automatically parse testreport output on Travis

Includes a python function to parse the testreport output and fail the Travis build as required.

Changes to .travis.yml to install required modules, pipe testreport output to file, and call python function.

Author: edoddridge

.travis.yml

@@ -10,7 +10,9 @@ before_install:
  - docker exec -i fc11-testreport ls -altr /MITgcm
  - docker exec -i fc11-testreport yum install python-pip
  - docker exec -i fc11-testreport yum install gcc-gfortran
+ - docker exec -i fc11-testreport yum install python-argparse
 
 script: 
  - echo `pwd`
- - docker exec -i fc11-testreport bash -c "cd /MITgcm/verification; ./testreport -t tutorial_barotropic_gyre"
+ - docker exec -i fc11-testreport bash -c "cd /MITgcm/verification; ./testreport -t tutorial_barotropic_gyre | tee tutorial_barotropic_gyre/testreport_out.txt"
+ - docker exec -i fc11-testreport bash -c "cd /MITgcm/verification; python verification_parser.py -filename 'tutorial_barotropic_gyre/testreport_out.txt' -threshold 15"

verification/verification_parser.py

@@ -0,0 +1,51 @@
+"""Parse verification outputs."""
+
+import re
+
+def verification_parser(filename, threshold):
+    with open(filename) as f:
+        lines = f.readlines()
+
+        first_match = True
+
+        # extract lines from output
+        for i, line in enumerate(lines):
+            if line[:5] == '2 d e':
+                if first_match:
+                    # skip the first match, since it doesn't contain output,
+                    # but set to false to catch next matches.
+                    first_match = False
+                else:
+                    test_results = lines[i+2]
+
+        # split test_results into a list with values for each number. 
+        # this uses spaces and the < > characters to separate the numbers.
+        test_results = re.split('[ ><]',test_results)
+        # ignore the Genmake, depend, make, and run checks, as well as
+        # the "pass" or "fail" and test name at the end of the line
+        test_results = test_results[4:-3]
+        # convert to floats
+        dp_similarity = []
+        for i, x in enumerate(test_results):
+            try:
+                dp_similarity.append(float(x))
+            except ValueError:
+                pass
+
+        assert all(elements > threshold for elements in dp_similarity)
+
+if __name__ == '__main__':
+
+    import argparse
+
+    parser = argparse.ArgumentParser(description='Check that verification simulation passed the test.')
+
+    parser.add_argument('-filename', type=str, 
+                        help='path to output file from the verification test')
+
+    parser.add_argument('-threshold', type=int, default=15, 
+                        help='number of decimal places of similarity required for test to pass')
+
+    args = parser.parse_args()
+
+    verification_parser(**vars(args))