def test_indent_arrow(self):
data = ("q4 0 -> P1 R q3\n" " 1 -> Px q4\n\n" " x -> P0 L q3\n\n")
expected = [ # The space is an indent token
"q4",
"0",
"->",
"P1",
"R",
"q3",
"\n",
" ",
"1",
"->",
"Px",
"q4",
"\n",
" ",
"x",
"->",
"P0",
"L",
"q3",
"\n",
]
assert list(tokenizer(data)) == expected
def test_single_mconf_in_first_line(self):
data = (
"A \n"
" b -> Pxx2 R E alpha\n"
)
expected = ["A", "b", "->", "Pxx2", "R", "E", "alpha", "\n",]
assert list(tokenizer(data)) == expected
def test_empty_lines(self):
data = (
"\n\nq312 0 -> P1 R\n\n\n"
" P0 L L q2\n\n"
)
expected = ["q312", "0", "->", "P1", "R", "P0", "L", "L", "q2", "\n"]
assert list(tokenizer(data)) == expected
def test_simple_multiline_rule(self):
data = (
"q3 0 -> P1 R\n"
" P0 R q2\n"
)
expected = ["q3", "0", "->", "P1", "R", "P0", "R", "q2", "\n"]
assert list(tokenizer(data)) == expected
def test_three_full_line_rules(self):
data = ("q1 1 -> P0 R q2\n" "q2 0 -> P1 q2\n" "q2 1 -> E L q1\n")
expected = [
"q1",
"1",
"->",
"P0",
"R",
"q2",
"\n",
"q2",
"0",
"->",
"P1",
"q2",
"\n",
"q2",
"1",
"->",
"E",
"L",
"q1",
"\n",
]
assert list(tokenizer(data)) == expected
def test_indent_arrow(self):
data = (
"q4 0 -> P1 R q3\n"
" 1 -> Px q4\n\n"
" x -> P0 L q3\n\n"
)
expected = [ # The space is an indent token
"q4", "0", "->", "P1", "R", "q3", "\n",
" ", "1", "->", "Px", "q4", "\n",
" ", "x", "->", "P0", "L", "q3", "\n",
]
assert list(tokenizer(data)) == expected
def test_three_full_line_rules(self):
data = (
"q1 1 -> P0 R q2\n"
"q2 0 -> P1 q2\n"
"q2 1 -> E L q1\n"
)
expected = [
"q1", "1", "->", "P0", "R", "q2", "\n",
"q2", "0", "->", "P1", "q2", "\n",
"q2", "1", "->", "E", "L", "q1", "\n",
]
assert list(tokenizer(data)) == expected
def test_single_mconf_in_first_line(self):
data = ("A \n" " b -> Pxx2 R E alpha\n")
expected = [
"A",
"b",
"->",
"Pxx2",
"R",
"E",
"alpha",
"\n",
]
assert list(tokenizer(data)) == expected
def test_square_brackets_in_symbols(self):
data = (
"q1 \n"
" [0 4] -> Pinf R aba\n"
" [1 '2' 3] -> P-1 k\n"
" [xTj'c] -> P0\nç \n\n"
" - -> +"
)
expected = [ # The space is an indent token
"q1", "[", "0", "4", "]", "->", "Pinf", "R", "aba", "\n",
" ", "[", "1", "'2'", "3", "]", "->", "P-1", "k", "\n",
" ", "[", "xTj'c", "]", "->", "P0", "ç", "\n",
" ", "-", "->", "+", "\n",
]
assert list(tokenizer(data)) == expected
def test_square_brackets_in_symbols(self):
data = ("q1 \n"
" [0 4] -> Pinf R aba\n"
" [1 '2' 3] -> P-1 k\n"
" [xTj'c] -> P0\nç \n\n"
" - -> +")
expected = [ # The space is an indent token
"q1",
"[",
"0",
"4",
"]",
"->",
"Pinf",
"R",
"aba",
"\n",
" ",
"[",
"1",
"'2'",
"3",
"]",
"->",
"P-1",
"k",
"\n",
" ",
"[",
"xTj'c",
"]",
"->",
"P0",
"ç",
"\n",
" ",
"-",
"->",
"+",
"\n",
]
assert list(tokenizer(data)) == expected
def test_no_token_at_all(self): # Empty files have a single token
assert list(tokenizer("\n\n\n")) == ["\n"] == \
list(tokenizer("")) == list(tokenizer("\n \n \n"))
def test_simple_full_line_rule(self):
data = "q1 1 -> P0 R q2" # Newline token is always implicit
expected = ["q1", "1", "->", "P0", "R", "q2", "\n"]
assert list(tokenizer(data)) == expected
def test_empty_lines(self):
data = ("\n\nq312 0 -> P1 R\n\n\n" " P0 L L q2\n\n")
expected = ["q312", "0", "->", "P1", "R", "P0", "L", "L", "q2", "\n"]
assert list(tokenizer(data)) == expected
def test_no_token_at_all(self): # Empty files have a single token
assert list(tokenizer("\n\n\n")) == ["\n"] == \
list(tokenizer("")) == list(tokenizer("\n \n \n"))
def test_simple_multiline_rule(self):
data = ("q3 0 -> P1 R\n" " P0 R q2\n")
expected = ["q3", "0", "->", "P1", "R", "P0", "R", "q2", "\n"]
assert list(tokenizer(data)) == expected
def test_simple_full_line_rule(self):
data = "q1 1 -> P0 R q2" # Newline token is always implicit
expected = ["q1", "1", "->", "P0", "R", "q2", "\n"]
assert list(tokenizer(data)) == expected