Mark geospatial tile sources.

This adds a check if a path is likely to be a geospatial file, and, if
so, prefers geospatial tile sources over non-geospatial sources.

This also adds a `canRead` function to the base library.

CHANGELOG.md

@@ -2,10 +2,14 @@
 
 ## Unreleased
 
+### Features
+- Added a `canRead` method to the core module (#512)
+
 ### Improvements
 - Better release bioformats resources (#502)
 - Better handling of tiff files with JPEG compression and RGB colorspace (#503)
 - The openjpeg tile source can decode with parallelism (#511)
+- Geospatial tile sources are preferred for geospatial files (#512)
 
 ### Bug Fixes
 - Harden updates of the item view after making a large image (#508)

large_image/__init__.py

@@ -19,7 +19,7 @@
 from pkg_resources import DistributionNotFound, get_distribution
 
 from . import tilesource  # noqa
-from .tilesource import getTileSource  # noqa
+from .tilesource import canRead, getTileSource  # noqa
 
 
 try:

large_image/tilesource/__init__.py

@@ -14,6 +14,30 @@
 AvailableTileSources = {}
 
 
+def isGeospatial(path):
+    """
+    Check if a path is likely to be a geospatial file.
+
+    :params path: The path to the file
+    :returns: True if geospatial.
+    """
+    try:
+        from osgeo import gdal
+        from osgeo import gdalconst
+    except ImportError:
+        # TODO: log a warning
+        return False
+    try:
+        ds = gdal.Open(path, gdalconst.GA_ReadOnly)
+    except Exception:
+        return False
+    if ds.GetProjection():
+        return True
+    if ds.GetDriver().ShortName in {'NITF', 'netCDF'}:
+        return True
+    return False
+
+
 def loadTileSources(entryPointName='large_image.source', sourceDict=AvailableTileSources):
     """
     Load all tilesources from entrypoints and add them to the
@@ -34,7 +58,7 @@ def loadTileSources(entryPointName='large_image.source', sourceDict=AvailableTil
             pass
 
 
-def getTileSourceFromDict(availableSources, pathOrUri, *args, **kwargs):
+def getSourceNameFromDict(availableSources, pathOrUri, *args, **kwargs):
     """
     Get a tile source based on a ordered dictionary of known sources and a path
     name or URI.  Additional parameters are passed to the tile source and can
@@ -43,12 +67,15 @@ def getTileSourceFromDict(availableSources, pathOrUri, *args, **kwargs):
     :param availableSources: an ordered dictionary of sources to try.
     :param pathOrUri: either a file path or a fixed source via
         large_image://<source>.
-    :returns: a tile source instance or and error.
+    :returns: the name of a tile source that can read the input, or None if
+        there is no such source.
     """
-    sourceObj = pathOrUri
     uriWithoutProtocol = pathOrUri.split('://', 1)[-1]
     isLargeImageUri = pathOrUri.startswith('large_image://')
     extensions = [ext.lower() for ext in os.path.basename(uriWithoutProtocol).split('.')[1:]]
+    properties = {
+        'geospatial': isGeospatial(pathOrUri),
+    }
     sourceList = []
     for sourceName in availableSources:
         sourceExtensions = availableSources[sourceName].extensions
@@ -60,10 +87,29 @@ def getTileSourceFromDict(availableSources, pathOrUri, *args, **kwargs):
             priority = SourcePriority.NAMED
         if priority >= SourcePriority.MANUAL:
             continue
-        sourceList.append((priority, sourceName))
-    for _priority, sourceName in sorted(sourceList):
-        if availableSources[sourceName].canRead(sourceObj, *args, **kwargs):
-            return availableSources[sourceName](sourceObj, *args, **kwargs)
+        propertiesClash = any(
+            getattr(availableSources[sourceName], k, False) != v
+            for k, v in properties.items())
+        sourceList.append((propertiesClash, priority, sourceName))
+    for _clash, _priority, sourceName in sorted(sourceList):
+        if availableSources[sourceName].canRead(pathOrUri, *args, **kwargs):
+            return sourceName
+
+
+def getTileSourceFromDict(availableSources, pathOrUri, *args, **kwargs):
+    """
+    Get a tile source based on a ordered dictionary of known sources and a path
+    name or URI.  Additional parameters are passed to the tile source and can
+    be used for properties such as encoding.
+
+    :param availableSources: an ordered dictionary of sources to try.
+    :param pathOrUri: either a file path or a fixed source via
+        large_image://<source>.
+    :returns: a tile source instance or and error.
+    """
+    sourceName = getSourceNameFromDict(availableSources, pathOrUri, *args, **kwargs)
+    if sourceName:
+        return availableSources[sourceName](pathOrUri, *args, **kwargs)
     raise TileSourceException('No available tilesource for %s' % pathOrUri)
 
 
@@ -79,10 +125,24 @@ def getTileSource(*args, **kwargs):
     return getTileSourceFromDict(AvailableTileSources, *args, **kwargs)
 
