Merge pull request #30 from jenshnielsen/test_sample_corpus

Add test of sample corpus to unittest suite

.travis.yml

@@ -50,7 +50,7 @@ script:
       if [ "$JYTHON" == "true" ]; then
         py.test
       else
-        py.test --cov=pyoracc --cov-report xml
+        py.test --cov=pyoracc --cov-report xml --runslow
       fi
 
   - pep8 --exclude=parsetab.py .

conftest.py

@@ -0,0 +1,6 @@
+import pytest
+
+
+def pytest_addoption(parser):
+    parser.addoption("--runslow", action="store_true",
+                     help="run slow tests")

pyoracc/test/fixtures/__init__.py

@@ -16,6 +16,23 @@ def tiny_corpus():
     return os.path.join(os.path.dirname(here), 'tiny_corpus')
 
 
+def sample_corpus():
+    return os.path.join(os.path.dirname(here), 'sample_corpus')
+
+
+def whole_corpus():
+    """
+    Return the path to the full oracc corpus. This is not bundled with
+    pyoracc and only works if the environmental variable oracc_corpus_path
+    is set to the correct directory.
+    """
+    try:
+        oracc_corpus_path = os.environ['oracc_corpus_path']
+    except KeyError:
+        oracc_corpus_path = None
+    return oracc_corpus_path
+
+
 def anzu():
     return sample_file("anzu")
 

pyoracc/test/model/test_corpus.py

@@ -1,9 +1,35 @@
+import pytest
+
 from ...model.corpus import Corpus
 
-from ..fixtures import tiny_corpus
+from ..fixtures import tiny_corpus, sample_corpus, whole_corpus
+
+
+slow = pytest.mark.skipif(
+    not pytest.config.getoption("--runslow"),
+    reason="need --runslow option to run"
+)
 
 
 def test_tiny():
     corpus = Corpus(source=tiny_corpus())
     assert corpus.successes == 1
     assert corpus.failures == 1
+
+
+@slow
+def test_sample():
+    corpus = Corpus(source=sample_corpus())
+    assert corpus.successes == 36
+    assert corpus.failures == 3
+
+
+@pytest.mark.skipif(not whole_corpus(),
+                    reason="Need to set oracc_corpus_path to point "
+                           "to the whole corpus, which is not bundled with "
+                           "pyoracc")
+@slow
+def test_whole():
+    corpus = Corpus(source=whole_corpus())
+    assert corpus.successes == 2474
+    assert corpus.failures == 394