update GNPSFormat to support gnps2 (#292)

Update GNPSFormat class and relevant function to support gnps2 data.

src/nplinker/metabolomics/gnps/__init__.py

@@ -5,7 +5,7 @@
 from .gnps_format import GNPSFormat
 from .gnps_format import gnps_format_from_archive
 from .gnps_format import gnps_format_from_file_mapping
-from .gnps_format import gnps_format_from_task_id
+from .gnps_format import gnps_format_from_gnps1_task_id
 from .gnps_molecular_family_loader import GNPSMolecularFamilyLoader
 from .gnps_spectrum_loader import GNPSSpectrumLoader
 
@@ -20,5 +20,5 @@
     "GNPSSpectrumLoader",
     "gnps_format_from_archive",
     "gnps_format_from_file_mapping",
-    "gnps_format_from_task_id",
+    "gnps_format_from_gnps1_task_id",
 ]

src/nplinker/metabolomics/gnps/gnps_downloader.py

@@ -4,7 +4,7 @@
 from typing_extensions import Self
 from nplinker.utils import download_url
 from .gnps_format import GNPSFormat
-from .gnps_format import gnps_format_from_task_id
+from .gnps_format import gnps_format_from_gnps1_task_id
 
 
 class GNPSDownloader:
@@ -42,7 +42,7 @@ def __init__(self, task_id: str, download_root: str | PathLike):
         Examples:
             >>> GNPSDownloader("c22f44b14a3d450eb836d607cb9521bb", "~/downloads")
         """
-        gnps_format = gnps_format_from_task_id(task_id)
+        gnps_format = gnps_format_from_gnps1_task_id(task_id)
         if gnps_format == GNPSFormat.Unknown:
             raise ValueError(
                 f"Unknown workflow type for GNPS task '{task_id}'."

src/nplinker/metabolomics/gnps/gnps_format.py

@@ -9,9 +9,6 @@
 from bs4 import BeautifulSoup
 
 
-GNPS_TASK_URL = "https://gnps.ucsd.edu/ProteoSAFe/status.jsp?task={}"
-
-
 @unique
 class GNPSFormat(Enum):
     """Enum class for GNPS formats or workflows.
@@ -24,20 +21,27 @@ class GNPSFormat(Enum):
     """
 
     # Format: ShortName = "GNPSWorkflowName"
+    # For GNPS1
     SNETS = "METABOLOMICS-SNETS"
     SNETSV2 = "METABOLOMICS-SNETS-V2"
     FBMN = "FEATURE-BASED-MOLECULAR-NETWORKING"
+    # For GNPS2
+    GNPS2CN = "classical_networking_workflow"
+    GNPS2FBMN = "feature_based_molecular_networking_workflow"
+    # Unknown format
     Unknown = "Unknown-GNPS-Workflow"
 
 
-def gnps_format_from_task_id(task_id: str) -> GNPSFormat:
-    """Detect GNPS format for the given task id.
+def gnps_format_from_gnps1_task_id(task_id: str) -> GNPSFormat:
+    """Detect GNPS format or workflow for the given GNPS1 task id.
+
+    GNPS1 tasks are those generated on the platform https://gnps.ucsd.edu.
 
     Args:
-        task_id: GNPS task id.
+        task_id: GNPS1 task id.
 
     Returns:
-        The format identified in the GNPS task.
+        The format identified in the task.
 
     Examples:
         >>> gnps_format_from_task_id("c22f44b14a3d450eb836d607cb9521bb")
@@ -49,7 +53,8 @@ def gnps_format_from_task_id(task_id: str) -> GNPSFormat:
         >>> gnps_format_from_task_id("0ad6535e34d449788f297e712f43068a")
         <GNPSFormat.Unknown: 'Unknown-GNPS-Workflow'>
     """
-    task_html = httpx.get(GNPS_TASK_URL.format(task_id))
+    gnps1_task_url = "https://gnps.ucsd.edu/ProteoSAFe/status.jsp?task={}"
+    task_html = httpx.get(gnps1_task_url.format(task_id))
     soup = BeautifulSoup(task_html.text, features="html.parser")
     try:
         # find the td tag that follows the th tag containing 'Workflow'

tests/unit/metabolomics/test_gnps_format.py

@@ -2,7 +2,7 @@
 from nplinker.metabolomics.gnps import GNPSFormat
 from nplinker.metabolomics.gnps import gnps_format_from_archive
 from nplinker.metabolomics.gnps import gnps_format_from_file_mapping
-from nplinker.metabolomics.gnps import gnps_format_from_task_id
+from nplinker.metabolomics.gnps import gnps_format_from_gnps1_task_id
 
 
 @pytest.mark.parametrize(
@@ -14,10 +14,10 @@
         ["0ad6535e34d449788f297e712f43068a", GNPSFormat.Unknown],
     ],
 )
-def test_gnps_format_from_task_id(task_id: str, expected: GNPSFormat, gnps_website_is_down):
+def test_gnps_format_from_gnps1_task_id(task_id: str, expected: GNPSFormat, gnps_website_is_down):
     if gnps_website_is_down:
         pytest.skip("GNPS website is down")
-    actual = gnps_format_from_task_id(task_id)
+    actual = gnps_format_from_gnps1_task_id(task_id)
     assert actual is expected