add explicit docker_image support across CLI, pipeline, and job layer

Author: JaskRendix

tests/test_cli.py

@@ -0,0 +1,139 @@
+import pytest
+
+from waters2mzml.cli import convert
+
+
+def test_docker_without_image_errors(tmp_path):
+    raw = tmp_path / "raw"
+    raw.mkdir()
+    out = tmp_path / "mzml"
+    out.mkdir()
+
+    with pytest.raises(Exception) as exc:
+        convert(
+            input=raw,
+            output=out,
+            centroid=False,
+            base_dir=tmp_path,
+            parallel=1,
+            docker=True,
+            docker_image=None,
+            retries=0,
+            log_level="INFO",
+        )
+
+    assert "docker-image" in str(exc.value).lower()
+
+
+def test_docker_with_image_passes_validation(tmp_path, monkeypatch):
+    called = {}
+
+    def fake_pipeline(**kwargs):
+        called.update(kwargs)
+
+    monkeypatch.setattr("waters2mzml.cli.run_pipeline", fake_pipeline)
+
+    raw = tmp_path / "raw"
+    raw.mkdir()
+    (raw / "sample.raw").mkdir()
+
+    out = tmp_path / "mzml"
+    out.mkdir()
+
+    convert(
+        input=raw,
+        output=out,
+        centroid=False,
+        base_dir=tmp_path,
+        parallel=1,
+        docker=True,
+        docker_image="my/image",
+        retries=0,
+        log_level="INFO",
+    )
+
+    assert called["use_docker"] is True
+    assert called["docker_image"] == "my/image"
+
+
+def test_parallel_pipeline_receives_docker_image(tmp_path, monkeypatch):
+    called = {}
+
+    def fake_parallel(**kwargs):
+        called.update(kwargs)
+
+    monkeypatch.setattr("waters2mzml.cli.run_pipeline_parallel", fake_parallel)
+
+    raw = tmp_path / "raw"
+    raw.mkdir()
+    (raw / "sample.raw").mkdir()
+
+    out = tmp_path / "mzml"
+    out.mkdir()
+
+    convert(
+        input=raw,
+        output=out,
+        centroid=False,
+        base_dir=tmp_path,
+        parallel=4,
+        docker=True,
+        docker_image="pwiz/msconvert",
+        retries=0,
+        log_level="INFO",
+    )
+
+    assert called["use_docker"] is True
+    assert called["docker_image"] == "pwiz/msconvert"
+    assert called["jobs"] == 4
+
+
+def test_native_mode_does_not_require_image(tmp_path, monkeypatch):
+    called = {}
+
+    def fake_pipeline(**kwargs):
+        called.update(kwargs)
+
+    monkeypatch.setattr("waters2mzml.cli.run_pipeline", fake_pipeline)
+
+    raw = tmp_path / "raw"
+    raw.mkdir()
+    (raw / "sample.raw").mkdir()
+
+    out = tmp_path / "mzml"
+    out.mkdir()
+
+    convert(
+        input=raw,
+        output=out,
+        centroid=False,
+        base_dir=tmp_path,
+        parallel=1,
+        docker=False,
+        docker_image=None,
+        retries=0,
+        log_level="INFO",
+    )
+
+    assert called["use_docker"] is False
+    assert called["docker_image"] is None
+
+
+def test_invalid_paths_fail_cleanly(tmp_path, caplog):
+    out = tmp_path / "mzml"
+    out.mkdir()
+
+    convert(
+        input=tmp_path / "does_not_exist",
+        output=out,
+        centroid=False,
+        base_dir=tmp_path,
+        parallel=1,
+        docker=False,
+        docker_image=None,
+        retries=0,
+        log_level="INFO",
+    )
+
+    # The pipeline should NOT raise — it should log a message
+    assert "no .raw folders found" in caplog.text.lower()

tests/test_msconvert.py

@@ -235,3 +235,45 @@ def fake_build(self):
     args = cfg.build_msconvert_args()
     assert "--foo" in args
     assert "--bar=123" in args
+
+
+def test_docker_image_required(tmp_path):
+    raw = tmp_path / "sample.raw"
+    raw.mkdir()
+
+    cfg = ConversionConfig(centroid=False, use_docker=True, docker_image=None)
+
+    with pytest.raises(MsconvertError):
+        _run_msconvert_docker(raw, cfg)
+
+
+def test_docker_command_construction(monkeypatch, tmp_path):
+    raw = tmp_path / "sample.raw"
+    raw.mkdir()
+
+    cfg = ConversionConfig(
+        centroid=True,
+        use_docker=True,
+        docker_image="my/image",
+    )
+
+    captured = {}
+
+    def fake_run(cmd, capture_output, text):
+        captured["cmd"] = cmd
+
+        class P:
+            returncode = 0
+            stderr = ""
+
+        return P()
+
+    monkeypatch.setattr(subprocess, "run", fake_run)
+
+    _run_msconvert_docker(raw, cfg)
+
+    cmd = captured["cmd"]
+    assert cmd[0] == "docker"
+    assert "my/image" in cmd
+    assert "/data/sample.raw" in cmd
+    assert "--outdir" in cmd

