Build a discrete colormap and normalization of the data.
(cmap, levels, extend, filled)
| 72 | |
| 73 | |
| 74 | def _build_discrete_cmap(cmap, levels, extend, filled): |
| 75 | """ |
| 76 | Build a discrete colormap and normalization of the data. |
| 77 | """ |
| 78 | import matplotlib as mpl |
| 79 | |
| 80 | if len(levels) == 1: |
| 81 | levels = [levels[0], levels[0]] |
| 82 | |
| 83 | if not filled: |
| 84 | # non-filled contour plots |
| 85 | extend = "max" |
| 86 | |
| 87 | if extend == "both": |
| 88 | ext_n = 2 |
| 89 | elif extend in ["min", "max"]: |
| 90 | ext_n = 1 |
| 91 | else: |
| 92 | ext_n = 0 |
| 93 | |
| 94 | n_colors = len(levels) + ext_n - 1 |
| 95 | pal = _color_palette(cmap, n_colors) |
| 96 | |
| 97 | new_cmap, cnorm = mpl.colors.from_levels_and_colors(levels, pal, extend=extend) |
| 98 | # copy the old cmap name, for easier testing |
| 99 | new_cmap.name = getattr(cmap, "name", cmap) |
| 100 | |
| 101 | # copy colors to use for bad, under, and over values in case they have been |
| 102 | # set to non-default values |
| 103 | if isinstance(cmap, mpl.colors.Colormap): |
| 104 | bad = cmap(np.nan) |
| 105 | |
| 106 | # Only update under and over if they were explicitly changed by the user |
| 107 | # (i.e. are different from the lowest or highest values in cmap). Otherwise |
| 108 | # leave unchanged so new_cmap uses its default values (its own lowest and |
| 109 | # highest values). |
| 110 | under = cmap(-np.inf) |
| 111 | if under == cmap(0): |
| 112 | under = None |
| 113 | |
| 114 | over = cmap(np.inf) |
| 115 | if over == cmap(cmap.N - 1): |
| 116 | over = None |
| 117 | |
| 118 | new_cmap = new_cmap.with_extremes(bad=bad, under=under, over=over) |
| 119 | |
| 120 | return new_cmap, cnorm |
| 121 | |
| 122 | |
| 123 | def _color_palette(cmap, n_colors): |
searching dependent graphs…