mirror of
https://github.com/python-pillow/Pillow.git
synced 2024-12-28 10:56:18 +03:00
4cd4adddc3
Follow Python's file object semantics. User code is responsible for closing resources (usually through a context manager) in a deterministic way. To achieve this, remove __del__ functions. These functions used to closed open file handlers in an attempt to silence Python ResourceWarnings. However, using __del__ has the following drawbacks: - __del__ isn't called until the object's reference count reaches 0. Therefore, resource handlers remain open or in use longer than necessary. - The __del__ method isn't guaranteed to execute on system exit. See the Python documentation: https://docs.python.org/3/reference/datamodel.html#object.__del__ > It is not guaranteed that __del__() methods are called for objects > that still exist when the interpreter exits. - Exceptions that occur inside __del__ are ignored instead of raised. This has the potential of hiding bugs. This is also in the Python documentation: > Warning: Due to the precarious circumstances under which __del__() > methods are invoked, exceptions that occur during their execution > are ignored, and a warning is printed to sys.stderr instead. Instead, always close resource handlers when they are no longer in use. This will close the file handler at a specified point in the user's code and not wait until the interpreter chooses to. It is always guaranteed to run. And, if an exception occurs while closing the file handler, the bug will not be ignored. Now, when code receives a ResourceWarning, it will highlight an area that is mishandling resources. It should not simply be silenced, but fixed by closing resources with a context manager. All warnings that were emitted during tests have been cleaned up. To enable warnings, I passed the `-Wa` CLI option to Python. This exposed some mishandling of resources in ImageFile.__init__() and SpiderImagePlugin.loadImageSeries(), they too were fixed.
287 lines
9.4 KiB
Python
287 lines
9.4 KiB
Python
import math
|
|
|
|
from PIL import Image
|
|
|
|
from .helper import PillowTestCase, hopper
|
|
|
|
|
|
class TestImageTransform(PillowTestCase):
|
|
def test_sanity(self):
|
|
from PIL import ImageTransform
|
|
|
|
im = Image.new("L", (100, 100))
|
|
|
|
seq = tuple(range(10))
|
|
|
|
transform = ImageTransform.AffineTransform(seq[:6])
|
|
im.transform((100, 100), transform)
|
|
transform = ImageTransform.ExtentTransform(seq[:4])
|
|
im.transform((100, 100), transform)
|
|
transform = ImageTransform.QuadTransform(seq[:8])
|
|
im.transform((100, 100), transform)
|
|
transform = ImageTransform.MeshTransform([(seq[:4], seq[:8])])
|
|
im.transform((100, 100), transform)
|
|
|
|
def test_extent(self):
|
|
im = hopper("RGB")
|
|
(w, h) = im.size
|
|
# fmt: off
|
|
transformed = im.transform(im.size, Image.EXTENT,
|
|
(0, 0,
|
|
w//2, h//2), # ul -> lr
|
|
Image.BILINEAR)
|
|
# fmt: on
|
|
|
|
scaled = im.resize((w * 2, h * 2), Image.BILINEAR).crop((0, 0, w, h))
|
|
|
|
# undone -- precision?
|
|
self.assert_image_similar(transformed, scaled, 23)
|
|
|
|
def test_quad(self):
|
|
# one simple quad transform, equivalent to scale & crop upper left quad
|
|
im = hopper("RGB")
|
|
(w, h) = im.size
|
|
# fmt: off
|
|
transformed = im.transform(im.size, Image.QUAD,
|
