Miscellaneous Changes

* Added a function in FederatedASDFDataSet to fetch unique location
  codes
* The correlator module now handles location codes and their preferences
  more gracefully
* Minor changes in cwb2asdf

seismic/ASDFdatabase/FederatedASDFDataSet.py

@@ -191,6 +191,20 @@ def get_waveforms(self, network, station, location, channel, starttime,
 
     # end func
 
+    def get_location_codes(self, network, station, starttime=None, endtime=None):
+        """
+        :param network: network code
+        :param station: station code
+        :param starttime: start time string in UTCDateTime format; can also be an instance of obspy.UTCDateTime
+        :param endtime: end time string in UTCDateTime format; can also be an instance of obspy.UTCDateTime
+
+        :return: a list containing unique location codes within the timeframe specified
+        """
+
+        return self.fds.get_location_codes(network, station, starttime=starttime, endtime=endtime)
+
+    # end func
+
     def stations_iterator(self, network_list=[], station_list=[]):
         """
         This function provides an iterator over the entire data volume contained in all the ASDF files listed in the

seismic/ASDFdatabase/_FederatedASDFDataSetImpl.py

@@ -589,6 +589,28 @@ def get_waveforms(self, network, station, location, channel, starttime,
         return s
     # end func
 
+    def get_location_codes(self, network, station, starttime=None, endtime=None):
+        st, et = self.get_global_time_range(network, station)
+
+        if(starttime):
+            starttime = UTCDateTime(starttime)
+            if(starttime > st): st = starttime
+        # end if
+
+        if(endtime):
+            endtime = UTCDateTime(endtime)
+            if(endtime < et): et = endtime
+        # end if
+
+        rows = self.get_stations(st, et, network=network, station=station)
+        uniqueLocCodes = set()
+        for row in rows:
+            uniqueLocCodes.add(row[2])
+        # end for
+
+        return sorted(list(uniqueLocCodes))
+    # end func
+
     def stations_iterator(self, network_list=[], station_list=[]):
         workload = None
         if(self.rank==0):

seismic/ASDFdatabase/cwb2asdf/cwb2asdf.py

@@ -51,14 +51,17 @@ def make_ASDF_tag(tr, tag):
 @click.argument('inventory', required=True,
                 type=click.Path(exists=True))
 @click.argument('output-file-name', required=True)
+@click.option('--file-pattern', type=str, default='*.mseed',
+              help="File pattern to be used while looking for data files")
 @click.option('--channels-to-extract', type=str, default=None, help="Channels to extract, within quotes and space- "
                                                                    "separated.")
 @click.option('--min-length-sec', type=int, default=None, help="Minimum length in seconds")
 @click.option('--merge-threshold', type=int, default=None, help="Merge traces if the number of traces fetched for an "
                                                                 "interval exceeds this threshold")
 @click.option('--ntraces-per-file', type=int, default=3600, help="Maximum number of traces per file; if exceeded, the "
                                                                  "file is ignored.")
-def process(input_folder, inventory, output_file_name, channels_to_extract, min_length_sec, merge_threshold,
+def process(input_folder, inventory, output_file_name, file_pattern,
+            channels_to_extract, min_length_sec, merge_threshold,
             ntraces_per_file):
     """
     INPUT_FOLDER: Path to input folder containing miniseed files \n
@@ -109,11 +112,12 @@ def _write(ds, ostream, inventory_dict, netsta_set):
 
         # generate a list of files
         paths = [i for i in os.listdir(input_folder) if os.path.isfile(os.path.join(input_folder, i))]
-        expr = re.compile(fnmatch.translate('*.mseed'), re.IGNORECASE)
+        expr = re.compile(fnmatch.translate(file_pattern), re.IGNORECASE)
         files = [os.path.join(input_folder, j) for j in paths if re.match(expr, j)]
 
         files = np.array(files)
         random.Random(nproc).shuffle(files)
+        #print(files); exit(0)
         #files = files[370:380]
 
         ustations = set()

seismic/xcorqc/xcorqc.py

@@ -638,14 +638,17 @@ def IntervalStackXCorr(refds, tempds,
                          'together is redundant')
     # end if
 
-    # setup logger
+    # get preferred location codes, or use the first available
     ref_loc = location_preferences_dict[ref_net_sta]
     temp_loc = location_preferences_dict[temp_net_sta]
-    if(ref_loc is None): ref_loc = ''
-    if(temp_loc is None): temp_loc = ''
+    if(ref_loc is None): ref_loc = refds.get_location_codes(*ref_net_sta.split('.'))[0]
+    if(temp_loc is None): temp_loc = tempds.get_location_codes(*temp_net_sta.split('.'))[0]
+
     stationPair = '%s.%s.%s.%s.%s.%s' % (ref_net_sta, ref_loc, ref_cha, temp_net_sta, temp_loc, temp_cha)
     fn = os.path.join(outputPath, '%s.log' % (stationPair if not tracking_tag else
                                               '.'.join([stationPair, tracking_tag])))
+
+    # setup logger
     logger = setup_logger(stationPair, fn)
 
     #########################################################################

tests/test_seismic/ASDFdatabase/test_federatedasdfdataset.py

@@ -84,6 +84,21 @@ def test_get_stations():
     assert len(station_set) == 8
 # end func
 
+def test_get_locaiton_codes():
+    fds = FederatedASDFDataSet(asdf_file_list)
+
+    rows = np.array(fds.get_stations('1900-01-01T00:00:00', '2100-01-01T00:00:00'))
+
+    location_set = set()
+    for n, s in rows[:, 0:2]: 
+        locs = fds.get_location_codes(n, s)
+        for loc in locs: location_set.add(loc)
+    # end for
+
+    # There is 1 unique locaiton code in the h5 file
+    assert len(location_set) == 1
+# end func
+
 def test_get_coordinates():
     fds = FederatedASDFDataSet(asdf_file_list)
 
@@ -121,7 +136,7 @@ def test_get_global_time_range():
 # end func
 
 
-def test_get_stations():
+def test_stations_iterator():
     fds = FederatedASDFDataSet(asdf_file_list)
 
     local_netsta_list = list(fds.stations_iterator())