[MRG] support loading of empty picklists

src/sourmash/picklist.py

@@ -140,7 +140,7 @@ def init(self, values=[]):
         self.pickset = set(values)
         return self.pickset
 
-    def load(self, pickfile, column_name):
+    def load(self, pickfile, column_name, *, allow_empty=False):
         "load pickset, return num empty vals, and set of duplicate vals."
         pickset = self.init()
 
@@ -150,6 +150,7 @@ def load(self, pickfile, column_name):
         n_empty_val = 0
         dup_vals = set()
         with open(pickfile, newline='') as csvfile:
+            self.pickfile = pickfile
             x = csvfile.readline()
 
             # skip leading comment line in case there's a manifest header
@@ -160,7 +161,10 @@ def load(self, pickfile, column_name):
 
             r = csv.DictReader(csvfile)
             if not r.fieldnames:
-                raise ValueError(f"empty or improperly formatted pickfile '{pickfile}'")
+                if not allow_empty:
+                    raise ValueError(f"empty or improperly formatted pickfile '{pickfile}'")
+                else:
+                    return 0, 0
 
             if column_name not in r.fieldnames:
                 raise ValueError(f"column '{column_name}' not in pickfile '{pickfile}'")
@@ -180,7 +184,6 @@ def load(self, pickfile, column_name):
                 else:
                     self.add(col)
 
-        self.pickfile = pickfile
         return n_empty_val, dup_vals
 
     def add(self, value):

tests/test-data/picklist/empty.csv

@@ -0,0 +1 @@
+

tests/test_picklist.py

@@ -0,0 +1,23 @@
+"""
+Tests for the picklist API.
+"""
+import pytest
+import sourmash
+
+import sourmash_tst_utils as utils
+from sourmash.picklist import SignaturePicklist
+
+
+def test_load_empty_picklist_fail():
+    empty = utils.get_test_data('picklist/empty.csv')
+
+    pl = SignaturePicklist('manifest')
+    with pytest.raises(ValueError):
+        pl.load(empty, 'foo', allow_empty=False)
+
+
+def test_load_empty_picklist_allow():
+    empty = utils.get_test_data('picklist/empty.csv')
+
+    pl = SignaturePicklist('manifest')
+    pl.load(empty, 'foo', allow_empty=True)