Handle tiff files with RGB colorspace JPEG compression.

When a TIFF is marked as having a photometric space of RGB and uses JPEG
compression, the JPEG data does not have to encode its colorspace;
since, once a tile is extracted, we don't decode the tile unless
necessary, ensure that it is marked with the appropriate colorspace.

CHANGELOG.md

@@ -4,6 +4,7 @@
 
 ### Improvements
 - Better release bioformats resources (#502)
+- Better handling of tiff files with JPEG compression and RGB colorspace (#503)
 
 ## Version 1.3.2
 

sources/tiff/large_image_source_tiff/tiff_reader.py

@@ -548,7 +548,21 @@ def _getJpegFrame(self, tileNum, entire=False):
                 frameStartPos = frameBuffer.raw.find(b'\xff\xc2', 2, -2)
                 if frameStartPos == -1:
                     raise IOTiffException('Missing JPEG Start Of Frame marker')
-
+        # If the photometric value is RGB and the JPEG component ids are just
+        # 0, 1, 2, change the component ids to R, G, B to ensure color space
+        # information is preserved.
+        if self._tiffInfo.get('photometric') == libtiff_ctypes.PHOTOMETRIC_RGB:
+            sof = frameBuffer.raw.find(b'\xff\xc0')
+            if sof == -1:
+                sof = frameBuffer.raw.find(b'\xff\xc2')
+            sos = frameBuffer.raw.find(b'\xff\xda')
+            if (sof >= frameStartPos and sos >= frameStartPos and
+                    frameBuffer[sof + 2:sof + 4] == b'\x00\x11' and
+                    frameBuffer[sof + 10:sof + 19:3] == b'\x00\x01\x02' and
+                    frameBuffer[sos + 5:sos + 11:2] == b'\x00\x01\x02'):
+                for idx, val in enumerate(b'RGB'):
+                    frameBuffer[sof + 10 + idx * 3] = val
+                    frameBuffer[sos + 5 + idx * 2] = val
         # Strip the Start / End Of Image markers
         tileData = frameBuffer.raw[frameStartPos:-2]
         return tileData

test/data/TCGA-AA-A02O-11A-01-BS1.8b76f05c-4a8b-44ba-b581-6b8b4f437367.svs.sha512

@@ -0,0 +1 @@
+1b75a4ec911017aef5c885760a3c6575dacf5f8efb59fb0e011108dce85b1f4e97b8d358f3363c1f5ea6f1c3698f037554aec1620bbdd4cac54e3d5c9c1da1fd

test/test_source_tiff.py

@@ -653,3 +653,11 @@ def testInternalMetadata():
     source = large_image_source_tiff.TiffFileTileSource(imagePath)
     metadata = source.getInternalMetadata()
     assert 'xml' in metadata
+
+
+def testFromTiffRGBJPEG():
+    imagePath = utilities.externaldata(
+        'data/TCGA-AA-A02O-11A-01-BS1.8b76f05c-4a8b-44ba-b581-6b8b4f437367.svs.sha512')
+    source = large_image_source_tiff.TiffFileTileSource(imagePath)
+    tile = source.getSingleTile()
+    assert list(tile['tile'][0, 0]) == [243, 243, 243]