use it in index CLI, fix bug for single-leaf SBT

sourmash/commands.py

@@ -12,6 +12,7 @@
 from . import signature as sig
 from . import sourmash_args
 from .logging import notify, error, print_results, set_quiet
+from .sbt import scaffold
 from .sbtmh import SearchMinHashesFindBest, SigLeaf
 
 from .sourmash_args import DEFAULT_LOAD_K, FileOutput
@@ -327,10 +328,11 @@ def index(args):
     set_quiet(args.quiet)
     moltype = sourmash_args.calculate_moltype(args)
 
+    tree = None
     if args.append:
         tree = load_sbt_index(args.sbt_name)
-    else:
-        tree = create_sbt_index(args.bf_size, n_children=args.n_children)
+    #else:
+    #    tree = create_sbt_index(args.bf_size, n_children=args.n_children)
 
     if args.sparseness < 0 or args.sparseness > 1.0:
         error('sparseness must be in range [0.0, 1.0].')
@@ -366,6 +368,7 @@ def index(args):
 
         # load all matching signatures in this file
         ss = None
+        sigs = []
         for ss in siglist:
             ksizes.add(ss.minhash.ksize)
             moltypes.add(sourmash_args.get_moltype(ss))
@@ -375,12 +378,19 @@ def index(args):
                 ss.minhash = ss.minhash.downsample(scaled=args.scaled)
             scaleds.add(ss.minhash.scaled)
 
-            tree.insert(ss)
+            if tree:
+                tree.insert(ss)
+            else:
+                sigs.append(ss)
             n += 1
 
         if not ss:
             continue
 
+        if not tree:
+            # TODO: init storage
+            tree = scaffold(sigs, None)
+
         # check to make sure we aren't loading incompatible signatures
         if len(ksizes) > 1 or len(moltypes) > 1:
             error('multiple k-mer sizes or molecule types present; fail.')

sourmash/sbt.py

@@ -1442,7 +1442,23 @@ def scaffold(datasets, storage):
 
     # next_round only contains the root of the SBT
     # Convert from binary tree to nodes/leaves
-    root = next_round.pop()
+    if total_datasets == 1:
+        d = datasets.pop()
+        if isinstance(d, SigLeaf):
+            common = set(d.data.minhash.hashes)
+        elif isinstance(d, Leaf):
+            # TODO: we can't build common as a set;
+            #       use Nodegraph instead?
+            raise NotImplementedError()
+        elif isinstance(d, SourmashSignature):
+            common = set(d.minhash.hashes)
+            d = SigLeaf(d.md5sum(), d)
+        else:
+            raise ValueError("Input not supported")
+
+        root = InternalNode(common, BinaryLeaf(d), None)
+    else:
+        root = next_round.pop()
     assert not next_round
 
     nodes = {}