1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472
|
"""
Procedure for unit-testing with images:
1. Run unit tests at least once; this initializes a git clone of
vispy/test-data in config['test_data_path']. This path is
`~/.vispy/test-data` unless the config variable has been modified.
The config file is located at `vispy/vispy/util/config.py`
2. Run individual test scripts with the --vispy-audit flag:
$ python vispy/visuals/tests/test_ellipse.py --vispy-audit
Any failing tests will
display the test results, standard image, and the differences between the
two. If the test result is bad, then press (f)ail. If the test result is
good, then press (p)ass and the new image will be saved to the test-data
directory.
3. After adding or changing test images, create a new commit:
$ cd ~/.vispy/test-data
$ git add ...
$ git commit -a
4. Look up the most recent tag name from the `test_data_tag` variable in
get_test_data_repo() below. Increment the tag name by 1 in the function
and create a new tag in the test-data repository:
$ git tag test-data-NNN
$ git push --tags origin master
This tag is used to ensure that each vispy commit is linked to a specific
commit in the test-data repository. This makes it possible to push new
commits to the test-data repository without interfering with existing
tests, and also allows unit tests to continue working on older vispy
versions.
Finally, update the tag name in ``get_test_data_repo`` to the new name.
"""
import time
import os
import sys
import inspect
import base64
from subprocess import check_call, CalledProcessError
import numpy as np
from ..ext.six import string_types
from ..ext.six.moves import http_client as httplib
from ..ext.six.moves import urllib_parse as urllib
from .. import scene, config
from ..io import read_png, write_png
from ..gloo.util import _screenshot
from ..util import run_subprocess
tester = None
def _get_tester():
global tester
if tester is None:
tester = ImageTester()
return tester
def assert_image_approved(image, standard_file, message=None, **kwargs):
"""Check that an image test result matches a pre-approved standard.
If the result does not match, then the user can optionally invoke a GUI
to compare the images and decide whether to fail the test or save the new
image as the standard.
This function will automatically clone the test-data repository into
~/.vispy/test-data. However, it is up to the user to ensure this repository
is kept up to date and to commit/push new images after they are saved.
Run the test with python <test-path> --vispy-audit-tests to bring up
the auditing GUI.
Parameters
----------
image : (h, w, 4) ndarray or 'screenshot'
The test result to check
standard_file : str
The name of the approved test image to check against. This file name
is relative to the root of the vispy test-data repository and will
be automatically fetched.
message : str
A string description of the image. It is recommended to describe
specific features that an auditor should look for when deciding whether
to fail a test.
Extra keyword arguments are used to set the thresholds for automatic image
comparison (see ``assert_image_match()``).
"""
if isinstance(image, string_types) and image == "screenshot":
image = _screenshot(alpha=True)
if message is None:
code = inspect.currentframe().f_back.f_code
message = "%s::%s" % (code.co_filename, code.co_name)
# Make sure we have a test data repo available, possibly invoking git
data_path = get_test_data_repo()
# Read the standard image if it exists
std_file = os.path.join(data_path, standard_file)
if not os.path.isfile(std_file):
std_image = None
else:
std_image = read_png(std_file)
# If the test image does not match, then we go to audit if requested.
try:
if image.shape != std_image.shape:
# Allow im1 to be an integer multiple larger than im2 to account
# for high-resolution displays
ims1 = np.array(image.shape).astype(float)
ims2 = np.array(std_image.shape).astype(float)
sr = ims1 / ims2
if (sr[0] != sr[1] or not np.allclose(sr, np.round(sr)) or
sr[0] < 1):
raise TypeError("Test result shape %s is not an integer factor"
" larger than standard image shape %s." %
(ims1, ims2))
sr = np.round(sr).astype(int)
image = downsample(image, sr[0], axis=(0, 1)).astype(image.dtype)
assert_image_match(image, std_image, **kwargs)
except Exception:
if standard_file in git_status(data_path):
print("\n\nWARNING: unit test failed against modified standard "
"image %s.\nTo revert this file, run `cd %s; git checkout "
"%s`\n" % (std_file, data_path, standard_file))
if config['audit_tests']:
sys.excepthook(*sys.exc_info())
_get_tester().test(image, std_image, message)
std_path = os.path.dirname(std_file)
print('Saving new standard image to "%s"' % std_file)
if not os.path.isdir(std_path):
os.makedirs(std_path)
write_png(std_file, image)
else:
if std_image is None:
raise Exception("Test standard %s does not exist." % std_file)
else:
if os.getenv('TRAVIS') is not None or \
os.getenv('APPVEYOR') is not None:
_save_failed_test(image, std_image, standard_file)
raise
def assert_image_match(im1, im2, min_corr=0.9, px_threshold=50.,
px_count=None, max_px_diff=None, avg_px_diff=None,
img_diff=None):
"""Check that two images match.
Images that differ in shape or dtype will fail unconditionally.
Further tests for similarity depend on the arguments supplied.
