Using Linspace

Linspace Parameters

Linspace uses start and end parameters, with a third parameter showing how many intervals we want. By default the start and end are included in the resulting array. The size of the number of generated samples (num) is inclusive, so add 1 to the difference between end and start to calculate num:

linspace(start, end, num)

Next generate the array using start and end points of our coordinate pairs. Use numpy's linspace function, this provides values at equal spacing. Linspace can work with coordinate sets to produce equally spaced coordinates. Use two linspace functions, the first to work with the start of the coordinate pairs, the second to work with the finish of the coordinate pairs. Join the two arrays produced by both linspaces using concatenate. Finally sort this final array. Use argsort which produces results directly as the desired column of coordinates:

start_arr = linspace((30, 10), (30, 70), 3)
end_arr = linspace((30, 24), (30, 84), 3)

both_arr = concatenate([start_arr, end_arr], axis=0)

arr = int_(both_arr[both_arr[:, 1].argsort()])

Sorting in Numpy

Both np.sort and np.argsort can sort two dimensional arrays:

>>> arr = np.array([[30, 10],
    [30, 24],
    [30, 40],
    [30, 54],
    [30, 70],
    [30, 84]])

>>> np.sort(arr, axis=1)
array([[10, 30],
    [24, 30],
    [30, 40],
    [30, 54],
    [30, 70],
    [30, 84]])

Notice how the second row has been switched, as you see it is correct but not what is needed. Only one column or the other requires sorting:

>>> arr[arr[:,1].sort()] # sorting on the 2nd column
array([[[30, 10],
    [30, 24],
    [30, 40],
    [30, 54],
    [30, 70],
    [30, 84]]])

The sort order can be changed by inverting the array:

>>> arr[::-1]
array([[30, 84],
    [30, 70],
    [30, 54],
    [30, 40],
    [30, 24],
    [30, 10]])

Show/Hide Code 04line_start_finish_linspace.py

from PIL import Image, ImageDraw, ImageFont
import sys
sys.path.append('../dims')

from DimLinesPIL import angled_text, dims
from numpy import linspace, concatenate, int_, astype

Font = ImageFont.truetype('consola.ttf', 12)

w, h = 100, 100
image = Image.new('RGB', (w,h), '#FFFFDD')

'''
both_arr =[(30, 10),
(30, 24),
(30, 40),
(30, 54),
(30, 70),
(30, 84)]
'''
start_arr = linspace((30, 10), (30, 70), 3)
end_arr = linspace((30, 24), (30, 84), 3)

both_arr = concatenate((start_arr, end_arr)) #, axis=0
#print(both_arr)

fin_arr = int_(both_arr[both_arr[:, 1].argsort()])
'''
 fin_arr [[30 10]
 [30 24]
 [30 40]
 [30 54]
 [30 70]
 [30 84]]
both_arr is different has comma between values
'''

print('len(fin_arr)', len(fin_arr), fin_arr)
nr_lines = len(fin_arr)

draw = ImageDraw.Draw(image)

[draw.line([tuple(fin_arr[n]), tuple(fin_arr[n+1])], width=1, fill='black')
            for n in range(0, nr_lines, 2)]

#width, height = Font.getsize(str(arr[4]))
unused1, unused2, width, height = Font.getbbox('(30, 84)') #Font.getbbox(str(arr[4]))

for i in range(nr_lines):
    angled_text(image, (fin_arr[i][0] + 10 + width//2, fin_arr[i][1]),text=str(fin_arr[i]),
            angle=0, fill='black', font=Font)

for j in range(nr_lines - 1):
    #TypeError: only length-1 arrays can be converted to Python scalars
    dims(image, draw, fin_arr[j], fin_arr[j+1], (8, 2), text='15', font=Font, fill='lightgreen',
        textorient='horizontal')
image.show()

The result should look exactly the same as with our array.