renamed save to _save on signature; changed e-mail to empty by default

sourmash

@@ -106,6 +106,7 @@ Commands can be:
                             help='number of hashes to use in each sketch')
         parser.add_argument('-f', '--force', action='store_true')
         parser.add_argument('-o', '--output', type=argparse.FileType('wt'))
+        parser.add_argument('--email', type=str, default='')
         args = parser.parse_args(args)
 
         print('computing signatures for files:', args.filenames,
@@ -151,7 +152,7 @@ Commands can be:
                         E.add_sequence(s, args.force)
 
             # convert into a signature
-            siglist = [ sig.SourmashSignature('titus@idyll.org', E,
+            siglist = [ sig.SourmashSignature(args.email, E,
                                               filename=filename) ]
 
             # save!
@@ -280,6 +281,7 @@ Commands can be:
         p.add_argument('mash_csvfile')
         p.add_argument('-o', '--output', type=argparse.FileType('wt'),
                        default=sys.stdout)
+        p.add_argument('--email', type=str, default='')
         args = p.parse_args(args)
 
         with open(args.mash_csvfile, 'r') as fp:
@@ -302,8 +304,7 @@ Commands can be:
                 e = sourmash_lib.Estimators(len(hashes), ksize)
                 for h in hashes:
                     e.mh.add_hash(h)
-                s = sig.SourmashSignature('titus@idyll.org',
-                                                  e, filename=name)
+                s = sig.SourmashSignature(args.email, e, filename=name)
                 siglist.append(s)
                 print('loaded signature:', name,
                       s.md5sum()[:8], file=sys.stderr)

sourmash_lib/signature.py

@@ -49,7 +49,7 @@ def name(self):
         else:
             return self.md5sum()[:8]
 
-    def save(self):
+    def _save(self):
         "Return metadata and a dictionary containing the sketch info."
         e = dict(self.d)
         estimator = self.estimator
@@ -65,8 +65,9 @@ def save(self):
             self.d.get('filename'), sketch
 
     def similarity(self, other):
-        "Compute similarity with the stored MinHash."
+        "Compute similarity with the other MinHash signature."
         return self.estimator.similarity(other.estimator)
+    jaccard = similarity
 
 
 def load_signatures(data, select_ksize=None, ignore_md5sum=False):
@@ -135,9 +136,7 @@ def save_signatures(siglist, fp=None):
     "Save multiple signatures into a YAML string (or into file handle 'fp')"
     top_records = {}
     for sig in siglist:
-        email, name, filename, sketch = sig.save()
-        if not email:
-            raise Exception('email must be non-unique')
+        email, name, filename, sketch = sig._save()
         k = (email, name, filename)
 
         x = top_records.get(k, [])
@@ -184,11 +183,24 @@ def test_roundtrip():
     assert sig2.similarity(sig) == 1.0
 
 
+def test_roundtrip_empty_email():
+    e = sourmash_lib.Estimators(n=1, ksize=20)
+    e.add("AT" * 10)
+    sig = SourmashSignature('', e)
+    s = save_signatures([sig])
+    siglist = load_signatures(s)
+    sig2 = siglist[0]
+    e2 = sig2.estimator
+
+    assert sig.similarity(sig2) == 1.0
+    assert sig2.similarity(sig) == 1.0
+
+
 def test_md5():
     e = sourmash_lib.Estimators(n=1, ksize=20)
     e.mh.add_hash(5)
     sig = SourmashSignature('titus@idyll.org', e)
-    print(sig.save())
+    print(sig._save())
     assert sig.md5sum() == 'eae27d77ca20db309e056e3d2dcd7d69', sig.md5sum()