Read some untiled tiffs using the tiff source

CHANGELOG.md

@@ -3,7 +3,8 @@
 ## 1.28.2
 
 ### Improvements
-- Improve uint16 image scaling  ([#1511](../../pull/1511))
+- Improve uint16 image scaling ([#1511](../../pull/1511))
+- Read some untiled tiffs using the tiff source ([#1512](../../pull/1512))
 
 ## 1.28.1
 

sources/tiff/large_image_source_tiff/__init__.py

@@ -73,6 +73,7 @@ class TiffFileTileSource(FileTileSource, metaclass=LruCacheMetaclass):
     }
 
     _maxAssociatedImageSize = 8192
+    _maxUntiledImage = 4096
 
     def __init__(self, path, **kwargs):  # noqa
         """
@@ -85,18 +86,18 @@ def __init__(self, path, **kwargs):  # noqa
 
         self._largeImagePath = str(self._getLargeImagePath())
 
+        lastException = None
         try:
             self._initWithTiffTools()
             return
         except Exception as exc:
             self.logger.debug('Cannot read with tifftools route; %r', exc)
+            lastException = exc
 
         alldir = []
         try:
             if hasattr(self, '_info'):
                 alldir = self._scanDirectories()
-            else:
-                lastException = 'Could not parse file with tifftools'
         except IOOpenTiffError:
             msg = 'File cannot be opened via tiff source.'
             raise TileSourceError(msg)
@@ -157,7 +158,7 @@ def __init__(self, path, **kwargs):  # noqa
             tifftools.constants.SampleFormat[sampleformat or 1].name,
             bitspersample,
         ))
-        self._bandCount = highest._tiffInfo.get('samplesperpixel')
+        self._bandCount = highest._tiffInfo.get('samplesperpixel', 1)
         # Sort the directories so that the highest resolution is the last one;
         # if a level is missing, put a None value in its place.
         self._tiffDirectories = [directories.get(key) for key in
@@ -252,8 +253,13 @@ def _levelFromIfd(self, ifd, baseifd):
         """
         sizeX = ifd['tags'][tifftools.Tag.ImageWidth.value]['data'][0]
         sizeY = ifd['tags'][tifftools.Tag.ImageLength.value]['data'][0]
-        tileWidth = baseifd['tags'][tifftools.Tag.TileWidth.value]['data'][0]
-        tileHeight = baseifd['tags'][tifftools.Tag.TileLength.value]['data'][0]
+        if tifftools.Tag.TileWidth.value in baseifd['tags']:
+            tileWidth = baseifd['tags'][tifftools.Tag.TileWidth.value]['data'][0]
+            tileHeight = baseifd['tags'][tifftools.Tag.TileLength.value]['data'][0]
+        else:
+            tileWidth = sizeX
+            tileHeight = baseifd['tags'][tifftools.Tag.RowsPerStrip.value]['data'][0]
+
         for tag in {
                 tifftools.Tag.SamplesPerPixel.value,
                 tifftools.Tag.BitsPerSample.value,
@@ -298,7 +304,7 @@ def _initWithTiffTools(self):  # noqa
         directories are the same size and format; all non-tiled directories are
         treated as associated images.
         """
-        dir0 = self.getTiffDir(0)
+        dir0 = self.getTiffDir(0, mustBeTiled=None)
         self.tileWidth = dir0.tileWidth
         self.tileHeight = dir0.tileHeight
         self.sizeX = dir0.imageWidth
@@ -312,12 +318,11 @@ def _initWithTiffTools(self):  # noqa
             tifftools.constants.SampleFormat[sampleformat or 1].name,
             bitspersample,
         ))
-        self._bandCount = dir0._tiffInfo.get('samplesperpixel')
+        self._bandCount = dir0._tiffInfo.get('samplesperpixel', 1)
         info = _cached_read_tiff(self._largeImagePath)
         self._info = info
         frames = []
         associated = []  # for now, a list of directories
-        curframe = -1
         for idx, ifd in enumerate(info['ifds']):
             # if not tiles, add to associated images
             if tifftools.Tag.tileWidth.value not in ifd['tags']:
@@ -326,7 +331,6 @@ def _initWithTiffTools(self):  # noqa
             level = self._levelFromIfd(ifd, info['ifds'][0])
             # if the same resolution as the main image, add a frame
             if level == self.levels - 1:
-                curframe += 1
                 frames.append({'dirs': [None] * self.levels})
                 frames[-1]['dirs'][-1] = (idx, 0)
                 try:
@@ -365,6 +369,35 @@ def _initWithTiffTools(self):  # noqa
                     else:
                         msg = 'Tile layers are in a surprising order'
                         raise TileSourceError(msg)
+        # If we have a single untiled ifd that is "small", use it
+        if tifftools.Tag.tileWidth.value not in info['ifds'][0]['tags']:
+            if (
+                self.sizeX > self._maxUntiledImage or self.sizeY > self._maxUntiledImage or
+                (len(info['ifds']) != 1 or tifftools.Tag.SubIfd.value in ifd['tags']) or
+                (tifftools.Tag.ImageDescription.value in ifd['tags'] and
+                 'ImageJ' in ifd['tags'][tifftools.Tag.ImageDescription.value]['data'])
+            ):
+                msg = 'A tiled TIFF is required.'
+                raise ValidationTiffError(msg)
+            associated = []
+            level = self._levelFromIfd(ifd, info['ifds'][0])
+            frames.append({'dirs': [None] * self.levels})
+            frames[-1]['dirs'][-1] = (idx, 0)
+            try:
+                frameMetadata = json.loads(
+                    ifd['tags'][tifftools.Tag.ImageDescription.value]['data'])
+                for key in {'channels', 'frame'}:
+                    if key in frameMetadata:
+                        frames[-1][key] = frameMetadata[key]
+            except Exception:
+                pass
+            if tifftools.Tag.ICCProfile.value in ifd['tags']:
+                if not hasattr(self, '_iccprofiles'):
+                    self._iccprofiles = []
+                while len(self._iccprofiles) < len(frames) - 1:
+                    self._iccprofiles.append(None)
+                self._iccprofiles.append(ifd['tags'][
+                    tifftools.Tag.ICCProfile.value]['data'])
         self._associatedImages = {}
         for dirNum in associated:
             self._addAssociatedImage(dirNum)

