Merge pull request #835 from girder/validate-cog

Add validateCOG method to GDALFileTileSource

large_image/exceptions.py

@@ -17,6 +17,10 @@ class TileSourceXYZRangeError(TileSourceError):
     pass
 
 
+class TileSourceInefficientError(TileSourceError):
+    pass
+
+
 class TileSourceFileNotFoundError(TileSourceError, FileNotFoundError):
     def __init__(self, *args, **kwargs):
         return super().__init__(errno.ENOENT, *args, **kwargs)

sources/gdal/large_image_source_gdal/__init__.py

@@ -41,7 +41,9 @@
 from large_image.constants import (TILE_FORMAT_IMAGE, TILE_FORMAT_NUMPY,
                                    TILE_FORMAT_PIL, SourcePriority,
                                    TileInputUnits, TileOutputMimeTypes)
-from large_image.exceptions import TileSourceError, TileSourceFileNotFoundError
+from large_image.exceptions import (TileSourceError,
+                                    TileSourceFileNotFoundError,
+                                    TileSourceInefficientError)
 from large_image.tilesource import FileTileSource
 from large_image.tilesource.utilities import getPaletteColors
 
@@ -1185,6 +1187,44 @@ def getRegion(self, format=(TILE_FORMAT_IMAGE, ), **kwargs):
             raise exc
         return pathlib.Path(outputPath), TileOutputMimeTypes['TILED']
 
+    def validateCOG(self, check_tiled=True, full_check=False, strict=True, warn=True):
+        """Check if this image is a valid Cloud Optimized GeoTiff.
+
+        This will raise a :class:`large_image.exceptions.TileSourceInefficientError`
+        if not a valid Cloud Optimized GeoTiff. Otherwise, returns True.
+
+        Requires the ``osgeo_utils`` package.
+
+        Parameters
+        ----------
+        check_tiled : bool
+            Set to False to ignore missing tiling.
+        full_check : bool
+            Set to True to check tile/strip leader/trailer bytes.
+            Might be slow on remote files
+        strict : bool
+            Enforce warnings as exceptions. Set to False to only warn and not
+            raise exceptions.
+        warn : bool
+            Log any warnings
+
+        """
+        from osgeo_utils.samples.validate_cloud_optimized_geotiff import validate
+
+        warnings, errors, details = validate(
+            self._largeImagePath,
+            check_tiled=check_tiled,
+            full_check=full_check
+        )
+        if errors:
+            raise TileSourceInefficientError(errors)
+        if strict and warnings:
+            raise TileSourceInefficientError(warnings)
+        if warn:
+            for warning in warnings:
+                self.logger.warning(warning)
+        return True
+
 
 def open(*args, **kwargs):
     """

test/datastore.py

@@ -75,6 +75,11 @@
     # Multi source file using different sources
     # Source: manually generated.
     'multi_source.yml': 'sha512:81d7768b06eca6903082daa5b91706beaac8557ba4cede7f826524303df69a33478d6bb205c56af7ee2b45cd7d75897cc4b5704f743ddbf71bb3537ed3b9e8a8',  # noqa
+    # Geospatial tiff - not cloud optimized
+    'TC_NG_SFBay_US_Geo.tif': 'sha512:da2e66528f77a5e10af5de9e496074b77277c3da81dafc69790189510e5a7e18dba9e966329d36c979f1b547f0d36a82fbc4cfccc65ae9ef9e2747b5a9ee77b0',  # noqa
+    # Geospatial tiff - cloud optimized
+    'TC_NG_SFBay_US_Geo_COG.tif': 'sha512:5e56cdb8fb1a02615698a153862c10d5292b1ad42836a6e8bce5627e93a387dc0d3c9b6cfbd539796500bc2d3e23eafd07550f8c214e9348880bbbc6b3b0ea0c',  # noqa
+
 }
 
 

test/test_source_base.py

@@ -42,16 +42,16 @@
     'openjpeg': {'read': r'\.(jp2)$'},
     'openslide': {
         'read': r'\.(ptif|svs|tif.*)$',
-        'noread': r'(oahu|DDX58_AXL|huron\.image2_jpeg2k|landcover_sample|d042-353\.crop)',
+        'noread': r'(oahu|DDX58_AXL|huron\.image2_jpeg2k|landcover_sample|d042-353\.crop|US_Geo\.)',
         'skipTiles': r'one_layer_missing'},
     'pil': {
         'read': r'\.(jpeg|png|tif.*)$',
-        'noread': r'(G10-3|JK-kidney|d042-353|huron|one_layer_missing)'},
+        'noread': r'(G10-3|JK-kidney|d042-353|huron|one_layer_missing|US_Geo)'},
     'test': {'any': True, 'skipTiles': r''},
     'tiff': {
         'read': r'\.(ptif|scn|svs|tif.*)$',
         'noread': r'(oahu|DDX58_AXL|G10-3_pelvis_crop|'
-                  r'd042-353\.crop\.small\.float|landcover_sample)',
+                  r'd042-353\.crop\.small\.float|landcover_sample|US_Geo\.)',
         'skipTiles': r'(sample_image\.ptif|one_layer_missing_tiles)'},
     'vips': {
         'read': r'',
@@ -66,7 +66,7 @@
     # Python 3.6 has an older version of PIL that won't read some of the
     # ome.tif files.
     SourceAndFiles['pil']['noread'] = \
-        r'(G10-3|JK-kidney|d042-353|huron|sample.*ome|one_layer_missing)'
+        r'(G10-3|JK-kidney|d042-353|huron|sample.*ome|one_layer_missing|US_Geo)'
 
 
 def testNearPowerOfTwo():

test/test_source_gdal.py

@@ -10,7 +10,7 @@
 import pytest
 
 from large_image import constants
-from large_image.exceptions import TileSourceError
+from large_image.exceptions import TileSourceError, TileSourceInefficientError
 
 from . import utilities
 from .datastore import datastore
@@ -550,3 +550,15 @@ def testHttpVfsPath():
     assert tileMetadata['bounds']['ymin'] == pytest.approx(4876273, 1)
     assert tileMetadata['bounds']['srs'] == 'epsg:3857'
     assert tileMetadata['geospatial']
+
+
+def testVfsCogValidation():
+    imagePath = datastore.get_url('TC_NG_SFBay_US_Geo_COG.tif')
+    source = large_image_source_gdal.open(
+        imagePath, projection='EPSG:3857', encoding='PNG')
+    assert source.validateCOG()
+    imagePath = datastore.get_url('TC_NG_SFBay_US_Geo.tif')
+    source = large_image_source_gdal.open(
+        imagePath, projection='EPSG:3857', encoding='PNG')
+    with pytest.raises(TileSourceInefficientError):
+        source.validateCOG()