def test_get_characters_omits_error_nodes():
with open(get_test_path(os.path.join("mathml",
"x_plus_error.xml"))) as mathml_file:
mathml = mathml_file.read()
characters = get_characters(mathml)
assert len(characters) == 1
assert Character("x", 0, 0, 1) in characters
def test_parse_equation():
with open(
get_test_path(os.path.join("mathml", "x_sub_t_sub_i.xml"))
) as mathml_file:
mathml = mathml_file.read()
symbols = parse_equation(mathml)
assert len(symbols) == 5
x_sub_t_sub_i = list(
filter(lambda s: "msub" in str(s.element) and "x" in str(s.element), symbols)
)[0]
t_sub_i = list(
filter(lambda s: "msub" in str(s.element) and s is not x_sub_t_sub_i, symbols)
)[0]
x = list(filter(lambda s: str(s.element) == "<mi>x</mi>", symbols))[0]
t = list(filter(lambda s: str(s.element) == "<mi>t</mi>", symbols))[0]
i = list(filter(lambda s: str(s.element) == "<mi>i</mi>", symbols))[0]
assert len(x_sub_t_sub_i.children) == 2
assert x in x_sub_t_sub_i.children
assert t_sub_i in x_sub_t_sub_i.children
assert len(t_sub_i.children) == 2
assert t in t_sub_i.children
assert i in t_sub_i.children
def test_get_characters():
with open(get_test_path(os.path.join("mathml",
"x_plus_y.xml"))) as mathml_file:
mathml = mathml_file.read()
characters = get_characters(mathml)
assert len(characters) == 2
assert Character("x", 0, 0, 1) in characters
assert Character("y", 2, 4, 5) in characters
def load_fragment_tag(filename: str) -> Tag:
" Read a MathML fragment from file and return a BeautifulSoup tag for it. "
with open(get_test_path(os.path.join("mathml-fragments", filename))) as file_:
mathml = file_.read()
# 'body.next' is used as the parser adds in 'html' and 'body' tags; this return just the child
# node of the body (the original node we were parsing)
return BeautifulSoup(mathml, "lxml").body.next
def test_find_bounding_box():
image = cv2.imread(
get_test_path(os.path.join("images", "rectangle_hue_40.png")))
boxes = find_boxes_with_color(image, (40 / 360.0))
assert len(boxes) == 1
assert boxes[0].left == 10
assert boxes[0].top == 10
assert boxes[0].width == 10
assert boxes[0].height == 10
def test_get_symbols_omits_error_nodes():
with open(get_test_path(os.path.join("mathml",
"x_plus_error.xml"))) as mathml_file:
mathml = mathml_file.read()
symbols = get_symbols(mathml)
assert len(symbols) == 1
x = list(filter(lambda s: s.characters == [0], symbols))[0]
assert x.mathml.strip() == "<mi>x</mi>"
def test_parse_consecutive_mi():
with open(get_test_path(os.path.join("mathml",
"relu.xml"))) as mathml_file:
mathml = mathml_file.read()
symbols = get_symbols(mathml)
assert len(symbols) == 1
relu = symbols[0]
assert len(relu.children) == 0
assert len(relu.characters) == 4
assert relu.mathml == "<mi>ReLU</mi>"
def test_get_symbols_for_subscript():
with open(get_test_path(os.path.join("mathml",
"x_sub_i.xml"))) as mathml_file:
mathml = mathml_file.read()
symbols = get_symbols(mathml)
assert len(symbols) == 3
x_sub_i = list(filter(lambda s: "msub" in s.mathml, symbols))[0]
assert re.search(r"<msub>\s*<mi>x</mi>\s*<mi>i</mi>\s*</msub>",
x_sub_i.mathml)
assert len(x_sub_i.characters) == 2
assert 0 in x_sub_i.characters
assert 1 in x_sub_i.characters
def test_get_symbols_for_characters():
with open(get_test_path(os.path.join("mathml",
"x_plus_y.xml"))) as mathml_file:
mathml = mathml_file.read()