sources/tiff/large_image_source_tiff/tiff_reader.py

@@ -207,8 +207,8 @@ def _validate(self):  # noqa
         # the create_image.py script, such as flatten or colourspace.  These
         # should only be done if necessary, which would require the conversion
         # job to check output and perform subsequent processing as needed.
-        if (not self._tiffInfo.get('samplesperpixel') or
-                self._tiffInfo.get('samplesperpixel') < 1):
+        if (not self._tiffInfo.get('samplesperpixel', 1) or
+                self._tiffInfo.get('samplesperpixel', 1) < 1):
             msg = 'Only RGB and greyscale TIFF files are supported'
             raise ValidationTiffError(msg)
 
@@ -607,7 +607,7 @@ def _getUncompressedTile(self, tileNum):
             self._tiffInfo.get('bitspersample'),
             self._tiffInfo.get('sampleformat') if self._tiffInfo.get(
                 'sampleformat') is not None else libtiff_ctypes.SAMPLEFORMAT_UINT)
-        image = np.empty((th, tw, self._tiffInfo['samplesperpixel']),
+        image = np.empty((th, tw, self._tiffInfo.get('samplesperpixel', 1)),
                          dtype=_ctypesFormattbl[format])
         imageBuffer = image.ctypes.data_as(ctypes.POINTER(ctypes.c_char))
         if self._tiffInfo.get('istiled'):
@@ -635,7 +635,7 @@ def _getUncompressedTile(self, tileNum):
             raise IOTiffError(
                 'Read an unexpected number of bytes from an encoded tile' if readSize >= 0 else
                 'Failed to read from an encoded tile')
-        if (self._tiffInfo.get('samplesperpixel') == 3 and
+        if (self._tiffInfo.get('samplesperpixel', 1) == 3 and
                 self._tiffInfo.get('photometric') == libtiff_ctypes.PHOTOMETRIC_YCBCR):
             if self._tiffInfo.get('bitspersample') == 16:
                 image = np.floor_divide(image, 256).astype(np.uint8)

