MRG: update to (prospective) sourmash r0.18.0

Cargo.toml

@@ -12,7 +12,8 @@ crate-type = ["cdylib"]
 pyo3 = { version = "0.23.3", features = ["extension-module", "anyhow"] }
 rayon = "1.10.0"
 serde = { version = "1.0.216", features = ["derive"] }
-sourmash = { version = "0.17.2", features = ["branchwater"] }
+#sourmash = { version = "0.17.2", features = ["branchwater"] }
+sourmash = { git = "https://github.com/sourmash-bio/sourmash.git", branch = "latest", features = ["branchwater"] }
 serde_json = "1.0.133"
 niffler = "2.4.0"
 log = "0.4.22"

src/python/tests/test_fastgather.py

@@ -427,40 +427,6 @@ def test_bad_against_2(runtmp, capfd):
     )
 
 
-def test_bad_against_3(runtmp, capfd):
-    # test with a bad against (a .sig.gz file renamed as zip file)
-    query = get_test_data("SRR606249.sig.gz")
-
-    sig2 = get_test_data("2.fa.sig.gz")
-    against_zip = runtmp.output("against.zip")
-    # cp sig2 into against_zip
-    with open(against_zip, "wb") as fp:
-        with open(sig2, "rb") as fp2:
-            fp.write(fp2.read())
-
-    g_output = runtmp.output("gather.csv")
-    p_output = runtmp.output("prefetch.csv")
-
-    with pytest.raises(utils.SourmashCommandFailed):
-        runtmp.sourmash(
-            "scripts",
-            "fastgather",
-            query,
-            against_zip,
-            "-o",
-            g_output,
-            "--output-prefetch",
-            p_output,
-            "-s",
-            "100000",
-        )
-
-    captured = capfd.readouterr()
-    print(captured.err)
-
-    assert "InvalidArchive" in captured.err
-
-
 @pytest.mark.xfail(reason="should work, bug")
 def test_against_multisigfile(runtmp, zip_against):
     # test against a sigfile that contains multiple sketches

src/python/tests/test_fastmultigather.py

@@ -561,34 +561,6 @@ def test_sig_query(runtmp, capfd, indexed):
         }.issubset(keys)
 
 
-def test_bad_query(runtmp, capfd, indexed):
-    # test with a bad query (a .sig.gz file renamed as zip file)
-    against_list = runtmp.output("against.txt")
-
-    sig2 = get_test_data("2.fa.sig.gz")
-    sig47 = get_test_data("47.fa.sig.gz")
-    sig63 = get_test_data("63.fa.sig.gz")
-
-    query_zip = runtmp.output("query.zip")
-    # cp sig2 into query_zip
-    with open(query_zip, "wb") as fp:
-        with open(sig2, "rb") as fp2:
-            fp.write(fp2.read())
-
-    make_file_list(against_list, [sig2, sig47, sig63])
-
-    if indexed:
-        against_list = index_siglist(runtmp, against_list, runtmp.output("db"))
-
-    with pytest.raises(utils.SourmashCommandFailed):
-        runtmp.sourmash("scripts", "fastmultigather", query_zip, against_list)
-
-    captured = capfd.readouterr()
-    print(captured.err)
-
-    assert "InvalidArchive" in captured.err
-
-
 def test_missing_query(runtmp, capfd, indexed):
     # test missing query
     query_list = runtmp.output("query.txt")
@@ -736,33 +708,6 @@ def test_bad_against(runtmp, capfd):
     )
 
 
-def test_bad_against_2(runtmp, capfd, zip_query):
-    # test with a bad against (a .sig.gz file renamed as zip file)
-    query = get_test_data("SRR606249.sig.gz")
-    query_list = runtmp.output("query.txt")
-    make_file_list(query_list, [query])
-
-    sig2 = get_test_data("2.fa.sig.gz")
-    against_zip = runtmp.output("against.zip")
-    # cp sig2 into query_zip
-    with open(against_zip, "wb") as fp:
-        with open(sig2, "rb") as fp2:
-            fp.write(fp2.read())
-
-    if zip_query:
-        query_list = zip_siglist(runtmp, query_list, runtmp.output("query.zip"))
-
-    with pytest.raises(utils.SourmashCommandFailed):
-        runtmp.sourmash(
-            "scripts", "fastmultigather", query_list, against_zip, "-s", "100000"
-        )
-
-    captured = capfd.readouterr()
-    print(captured.err)
-
-    assert "InvalidArchive" in captured.err
-
-
 def test_empty_against(runtmp, capfd):
     # test bad 'against' file - in this case, an empty one
     query = get_test_data("SRR606249.sig.gz")

src/python/tests/test_index.py

@@ -429,27 +429,6 @@ def test_index_zipfile_multiparam(runtmp, capfd, toggle_internal_storage):
         runtmp.sourmash("scripts", "index", zipf, "-o", output, toggle_internal_storage)
 
 
-def test_index_zipfile_bad(runtmp, capfd):
-    # test with a bad input zipfile (a .sig.gz file renamed as zip file)
-    sig2 = get_test_data("2.fa.sig.gz")
-
-    query_zip = runtmp.output("query.zip")
-    # cp sig2 into query_zip
-    with open(query_zip, "wb") as fp:
-        with open(sig2, "rb") as fp2:
-            fp.write(fp2.read())
-
-    output = runtmp.output("out.csv")
-
-    with pytest.raises(utils.SourmashCommandFailed):
-        runtmp.sourmash("scripts", "index", query_zip, "-o", output)
-
-    captured = capfd.readouterr()
-    print(captured.err)
-
-    assert "Couldn't find End Of Central Directory Record" in captured.err
-
-
 def test_index_check(runtmp, toggle_internal_storage):
     # test check index
     siglist = runtmp.output("db-sigs.txt")
@@ -506,3 +485,27 @@ def test_index_subdir(runtmp, toggle_internal_storage):
     print(runtmp.last_result.err)
 
     runtmp.sourmash("scripts", "check", output)
+
+
+def test_index_misnamed_zipfile(runtmp, capfd):
+    # test with a misnamed input zipfile (a .sig.gz file renamed as zip file)
+    # (This is a generic test that checks to make sure misnamed zip files
+    # can be loaded. It's not really specific to index. See
+    # https://github.com/sourmash-bio/sourmash_plugin_branchwater/issues/551)
+    sig2 = get_test_data("2.fa.sig.gz")
+
+    query_zip = runtmp.output("query.zip")
+    # cp sig2 into query_zip
+    with open(query_zip, "wb") as fp:
+        with open(sig2, "rb") as fp2:
+            fp.write(fp2.read())
+
+    output = runtmp.output("out.rocksdb")
+
+    runtmp.sourmash("scripts", "index", query_zip, "-o", output)
+
+    captured = capfd.readouterr()
+    print(captured.err)
+
+    assert os.path.exists(output)
+    assert os.path.isdir(output)