make sure md5 selector is unique in collection

sourmash/sourmash_args.py

@@ -74,11 +74,20 @@ def load_query_signature(filename, ksize, select_moltype, select_md5=None):
         sys.exit(-1)
 
     if len(sl) and select_md5:
+        found_sig = None
         for sig in sl:
             sig_md5 = sig.md5sum()
             if sig_md5.startswith(select_md5.lower()):
-                sl = [sig]
-                break
+                # make sure we pick only one --
+                if found_sig is not None:
+                    error("Error! Multiple signatures start with md5 '{}'",
+                          select_md5)
+                    error("Please use a longer --md5 selector.")
+                    sys.exit(-1)
+                else:
+                    found_sig = sig
+
+            sl = [found_sig]
 
     if len(sl) and ksize is None:
         ksizes = set([ ss.minhash.ksize for ss in sl ])

tests/test_sourmash.py

@@ -773,6 +773,18 @@ def test_gather_query_db_md5(c):
     assert '340.9 kbp    100.0%  100.0%    ...01593925.1_ASM159392v1_protein.faa.gz' in str(c)
 
 
+@utils.in_thisdir
+def test_gather_query_db_md5_ambiguous(c):
+    # what if we give an ambiguous md5 prefix?
+    db = utils.get_test_data('prot/protein.sbt.zip')
+
+    with pytest.raises(ValueError) as exc:
+        c.run_sourmash('gather', db, db, '--md5', '1')
+
+    err = c.last_result.err
+    assert "Error! Multiple signatures start with md5 '1'" in err
+
+
 @utils.in_tempdir
 def test_gather_lca_db(c):
     # can we do a 'sourmash gather' on an LCA database?