gstlal_feature_combiner: allow saving features to a separate directory,...

gstlal_feature_combiner: allow saving features to a separate directory, optionally pass in start, end times

gstlal-burst/bin/gstlal_feature_combiner

@@ -36,6 +36,9 @@ import shutil
 import h5py
 import numpy
 
+from ligo.segments import segment, segmentlist
+
+from gstlal import aggregator
 from gstlal.fxtools import utils
 
 # =============================================================================
@@ -52,11 +55,14 @@ def parse_command_line():
 
 	group = optparse.OptionGroup(parser, "Combiner Options", "General settings for configuring the file combiner.")
 	group.add_option("-v","--verbose", default=False, action="store_true", help = "Print to stdout in addition to writing to automatically generated log.")
+	group.add_option("--start-time", type = "int", help = "Set the start time to combine features.")
+	group.add_option("--end-time", type = "int", help = "Set the end time to combine features.")
 	group.add_option("--log-level", type = "int", default = 10, help = "Sets the verbosity of logging. Default = 10.")
 	group.add_option("--rootdir", metavar = "path", default = ".", help = "Sets the root directory where features, logs, and metadata are stored.")
 	group.add_option("--basename", metavar = "string", default = "GSTLAL_IDQ_FEATURES", help = "Sets the basename for files written to disk. Default = GSTLAL_IDQ_FEATURES")
 	group.add_option("--instrument", metavar = "string", default = "H1", help = "Sets the instrument for files written to disk. Default = H1")
 	group.add_option("--tag", metavar = "string", default = "test", help = "Sets the name of the tag used. Default = 'test'")
+	group.add_option("--outdir", metavar = "path", help = "If set, chooses an alternate directory to save the features to. Default = --rootdir")
 	parser.add_option_group(group)
 
 	opts, args = parser.parse_args()
@@ -81,6 +87,11 @@ if __name__ == "__main__":
 		verbose=options.verbose
 	)
 
+	### define gps bounds to grab features
+	start_time = options.start_time if options.start_time else -numpy.inf
+	end_time = options.end_time if options.end_time else numpy.inf
+	file_segs = segmentlist([segment(start_time, end_time)])
+
 	### get base temp directory
 	if '_CONDOR_SCRATCH_DIR' in os.environ:
 		tmp_dir = os.environ['_CONDOR_SCRATCH_DIR']
@@ -93,6 +104,11 @@ if __name__ == "__main__":
 		ifo=options.instrument[0],
 	)
 	cache = sorted(utils.path2cache(options.rootdir, pattern), key=lambda x: x.segment)
+
+	### filter cache with segments
+	cache = [entry for entry in cache if file_segs.intersects_segment(entry.segment)]
+
+	### group by segment
 	grouped_cache = [(seg, list(group)) for seg, group in itertools.groupby(cache, key=lambda x: x.segment)]
 
 	### combine features in each stride
@@ -122,7 +138,16 @@ if __name__ == "__main__":
 			for dset in features[channel].keys():
 				utils.create_new_dataset(tmp_dir, filename, features[channel][dset], name=dset, group=channel, tmp=True, metadata=metadata)
 
-		final_path = os.path.join(dirname, filename)+".h5"
+		### determine final location for features
+		if options.outdir:
+			start_time = int(filename.split('-')[2])
+			basename = '-'.join([options.instrument[0], options.basename])
+			base_path = utils.to_trigger_path(options.outdir, basename, start_time)
+			aggregator.makedir(base_path)
+			final_path = os.path.join(base_path, filename)+".h5"
+		else:
+			final_path = os.path.join(dirname, filename)+".h5"
+
 		tmp_path = os.path.join(tmp_dir, filename)+".h5.tmp"
 		logger.info('saving features to: {}'.format(final_path))
 		shutil.move(tmp_path, final_path)