Have PrefixComparer and SuffixComparer take min_overlap into account

The locate() methods should return None if the min_overlap is not fulfilled.
The Prefix-/SuffixComparer did not do so.

Closes #376

CHANGES.rst

@@ -2,9 +2,11 @@
 Changes
 =======
 
-v2.2 (2019-03-20)
------------------
+v2.2 (in development)
+---------------------
 
+* :issue:`376`: Fix a crash when using anchored 5' adapters together with
+  ``--no-indels`` and trying to trim an empty read.
 * :issue:`369`: Fix a crash when attempting to trim an empty read using a ``-g``
   adapter with wildcards.
 

src/cutadapt/_align.pyx

@@ -549,6 +549,7 @@ cdef class PrefixComparer:
         int m
         int max_k  # max. number of errors
         readonly int effective_length
+        int min_overlap
 
     # __init__ instead of __cinit__ because we need to override this in SuffixComparer
     def __init__(
@@ -557,6 +558,7 @@ cdef class PrefixComparer:
         double max_error_rate,
         bint wildcard_ref=False,
         bint wildcard_query=False,
+        int min_overlap=1,
     ):
         self.wildcard_ref = wildcard_ref
         self.wildcard_query = wildcard_query
@@ -570,6 +572,9 @@ cdef class PrefixComparer:
             raise ValueError("max_error_rate must be between 0 and 1")
         self.max_k = int(max_error_rate * self.effective_length)
         self.reference = reference.encode('ascii').upper()
+        if min_overlap < 1:
+            raise ValueError("min_overlap must be at least 1")
+        self.min_overlap = min_overlap
         if self.wildcard_ref:
             self.reference = self.reference.translate(IUPAC_TABLE)
         elif self.wildcard_query:
@@ -621,7 +626,7 @@ cdef class PrefixComparer:
                     matches += 1
 
         errors = length - matches
-        if errors > self.max_k:
+        if errors > self.max_k or length < self.min_overlap:
             return None
         return (0, length, 0, length, matches, length - matches)
 
@@ -634,8 +639,9 @@ cdef class SuffixComparer(PrefixComparer):
         double max_error_rate,
         bint wildcard_ref=False,
         bint wildcard_query=False,
+        int min_overlap=1,
     ):
-        super().__init__(reference[::-1], max_error_rate, wildcard_ref, wildcard_query)
+        super().__init__(reference[::-1], max_error_rate, wildcard_ref, wildcard_query, min_overlap)
 
     def locate(self, str query):
         cdef int n = len(query)

src/cutadapt/adapters.py

@@ -710,8 +710,13 @@ def __init__(self, sequence, where, remove=None, max_error_rate=0.1, min_overlap
         self.indels = indels
         if self.is_anchored and not self.indels:
             aligner_class = align.PrefixComparer if self.where is Where.PREFIX else align.SuffixComparer
-            self.aligner = aligner_class(self.sequence, self.max_error_rate,
-                wildcard_ref=self.adapter_wildcards, wildcard_query=self.read_wildcards)
+            self.aligner = aligner_class(
+                self.sequence,
+                self.max_error_rate,
+                wildcard_ref=self.adapter_wildcards,
+                wildcard_query=self.read_wildcards,
+                min_overlap=self.min_overlap
+            )
         else:
             # TODO
             # Indels are suppressed by setting their cost very high, but a different algorithm

tests/test_adapters.py

@@ -328,3 +328,11 @@ def test_anywhere_parameter():
     cutter = AdapterCutter([adapter])
     trimmed_read = cutter(read, [])
     assert trimmed_read.sequence == ''
+
+
+@pytest.mark.parametrize("where", [Where.PREFIX, Where.SUFFIX])
+def test_no_indels_empty_read(where):
+    # Issue #376
+    adapter = Adapter('ACGT', where=where, indels=False)
+    empty = Sequence('name', '')
+    adapter.match_to(empty)