upd with test

Author: ctb

src/utils/mod.rs

@@ -124,6 +124,16 @@ pub fn load_gbassembly_info(input_csv: String) -> Result<(Vec<GBAssemblyData>, u
     let header = rdr.headers()?;
     let expected_header = vec!["accession", "name"];
 
+    let header_v: Vec<_> = header.iter().collect();
+
+    if header_v[0..2].to_vec() != expected_header {
+        return Err(anyhow!(
+            "Invalid column names in CSV file. Columns should be: {:?}, in that order",
+            expected_header
+        ));
+    }
+
+/*  Not needed for the moment, but leaving the code around ;).
     for h in expected_header.iter() {
         if !header.iter().any(|e| *h == e) {
             return Err(anyhow!(
@@ -133,7 +143,7 @@ pub fn load_gbassembly_info(input_csv: String) -> Result<(Vec<GBAssemblyData>, u
             ));
         }
     }
-
+*/
     for h in header.iter() {
         if !expected_header.iter().any(|e| h == *e) {
             eprintln!("WARNING: extra column '{}' in CSV file. Ignoring.", h);

tests/test-data/acc-bad-header-order.csv

@@ -0,0 +1,3 @@
+name,accession
+GCA_000961135.2,GCA_000961135.2 Candidatus Aramenus sulfurataquae isolate AZ1-454
+GCA_000175535.1,GCA_000175535.1 Chlamydia muridarum MopnTet14 (agent of mouse pneumonitis) strain=MopnTet14

tests/test_gbsketch.py

@@ -400,6 +400,26 @@ def test_gbsketch_extra_column(runtmp, capfd):
     assert "WARNING: extra column 'extra' in CSV file. Ignoring." in captured.err
 
 
+def test_gbsketch_bad_column_order(runtmp, capfd):
+    acc_csv = get_test_data('acc-bad-header-order.csv')
+
+    output = runtmp.output('simple.zip')
+    failed = runtmp.output('failed.csv')
+    ch_fail = runtmp.output('checksum_dl_failed.csv')
+
+    with pytest.raises(utils.SourmashCommandFailed):
+        runtmp.sourmash('scripts', 'gbsketch', acc_csv, '-o', output,
+                        '--failed', failed, '-r', '3', '--checksum-fail', ch_fail,
+                        '--param-str', "dna,k=31,scaled=1000", '-p', "protein,k=10,scaled=200")
+
+    assert not os.path.exists(output)
+    assert not runtmp.last_result.out # stdout should be empty
+    captured = capfd.readouterr()
+    print(captured.err)
+
+    assert "Invalid column names in CSV file." in captured.err
+
+
 def test_gbsketch_missing_accfile(runtmp, capfd):
     acc_csv = runtmp.output('acc1.csv')
     output = runtmp.output('simple.zip')