symbols = get_symbols(mathml)
assert len(symbols) == 2
x = list(filter(lambda s: s.characters == [0], symbols))[0]
assert x.mathml.strip() == "<mi>x</mi>"
y = list(filter(lambda s: s.characters == [2], symbols))[0]
assert y.mathml.strip() == "<mi>y</mi>"
def test_get_symbol_sub_super_with_numeric_child():
with open(get_test_path(os.path.join(
"mathml", "x_sub_four_sub_three.xml"))) as mathml_file:
mathml = mathml_file.read()
symbols = get_symbols(mathml)
assert len(symbols) == 3
x_sub_four_sub_three = list(
filter(lambda s: "msub" in s.mathml and "x" in s.mathml, symbols))[0]
x = list(filter(lambda s: s.mathml == "<mi>x</mi>", symbols))[0]
four_sub_three = list(
filter(lambda s: s.mathml == "<msub>\n<mn>4</mn>\n<mn>3</mn>\n</msub>",
symbols))[0]
assert len(x_sub_four_sub_three.children) == 2
assert x in x_sub_four_sub_three.children
assert four_sub_three in x_sub_four_sub_three.children
def test_split_into_lines():
image = cv2.imread(
get_test_path(os.path.join("images", "two_lines_hue_40.png")))
boxes = find_boxes_with_color(image, (40 / 360.0))
assert len(boxes) == 2
# The two boxes that are horizontally aligned but separated get merged.
assert boxes[0].left == 5
assert boxes[0].top == 10
assert boxes[0].width == 30
assert boxes[0].height == 10
# The box at a different vertical location doesn't get merged.
assert boxes[1].left == 5
assert boxes[1].top == 21
assert boxes[1].width == 10
assert boxes[1].height == 10
def test_find_boxes_within_masks():
image = cv2.imread(
get_test_path(os.path.join("images", "two_lines_hue_40.png")))
boxes = find_boxes_with_color(
image,
(40 / 360.0),
masks=((
Rectangle(left=5, top=10, width=10, height=10),
Rectangle(left=25, top=10, width=10, height=10),
)),
)
# The masks will split up two colorized regions that would normally be merged.
assert len(boxes) == 2
assert boxes[0].left == 5
assert boxes[0].top == 10
assert boxes[0].width == 10
assert boxes[0].height == 10
assert boxes[1].left == 25
assert boxes[1].top == 10
assert boxes[1].width == 10
assert boxes[1].height == 10
def test_detect_black_pixels():
image = cv2.imread(
get_test_path(os.path.join("images", "black_letter_40.png")))
assert contains_black_pixels(image)
def test_no_detect_saturated_pixels_as_black():
image = cv2.imread(
get_test_path(os.path.join("images", "yellow_letter_40.png")))
assert not contains_black_pixels(image)
def test_no_detect_black_pixels_in_blank_image():
image = cv2.imread(get_test_path(os.path.join("images", "blank_40.png")))
assert not contains_black_pixels(image)
def test_no_detect_shift_for_other_hue():
black = cv2.imread(get_test_path(os.path.join("images", "black_letter_40.png")))
pink_shifted = cv2.imread(
get_test_path(os.path.join("images", "pink_letter_40_shifted.png"))
)
assert not has_hue_shifted(black, pink_shifted, YELLOW_HUE)
def test_detect_shift():
black = cv2.imread(get_test_path(os.path.join("images", "black_letter_40.png")))
yellow_shifted = cv2.imread(
get_test_path(os.path.join("images", "yellow_letter_40_shifted.png"))
)
assert has_hue_shifted(black, yellow_shifted, YELLOW_HUE)
def test_no_detect_shift_when_pixels_aligned():
black = cv2.imread(get_test_path(os.path.join("images", "black_letter_40.png")))
yellow = cv2.imread(get_test_path(os.path.join("images", "yellow_letter_40.png")))
assert not has_hue_shifted(black, yellow, YELLOW_HUE)
