_name, read_unknown=False):
         ifd_dict = {}
-        tag_count = unpack_from(self.endian_mark + "H",
-                                self.tiftag, pointer)[0]
+        if pointer > len(self.tiftag) - 2:
+            return {}
+        tag_count = self._unpack_from("H", pointer)[0]
+        ifd_length = 2 + 12 * tag_count
+        if pointer > len(self.tiftag) - ifd_length:
+            # Truncate IFD
+            tag_count = (len(self.tiftag) - 2) // 12
         offset = pointer + 2
         if ifd_name in ["0th", "1st"]:
             t = "Image"
         else:
             t = ifd_name
         for x in range(tag_count):
             pointer = offset + 12 * x
-            tag, value_type, value_num = unpack_from(
-                self.endian_mark + "HHL", self.tiftag, pointer)
-            value = self.tiftag[pointer+8: pointer+12]
-            v_set = (value_type, value_num, value, tag)
+            try:
+                tag, value_type, values = self._read_tag(pointer)
+            except struct.error:
+                # Skip broken tags
+                continue
             if tag in TAGS[t]:
-                converted = self.convert_value(v_set)
                 expected_value_type = TAGS[t][tag]['type']
                 if value_type != expected_value_type:
                     try:
-                        converted = coerce(converted, value_type, expected_value_type)
+                        values = coerce(values, value_type, expected_value_type)
                     except ValueError:
                         # Skip if coercion failed
                         continue
-                if isinstance(converted, tuple) and (len(converted) == 1):
-                    converted = converted[0]
-                ifd_dict[tag] = converted
+                if len(values) == 1:
+                    values = values[0]
+                ifd_dict[tag] = values
             elif read_unknown:
-                ifd_dict[tag] = (v_set[0], v_set[1], v_set[2], self.tiftag)
+                value_num, = self._unpack_from("L", pointer + 4)
+                pointer_or_value = self.tiftag[pointer + 8: pointer + 12]
+                ifd_dict[tag] = value_type, value_num, pointer_or_value, self.tiftag
             else:
                 pass
 
@@ -140,96 +179,6 @@ def get_ifd_dict(self, pointer, ifd_name, read_unknown=False):
             ifd_dict["first_ifd_pointer"] = self.tiftag[pointer:pointer + 4]
         return ifd_dict
 
-    def convert_value(self, val):
-        data = None
-        t = val[0]
-        length = val[1]
-        value = val[2]
-
-        if t == TYPES.Byte: # BYTE
-            if length > 4:
-                pointer = unpack_from(self.endian_mark + "L", value)[0]
-                data = unpack_from("B" * length, self.tiftag, pointer)
-            else:
-                data = unpack_from("B" * length, value)
-        elif t == TYPES.Ascii: # ASCII
-            if length > 4:
-                pointer = unpack_from(self.endian_mark + "L", value)[0]
-                data = self.tiftag[pointer: pointer+length - 1]
-            else:
-                data = value[0: length - 1]
-        elif t == TYPES.Short: # SHORT
-            if length > 2:
-                pointer = unpack_from(self.endian_mark + "L", value)[0]
-                data = unpack_from(self.endian_mark + "H" * length,
-                                   self.tiftag, pointer)
-            else:
-                data = unpack_from(self.endian_mark + "H" * length, value)
-        elif t == TYPES.Long: # LONG
-            if length > 1:
-                pointer = unpack_from(self.endian_mark + "L", value)[0]
-                data = unpack_from(self.endian_mark + "L" * length,
-                                   self.tiftag, pointer)
-            else:
-                data = unpack_from(self.endian_mark + "L" * length, value)
-        elif t == TYPES.Rational: # RATIONAL
-            pointer = unpack_from(self.endian_mark + "L", value)[0]
-            data = tuple(
-                unpack_from(self.endian_mark + "LL",
-                            self.tiftag, pointer + x * 8)
-                for x in range(length)
-            )
-        elif t == TYPES.SByte: # SIGNED BYTES
-            if length > 4:
-                pointer = unpack_from(self.endian_mark + "L", value)[0]
-                data = unpack_from("b" * length, self.tiftag, pointer)
-            else:
-                data = unpack_from("b" * length, value)
-        elif t == TYPES.Undefined: # UNDEFINED BYTES
-            if length > 4:
-                pointer = unpack_from(self.endian_mark + "L", value)[0]
-                data = self.tiftag[pointer: pointer+length]
-            else:
-                data = value[0: length]
-        elif t == TYPES.SShort: # SIGNED SHORT
-            if length > 2:
-                pointer = unpack_from(self.endian_mark + "L", value)[0]
-                data = unpack_from(self.endian_mark + "h" * length,
-                                   self.tiftag, pointer)
-            else:
-                data = unpack_from(self.endian_mark + "h" * length, value)
-        elif t == TYPES.SLong: # SLONG
-            if length > 1:
-                pointer = unpack_from(self.endian_mark + "L", value)[0]
-                data = unpack_from(self.endian_mark + "l" * length,
-                                   self.tiftag, pointer)
-            else:
-                data = unpack_from(self.endian_mark + "l" * length, value)
-        elif t == TYPES.SRational: # SRATIONAL
-            pointer = unpack_from(self.endian_mark + "L", value)[0]
-            data = tuple(
-              unpack_from(self.endian_mark + "ll",
-                          self.tiftag, pointer + x * 8)
-              for x in range(length)
-            )
-        elif t == TYPES.Float: # FLOAT
-            if length > 1:
-                pointer = unpack_from(self.endian_mark + "L", value)[0]
-                data = unpack_from(self.endian_mark + "f" * length,
-                                   self.tiftag, pointer)
-            else:
-                data = unpack_from(self.endian_mark + "f" * length, value)
-        elif t == TYPES.DFloat: # DOUBLE
-            pointer = unpack_from(self.endian_mark + "L", value)[0]
-            data = unpack_from(self.endian_mark + "d" * length,
-                               self.tiftag, pointer)
-        else:
-            raise ValueError("Exif might be wrong. Got incorrect value " +
-                             "type to decode.\n" +
-                             "tag: " + str(val[3]) + "\ntype: " + str(t))
-
-        return data
-
 
