Merge pull request #223 from maksteel/unicode_219

ignores non-utf8 chars in the header

Author: Lucas-Mc

sample-data/test_generator_2.dat

@@ -0,0 +1,13 @@
+test_generator_2 12 1 600 12:44:02 10/12/2009
+test_generator_2.dat 16x200 32.7675/uV 15 0 3276 0 0 squarewave
+test_generator_2.dat 16x200 32.7675/uV 15 0 -3276 480 0 ramp
+test_generator_2.dat 16x200 32.7675/uV 15 0 3276 -1920 0 pulse
+test_generator_2.dat 16x200 32.7675/uV 15 0 0 21000 0 ECG
+test_generator_2.dat 16x200 32.7675/uV 15 0 2752 -12878 0 noise
+test_generator_2.dat 16x200 32.7675/uV 15 0 102 0 0 sine 1 Hz
+test_generator_2.dat 16x200 32.7675/uV 15 0 814 0 0 sine 8 Hz
+test_generator_2.dat 16x200 32.7675/uV 15 0 864 0 0 sine 8.5 Hz
+test_generator_2.dat 16x200 32.7675/� 15 0 1487 0 0 sine 15 Hz
+test_generator_2.dat 16x200 32.7675/� 15 0 1668 0 0 sine 17 Hz
+test_generator_2.dat 16x200 32.7675/uV 15 0 3276 0 0 sine 50 Hz
+test_generator_2.dat 16x51 32767.5 15 0 12331 29678 0 EDF Annotations

sample-data/test_generator_2.edf

@@ -411,6 +411,13 @@ def test_4d(self):
 
         assert np.array_equal(sig_round, sig_target)
 
+    def test_header_with_non_utf8(self):
+        """
+        Ignores non-utf8 characters in the header part.
+        """
+        record = wfdb.rdrecord("sample-data/test_generator_2")
+        sig_units_target = ['uV', 'uV', 'uV', 'uV', 'uV', 'uV', 'uV', 'uV', 'mV', 'mV', 'uV', 'mV']
+        assert record.units.__eq__(sig_units_target)
 
     @classmethod
     def tearDownClass(cls):

sample-data/test_generator_2.hea

@@ -826,7 +826,7 @@ def _read_header_lines(base_record_name, dir_name, pn_dir):
 
     # Read local file
     if pn_dir is None:
-        with open(os.path.join(dir_name, file_name), 'r') as fp:
+        with open(os.path.join(dir_name, file_name), 'r', errors='ignore') as fp:
             # Record line followed by signal/segment lines if any
             header_lines = []
             # Comment lines