|
(0, 0, 0, h//2,
|
|
# ul -> ccw around quad:
|
|
w//2, h//2, w//2, 0),
|
|
Image.BILINEAR)
|
|
# fmt: on
|
|
|
|
scaled = im.transform(
|
|
(w, h), Image.AFFINE, (0.5, 0, 0, 0, 0.5, 0), Image.BILINEAR
|
|
)
|
|
|
|
self.assert_image_equal(transformed, scaled)
|
|
|
|
def test_fill(self):
|
|
for mode, pixel in [
|
|
["RGB", (255, 0, 0)],
|
|
["RGBA", (255, 0, 0, 255)],
|
|
["LA", (76, 0)],
|
|
]:
|
|
im = hopper(mode)
|
|
(w, h) = im.size
|
|
transformed = im.transform(
|
|
im.size,
|
|
Image.EXTENT,
|
|
(0, 0, w * 2, h * 2),
|
|
Image.BILINEAR,
|
|
fillcolor="red",
|
|
)
|
|
|
|
self.assertEqual(transformed.getpixel((w - 1, h - 1)), pixel)
|
|
|
|
def test_mesh(self):
|
|
# this should be a checkerboard of halfsized hoppers in ul, lr
|
|
im = hopper("RGBA")
|
|
(w, h) = im.size
|
|
# fmt: off
|
|
transformed = im.transform(im.size, Image.MESH,
|
|
[((0, 0, w//2, h//2), # box
|
|
(0, 0, 0, h,
|
|
w, h, w, 0)), # ul -> ccw around quad
|
|
((w//2, h//2, w, h), # box
|
|
(0, 0, 0, h,
|
|
w, h, w, 0))], # ul -> ccw around quad
|
|
Image.BILINEAR)
|
|
# fmt: on
|
|
|
|
scaled = im.transform(
|
|
(w // 2, h // 2), Image.AFFINE, (2, 0, 0, 0, 2, 0), Image.BILINEAR
|
|
)
|
|
|
|
checker = Image.new("RGBA", im.size)
|
|
checker.paste(scaled, (0, 0))
|
|
checker.paste(scaled, (w // 2, h // 2))
|
|
|
|
self.assert_image_equal(transformed, checker)
|
|
|
|
# now, check to see that the extra area is (0, 0, 0, 0)
|
|
blank = Image.new("RGBA", (w // 2, h // 2), (0, 0, 0, 0))
|
|
|
|
self.assert_image_equal(blank, transformed.crop((w // 2, 0, w, h // 2)))
|
|
self.assert_image_equal(blank, transformed.crop((0, h // 2, w // 2, h)))
|
|
|
|
def _test_alpha_premult(self, op):
|
|
# create image with half white, half black,
|
|
# with the black half transparent.
|
|
# do op,
|
|
# there should be no darkness in the white section.
|
|
im = Image.new("RGBA", (10, 10), (0, 0, 0, 0))
|
|
im2 = Image.new("RGBA", (5, 10), (255, 255, 255, 255))
|
|
im.paste(im2, (0, 0))
|
|
|
|
im = op(im, (40, 10))
|
|
im_background = Image.new("RGB", (40, 10), (255, 255, 255))
|
|
im_background.paste(im, (0, 0), im)
|
|
|
|
hist = im_background.histogram()
|
|
self.assertEqual(40 * 10, hist[-1])
|
|
|
|
def test_alpha_premult_resize(self):
|
|
def op(im, sz):
|
|
return im.resize(sz, Image.BILINEAR)
|
|
|
|
self._test_alpha_premult(op)
|
|
|
|
def test_alpha_premult_transform(self):
|
|
def op(im, sz):
|
|
(w, h) = im.size
|
|
return im.transform(sz, Image.EXTENT, (0, 0, w, h), Image.BILINEAR)
|
|
|
|
self._test_alpha_premult(op)
|
|
|
|
def test_blank_fill(self):
|
|
# attempting to hit
|
|
# https://github.com/python-pillow/Pillow/issues/254 reported
|
|
#
|
|
# issue is that transforms with transparent overflow area
|
|
# contained junk from previous images, especially on systems with
|
|
# constrained memory. So, attempt to fill up memory with a
|
|
# pattern, free it, and then run the mesh test again. Using a 1Mp
|
|
# image with 4 bands, for 4 megs of data allocated, x 64. OMM (64
|
|
# bit 12.04 VM with 512 megs available, this fails with Pillow <
|
|
# a0eaf06cc5f62a6fb6de556989ac1014ff3348ea
|
|
#
|
|
# Running by default, but I'd totally understand not doing it in
|
|
# the future
|
|
|
|
pattern = [Image.new("RGBA", (1024, 1024), (a, a, a, a)) for a in range(1, 65)]
