def download(self, keyword):
source = Links()
print('Collecting downloadable links of {}...'.format(keyword))
links = source.collect(keyword)
print(
'Downloading images of {} from collected links...'.format(keyword))
self.mkdir('{}/{}'.format(self.path, keyword))
n_links = len(links)
for index, link in enumerate(links):
try:
print(
'Downloading this image based on the keyword {} from {}: {}/{}'
.format(keyword, link, index + 1, n_links))
response = requests.get(link, stream=True)
ext = self.get_extension(link)
raw_path = '{}/{}/{}'.format(self.path, keyword,
str(index).zfill(4))
path = raw_path + '.' + ext
self.save(response, path)
del response
print("Validating image file")
ext2 = self.validate(path)
if ext2 is None:
print('Unreadable file - {}'.format(link))
os.remove(path)
else:
if ext != ext2:
path2 = raw_path + '.' + ext2
os.rename(path, path2)
print('Renaming extension {} -> {}'.format(ext, ext2))
except Exception as e:
print('Download failed.', e)
continue
