Loosely coupling a semantic brushstroke to a sentence

token_primer_test.py

import unittest
from token_primer import Tokens                                                                        
import re

class TokenRef(unittest.TestCase):

    def test_search_generators(self):

        # two canvasses, both pointing to a name with a colon
        canvas_A = "His hobby is: spinning gold.  His name is: Rumplestiltskin"
        canvas_B = "Hos hobby is: spinning gold.  His name is kind of funny: Rumplestiltskin"

        # semantic paint for either canvas:
        word_mask = re.compile(r'\b\w+\b')
        def is_name(candidate):
            try:
                if next(candidate.before()).suffix == ': ':
                    for preceeding_token in candidate.before():
                        if preceeding_token.string == 'name':
                            return True
            except StopIteration:
                pass
            return False

        # semantic paint for canvas A:
        A = Tokens(canvas_A, word_mask)
        name_A = list(filter(is_name, A))[0]

        # semantic paint for canvas B:
        B = Tokens(canvas_B, word_mask)
        name_B = list(filter(is_name, B))[0]

        # test code: the above paint should indicate the same name, despite the canvas differences

        self.assertEqual(name_A.string, "Rumplestiltskin")
        self.assertEqual(name_B.string, "Rumplestiltskin")

    def test_search_slices(self):

        # two canvasses, both pointint to a name with a colon
        canvas_A = "His hobby is: spinning gold.  His name is: Rumplestiltskin"
        canvas_B = "Hos hoppy is: spinning gold.  His name is kind of funny: Rumplestiltskin"

        word_mask = re.compile(r'\b\w+\b')

        # semantic paint for canvas A:
        A = Tokens(canvas_A, word_mask)
        name_A = next((y for (x, y) in zip(A['name'::'i'], A['name'::'e']) if x.suffix[0] == ':'))


        # semantic paint for canvas B:
        B = Tokens(canvas_B, word_mask)
        name_B = next((y for (x, y) in zip(A['name'::'i'], A['name'::'e']) if x.suffix[0] == ':'))

        # test code: the above paint should indicate the same name, despite the canvas differences

        self.assertEqual(name_A.string, "Rumplestiltskin")
        self.assertEqual(name_B.string, "Rumplestiltskin")