def rebuild_if_needed(lc_dir, db_dir, n_vps=20, n_lcs=1000):
"""
Helper that checks if needed time series files & vantage point databases already exists
recreates them if they don't.
"""
if (simsearch.need_to_rebuild(lc_dir, db_dir)):
simsearch.rebuild_lcs_dbs(lc_dir, db_dir, n_vps=n_vps, n_lcs=n_lcs)
def test_makelcs():
lc_temp_dir = TEMP_DIR + LIGHT_CURVES_DIR
db_temp_dir = TEMP_DIR + DB_DIR
# Before we've generated anything, confirm that our rebuild detector method properly detects that we need to rebuild
assert (simsearch.need_to_rebuild(lc_temp_dir, db_temp_dir) == True)
makelcs.make_lcs_wfm(5, lc_temp_dir)
assert (simsearch.need_to_rebuild(lc_temp_dir, db_temp_dir) == True)
clear_dir(TEMP_DIR, recreate=False)
# Generate 100 light curves in temp dir
makelcs.make_lcs_wfm(100, lc_temp_dir)
assert (simsearch.need_to_rebuild(lc_temp_dir, db_temp_dir) == True)
# load that back in from disk
lc_dict = genvpdbs.load_ts_fsm(lc_temp_dir)
assert (len(lc_dict) == 100)
def test_genvpdbs():
lc_temp_dir = TEMP_DIR + LIGHT_CURVES_DIR
db_temp_dir = TEMP_DIR + DB_DIR
# Create 10 vantage point database files
genvpdbs.create_vpdbs(10, lc_temp_dir, db_temp_dir)
# With vantage points created, we should have fully created dataset
assert (simsearch.need_to_rebuild(lc_temp_dir, db_temp_dir) == False)
# Load vantage points back in from disk
vp_dict = simsearch.load_vp_lcs(db_temp_dir, lc_temp_dir)
assert (len(vp_dict) == 10)
clear_dir(TEMP_DIR, recreate=False)
def cmd_line_util():
"""
Main program loop. Determines which flags were submitted, confirms that lc files and db files
exist (Recreates them if they don't) before kicking off similarity search.
"""
# Default conditions
rebuild = need_to_rebuild(LIGHT_CURVES_DIR, DB_DIR)
need_help = False
input_fpath = False
plot = False
demo = False
while True:
if len(sys.argv) <= 1:
print("No arguments provided")
print(USAGE)
break
# First, identify which flags were included
for arg in sys.argv[1:]:
if arg.lower() in ['-h','--help', 'help']: need_help = True
elif '.txt' in arg.lower() or '.dat_folded' in arg.lower():
input_fpath = arg
elif arg.lower() in ['-r', '--rebuild']: rebuild = True
elif arg.lower() in ['-d', '--demo']: demo = True
elif arg.lower() in ['-p', '--plot']: plot = True
# Execute selected options
if need_help:
print(HELP_MESSAGE)
break
elif rebuild:
rebuild_lcs_dbs(LIGHT_CURVES_DIR, DB_DIR)
if demo:
run_demo(DB_DIR, LIGHT_CURVES_DIR, plot)
break
elif(input_fpath is not False):
sim_search(input_fpath, DB_DIR, LIGHT_CURVES_DIR, plot)
break
else:
print("Error: no compatible time series or light curve file provided")
print(USAGE)
break
def test_simsearch():
lc_temp_dir = TEMP_DIR + LIGHT_CURVES_DIR
db_temp_dir = TEMP_DIR + DB_DIR
simsearch.rebuild_lcs_dbs(lc_temp_dir, db_temp_dir, 10, 100)
# Repeat checks from above on re-generated data
lc_dict = genvpdbs.load_ts_fsm(lc_temp_dir)
assert (len(lc_dict) == 100)
vp_dict = simsearch.load_vp_lcs(db_temp_dir, lc_temp_dir)
assert (len(vp_dict) == 10)
assert (simsearch.need_to_rebuild(lc_temp_dir, db_temp_dir) == False)
demo_ts_fn = random.choice(
os.listdir('cs207project/tsrbtreedb/' + SAMPLE_DIR))
demo_fp = 'cs207project/tsrbtreedb/' + SAMPLE_DIR + demo_ts_fn
simsearchutil.sim_search(demo_fp, db_temp_dir, lc_temp_dir, False)