Parameters
----------
im1 : (h, w, 4) ndarray
Test output image
im2 : (h, w, 4) ndarray
Test standard image
min_corr : float or None
Minimum allowed correlation coefficient between corresponding image
values (see numpy.corrcoef)
px_threshold : float
Minimum value difference at which two pixels are considered different
px_count : int or None
Maximum number of pixels that may differ
max_px_diff : float or None
Maximum allowed difference between pixels
avg_px_diff : float or None
Average allowed difference between pixels
img_diff : float or None
Maximum allowed summed difference between images
"""
assert im1.ndim == 3
assert im1.shape[2] == 4
assert im1.dtype == im2.dtype
diff = im1.astype(float) - im2.astype(float)
if img_diff is not None:
assert np.abs(diff).sum() <= img_diff
pxdiff = diff.max(axis=2) # largest value difference per pixel
mask = np.abs(pxdiff) >= px_threshold
if px_count is not None:
assert mask.sum() <= px_count
masked_diff = diff[mask]
if max_px_diff is not None and masked_diff.size > 0:
assert masked_diff.max() <= max_px_diff
if avg_px_diff is not None and masked_diff.size > 0:
assert masked_diff.mean() <= avg_px_diff
if min_corr is not None:
with np.errstate(invalid='ignore'):
corr = np.corrcoef(im1.ravel(), im2.ravel())[0, 1]
assert corr >= min_corr
def _save_failed_test(data, expect, filename):
from ..io import _make_png
commit, error = run_subprocess(['git', 'rev-parse', 'HEAD'])
name = filename.split('/')
name.insert(-1, commit.strip())
filename = '/'.join(name)
host = 'data.vispy.org'
# concatenate data, expect, and diff into a single image
ds = data.shape
es = expect.shape
shape = (max(ds[0], es[0]) + 4, ds[1] + es[1] + 8 + max(ds[1], es[1]), 4)
img = np.empty(shape, dtype=np.ubyte)
img[..., :3] = 100
img[..., 3] = 255
img[2:2+ds[0], 2:2+ds[1], :ds[2]] = data
img[2:2+es[0], ds[1]+4:ds[1]+4+es[1], :es[2]] = expect
diff = make_diff_image(data, expect)
img[2:2+diff.shape[0], -diff.shape[1]-2:-2] = diff
png = _make_png(img)
conn = httplib.HTTPConnection(host)
req = urllib.urlencode({'name': filename,
'data': base64.b64encode(png)})
conn.request('POST', '/upload.py', req)
response = conn.getresponse().read()
conn.close()
print("\nImage comparison failed. Test result: %s %s Expected result: "
"%s %s" % (data.shape, data.dtype, expect.shape, expect.dtype))
print("Uploaded to: \nhttp://%s/data/%s" % (host, filename))
if not response.startswith(b'OK'):
print("WARNING: Error uploading data to %s" % host)
print(response)
def make_diff_image(im1, im2):
"""Return image array showing the differences between im1 and im2.
Handles images of different shape. Alpha channels are not compared.
"""
ds = im1.shape
es = im2.shape
diff = np.empty((max(ds[0], es[0]), max(ds[1], es[1]), 4), dtype=int)
diff[..., :3] = 128
diff[..., 3] = 255
diff[:ds[0], :ds[1], :min(ds[2], 3)] += im1[..., :3]
diff[:es[0], :es[1], :min(es[2], 3)] -= im2[..., :3]
diff = np.clip(diff, 0, 255).astype(np.ubyte)
return diff
def downsample(data, n, axis=0):
"""Downsample by averaging points together across axis.
If multiple axes are specified, runs once per axis.
"""
if hasattr(axis, '__len__'):
if not hasattr(n, '__len__'):
n = [n]*len(axis)
for i in range(len(axis)):
data = downsample(data, n[i], axis[i])
return data
if n <= 1:
return data
nPts = int(data.shape[axis] / n)
s = list(data.shape)
s[axis] = nPts
s.insert(axis+1, n)
sl = [slice(None)] * data.ndim
sl[axis] = slice(0, nPts*n)
d1 = data[tuple(sl)]
d1.shape = tuple(s)
d2 = d1.mean(axis+1)
return d2
class ImageTester(scene.SceneCanvas):
"""Graphical interface for auditing image comparison tests.
