Add beginnings of unit test coverage of align

.gitignore

@@ -21,7 +21,9 @@ dist/
 *.pyc
 __pycache__/
 .cache/
+.coverage
 augur.egg-info/
+.mypy_cache/
 nextstrain_augur.egg-info/
 env/
 .pytest_cache

pytest.python3.ini

@@ -1,2 +1,14 @@
 [pytest]
-addopts = --doctest-modules tests/python3/ augur/
+addopts =
+	# do not capture any output---necessary for interactive breakpoints
+	-s
+
+	# run any doctests found in .py modules
+	--doctest-modules
+
+	# calculate test coverage for the `augur` module
+	--cov=augur
+
+testpaths =
+	augur/
+	tests/

run_tests.sh

@@ -0,0 +1,16 @@
+#!/bin/sh
+
+case "$@" in
+	*-k*) partial_test=1 ;;
+esac
+
+if [ "$partial_test" = 1 ]; then
+	# skip test coverage when running a subset of the test suite
+	coverage_arg='--no-cov'
+else
+	coverage_arg=''
+fi
+
+python3 -m pytest -c pytest.python3.ini $coverage_arg "$@"
+
+exit $?

setup.py

@@ -49,7 +49,8 @@
         ],
         'dev': [
             "pylint >=1.7.6, ==1.7.*",
-            "pytest >=3.2.1, ==3.*",
+            "pytest >=5.4.1, ==5.4.*",
+            "pytest-cov >=2.8.1, ==2.8.*",
             "recommonmark >=0.5.0, ==0.*",
             "Sphinx >=2.0.1, ==2.*",
             "sphinx-argparse >=0.2.5, ==0.*",

tests/test_align.py

@@ -0,0 +1,75 @@
+from Bio.Align import MultipleSeqAlignment
+from Bio.Seq import Seq
+from Bio.SeqRecord import SeqRecord
+
+from augur import align
+
+import pytest
+
+
+class TestAlign:
+    def test_make_gaps_ambiguous(self):
+        alignment = MultipleSeqAlignment(
+            [SeqRecord(Seq("G-AC")), SeqRecord(Seq("----")), SeqRecord(Seq("TAGC"))]
+        )
+
+        align.make_gaps_ambiguous(alignment)
+
+        assert [seq_record.seq for seq_record in alignment] == [
+            Seq("GNAC"),
+            Seq("NNNN"),
+            Seq("TAGC"),
+        ]
+
+    def test_check_duplicates_no_arguments(self):
+        assert align.check_duplicates() is None
+
+    def test_check_duplicates_strings_with_no_duplicates(self):
+        assert align.check_duplicates("GTAC", "CGTT") is None
+
+    def test_check_duplicates_MSA_with_no_duplicates(self):
+        alignment = MultipleSeqAlignment(
+            [
+                SeqRecord(Seq("GTAC"), name="seq1"),
+                SeqRecord(Seq("CGTT"), name="seq2"),
+                SeqRecord(Seq("TAGC"), name="seq3"),
+            ]
+        )
+        assert align.check_duplicates(alignment) is None
+
+    def test_check_duplicates_MSA_and_string_with_no_duplicates(self):
+        alignment = MultipleSeqAlignment(
+            [
+                SeqRecord(Seq("GTAC"), name="seq1"),
+                SeqRecord(Seq("CGTT"), name="seq2"),
+                SeqRecord(Seq("TAGC"), name="seq3"),
+            ]
+        )
+        assert align.check_duplicates(alignment, "TGTT") is None
+
+    def test_check_duplicates_string_with_duplicates(self):
+        with pytest.raises(align.AlignmentError):
+            assert align.check_duplicates("GTAC", "CGTT", "CGTT")
+
+    def test_check_duplicates_MSA_with_duplicates(self):
+        alignment = MultipleSeqAlignment(
+            [
+                SeqRecord(Seq("GTAC"), name="seq1"),
+                SeqRecord(Seq("CGTT"), name="seq2"),
+                SeqRecord(Seq("TAGC"), name="seq3"),
+                SeqRecord(Seq("TAGC"), name="seq3"),
+            ]
+        )
+        with pytest.raises(align.AlignmentError):
+            assert align.check_duplicates(alignment)
+
+    def test_check_duplicates_MSA_and_string_with_duplicates(self):
+        alignment = MultipleSeqAlignment(
+            [
+                SeqRecord(Seq("GTAC"), name="seq1"),
+                SeqRecord(Seq("CGTT"), name="seq2"),
+                SeqRecord(Seq("TAGC"), name="seq3"),
+            ]
+        )
+        with pytest.raises(align.AlignmentError):
+            assert align.check_duplicates(alignment, "seq3")