MRG: support multiple input files for singlesketch

doc/README.md

@@ -5,7 +5,7 @@
 | command | functionality | docs |
 | -------- | -------- | -------- |
 | `manysketch` | Rapidly build sketches for many input files     | [link](#Running-manysketch)     |
-| `singlesketch` | Sketch a single sequence file | [link](#Running-singlesketch)
+| `singlesketch` | Sketch a single sample | [link](#Running-singlesketch)
 | `fastgather` | Multithreaded `gather` of **one** metagenome against a database| [link](#Running-fastgather)
 | `fastmultigather` | Multithreaded `gather` of **multiple** metagenomes against a database | [link](#Running-fastmultigather)
 | `manysearch` | Multithreaded containment search for many queries in many large metagenomes | [link](#Running-manysearch)
@@ -259,19 +259,21 @@ In this case, three sketches of `protein`, `dayhoff`, and `hp` moltypes were mad
 
 ## Running `singlesketch`
 
-The `singlesketch` command generates a sketch for a single sequence file.
+The `singlesketch` command generates a sketch for a single sample, with one or more input FASTA/FASTQ files.
 
 ### Basic Usage
+
 ```bash
 sourmash scripts singlesketch input.fa -p k=21,scaled=1000,dna -o output.sig --name signature_name
 ```
+
 ### Using `stdin/stdout`
+
 You can use `-` for `stdin` and output the result to `stdout`:
 ```bash
 cat input.fa | sourmash scripts singlesketch - -o -
 ```
 
-
 ### Running `multisearch` and `pairwise`
 
 The `multisearch` command compares one or more query genomes, and one or more subject genomes. It differs from `manysearch` because it loads everything into memory.

src/lib.rs

@@ -324,15 +324,15 @@ fn do_manysketch(
 }
 
 #[pyfunction]
-#[pyo3(signature = (input_filename, input_moltype, param_str, output, name))]
+#[pyo3(signature = (input_filenames, input_moltype, param_str, output, name))]
 fn do_singlesketch(
-    input_filename: String,
+    input_filenames: Vec<String>,
     input_moltype: String,
     param_str: String,
     output: String,
     name: String,
 ) -> anyhow::Result<u8> {
-    match singlesketch::singlesketch(input_filename, input_moltype, param_str, output, name) {
+    match singlesketch::singlesketch(input_filenames, input_moltype, param_str, output, name) {
         Ok(_) => Ok(0),
         Err(e) => {
             eprintln!("Error: {e}");

src/python/sourmash_plugin_branchwater/__init__.py

@@ -602,11 +602,13 @@ def main(self, args):
 
 class Branchwater_SingleSketch(CommandLinePlugin):
     command = "singlesketch"
-    description = "sketch a single sequence file"
+    description = "sketch a single sample"
 
     def __init__(self, p):
         super().__init__(p)
-        p.add_argument("input_filename", help="input FASTA file or '-' for stdin")
+        p.add_argument(
+            "input_filenames", help="input file(s); use '-' for stdin", nargs="+"
+        )
         p.add_argument(
             "-o",
             "--output",
@@ -660,19 +662,19 @@ def main(self, args):
             args.name
             if args.name
             else (
-                os.path.basename(args.input_filename)
-                if args.input_filename != "-"
+                os.path.basename(args.input_filenames[0])
+                if args.input_filenames[0] != "-"
                 else ""
             )
         )
 
         notify(
-            f"sketching file '{args.input_filename}' ({args.input_moltype}) with params '{args.param_string}' and name '{signature_name}' using a single thread"
+            f"sketching {len(args.input_filenames)} files ({args.input_moltype}) with params '{args.param_string}' and name '{signature_name}' using a single thread"
         )
 
         super().main(args)
         status = sourmash_plugin_branchwater.do_singlesketch(
-            args.input_filename,
+            args.input_filenames,
             args.input_moltype,
             args.param_string,
             args.output,

src/python/tests/test_sketch.py

@@ -1630,3 +1630,55 @@ def test_singlesketch_stdin(runtmp):
     runtmp.sourmash("sketch", "dna", fa1, "-o", output2, "-p", "dna,scaled=10")
     sig2 = sourmash.load_one_signature(output2)
     assert sig.minhash.hashes == sig2.minhash.hashes
+
+
+def test_singlesketch_multifiles(runtmp, capfd):
+    # multiple input files to singlesketch
+    fa_csv = runtmp.output("db-fa.csv")
+
+    fa1 = get_test_data("short.fa")
+    fa2 = get_test_data("short2.fa")
+
+    output = runtmp.output("db.zip")
+
+    runtmp.sourmash(
+        "scripts",
+        "singlesketch",
+        fa1,
+        fa2,
+        "-o",
+        output,
+        "--param-str",
+        "dna,k=31,scaled=1",
+    )
+
+    assert os.path.exists(output)
+    assert not runtmp.last_result.out  # stdout should be empty
+    captured = capfd.readouterr()
+    print(captured.out)
+    print(captured.err)
+    assert "calculated 1 signatures for 2 sequences in 2 files" in captured.err
+
+    idx = sourmash.load_file_as_index(output)
+    sigs = list(idx.signatures())
+    print(sigs)
+    assert len(sigs) == 1
+    made_sig = sigs[0]
+    assert made_sig.name == "short.fa"
+
+    s1 = runtmp.output("short.sig")
+    runtmp.sourmash(
+        "sketch",
+        "dna",
+        fa1,
+        fa2,
+        "-o",
+        s1,
+        "--param-str",
+        "k=31,scaled=1",
+        "--name",
+        "short.fa",
+    )
+    sig1 = sourmash.load_one_signature(s1)
+
+    assert made_sig == sig1

src/singlesketch.rs

@@ -2,7 +2,7 @@ use crate::utils::buildutils::BuildCollection;
 use anyhow::{bail, Result};
 
 pub fn singlesketch(
-    input_filename: String,
+    input_filenames: Vec<String>,
     input_moltype: String,
     param_str: String,
     output: String,
@@ -26,14 +26,20 @@ pub fn singlesketch(
         bail!("No signatures to build for the given parameters.");
     }
 
-    let sequence_count =
-        sigs.build_sigs_from_file_or_stdin(&input_moltype, name, input_filename.clone())?;
+    let mut sequence_count = 0;
+    for input_filename in input_filenames.iter() {
+        sequence_count += sigs.build_sigs_from_file_or_stdin(
+            &input_moltype,
+            name.clone(),
+            input_filename.clone(),
+        )?;
+    }
 
     eprintln!(
-        "calculated {} signatures for {} sequences in {}",
+        "calculated {} signatures for {} sequences in {} files",
         sigs.size(),
         sequence_count,
-        input_filename
+        input_filenames.len(),
     );
 
     // Write signatures to stdout or output file