def test_validate_arches(self):
got = {}
for arch_name in arches:
cwd = os.path.dirname(os.path.realpath(__file__))
archPath = os.path.join(cwd, 'test_ptree_base', arch_name)
arch = pArch.PhashArchive(archPath)
archHashes = list(arch.iterHashes())
for item in archHashes:
del item[1]['cont']
got[arch_name] = archHashes
expect_keys = list(expect.keys())
got_keys = list(got.keys())
expect_keys.sort()
got_keys.sort()
self.assertEqual(expect_keys, got_keys)
for key in expect_keys:
if got[key] != expect_keys:
print("Key:", key)
pprint.pprint(got[key])
pprint.pprint(expect[key])
self.assertEqual(expect[key], got[key])
def __init__(self, archPath, pathFilter=None):
'''
Params:
pathFilter (list): default =``['']``
Basically, if you pass a list of valid path prefixes, any matches not
on any of those path prefixes are not matched.
Default is [''], which matches every path, because "anything".startswith('') is true
'''
super().__init__()
self.maskedPaths = pathFilter or ['']
self.archPath = archPath
self.arch = pArch.PhashArchive(archPath)
self.log = logging.getLogger("Main.Deduper")
self.log.info("ArchChecker Instantiated on '%s'", archPath)
def test_pArch_2(self):
cwd = os.path.dirname(os.path.realpath(__file__))
archPath = os.path.join(cwd, 'testArches', 'testArches.zip')
arch = pArch.PhashArchive(archPath)
ret = arch.getHashInfo('dangerous-to-go-alone.jpg')
ret.pop('cont')
expect = {
# 'dHash': 4576150637722077151,
'hexHash': 'dcd6097eeac911efed3124374f44085b',
'imX': 325,
'pHash': -149413575039568585,
'type': 'image/jpeg',
'imY': 307
}
self.assertEqual(ret, expect)
def __init__(self,
archPath,
pathNegativeFilter=None,
pathPositiveFilter=None,
negativeKeywords=None):
'''
Params:
pathNegativeFilter (list): default =``[]``
List of paths to exclude from matching.
By default, and empty list, leading to all possible paths being used.
'''
super().__init__()
self.negativeMaskedPaths = pathNegativeFilter or []
self.positiveMaskedPaths = pathPositiveFilter or []
self.negativeKeywords = negativeKeywords or []
self.archPath = archPath
self.arch = pArch.PhashArchive(archPath)
self.log = logging.getLogger("Main.Deduper")
self.log.info("ArchChecker Instantiated on '%s'", archPath)
def test_pArch_4(self):
cwd = os.path.dirname(os.path.realpath(__file__))
archPath = os.path.join(cwd, 'testArches', 'testArches.7z')
arch = pArch.PhashArchive(archPath)
ret = arch.getHashInfo('dangerous-to-go-alone.jpg')
# We're not confirming image contents here, so pop that from the list because it's
# enormous, and I don't want to insert that into the file.
ret.pop('cont')
expect = {
# 'dHash': 4576150637722077151,
'hexHash': 'dcd6097eeac911efed3124374f44085b',
'imX': 325,
'pHash': -149413575039568585,
'type': 'image/jpeg',
'imY': 307
}
self.assertEqual(ret, expect)
def test_pArch_1(self):
cwd = os.path.dirname(os.path.realpath(__file__))
archPath = os.path.join(cwd, 'testArches', 'testArches.zip')
arch = pArch.PhashArchive(archPath)
match = [
(
'Lolcat_this_is_mah_job.jpg',
{
'hexHash': 'd9ceeb6b43c2d7d096532eabfa6cf482',
'type': 'image/jpeg',
'imY': 389,
'pHash': 27427800275512429,
# 'dHash' : -4504585791368671746,
'imX': 493
}),
(
'Lolcat_this_is_mah_job.png',
{
'hexHash': '1268e704908cc39299d73d6caafc23a0',
'type': 'image/png',
'imY': 389,
'pHash': 27427800275512429,
# 'dHash' : -4504585791368671746,
'imX': 493
}),
(
'Lolcat_this_is_mah_job_small.jpg',
{
'hexHash': '40d39c436e14282dcda06e8aff367307',
'type': 'image/jpeg',
'imY': 237,
'pHash': 27427800275512429,
# 'dHash' : -4504585791368671746,
'imX': 300
}),
(
'dangerous-to-go-alone.jpg',
{
'hexHash': 'dcd6097eeac911efed3124374f44085b',
'type': 'image/jpeg',
'imY': 307,
'pHash': -149413575039568585,
# 'dHash' : 4576150637722077151,
'imX': 325
}),
(
'lolcat-crocs.jpg',
{
'hexHash': '6d0a977694630ac9d1d33a7f068e10f8',
'type': 'image/jpeg',
'imY': 363,
'pHash': -5569898607211671279,
# 'dHash' : 167400391896309758,
'imX': 500
}),
(
'lolcat-oregon-trail.jpg',
{
'hexHash': '7227289a017988b6bdcf61fd4761f6b9',
'type': 'image/jpeg',
'imY': 356,
'pHash': -4955310669995365332,
# 'dHash' : -8660145558008088574,
'imX': 501
})
]
archHashes = list(arch.iterHashes())
for item in archHashes:
del item[1]['cont']
print(archHashes)
self.assertEqual(archHashes, match)