refac

emzed/ext/mzmine2/installers.py

@@ -127,22 +127,3 @@ def get_mzmine2_home(target_folder=_default_mzmine2_folder()):
         raise RuntimeError(f"file {mzmine2_path} contains invalid path {path}")
 
     return path
-
-
-def construct_java_call_cli_args(
-    jre_root_folder=_default_jre_folder(), mzmine2_root_folder=_default_mzmine2_folder()
-):
-
-    java_bin = find_java_bin(jre_root_folder)
-    mzmine2_home = get_mzmine2_home(mzmine2_root_folder)
-
-    here = os.path.dirname(os.path.abspath(__file__))
-
-    class_path = ":".join(
-        [
-            os.path.join(mzmine2_home, "lib", "*"),
-            os.path.join(here, "java", "extensions.jar"),
-        ]
-    )
-
-    return [java_bin, "-Djava.awt.headless=true", "-ea", "-cp", '"' + class_path + '"']

emzed/ext/mzmine2/java/RemoveShoulderPeaks.java

@@ -139,7 +139,7 @@ public class RemoveShoulderPeaks {
 
     RawDataFileImpl rdf_out = new RawDataFileImpl(rmp.outPath);
 
-    for (int i=1; i <= rdf.getNumOfScans(); i++){
+    for (int i: rdf.getScanNumbers(1)) {
         Scan s = rdf.getScan(i);
         MassList ml = s.getMassList("_mass_list ");
         SimpleScan new_scan = new SimpleScan(s);

emzed/ext/mzmine2/remove_shoulder_peaks.py

@@ -9,7 +9,7 @@ import emzed.lib
 
 from .exceptions import MzMine2Exception
 from .parameters import ParameterBase
-from .installers import construct_java_call_cli_args
+from .java_utils import construct_java_call_cli_args, run_java
 
 
 class RemoveShoulderPeaksParameters(ParameterBase):
@@ -21,7 +21,6 @@ class RemoveShoulderPeaksParameters(ParameterBase):
         return value > 0, "value must be positive"
 
 
-
 def remove_shoulder_peaks(peakmap, parameters):
     assert isinstance(peakmap, emzed.lib.PeakMap)
     assert isinstance(parameters, RemoveShoulderPeaksParameters)
@@ -33,28 +32,19 @@ def remove_shoulder_peaks(peakmap, parameters):
     temp_file_out = tempfile.mktemp(".mzML")
     peakmap.store(temp_file_in)
 
-    cli = construct_java_call_cli_args() + [
-        "ch.ethz.emzed.RemoveShoulderPeaks",
+    run_java(
+        "RemoveShoulderPeaks",
         temp_file_in,
-        str(parameters.resolution),
+        parameters.resolution,
         parameters.peak_model,
         temp_file_out,
-    ]
-
-    p = subprocess.Popen(
-        " ".join(cli), shell=True, stderr=subprocess.STDOUT, stdout=subprocess.PIPE
     )
-    for line in iter(p.stdout.readline, b""):
-        line = str(line, "utf-8").rstrip()
-        if line.startswith("!!!ERROR"):
-            raise MzMine2Exception("calling java failed")
-        if line.startswith("!!!DONE"):
-            break
-        print(line)
 
     peakmap_back = emzed.lib.PeakMap.load(temp_file_out)
 
     # the mzXML parser of mzedmin2 rounds rt values, we fix this here:
     for s, s_back in zip(peakmap.spectra, peakmap_back.spectra):
-        s_back.rt = s.rt
+        s_back._set_rt(s.rt)
+
+    s_back._conn.commit()
     return peakmap_back

tests/test_remove_shoulder_peaks.py

@@ -5,8 +5,7 @@ import multiprocessing
 
 import dill
 
-from emzed.ext.mzmine2 import (RemoveShoulderPeaksParameters,
-                               remove_shoulder_peaks)
+from emzed.ext.mzmine2 import RemoveShoulderPeaksParameters, remove_shoulder_peaks
 from emzed.lib import PeakMap
 
 
@@ -16,19 +15,20 @@ def test_remove_shoulder_peaks(data_path):
     parameters.resolution = 10000.0
     parameters.peak_model = "GAUSS"
 
-    pm = PeakMap.load(data_path("test_smaller.mzXML"))
+    pm = PeakMap.load(data_path("test_smallest.mzXML"))
     pm_back = remove_shoulder_peaks(pm, parameters)
 
     assert pm != pm_back
-    assert pm_back.spectra[0].mzs.shape == (241,)
-    assert pm_back.spectra[1].mzs.shape == (259,)
+    assert pm_back.spectra[0].mzs.shape == (303,)
+    assert pm_back.spectra[1].mzs.shape == (319,)
 
-    assert pm_back.spectra[0].rt == pm.spectra[0].rt
+    assert all(s.rt == s_back.rt for (s, s_back) in zip(pm.spectra, pm_back.spectra))
 
 
 def execute(what):
     f, *args = dill.loads(what)
-    return f(*args)
+    result = f(*args)
+    return result
 
 
 def test_remove_shoulder_peaks_parallel(data_path, tmpdir):
@@ -37,13 +37,16 @@ def test_remove_shoulder_peaks_parallel(data_path, tmpdir):
     parameters.resolution = 10000.0
     parameters.peak_model = "GAUSS"
 
-    pm = PeakMap.load(data_path("test_smaller.mzXML"))
+    pm = PeakMap.load(data_path("test_smallest.mzXML"))
 
-    multiprocessing.set_start_method("spawn")
+    # multiprocessing.set_start_method("spawn")
     with multiprocessing.Pool(2) as p:
         args = dill.dumps((remove_shoulder_peaks, pm, parameters))
         results = p.map(execute, [args, args])
 
     for pm_back in results:
-        assert pm_back.spectra[0].mzs.shape == (241,)
-        assert pm_back.spectra[1].mzs.shape == (259,)
+        assert pm_back.spectra[0].mzs.shape == (303,)
+        assert pm_back.spectra[1].mzs.shape == (319,)
+        assert all(
+            s.rt == s_back.rt for (s, s_back) in zip(pm.spectra, pm_back.spectra)
+        )