WIP: Implement new style image class

Author: ksunden

data_prototype/image.py

@@ -0,0 +1,78 @@
+from typing import Sequence
+
+import matplotlib.cm as mcm
+import matplotlib.colors as mcolors
+import matplotlib.transforms as mtransforms
+import numpy as np
+
+from .artist import Artist
+from .description import Desc, desc_like
+from .conversion_edge import Edge, FuncEdge, Graph, CoordinateEdge, DefaultEdge
+
+
+class Image(Artist):
+    def __init__(self, container, edges=None, **kwargs):
+        super().__init__(container, edges, **kwargs)
+
+        strdesc = Desc((), str)
+        arrdesc = Desc(("M", "N"), np.float64)
+        normdesc = Desc((), mcolors.Normalize, "norm")
+        cmapdesc = Desc((), mcolors.Colormap, "cmap")
+
+        self._edges += [
+            CoordinateEdge.from_coords(
+                "A_pcolor", {"A": Desc(("M", "N"), np.float64)}, "data"
+            ),
+            CoordinateEdge.from_coords(
+                "A_rgb", {"A": Desc(("M", "N", 3), np.float64)}, "rgb"
+            ),
+            CoordinateEdge.from_coords(
+                "A_rgba", {"A": Desc(("M", "N", 4), np.float64)}, "rgba"
+            ),
+            FuncEdge.from_func(
+                "norm_obj",
+                lambda norm: mcm._auto_norm_from_scale(norm),
+                {"norm": desc_like(strdesc, coordinates="norm")},
+                {"norm": normdesc},
+            ),
+            FuncEdge.from_func(
+                "cmap_obj",
+                lambda cmap: mcm._colormaps[cmap],
+                {"cmap": desc_like(strdesc, coordinates="cmap")},
+                {"cmap": cmapdesc},
+            ),
+            FuncEdge.from_func(
+                "A_norm",
+                lambda A, norm: norm(A),
+                {"A": desc_like(arrdesc), "norm": normdesc},
+                {"A": desc_like(arrdesc, coordinates="norm")},
+            ),
+            FuncEdge.from_func(
+                "A_cmap",
+                lambda A, cmap: cmap(A),
+                {"A": desc_like(arrdesc, coordinates="norm"), "cmap": cmapdesc},
+                {"A": Desc(("M", "N", 4), np.float64, coordinates="rgba")},
+            ),
+            DefaultEdge.from_default_value("cmap_def", "cmap", strdesc, "viridis"),
+            DefaultEdge.from_default_value("norm_def", "norm", strdesc, "linear"),
+            CoordinateEdge.from_coords(
+                "A_rgba", {"A": Desc(("M", "N", 4), np.float64, "rgba")}, "display"
+            ),
+        ]
+
+    def draw(self, renderer, edges: Sequence[Edge]) -> None:
+        import matplotlib.pyplot as plt
+
+        g = Graph(list(edges) + self._edges)
+        g.visualize()
+        plt.show()
+        conv = g.evaluator(
+            self._container.describe(),
+            {"A": Desc(("M", "N", 4), np.float64, "display")},
+        )
+        query, _ = self._container.query(g)
+        image = conv.evaluate(query)["A"]
+
+        gc = renderer.new_gc()
+        renderer.draw_image(gc, image, mtransforms.IdentityTransform())
+        ...

examples/subsample.py

@@ -21,7 +21,7 @@
 import numpy as np
 
 from data_prototype.description import Desc, desc_like
-from data_prototype.wrappers import ImageWrapper
+from data_prototype.artist import Image, CompatibilityArtist as CA
 
 from skimage.transform import downscale_local_mean
 
@@ -74,10 +74,10 @@ def query(
 sub = Subsample()
 cmap = mpl.colormaps["coolwarm"]
 norm = Normalize(-2.2, 2.2)
-im = ImageWrapper(sub, cmap=cmap, norm=norm)
+im = Image(sub, cmap=cmap, norm=norm)
 
 fig, ax = plt.subplots()
-ax.add_artist(im)
+ax.add_artist(CA(im))
 ax.set_xlim(-3, 3)
 ax.set_ylim(-3, 3)
 plt.show()