Add sieve and fix a bug in it.

Sieving of images read directly from GDAL was broken, but is now
fixed, with tests.

docs/cli.rst

@@ -91,7 +91,7 @@ calc
 
 The calc command reads files as arrays, evaluates lisp-like expressions in
 their context, and writes the result as a new file. Members of the numpy
-module and arithmetic and logical operators are surfaced as builtin functions
+module and arithmetic and logical operators are available builtin functions
 and operators. It is intended for simple calculations; any calculations
 requiring multiple steps is better done in Python using the Rasterio and Numpy
 APIs.
@@ -123,8 +123,9 @@ collects a sequence of 2-D arrays into a 3-D array for output.
     $ rio calc "(asarray (take a 1) (* (take a 2) (/ (mean (take a 1)) (mean (take a 2)))) (* (take a 3) (/ (mean (take a 1)) (mean (take a 3)))))" \
     > --name a=tests/data/RGB.byte.tif --dtype ubyte /tmp/out.rgb.tif
 
-The command above is also an example of a calculation that is a bit beyond the design of the calc command and something that could be done much more efficiently in
-Python.
+The command above is also an example of a calculation that is far beyond the
+design of the calc command and something that could be done much more
+efficiently in Python.
 
 info
 ----

rasterio/_features.pyx

@@ -173,11 +173,12 @@ def _sieve(image, size, output, mask, connectivity):
         in_band = in_mem_ds.band
     elif isinstance(image, tuple):
         rdr = image.ds
-        hband = rdr.band(image.bidx)
+        in_band = rdr.band(image.bidx)
     else:
         raise ValueError("Invalid source image")
 
     if isinstance(output, np.ndarray):
+        log.debug("Output array: %r", output)
         out_mem_ds = InMemoryRaster(output)
         out_band = out_mem_ds.band
     elif isinstance(output, tuple):

rasterio/features.py

@@ -127,7 +127,7 @@ def sieve(image, size, out=None, output=None, mask=None, connectivity=4):
     if np.dtype(image.dtype).name not in valid_dtypes:
         valid_types_str = ', '.join(('rasterio.{0}'.format(t) for t
                                      in valid_dtypes))
-        raise ValueError('image dtype must be one of: %' % valid_types_str)
+        raise ValueError('image dtype must be one of: %s' % valid_types_str)
 
     if size <= 0:
         raise ValueError('size must be greater than 0')
@@ -155,7 +155,10 @@ def sieve(image, size, out=None, output=None, mask=None, connectivity=4):
     
     out = out if out is not None else output
     if out is None:
-        out = np.zeros_like(image)
+        if isinstance(image, tuple):
+            out = np.zeros(image.shape, image.dtype)
+        else:
+            out = np.zeros_like(image)
     else:
         if np.dtype(image.dtype).name != np.dtype(out.dtype).name:
             raise ValueError('output must match dtype of image')

rasterio/rio/calc.py

@@ -12,6 +12,8 @@ import snuggs
 
 import rasterio
 from rasterio.fill import fillnodata
+from rasterio.features import sieve
+
 from rasterio.rio.cli import cli
 
 
@@ -25,6 +27,14 @@ def get_bands(inputs, d, i=None):
         return [rasterio.band(src, i) for i in src.indexes]
 
 
+def read_array(ix, subix=None, dtype=None):
+    """Change the type of a read array"""
+    arr = snuggs._ctx.lookup(ix, subix)
+    if dtype:
+        arr = arr.astype(dtype)
+    return arr
+
+
 @cli.command(short_help="Raster data calculator.")
 @click.argument('command')
 @click.argument(
@@ -115,9 +125,11 @@ def calc(ctx, command, files, name, dtype):
                         src.read(), 'float64', copy=False)
 
             # Extend snuggs.
+            snuggs.func_map['read'] = read_array
             snuggs.func_map['band'] = lambda d, i: get_bands(inputs, d, i)
             snuggs.func_map['bands'] = lambda d: get_bands(inputs, d)
             snuggs.func_map['fillnodata'] = lambda *args: fillnodata(*args)
+            snuggs.func_map['sieve'] = lambda *args: sieve(*args)
 
             res = snuggs.eval(command, **ctxkwds)
 

requirements-dev.txt

@@ -5,7 +5,7 @@ cython==0.21.2
 delocate
 enum34
 numpy>=1.8.0
-snuggs>=1.0
+snuggs>=1.1
 pytest
 setuptools>=0.9.8
 wheel

tests/test_features_sieve.py

@@ -139,3 +139,10 @@ def test_dtypes():
                 image = numpy.zeros((rows, cols), dtype=dtype)
                 image[2:5, 2:5] = test_value
                 sieved_image = ftrz.sieve(image, 2)
+
+
+def test_sieve_shade():
+    with rasterio.drivers():
+        with rasterio.open('tests/data/shade.tif') as src:
+            sieved_image = ftrz.sieve(rasterio.band(src, 1), 42)
+            assert sieved_image.shape == (1024, 1024)

tests/test_rio_calc.py

@@ -154,3 +154,18 @@ def test_fillnodata_map(tmpdir):
         assert src.meta['dtype'] == 'uint8'
         data = src.read()
         assert round(data.mean(), 1) == 58.6
+
+
+def test_sieve_band(tmpdir):
+    outfile = str(tmpdir.join('out.tif'))
+    runner = CliRunner()
+    result = runner.invoke(calc, [
+                    '(sieve (band 1 1) 42)',
+                    '--dtype', 'uint8',
+                    'tests/data/shade.tif',
+                    outfile],
+                catch_exceptions=False)
+    assert result.exit_code == 0
+    with rasterio.open(outfile) as src:
+        assert src.count == 1
+        assert src.meta['dtype'] == 'uint8'