 def _get_key_name_dict(exif_dict):
     new_dict = {
@@ -246,7 +195,7 @@ def coerce(value, type, target):
     if target == TYPES.Undefined:
         if type == TYPES.Byte:
             # Interpret numbers as byte values, to fit Pillow behaviour
-            return b''.join(min(x, 255).to_bytes(1, 'big') for x in value)
+            return ( b''.join(min(x, 255).to_bytes(1, 'big') for x in value), )
     elif target in SIMPLE_NUMERICS:
         if type in SIMPLE_NUMERICS:
             return value

tests/s_test.py

@@ -644,7 +644,7 @@ def _compare_piexifDict_PILDict(self, piexifDict, pilDict, p=True):
 class UTests(unittest.TestCase):
     def test_ExifReader_return_unknown(self):
         b1 = b"MM\x00\x2a\x00\x00\x00\x08"
-        b2 = b"\x00\x01" + b"\xff\xff\x00\x00\x00\x00" + b"\x00\x00\x00\x00"
+        b2 = b"\x00\x01" + b"\xff\xff\x00\x00\x00\x00\x00\x00" + b"\x00\x00\x00\x00"
         er = piexif._load._ExifReader(b1 + b2)
         if er.tiftag[0:2] == b"II":
             er.endian_mark = "<"
@@ -653,12 +653,23 @@ def test_ExifReader_return_unknown(self):
         ifd = er.get_ifd_dict(8, "0th", True)
         self.assertEqual(ifd[65535][0], 0)
         self.assertEqual(ifd[65535][1], 0)
-        self.assertEqual(ifd[65535][2], b"\x00\x00")
+        self.assertEqual(ifd[65535][2], b"\x00\x00\x00\x00")
 
-    def test_ExifReader_convert_value_fail(self):
-        er = piexif._load._ExifReader(I1)
-        with self.assertRaises(ValueError):
-            er.convert_value((None, None, None, None))
+    def test_truncated_ifd(self):
+        b1 = b"MM\x00\x2a\x00\x00\x00\x08"
+        b2 = b"\xff\xff" + b"\x00\x0b\x00\x02\x00\x00\x00\x04" + b"FOO\x00"
+        er = piexif._load._ExifReader(b1 + b2)
+        er.endian_mark = ">"
+        ifd = er.get_ifd_dict(8, "0th", True)
+        self.assertEqual(ifd[ImageIFD.ProcessingSoftware], b"FOO")
+
+    def test_ascii_zero(self):
+        b1 = b"MM\x00\x2a\x00\x00\x00\x08"
+        b2 = b"\x00\x01" + b"\x00\x0b\x00\x02\x00\x00\x00\x04" + b"F\x00OO"
+        er = piexif._load._ExifReader(b1 + b2)
+        er.endian_mark = ">"
+        ifd = er.get_ifd_dict(8, "0th", True)
+        self.assertEqual(ifd[ImageIFD.ProcessingSoftware], b"F")
 
     def test_split_into_segments_fail1(self):
         with self.assertRaises(InvalidImageDataError):