tests/test_pipeline_integration.py

@@ -69,3 +69,40 @@ def test_full_pipeline(tmp_path, monkeypatch):
     assert "scan=1" in text
     assert "scan=2" in text
     assert 'value="2"' in text
+
+
+def test_pipeline_passes_docker_image(monkeypatch, tmp_path):
+    called = {}
+
+    def fake_process(**kwargs):
+        called.update(kwargs)
+
+        class R:
+            success = True
+            warnings = []
+            qc = None
+            mzml_path = tmp_path / "x.mzML"
+            raw_dir = tmp_path / "raw/sample.raw"
+            error = None
+
+        return R()
+
+    monkeypatch.setattr("waters2mzml.pipeline.process_single_raw", fake_process)
+
+    raw = tmp_path / "raw"
+    raw.mkdir()
+    (raw / "sample.raw").mkdir()
+
+    out = tmp_path / "mzml"
+
+    run_pipeline(
+        base_dir=tmp_path,
+        input_dir=raw,
+        output_dir=out,
+        centroid=False,
+        use_docker=True,
+        docker_image="img",
+    )
+
+    assert called["use_docker"] is True
+    assert called["docker_image"] == "img"

waters2mzml/cli.py

@@ -56,7 +56,12 @@ def convert(
     docker: bool = typer.Option(
         False,
         "--docker",
-        help="Run msconvert inside a Docker container (requires a user-provided image)",
+        help="Run msconvert inside a Docker container",
+    ),
+    docker_image: str | None = typer.Option(
+        None,
+        "--docker-image",
+        help="Docker image containing msconvert (required with --docker)",
     ),
     retries: int = typer.Option(
         0,
@@ -76,6 +81,9 @@ def convert(
     """
     setup_logging(log_level)
 
+    if docker and not docker_image:
+        raise typer.BadParameter("You must specify --docker-image when using --docker")
+
     if parallel <= 1:
         # Sequential pipeline
         run_pipeline(
@@ -84,6 +92,7 @@ def convert(
             output_dir=output,
             centroid=centroid,
             use_docker=docker,
+            docker_image=docker_image,
         )
     else:
         # Parallel pipeline
@@ -95,6 +104,7 @@ def convert(
             jobs=parallel,
             use_docker=docker,
             retries=retries,
+            docker_image=docker_image,
         )
 
 

waters2mzml/config.py

@@ -8,7 +8,7 @@ class ConversionConfig:
     ms_level_filter: str = "1-2"
     compression: str = "--zlib --32"
     use_docker: bool = False
-    docker_image: str = "proteowizard/msconvert:latest"
+    docker_image: str | None = None
 
     def build_msconvert_args(self) -> str:
         # Mirrors original config strings

waters2mzml/job.py

@@ -58,6 +58,7 @@ def process_single_raw(
     output_dir: Path,
     centroid: bool,
     use_docker: bool = False,
+    docker_image: str | None = None,
     do_postprocess: bool = True,
 ) -> JobResult:
     """
@@ -77,7 +78,11 @@ def process_single_raw(
 
         # 2) Convert with msconvert
         logger.debug(f"Converting {raw_dir} with msconvert")
-        config = ConversionConfig(centroid=centroid, use_docker=use_docker)
+        config = ConversionConfig(
+            centroid=centroid,
+            use_docker=use_docker,
+            docker_image=docker_image,
+        )
         mzml_path = run_msconvert(msconvert_path, job_raw, config)
 
         # 3) Post-process

waters2mzml/pipeline.py

@@ -20,6 +20,7 @@ def run_pipeline(
     centroid: bool,
     skip_cleanup: bool = False,
     use_docker: bool = False,
+    docker_image: str | None = None,
     do_postprocess: bool = True,
 ) -> None:
     paths = default_paths(base_dir)
@@ -50,6 +51,7 @@ def run_pipeline(
             output_dir=paths.mzml_dir,
             centroid=centroid,
             use_docker=use_docker,
+            docker_image=docker_image,
             do_postprocess=do_postprocess,
         )
 
@@ -81,15 +83,12 @@ def run_pipeline_parallel(
     jobs: int,
     skip_cleanup: bool = False,
     use_docker: bool = False,
+    docker_image: str | None = None,
     do_postprocess: bool = True,
     retries: int = 0,
 ) -> None:
     """
     Parallel version of run_pipeline using run_parallel.
-
-    - Same behavior as run_pipeline, but processes all RAW folders concurrently.
-    - Uses the redesigned run_parallel with per-job isolation and retry logic.
-    - Output is deterministic and matches the sequential pipeline's reporting style.
     """
     paths = default_paths(base_dir)
     if input_dir is not None:
@@ -116,6 +115,7 @@ def run_pipeline_parallel(
         centroid=centroid,
         jobs=jobs,
         use_docker=use_docker,
+        docker_image=docker_image,
         retries=retries,
         do_postprocess=do_postprocess,
     )