check for directories, fail appropriately

src/python/tests/test_pairwise.py

@@ -20,6 +20,16 @@ def test_installed(runtmp):
     assert "usage:  pairwise" in runtmp.last_result.err
 
 
+def test_on_dir(runtmp, capfd):
+    with pytest.raises(utils.SourmashCommandFailed):
+        runtmp.sourmash("scripts", "pairwise", runtmp.output(""),
+                        "-o", runtmp.output('xxx.csv'))
+
+    captured = capfd.readouterr()
+    print(captured.err)
+    assert "arbitrary directories are not supported" in captured.err
+
+
 def test_simple_no_ani(runtmp, capfd, zip_query, indexed):
     # test basic execution!
     query_list = runtmp.output("query.txt")

src/utils/mod.rs

@@ -14,7 +14,7 @@ use serde::{Deserialize, Serialize};
 // use rust_decimal::{MathematicalOps, Decimal};
 use std::cmp::{Ordering, PartialOrd};
 use std::collections::BinaryHeap;
-use std::fs::{create_dir_all, File};
+use std::fs::{create_dir_all, metadata, File};
 use std::io::{BufWriter, Write};
 use std::panic;
 use std::sync::atomic;
@@ -555,6 +555,14 @@ pub fn load_collection(
         }
     });
 
+    // we support RocksDB directory paths, but nothing else, unlike sourmash.
+    if collection.is_none() {
+        let path_metadata = metadata(sigpath.clone()).expect("getting path metadata failed");
+        if path_metadata.is_dir() {
+            bail!("arbitrary directories are not supported as input");
+        }
+    }
+
     let collection =
         collection.or_else(
             || match MultiCollection::from_standalone_manifest(&sigpath) {