init

src/directsketch.rs

@@ -210,9 +210,13 @@ async fn sketch_data(
     moltype: String,
 ) -> Result<Vec<Signature>> {
     tokio::task::spawn_blocking(move || {
+
         let cursor = Cursor::new(compressed_data);
-        let mut fastx_reader =
-            parse_fastx_reader(cursor).context("Failed to parse FASTA/FASTQ data")?;
+        // use niffler to get decompressed reader
+        let (mut reader, compression) = niffler::get_reader(Box::new(cursor))?;
+        let mut fastx_reader = parse_fastx_reader(&mut reader).context("Failed to parse FASTA/FASTQ data")?;
+        // let mut fastx_reader =
+        //     parse_fastx_reader(cursor).context("Failed to parse FASTA/FASTQ data")?;
 
         let mut set_name = false;
         while let Some(record) = fastx_reader.next() {

tests/test-data/acc-url-xz.csv

@@ -0,0 +1,2 @@
+accession,name,moltype,md5sum,download_filename,url
+achromobacter_xylosoxidans,achromobacter_xylosoxidans,dna,,achromobacter_xylosoxidans__01.asm.genomic.fna.tar.xz,https://ftp.ebi.ac.uk/pub/databases/AllTheBacteria/Releases/0.2/assembly/achromobacter_xylosoxidans__01.asm.tar.xz

tests/test_urlsketch.py

@@ -318,4 +318,26 @@ def test_urlsketch_from_gbsketch_failed(runtmp, capfd):
             assert url == "https://ftp.ncbi.nlm.nih.gov/genomes/all/GCA/000/175/535/GCA_000175535.1_ASM17553v1/GCA_000175535.1_ASM17553v1_protein.faa.gz"
 
 
-# def test_urlsketch_from_urlsketch_failed(runtmp, capfd):
+def test_urlsketch_tarxz(runtmp):
+    acc_csv = get_test_data('acc-url-xz.csv')
+    output = runtmp.output('simple.zip')
+    failed = runtmp.output('failed.csv')
+
+    # sig3 = get_test_data('GCA_000961135.2.protein.sig.gz')
+    # ss1 = sourmash.load_one_signature(sig1, ksize=31)
+
+    runtmp.sourmash('scripts', 'urlsketch', acc_csv, '-o', output,
+                    '--failed', failed, '-r', '1',
+                    '--param-str', "dna,k=31,scaled=1000", '-p', "protein,k=10,scaled=200")
+
+    assert os.path.exists(output)
+    assert not runtmp.last_result.out # stdout should be empty
+
+    # idx = sourmash.load_file_as_index(output)
+    # sigs = list(idx.signatures())
+
+    # assert len(sigs) == 1
+    # for sig in sigs:
+    #     if 'GCA_000175535.1' in sig.name:
+    #         assert sig.name == ss1.name
+    #         assert sig.md5sum() == ss1.md5sum()