How to assign specific color to a category in an array

I want to show the land use classes of a region.

I have different raster files that contain information of land use. Each value of the pixel correspond to a specific land use class. I want to assign to each class a color as shown below. This is what I am doing:

from matplotlib.colors import ListedColormap
import rasterio
import numpy
import earthpy.plot as ep

cmap_values = [0, 11, 22, 33, 40, 55, 66, 77]
cmap_colors = ['white', ## 0: No Data
               'black', ## 11: urban 
               'darkorange', ## 22: Cropland
               'brown', ## 33: Pasture
               'darkgreen', ## 40: Forest
               'purple', ## 55: Grass
               'gray', ## 66: Other land
               'blue'  ## 77: water
              ] 
cmap = ListedColormap(cmap_colors)
value_text = ['No Data', 'Urban', 'Cropland', 'Pasture', 
             'Forest', 'Grass/shrubland', 
             'Other land', 'Water']

In some cases I do not have all the classes in the raster. For instance in the following example I have only

src = rasterio.open('myFile.tif')
data = src.read(1)
print(np.unique(data))
array([11, 22, 33, 40, 55, 77], dtype=uint8)

If I try to show the image it seems that it does not assign the right values to the colorbar but is assigns the color starting from the first value. In the figure below the color white is for the urban class, the black for the cropland, the darkorange for the pasture and so on. How can I keep the same color-class for each case that I encounter.

f,ax=plt.subplots()
im = ax.imshow(data, cmap=cmap)
ep.draw_legend(im, titles=value_text, classes=cmap_values)

You can change the actual data by mapping the values to the indices.

Create a mapping:

func = np.vectorize(lambda x: cmap_values.index(x))

Once you get your data, convert it with the mapping:

data = func(data)

In the imshow, add vmin and vmax, corresponding to the min and max indices of your colours (avoiding the need for your own normalization rule):

ax.imshow(data, cmap=cmap, vmin = 0, vmax = len(cmap_colours)-1)

Note that I am assuming no edge/weird cases, such as your data having a value outside of the ones listed in cmap_values. Though, a more complex function than the lambda one used, may be able to do the job, if cases like that exist.