Merge branch 'fix-h5parm-compare' into 'master'

Fix h5parm compare

See merge request eosc/prefactor3-cwl!42

test_jobs/check_workflow_results.py

@@ -4,6 +4,7 @@ Checks the outputs of a workflow run
 """
 import argparse
 from argparse import RawTextHelpFormatter
+from filecmp import dircmp
 import h5py
 import sys
 import numpy as np
@@ -14,16 +15,30 @@ def main(check, control):
 
     Parameters
     ----------
-    results_path : str
+    check_path : str
         Path to output results directory of job
     control_path : str
         Path to control results directory
     """
+    error = False
+
+    #Check that all expected output files are present
+    dcmp = dircmp(check, control)
+    if len(dcmp.left_only) > 0 or len(dcmp.right_only) > 0:
+        if len(dcmp.left_only) > 0:
+            print('ERROR: The following files are present in the output but not in the '
+                  'control: {}'.format(dcmp.left_only))
+        if len(dcmp.right_only) > 0:
+            print('ERROR: The following files are present in the control but not in the '
+                  'output: {}'.format(dcmp.right_only))
+        error = True
+
     check_path = check + '/cal_values/cal_solutions.h5'
     control_path = control + '/cal_values/cal_solutions.h5'
     check_h5 = h5py.File(check_path, 'r')
     control_h5 = h5py.File(control_path, 'r')
 
+    # Check that the calibration solutions match the control ones for matching frequencies
     for solsetname in check_h5:
         soltabnames = [name for name in check_h5[solsetname] 
                             if name not in ('source', 'antenna')]
@@ -41,7 +56,7 @@ def main(check, control):
             check_soltabval = check_h5[solsetname][soltabname]['val']
             control_soltabval = control_h5[solsetname][soltabname]['val']
             matching_vals = control_soltabval
-            if 'freq' in check_h5[solsetname][soltabname].keys():
+            if 'freq' in check_h5[solsetname][soltabname].keys() and args.allow-frequency-subset:
                 check_axes = check_soltabval.attrs['AXES'].decode('utf-8').split(',')
                 freq_axis_index = check_axes.index('freq')
                 check_soltabfreq = check_h5[solsetname][soltabname]['freq'][:]
@@ -53,6 +68,7 @@ def main(check, control):
                                         axis=freq_axis_index)
             if not np.allclose(check_soltabval, matching_vals, 
                                rtol=1e-03, atol=1e-03, equal_nan=True):
+                error = True
                 print("Val array of soltab {} does not match the control".format(soltabname))
                 with open("check_soltab.{}.val".format(soltabname), "w") as f:
                     f.write(str(check_soltabval[:]))
@@ -62,14 +78,17 @@ def main(check, control):
                 print("Val array of soltab {} matches the control".format(soltabname))
     check_h5.close()
     control_h5.close()
-    sys.exit(0)
+    sys.exit(error)
 
 if __name__ == '__main__':
     descriptiontext = "Checks the output of a workflow run.\n"
 
     parser = argparse.ArgumentParser(description=descriptiontext, formatter_class=RawTextHelpFormatter)
-    parser.add_argument('results_path', help='Path to output results')
+    parser.add_argument('check_path', help='Path to output results')
     parser.add_argument('control_path', help='Path to control results')
+    parser.add_argument('--allow-frequency-subset',
+                        help='Use if frequency range in control differs from those in output',
+                        action='store_true')
     args = parser.parse_args()
 
-    main(args.results_path, args.control_path)
+    main(args.check_path, args.control_path)