Cigar tools: fix strips of empty queries

cfe-lab · Nov 14, 2023 · 38f8833 · 38f8833
1 parent 98e9240
commit 38f8833
Show file tree

Hide file tree

Showing 2 changed files with 31 additions and 2 deletions.
diff --git a/micall/tests/test_cigar_tools.py b/micall/tests/test_cigar_tools.py
@@ -341,6 +341,35 @@ def test_cigar_hit_strip_combines_with_connect(hit, cut_point):
     assert left.connect(right).coordinate_mapping == hit.coordinate_mapping
 
 
+@pytest.mark.parametrize('hit', [x[0] for x in cigar_hit_ref_cut_cases])
+def test_cigar_hit_strip_never_crashes(hit):
+    hit.rstrip_query().lstrip_query()
+    hit.lstrip_query().rstrip_query()
+    hit.lstrip_query().lstrip_query()
+    hit.rstrip_query().rstrip_query()
+
+
+@pytest.mark.parametrize('hit', [x[0] for x in cigar_hit_ref_cut_cases])
+def test_cigar_hit_strip_is_idempotent(hit):
+    h1 = hit.rstrip_query()
+    assert h1 == h1.rstrip_query() == h1.rstrip_query().rstrip_query()
+
+    h1 = hit.lstrip_query()
+    assert h1 == h1.lstrip_query() == h1.lstrip_query().lstrip_query()
+
+    h1 = hit.lstrip_query().rstrip_query()
+    assert h1 == h1.lstrip_query() == h1.rstrip_query()
+
+    h1 = hit.rstrip_query().lstrip_query()
+    assert h1 == h1.rstrip_query() == h1.lstrip_query()
+
+
+@pytest.mark.parametrize('hit', [x[0] for x in cigar_hit_ref_cut_cases])
+def test_cigar_hit_strips_are_commutative(hit):
+    assert hit.rstrip_query().lstrip_query() \
+        == hit.lstrip_query().rstrip_query()
+
+
 @pytest.mark.parametrize('hit, cut_point', [(x[0], x[1]) for x in cigar_hit_ref_cut_cases
                                             if not isinstance(x[2], Exception)
                                             and not 'N' in str(x[0].cigar)])

diff --git a/micall/utils/cigar_tools.py b/micall/utils/cigar_tools.py
@@ -580,7 +580,7 @@ def cut_reference(self, cut_point: float) -> Tuple['CigarHit', 'CigarHit']:
     def lstrip_query(self) -> 'CigarHit':
         """ Return a copy of the CigarHit with leading (unmatched) query elements removed. """
 
-        if self.query_length == 0:
+        if len(self.coordinate_mapping.ref_to_query) == 0:
             return self
 
         closest_ref = self.coordinate_mapping.ref_to_query.closest_key(self.r_st - 1)
@@ -591,7 +591,7 @@ def lstrip_query(self) -> 'CigarHit':
     def rstrip_query(self) -> 'CigarHit':
         """ Return a copy of the CigarHit with trailing (unmatched) query elements removed. """
 
-        if self.query_length == 0:
+        if len(self.coordinate_mapping.ref_to_query) == 0:
             return self
 
         closest_ref = self.coordinate_mapping.ref_to_query.closest_key(self.r_ei + 1)