From 234df70530db680bf0134e3e3f06fbf2896f0e80 Mon Sep 17 00:00:00 2001
From: "C. Titus Brown" <titus@idyll.org>
Date: Mon, 7 Mar 2022 14:10:23 -0800
Subject: [PATCH] [MRG] Fix `sourmash prefetch` for multiple ksizes in database
 (#1866)

* add test that breaks prefetch b/c of multiple ksizes

* fix prefetch
---
 src/sourmash/commands.py |  1 +
 tests/test_prefetch.py   | 15 +++++++++++++++
 2 files changed, 16 insertions(+)

diff --git a/src/sourmash/commands.py b/src/sourmash/commands.py
index 16bd1e9672..c29ad99b3a 100644
--- a/src/sourmash/commands.py
+++ b/src/sourmash/commands.py
@@ -1176,6 +1176,7 @@ def prefetch(args):
         sys.exit(-1)
 
     query.minhash = query_mh
+    ksize = query_mh.ksize
 
     # set up CSV output, write headers, etc.
     csvout_fp = None
diff --git a/tests/test_prefetch.py b/tests/test_prefetch.py
index 84ec48a12c..9de8363d06 100644
--- a/tests/test_prefetch.py
+++ b/tests/test_prefetch.py
@@ -37,6 +37,21 @@ def test_prefetch_basic(runtmp, linear_gather):
     assert "a total of 0 query hashes remain unmatched." in c.last_result.err
 
 
+def test_prefetch_select_query_ksize(runtmp, linear_gather):
+    # test prefetch where query and subject db both have multiple ksizes
+    c = runtmp
+
+    ss = utils.get_test_data('GCF_000005845.2_ASM584v2_genomic.fna.gz.sig')
+
+    c.run_sourmash('prefetch', ss, ss, linear_gather)
+    print(c.last_result.status)
+    print(c.last_result.out)
+    print(c.last_result.err)
+
+    assert c.last_result.status == 0
+    assert 'of 4476 distinct query hashes, 4476 were found in matches above threshold.' in c.last_result.err
+
+
 def test_prefetch_query_abund(runtmp, linear_gather):
     c = runtmp