adding first salinity tests

tests/inputs/test_sample_info.yaml

@@ -39,4 +39,47 @@ tests:
       id: TEST:captest1
       text: bill clinton
     output:
-      text: Bill Clinton
+      text: Bill Clinton
+  - description: missing space 1
+    sample:
+      id: TEST:missing_space1
+      text: 400g/L
+    output:
+      text: 400 g/L
+  - description: missing space 2
+    sample:
+      id: TEST:missing_space2
+      text: 1%
+    output:
+      text: 1 %
+  - description: categorical vs numerical
+    sample:
+      id: TEST:cat_vs_num
+      text: Halophile
+    output:
+      text: NaN
+  - description: capital units
+    sample:
+      id: TEST:cap_units
+      text: 0.2 psu
+    output:
+      text: 0.2 PSU
+  - description: capital units and space
+    sample:
+      id: TEST:cap_units_and_space
+      text: 0.2psu
+    output:
+      text: 0.2 PSU
+  - description: plus minus error
+    sample:
+      id: TEST:plus_min_error
+      text: 17.3 +/- 2.0 PPT
+    output:
+      text: NaN
+  - description: with unit description
+    sample:
+      id: TEST:with_unit_descr
+      text: 40 PSU (practical salinity units)
+    output:
+      text: 40 PSU
+

tests/test_salinity.py

@@ -0,0 +1,35 @@
+# -*- coding: utf-8 -*-
+import os
+import unittest
+
+import yaml
+
+from sample_annotator import capitalizer
+# MODEL_DIR, INPUT_DIR, OUTPUT_DIR
+from tests import INPUT_DIR
+
+"""Test the ability to capitalize a text slot."""
+
+"""Run as follows to get see test-time printouts:"""
+
+"""python -m pytest -sv  tests/test_salinity.py"""
+
+# INPUT_DIR comes from __init__.py
+PWD = os.path.dirname(os.path.realpath(__file__))
+TEST_DATA = os.path.join(INPUT_DIR, 'test_sample_info.yaml')
+
+
+class TestSalinity(unittest.TestCase):
+    """salinity unit tests."""
+
+    def test_missing_space1(self):
+        with open(TEST_DATA) as stream:
+            test_obj = yaml.load(stream, Loader=yaml.FullLoader)
+        for t in test_obj.get('tests'):
+            desc = t.get('description', None)
+            # pdb.set_trace()
+            if desc == 'missing space 1':
+                current_input = t['sample']['text']
+                processed_input = capitalizer.capitalizer(current_input)
+                expected_output = t['output']['text']
+                assert processed_input == expected_output

tests/test_salinity_annotate.py

@@ -0,0 +1,40 @@
+# -*- coding: utf-8 -*-
+import os
+import yaml
+import unittest
+from sample_annotator.measurements.measurements import MeasurementEngine
+from sample_annotator.report_model import AnnotationReport
+
+# MODEL_DIR, INPUT_DIR, OUTPUT_DIR
+from tests import INPUT_DIR
+
+"""Test the ability to capitalize a text slot."""
+
+"""Run as follows to get see test-time printouts:"""
+
+"""python -m pytest -sv  tests/test_salinity.py"""
+
+# INPUT_DIR comes from __init__.py
+PWD = os.path.dirname(os.path.realpath(__file__))
+TEST_DATA = os.path.join(INPUT_DIR, 'test_sample_info.yaml')
+
+
+class TestSalinityAnnotate(unittest.TestCase):
+    """salinity unit tests."""
+
+    report = AnnotationReport(messages=[])
+    m = MeasurementEngine()
+
+    with open(TEST_DATA) as stream:
+        test_obj = yaml.load(stream, Loader=yaml.FullLoader)
+    for t in test_obj.get('tests'):
+        desc = t.get('description', None)
+        # pdb.set_trace()
+        if desc == 'missing space 1':
+            current_input = t['sample']['text']
+
+            processed_input = m.repair(current_input, report=report)
+            print(processed_input)
+
+            expected_output = t['output']['text']
+            assert processed_input == expected_output