|
|
|
|
# Yeah. Watch some JIT optimize this out.
|
|
pattern = None # noqa: F841
|
|
|
|
self.test_mesh()
|
|
|
|
def test_missing_method_data(self):
|
|
with hopper() as im:
|
|
self.assertRaises(ValueError, im.transform, (100, 100), None)
|
|
|
|
def test_unknown_resampling_filter(self):
|
|
with hopper() as im:
|
|
(w, h) = im.size
|
|
for resample in (Image.BOX, "unknown"):
|
|
self.assertRaises(
|
|
ValueError,
|
|
im.transform,
|
|
(100, 100),
|
|
Image.EXTENT,
|
|
(0, 0, w, h),
|
|
resample,
|
|
)
|
|
|
|
|
|
class TestImageTransformAffine(PillowTestCase):
|
|
transform = Image.AFFINE
|
|
|
|
def _test_image(self):
|
|
im = hopper("RGB")
|
|
return im.crop((10, 20, im.width - 10, im.height - 20))
|
|
|
|
def _test_rotate(self, deg, transpose):
|
|
im = self._test_image()
|
|
|
|
angle = -math.radians(deg)
|
|
matrix = [
|
|
round(math.cos(angle), 15),
|
|
round(math.sin(angle), 15),
|
|
0.0,
|
|
round(-math.sin(angle), 15),
|
|
round(math.cos(angle), 15),
|
|
0.0,
|
|
0,
|
|
0,
|
|
]
|
|
matrix[2] = (1 - matrix[0] - matrix[1]) * im.width / 2
|
|
matrix[5] = (1 - matrix[3] - matrix[4]) * im.height / 2
|
|
|
|
if transpose is not None:
|
|
transposed = im.transpose(transpose)
|
|
else:
|
|
transposed = im
|
|
|
|
for resample in [Image.NEAREST, Image.BILINEAR, Image.BICUBIC]:
|
|
transformed = im.transform(
|
|
transposed.size, self.transform, matrix, resample
|
|
)
|
|
self.assert_image_equal(transposed, transformed)
|
|
|
|
def test_rotate_0_deg(self):
|
|
self._test_rotate(0, None)
|
|
|
|
def test_rotate_90_deg(self):
|
|
self._test_rotate(90, Image.ROTATE_90)
|
|
|
|
def test_rotate_180_deg(self):
|
|
self._test_rotate(180, Image.ROTATE_180)
|
|
|
|
def test_rotate_270_deg(self):
|
|
self._test_rotate(270, Image.ROTATE_270)
|
|
|
|
def _test_resize(self, scale, epsilonscale):
|
|
im = self._test_image()
|
|
|
|
size_up = int(round(im.width * scale)), int(round(im.height * scale))
|
|
matrix_up = [1 / scale, 0, 0, 0, 1 / scale, 0, 0, 0]
|
|
matrix_down = [scale, 0, 0, 0, scale, 0, 0, 0]
|
|
|
|
for resample, epsilon in [
|
|
(Image.NEAREST, 0),
|
|
(Image.BILINEAR, 2),
|
|
(Image.BICUBIC, 1),
|
|
]:
|
|
transformed = im.transform(size_up, self.transform, matrix_up, resample)
|
|
transformed = transformed.transform(
|
|
im.size, self.transform, matrix_down, resample
|
|
)
|
|
self.assert_image_similar(transformed, im, epsilon * epsilonscale)
|
|
|
|
def test_resize_1_1x(self):
|
|
self._test_resize(1.1, 6.9)
|
|
|
|
def test_resize_1_5x(self):
|
|
self._test_resize(1.5, 5.5)
|
|
|
|
def test_resize_2_0x(self):
|
|
self._test_resize(2.0, 5.5)
|
|
|
|
def test_resize_2_3x(self):
|
|
self._test_resize(2.3, 3.7)
|
|
|
|
def test_resize_2_5x(self):
|
|
self._test_resize(2.5, 3.7)
|
|
|
|
def _test_translate(self, x, y, epsilonscale):
|
|
im = self._test_image()
|
|
|
|
size_up = int(round(im.width + x)), int(round(im.height + y))
|
|
matrix_up = [1, 0, -x, 0, 1, -y, 0, 0]
|
|
matrix_down = [1, 0, x, 0, 1, y, 0, 0]
|
|
|
|
for resample, epsilon in [
|
|
(Image.NEAREST, 0),
|
|
(Image.BILINEAR, 1.5),
|
|
(Image.BICUBIC, 1),
|
|
]:
|
|
transformed = im.transform(size_up, self.transform, matrix_up, resample)
|
|
transformed = transformed.transform(
|
|
im.size, self.transform, matrix_down, resample
|
|
)
|
|
self.assert_image_similar(transformed, im, epsilon * epsilonscale)
|
|
|
|
def test_translate_0_1(self):
|
|
self._test_translate(0.1, 0, 3.7)
|
|
|
|
def test_translate_0_6(self):
|
|
self._test_translate(0.6, 0, 9.1)
|
|
|
|
def test_translate_50(self):
|
|
self._test_translate(50, 50, 0)
|
|
|
|
|
|
class TestImageTransformPerspective(TestImageTransformAffine):
|
|
# Repeat all tests for AFFINE transformations with PERSPECTIVE
|
|
transform = Image.PERSPECTIVE
|