takino/ordigi
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Fix image.py

Browse Source
This commit is contained in:
Cédric Leporcq 2021-11-12 09:36:09 +01:00
parent 11291a582e
commit 5d7e0c4f46
1 changed files with 16 additions and 22 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

38
ordigi/images.py
View File

@ -54,11 +54,11 @@ class Image:
# things like mode, size, and other properties required to decode the file, # things like mode, size, and other properties required to decode the file,
# but the rest of the file is not processed until later. # but the rest of the file is not processed until later.
try: try:
im = img.open(self.img_path) image = img.open(self.img_path)
except (IOError, UnidentifiedImageError): except (IOError, UnidentifiedImageError):
return False return False
if im.format is None: if image.format is None:
return False return False
return True return True
@ -93,26 +93,20 @@ class Images:
'rw2', 'rw2',
) )
def __init__(self, images=set(), hash_size=8): def __init__(self, images, hash_size=8):
self.images = images self.images = images
self.duplicates = [] self.duplicates = []
self.hash_size = hash_size self.hash_size = hash_size
self.log = LOG.getChild(self.__class__.__name__) self.log = LOG.getChild(self.__class__.__name__)
if PYHEIF == False: if not PYHEIF:
self.log.info("No module named 'pyheif_pillow_opener'") self.log.info("No module named 'pyheif_pillow_opener'")
def add_images(self, file_paths):
for img_path in file_paths:
image = Image(img_path)
if image.is_image():
self.images.add(image)
def get_images_hashes(self): def get_images_hashes(self):
"""Get image hashes""" """Get image hashes"""
# Searching for duplicates. # Searching for duplicates.
for image in self.images: for image in self.images:
with img.open(image.img_path) as img: with img.open(image.img_path) as i:
yield imagehash.average_hash(img, self.hash_size) yield imagehash.average_hash(i, self.hash_size)
def find_duplicates(self, img_path): def find_duplicates(self, img_path):
"""Find duplicates""" """Find duplicates"""
@ -159,27 +153,27 @@ class Images:
return similarity_img return similarity_img
def find_similar(self, image, similarity=80): def find_similar(self, image0, similarity=80):
""" """
Find similar images Find similar images
:returns: img_path generator :returns: img_path generator
""" """
hash1 = image.get_hash() hash1 = image0.get_hash()
if hash1 is None: if hash1 is None:
return None return
self.log.info(f'Finding similar images to {image.img_path}') self.log.info(f"Finding similar images to {image0.img_path}")
threshold = 1 - similarity / 100 threshold = 1 - similarity / 100
diff_limit = int(threshold * (self.hash_size ** 2)) diff_limit = int(threshold * (self.hash_size ** 2))
for img in self.images: for image in self.images:
if not img.img_path.is_file(): if not image.img_path.is_file():
continue continue
if img.img_path == image.img_path: if image.img_path == image0.img_path:
continue continue
hash2 = img.get_hash() hash2 = image.get_hash()
# Be sure that hash are not None # Be sure that hash are not None
if hash2 is None: if hash2 is None:
continue continue
@ -188,6 +182,6 @@ class Images:
if img_diff <= diff_limit: if img_diff <= diff_limit:
similarity_img = self.similarity(img_diff) similarity_img = self.similarity(img_diff)
self.log.info( self.log.info(
f'{img.img_path} image found {similarity_img}% similar to {image}' f"{image.img_path} image found {similarity_img}% similar to {image0.img_path}"
) )
yield img.img_path yield image.img_path