class TestImageCache(unittest.TestCase):
def setUp(self):
self.ic = ImageCache()
def tearDown(self):
pass
def test_ic_construction(self):
self.assertIsInstance(self.ic, ImageCache)
def test_ic_get_table_name(self):
self.assertIsInstance(self.ic.get_table(), str)
async def gen_database(path: str, fast: bool) -> None:
"""
Takes in a target directory and computes information about
the images contained therin
"""
ic = ImageCache(fast=fast)
await ic.gen_cache_from_directory(path)
report = {}
queries = {
#"all_data": "SELECT * FROM {};",
"image_types": "SELECT COUNT(DISTINCT img_type) FROM {};",
"total_images": "SELECT COUNT(*) FROM {};",
"average_size": "SELECT AVG(size) FROM {};",
"total_size": "SELECT SUM(size) FROM {};",
}
for k, v in queries.items():
rows = ic.query(v.format(ic.get_table()))
report[k] = rows[0][0]
# Get duplicate and ambiguous images
report['duplicates'] = ic.get_duplicates()
report['ambiguous'] = ic.get_ambiguous()
report["process_time"] = ic.processing_time
pp = pprint.PrettyPrinter(indent=2, compact=False)
pp.pprint(report)
logger.info("Completed database generation.")
logger.info(
f"Processed {ic.get_count()} images in {ic.processing_time} seconds.")
logger.info(f"Encountered {len(report['duplicates'])} duplicate images.")
tstamp = datetime.datetime.now().strftime("gen_database_%Y-%m-%d.json")
with open(tstamp, 'w') as fout:
fout.write(json.dumps(report))
logger.info(f"Report written to {tstamp}")
