def _update_cache(self, start: int, end: int):
fetch_set = set(set(range(start, end)) - self._cached_items)
while len(fetch_set) != 0:
fetch_range = next(range_in_set(fetch_set))
page = int(fetch_range.start / self._page_size) + 1
self._cached_items.update(
range(*self._fetch_range(page, self._page_size)))
fetch_set = set(set(range(start, end)) - self._cached_items)
def v_split(img):
"""
A helper function to crop images vertically.
:param img: input image.
:return cropped images.
"""
row_filled = {
row if LegacyRecognizer.row_not_empty(img, row) else None
for row in range(img.height)
}
row_filled.remove(None)
segments = list(range_in_set(row_filled))
top = min(segment.start for segment in segments)
bottom = max(segment.stop for segment in segments)
return img.crop((0, top, img.width, bottom))
def h_split(img):
"""
A helper function to split captcha into segments horizontally using projection.
:param img: input image.
:return segmented images.
"""
col_filled = {
col if LegacyRecognizer.col_not_empty(img, col) else None
for col in range(img.width)
}
col_filled.remove(None)
segments = range_in_set(col_filled)
rtn = []
for segment in segments:
rtn.append(
img.crop((segment.start, 0, segment.stop - 1, img.height)))
return rtn
def test_range_in_set():
assert len(list(range_in_set({}))) == 0
ranges = range_in_set({1, 3, 4, 5, 7, 8, 9, 10})
assert isinstance(ranges, Generator)
assert list(ranges) == [range(1, 2), range(3, 6), range(7, 11)]