"""
def __init__(self):
self.grid = None
self.views = None
self.console = None
self.last_key = None
scene.SceneCanvas.__init__(self, size=(1000, 800))
self.bgcolor = (0.1, 0.1, 0.1, 1)
self.grid = self.central_widget.add_grid()
border = (0.3, 0.3, 0.3, 1)
self.views = (self.grid.add_view(row=0, col=0, border_color=border),
self.grid.add_view(row=0, col=1, border_color=border),
self.grid.add_view(row=0, col=2, border_color=border))
label_text = ['test output', 'standard', 'diff']
for i, v in enumerate(self.views):
v.camera = 'panzoom'
v.camera.aspect = 1
v.camera.flip = (False, True)
# unfreeze it to set the image and label on the view
# this is slightly hacky, but it is simpler than
# creating another class/storing as a dict or a tuple
v.unfreeze()
v.image = scene.Image(parent=v.scene)
v.label = scene.Text(label_text[i], parent=v, color='yellow',
anchor_x='left', anchor_y='top')
v.freeze()
self.views[1].camera.link(self.views[0].camera)
self.views[2].camera.link(self.views[0].camera)
self.console = scene.Console(text_color='white', border_color=border)
self.grid.add_widget(self.console, row=1, col=0, col_span=3)
def test(self, im1, im2, message):
self.show()
self.console.write('------------------')
self.console.write(message)
if im2 is None:
self.console.write('Image1: %s %s Image2: [no standard]' %
(im1.shape, im1.dtype))
im2 = np.zeros((1, 1, 3), dtype=np.ubyte)
else:
self.console.write('Image1: %s %s Image2: %s %s' %
(im1.shape, im1.dtype, im2.shape, im2.dtype))
self.console.write('(P)ass or (F)ail this test?')
self.views[0].image.set_data(im1)
self.views[1].image.set_data(im2)
diff = make_diff_image(im1, im2)
self.views[2].image.set_data(diff)
self.views[0].camera.set_range()
while True:
self.app.process_events()
if self.last_key is None:
pass
elif self.last_key.lower() == 'p':
self.console.write('PASS')
break
elif self.last_key.lower() in ('f', 'esc'):
self.console.write('FAIL')
raise Exception("User rejected test result.")
time.sleep(0.03)
for v in self.views:
v.image.set_data(np.zeros((1, 1, 3), dtype=np.ubyte))
def on_key_press(self, event):
self.last_key = event.key.name
def get_test_data_repo():
"""Return the path to a git repository with the required commit checked
out.
If the repository does not exist, then it is cloned from
https://github.com/vispy/test-data. If the repository already exists
then the required commit is checked out.
"""
# This tag marks the test-data commit that this version of vispy should
# be tested against. When adding or changing test images, create
# and push a new tag and update this variable.
test_data_tag = 'test-data-10'
data_path = config['test_data_path']
git_path = 'http://github.com/vispy/test-data'
gitbase = git_cmd_base(data_path)
if os.path.isdir(data_path):
# Already have a test-data repository to work with.
# Get the commit ID of test_data_tag. Do a fetch if necessary.
try:
tag_commit = git_commit_id(data_path, test_data_tag)
except NameError:
cmd = gitbase + ['fetch', '--tags', 'origin']
print(' '.join(cmd))
check_call(cmd)
try:
tag_commit = git_commit_id(data_path, test_data_tag)
except NameError:
raise Exception("Could not find tag '%s' in test-data repo at"
" %s" % (test_data_tag, data_path))
except Exception:
if not os.path.exists(os.path.join(data_path, '.git')):
raise Exception("Directory '%s' does not appear to be a git "
"repository. Please remove this directory." %
data_path)
else:
raise
# If HEAD is not the correct commit, then do a checkout
if git_commit_id(data_path, 'HEAD') != tag_commit:
print("Checking out test-data tag '%s'" % test_data_tag)
check_call(gitbase + ['checkout', test_data_tag])
else:
print("Attempting to create git clone of test data repo in %s.." %
data_path)
parent_path = os.path.split(data_path)[0]
if not os.path.isdir(parent_path):
os.makedirs(parent_path)
if os.getenv('TRAVIS') is not None:
# Create a shallow clone of the test-data repository (to avoid
# downloading more data than is necessary)
os.makedirs(data_path)
cmds = [
gitbase + ['init'],
gitbase + ['remote', 'add', 'origin', git_path],
gitbase + ['fetch', '--tags', 'origin', test_data_tag,
'--depth=1'],
gitbase + ['checkout', '-b', 'master', 'FETCH_HEAD'],
]
else:
# Create a full clone
cmds = [['git', 'clone', git_path, data_path]]
for cmd in cmds:
print(' '.join(cmd))
rval = check_call(cmd)
if rval == 0:
continue
raise RuntimeError("Test data path '%s' does not exist and could "
"not be created with git. Either create a git "
"clone of %s or set the test_data_path "
"variable to an existing clone." %
(data_path, git_path))
return data_path
def git_cmd_base(path):
return ['git', '--git-dir=%s/.git' % path, '--work-tree=%s' % path]
def git_status(path):
"""Return a string listing all changes to the working tree in a git
repository.
"""
cmd = git_cmd_base(path) + ['status', '--porcelain']
return run_subprocess(cmd, stderr=None, universal_newlines=True)[0]
def git_commit_id(path, ref):
"""Return the commit id of *ref* in the git repository at *path*.
"""
cmd = git_cmd_base(path) + ['show', ref]
try:
output = run_subprocess(cmd, stderr=None, universal_newlines=True)[0]
except CalledProcessError:
raise NameError("Unknown git reference '%s'" % ref)
commit = output.split('\n')[0]
assert commit[:7] == 'commit '
return commit[7:]
|