test/datastore.py

@@ -114,6 +114,8 @@
     'a131592c-a069-4aa7-8031-398654aa8a3d.dcm': 'sha512:99bd3da4b8e11ce7b4f7ed8a294ed0c37437320667a06c40c383f4b29be85fe8e6094043e0600bee0ba879f2401de4c57285800a4a23da2caf2eb94e5b847ee0',  # noqa
     # Synthetic newer ndpi with binary data and nonblank image labelled as RGB
     'synthetic_ndpi_2024.ndpi': 'sha512:192cdcf551a824277ef70358b8ed6225dca0c5b5d0817fe0d800b72638e0ad9416cd5bc40cb186219da60fd324b676d1b32cc05a276a33b71b485d665a31e42e',  # noqa
+    # Synthetic uint16 untiled tiff that can be read with the tiff source
+    'synthetic_untiled_16.tiff': 'sha512:f4773fcfa749ba9c2db25319c9e8ad8586dd148de4366dae0393a3703906dace9f11233eafdb24418b598170d6372ef1ca861bf8d7a8212cac21a0eb8636ee77',  # noqa
 }
 
 

test/test_source_base.py

@@ -62,7 +62,7 @@
     'openjpeg': {'read': r'\.(jp2)$'},
     'openslide': {
         'read': r'\.(ptif|svs|ndpi|tif.*|qptiff|dcm)$',
-        'noread': r'(oahu|DDX58_AXL|huron\.image2_jpeg2k|landcover_sample|d042-353\.crop|US_Geo\.|extraoverview|imagej|bad_axes)',  # noqa
+        'noread': r'(oahu|DDX58_AXL|huron\.image2_jpeg2k|landcover_sample|d042-353\.crop|US_Geo\.|extraoverview|imagej|bad_axes|synthetic_untiled)',  # noqa
         'skip': r'nokeyframe\.ome\.tiff$',
         'skipTiles': r'one_layer_missing',
     },
@@ -78,9 +78,7 @@
     'test': {'any': True, 'skipTiles': r''},
     'tiff': {
         'read': r'(\.(ptif|scn|svs|tif.*|qptiff)|[-0-9a-f]{36}\.dcm)$',
-        'noread': r'(oahu|DDX58_AXL|G10-3_pelvis_crop|'
-                  r'd042-353\.crop\.small\.float|landcover_sample|US_Geo\.|'
-                  r'imagej|bad_axes|nokeyframe\.ome\.tiff$)',
+        'noread': r'(DDX58_AXL|G10-3_pelvis_crop|landcover_sample|US_Geo\.|imagej)',
         'skipTiles': r'(sample_image\.ptif|one_layer_missing_tiles)'},
     'tifffile': {
         'read': r'',
@@ -91,7 +89,7 @@
     'vips': {
         'read': r'',
         'noread': r'\.(nc|nd2|yml|yaml|json|czi|png|svs|scn|zarr\.db|zarr\.zip)$',
-        'skipTiles': r'(sample_image\.ptif|one_layer_missing_tiles|JK-kidney_B-gal_H3_4C_1-500sec\.jp2|extraoverview)'  # noqa
+        'skipTiles': r'(sample_image\.ptif|one_layer_missing_tiles|JK-kidney_B-gal_H3_4C_1-500sec\.jp2|extraoverview|synthetic_untiled)'  # noqa
     },
     'zarr': {'read': r'\.(zarr|zgroup|zattrs|db|zarr\.zip)$'},
 }