Update tests & constrain behavior in various ways. (#54)

* require that Record objects have name, sequence
* new file for Record-specific tests
* remove 'quality' if set to None

screed/fasta.py

@@ -18,7 +18,7 @@ def fasta_iter(handle, parse_description=False, line=None):
         line = handle.readline()
 
     while line:
-        data = Record()
+        data = {}
 
         line = to_str(line.strip())
         if not line.startswith('>'):
@@ -45,4 +45,4 @@ def fasta_iter(handle, parse_description=False, line=None):
             line = to_str(handle.readline())
 
         data['sequence'] = ''.join(sequenceList)
-        yield data
+        yield Record(**data)

screed/fastq.py

@@ -19,7 +19,7 @@ def fastq_iter(handle, line=None, parse_description=False):
         line = handle.readline()
     line = to_str(line.strip())
     while line:
-        data = Record()
+        data = {}
 
         if line and not line.startswith('@'):
             raise IOError("Bad FASTQ format: no '@' at beginning of line")
@@ -60,4 +60,4 @@ def fastq_iter(handle, line=None, parse_description=False):
             raise IOError('sequence and quality strings must be '
                           'of equal length')
 
-        yield data
+        yield Record(**data)

screed/screedRecord.py

@@ -18,8 +18,22 @@ class Record(MutableMapping):
     Simple dict-like record interface with bag behavior.
     """
 
-    def __init__(self, *args, **kwargs):
-        self.d = dict(*args, **kwargs)
+    def __init__(self, name=None, sequence=None, **kwargs):
+        d = dict()
+        if name is not None:
+            d['name'] = name
+        if sequence is not None:
+            d['sequence'] = sequence
+
+        d.update(kwargs)
+
+        if 'name' not in d:
+            raise TypeError("'name' must be specified")
+        if 'sequence' not in d:
+            raise TypeError("'sequence' must be specified")
+        if 'quality' in d and d['quality'] is None:
+            del d['quality']
+        self.d = d
 
     def __setitem__(self, name, value):
         self.d[name] = value
@@ -38,11 +52,11 @@ def keys(self):
 
     def __getitem__(self, idx):
         if isinstance(idx, slice):
-            trimmed = Record(self.d)
+            trimmed = dict(self.d)
             trimmed['sequence'] = trimmed['sequence'][idx]
             if 'quality' in trimmed:
                 trimmed['quality'] = trimmed['quality'][idx]
-            return Record(trimmed)
+            return Record(**trimmed)
         return self.d[idx]
 
     def __delitem__(self, key):
@@ -178,7 +192,7 @@ def _buildRecord(fieldTuple, dbObj, rowName, queryBy):
         else:
             hackedResult.append((key, value))
 
-    return Record(hackedResult)
+    return Record(**dict(hackedResult))
 
 
 def write_fastx(record, fileobj):

screed/tests/test_fasta.py

@@ -24,6 +24,7 @@ def test_new_record():
     records = list(iter(screed.fasta.fasta_iter(s)))
     assert records[0]['name'] == '1'
     assert records[1]['name'] == '2'
+    assert not hasattr(records[0], 'accuracy')   # check for legacy attribute
 
 
 class Test_fasta(object):

screed/tests/test_record.py

@@ -0,0 +1,48 @@
+from __future__ import absolute_import, unicode_literals, print_function
+from screed import Record
+import pytest
+
+
+@pytest.mark.xfail(raises=TypeError)
+def test_create_noname():
+    r = Record(sequence='ATGGAC')
+
+
+@pytest.mark.xfail(raises=TypeError)
+def test_create_noseq():
+    r = Record(name='somename')
+
+
+def test_create_quality_none():
+    r = Record(name='foo', sequence='ATGACG', quality=None)
+    assert not hasattr(r, 'quality')
+
+
+def test_len():
+    r = Record(name='foo', sequence='ATGACG')
+    assert len(r) == 6
+
+
+# copied over from khmer tests/test_read_parsers.py
+def test_read_type_basic():
+    # Constructing without mandatory arguments should raise an exception
+    with pytest.raises(TypeError):
+        Record()
+
+    name = "895:1:1:1246:14654 1:N:0:NNNNN"
+    sequence = "ACGT"
+    r = Record(name, sequence)
+
+    assert r.name == name
+    assert r.sequence == sequence
+    assert not hasattr(r, 'quality'), x
+    assert not hasattr(r, 'annotations'), x
+
+
+# copied over from khmer tests/test_read_parsers.py
+def test_read_type_attributes():
+    r = Record(sequence='ACGT', quality='good', name='1234', annotations='ann')
+    assert r.sequence == 'ACGT'
+    assert r.quality == 'good'
+    assert r.name == '1234'
+    assert r.annotations == 'ann'