fixed the scan capture

quantmsutils/mzml/mzml_statistics.py

@@ -204,7 +204,7 @@ def __init__(
         self.parquet_writer = None
         self.id_parquet_writer = None
         self.acquisition_datetime = None
-        self.scan_pattern = re.compile(r"[scan|spectrum]=(\d+)")
+        self.scan_pattern = re.compile(r'(?:spectrum|scan)=(\d+)')
 
     def transform_mzml_spectrum(
         self,
@@ -320,10 +320,22 @@ def transform_mzml_spectrum(
             self._write_batch()
 
     def _extract_scan_id(self, spectrum: oms.MSSpectrum) -> str:
-        """Extract scan ID from native ID using regex pattern"""
-        match = self.scan_pattern.findall(spectrum.getNativeID())
+        """
+        Extracts the scan ID from a given spectrum's native ID.
+
+        Parameters
+        ----------
+        spectrum : oms.MSSpectrum
+          The spectrum from which to extract the scan ID.
+
+        Returns
+        -------
+        str
+           The extracted scan ID if found, otherwise the original native ID.
+        """
+        match = re.search(r'(?:spectrum|scan)=(\d+)', spectrum.getNativeID())
         if match:
-            return match[0]
+            return match.group(1)
         return spectrum.getNativeID()
 
     def _extract_first_precursor_data(

tests/test_commands.py

@@ -1,6 +1,7 @@
 from pathlib import Path
 
 import pandas as pd
+import pytest
 from click.testing import CliRunner
 from quantmsutils.quantmsutilsc import cli
 
@@ -142,3 +143,19 @@ def test_mzml_statistics():
     assert len(table2) == len(table1)
 
     assert result.exit_code == 0
+
+def test_mzml_statistics_local():
+    runner = CliRunner()
+
+    mzml_path = TESTS_DIR / "test_data" / "TMT_Erwinia_1uLSike_Top10HCD_isol2_45stepped_60min_01.mzML"
+
+    # check if the file exist, delete it
+    ms_info_path = TESTS_DIR / "test_data" / "TMT_Erwinia_1uLSike_Top10HCD_isol2_45stepped_60min_01_ms_info.parquet"
+    if ms_info_path.exists():
+        ms_info_path.unlink()
+
+    result = runner.invoke(cli, ["mzmlstats", "--id_only", "--ms_path", mzml_path])
+    table = pd.read_parquet(ms_info_path)
+
+    assert len(table) > 0
+    assert result.exit_code == 0