 
+def canRead(*args, **kwargs):
+    """
+    Check if large_image can read a path or uri.
+
+    :returns: True if any appropriate source reports it can read the path or
+        uri.
+    """
+    if not len(AvailableTileSources):
+        loadTileSources()
+    if getSourceNameFromDict(AvailableTileSources, *args, **kwargs):
+        return True
+    return False
+
+
 __all__ = [
     'TileSource', 'FileTileSource',
     'exceptions', 'TileGeneralException', 'TileSourceException', 'TileSourceAssetstoreException',
     'TileOutputMimeTypes', 'TILE_FORMAT_IMAGE', 'TILE_FORMAT_PIL', 'TILE_FORMAT_NUMPY',
-    'AvailableTileSources', 'getTileSource', 'nearPowerOfTwo',
+    'AvailableTileSources', 'getTileSource', 'canRead', 'getSourceNameFromDict', 'nearPowerOfTwo',
     'etreeToDict', 'dictToEtree',
 ]

sources/gdal/large_image_source_gdal/__init__.py

@@ -85,6 +85,7 @@ class GDALFileTileSource(FileTileSource):
         'image/tiff': SourcePriority.LOW,
         'image/x-tiff': SourcePriority.LOW,
     }
+    geospatial = True
 
     def __init__(self, path, projection=None, unitsPerPixel=None, **kwargs):
         """

sources/mapnik/large_image_source_mapnik/__init__.py

@@ -65,6 +65,7 @@ class MapnikFileTileSource(GDALFileTileSource):
         'image/tiff': SourcePriority.LOWER,
         'image/x-tiff': SourcePriority.LOWER,
     }
+    geospatial = True
 
     def __init__(self, path, projection=None, unitsPerPixel=None, **kwargs):
         """

test/data/04091217_ruc.nc.sha512

@@ -0,0 +1 @@
+3380c8de64afc46a5cfe764921e0fd5582380b1065754a6e6c4fa506625ee26edb1637aeda59c1d2a2dc245364b191563d8572488c32dcafe9e706e208fd9939

test/test_source_base.py

@@ -1,7 +1,11 @@
 # -*- coding: utf-8 -*-
+import os
 
+import large_image
 from large_image.tilesource import nearPowerOfTwo
 
+from . import utilities
+
 
 def testNearPowerOfTwo():
     assert nearPowerOfTwo(45808, 11456)
@@ -11,3 +15,12 @@ def testNearPowerOfTwo():
     assert not nearPowerOfTwo(45808, 11400, 0.005)
     assert nearPowerOfTwo(45808, 11500)
     assert not nearPowerOfTwo(45808, 11500, 0.005)
+
+
+def testCanRead():
+    testDir = os.path.dirname(os.path.realpath(__file__))
+    imagePath = os.path.join(testDir, 'test_files', 'yb10kx5k.png')
+    assert large_image.canRead(imagePath) is False
+
+    imagePath = utilities.externaldata('data/sample_image.ptif.sha512')
+    assert large_image.canRead(imagePath) is True

test/test_source_mapnik.py

@@ -8,6 +8,7 @@
 import pytest
 import six
 
+import large_image
 from large_image.exceptions import TileSourceException
 
 import large_image_source_mapnik
@@ -332,3 +333,36 @@ def testGuardAgainstBadLatLong():
     assert bounds['xmax'] == 179.99583333
     assert bounds['ymin'] == -89.99583333
     assert bounds['ymax'] == 90
+
+
+def testTileFromNetCDF():
+    imagePath = utilities.externaldata('data/04091217_ruc.nc.sha512')
+    source = large_image_source_mapnik.MapnikFileTileSource(imagePath)
+    tileMetadata = source.getMetadata()
+
+    assert tileMetadata['tileWidth'] == 256
+    assert tileMetadata['tileHeight'] == 256
+    assert tileMetadata['sizeX'] == 93
+    assert tileMetadata['sizeY'] == 65
+    assert tileMetadata['levels'] == 1
+    assert tileMetadata['bounds']['srs'].strip() == '+init=epsg:4326'
+    assert tileMetadata['geospatial']
+
+    # Getting the metadata with a specified projection will be different
+    source = large_image_source_mapnik.MapnikFileTileSource(
+        imagePath, projection='EPSG:3857')
+    tileMetadata = source.getMetadata()
+
+    assert tileMetadata['tileWidth'] == 256
+    assert tileMetadata['tileHeight'] == 256
+    assert tileMetadata['sizeX'] == 512
+    assert tileMetadata['sizeY'] == 512
+    assert tileMetadata['levels'] == 2
+    assert tileMetadata['bounds']['srs'] == '+init=epsg:3857'
+    assert tileMetadata['geospatial']
+
+
+def testTileSourceFromNetCDF():
+    imagePath = utilities.externaldata('data/04091217_ruc.nc.sha512')
+    ts = large_image.getTileSource(imagePath)
+    assert 'mapnik' in ts.name