Check if embedded RAF image is really a TIFF (backport #1796) (#1851)

* Check if embedded RAF image is really a TIFF

(cherry picked from commit be12ae6f9e2f053bf101902865deadfe7c937f8d)

* Clarify comment on old vs new RAF

Co-authored-by: Christoph Hasse <hassec@users.noreply.github.com>
(cherry picked from commit 30f39ac98c1671547aa0b4021081e374864f1f04)

* Check I/O read result on RAF inspection

Co-authored-by: Kevin Backhouse <kevinbackhouse@github.com>
(cherry picked from commit 09de8dc901a38555151bd87d89ada3bd931451ea)

* Add test

(cherry picked from commit 9c3db7f3c3cb48fe0669061e627174402f366816)

Co-authored-by: Miloš Komarčević <miloskomarcevic@aim.com>
Co-authored-by: Miloš Komarčević <4973094+kmilos@users.noreply.github.com>

src/rafimage.cpp

@@ -349,8 +349,17 @@ namespace Exiv2 {
         // sanity check.  Does tiff lie inside the file?
         enforce(Safe::add(tiffOffset, tiffLength) <= io_->size(), kerCorruptedMetadata);
 
-        DataBuf  tiff(tiffLength);
         if (io_->seek(tiffOffset, BasicIo::beg) != 0) throw Error(kerFailedToReadImageData);
+
+        // Check if this really is a tiff and then call the tiff parser.
+        // Check is needed because some older models just embed a raw bitstream.
+        // For those files we skip the parsing step. 
+        if (io_->read(readBuff, 4) != 4) { throw Error(kerFailedToReadImageData); }
+        io_->seek(-4, BasicIo::cur);
+        if (memcmp(readBuff, "\x49\x49\x2A\x00", 4) == 0 ||
+            memcmp(readBuff, "\x4D\x4D\x00\x2A", 4) == 0)
+        {
+            DataBuf  tiff(tiffLength);
             io_->read(tiff.pData_, tiff.size_);
 
             if (!io_->error() && !io_->eof())
@@ -361,6 +370,7 @@ namespace Exiv2 {
                                    tiff.pData_,
                                    tiff.size_);
             }
+        }
     } // RafImage::readMetadata
 
     void RafImage::writeMetadata()

tests/bugfixes/github/test_issue_1791.py

@@ -0,0 +1,24 @@
+# -*- coding: utf-8 -*-
+
+from system_tests import CaseMeta, path
+
+
+class RafNoTiffRegression(metaclass=CaseMeta):
+    """
+    Regression test for the bug described in:
+    https://github.com/Exiv2/exiv2/issues/1791
+    """
+    url = "https://github.com/Exiv2/exiv2/issues/1791"
+
+    filename1 = path("$data_path/issue_1791_old.raf")
+    filename2 = path("$data_path/issue_1791_new.raf")
+    commands = ["$exiv2 -pa $filename1", "$exiv2 -pa $filename2"]
+    stdout = ["""Exif.Image2.JPEGInterchangeFormat            Long        1  104
+Exif.Image2.JPEGInterchangeFormatLength      Long        1  12
+""",
+              """Exif.Image2.JPEGInterchangeFormat            Long        1  104
+Exif.Image2.JPEGInterchangeFormatLength      Long        1  12
+Exif.Image.NewSubfileType                    Long        1  Primary image
+"""]
+    stderr = ["", ""]
+    retval = [0, 0]