refactor wget-xlsx

utils/wget-xlsx-plugin/README.md → utils/pre-process/data-download/wget-xlsx-tool/README.md

@@ -9,5 +9,5 @@ This plugin takes     2     input arguments and 1 output argument:
 | Name          | Description             | I/O    | Type   | Default |
 |---------------|-------------------------|--------|--------|---------|
 | url | url, Type: string | Input | string | string |
-| output_xlsx_path |  | Input | string | string |
-| output_xlsx_path |  | Output | File | File |
+| output_xlsx_path | Output xlsx filename | Input | string | string |
+| output_xlsx_path | Output xlsx file | Output | File | File |

utils/wget-xlsx-plugin/ict.yml → utils/pre-process/data-download/wget-xlsx-tool/ict.yml

@@ -1,12 +1,12 @@
 specVersion: "0.1.0"
 name: wget_xlsx
 version: 0.1.0
-container: wget-xlsx-plugin
+container: wget-xlsx-tool
 entrypoint:
 title: wget_xlsx
 description: Download an xlsx file
-author: Data Scientist
-contact: [email protected]
+author: Brandon Walker, Nazanin Donyapour
+contact: [email protected], [email protected]
 repository:
 documentation:
 citation:

utils/wget-xlsx-plugin/pyproject.toml → utils/pre-process/data-download/wget-xlsx-tool/pyproject.toml

@@ -7,9 +7,7 @@ readme = "README.md"
 
 [tool.poetry.dependencies]
 python = ">=3.9,<3.12"
-typer = "^0.7.0"
-cwl-utils = "0.33"
-cwltool = "3.1.20240404144621"
+sophios = "0.1.1"
 
 [tool.poetry.group.dev.dependencies]
 bump2version = "^1.0.1"

utils/pre-process/data-download/wget-xlsx-tool/tests/test_wget_xlsx.py

@@ -0,0 +1,28 @@
+"""Test the wget_xlsx tool."""
+from pathlib import Path
+
+from sophios.api.pythonapi import Step
+from sophios.api.pythonapi import Workflow
+
+
+def test_wget_xlsx() -> None:
+    """Test wget_xlsx."""
+    cwl_file_str = "wget_xlsx_0@[email protected]"
+    cwl_file = Path(__file__).resolve().parent.parent / Path(cwl_file_str)
+
+    wget_xlsx = Step(clt_path=cwl_file)
+    wget_xlsx.url = "https://smacc.mml.unc.edu/ncats_target_based_curated.xlsx"
+    wget_xlsx.output_xlsx_path = "system.xlsx"
+
+    steps = [wget_xlsx]
+    filename = "wget_xlsx"
+    viz = Workflow(steps, filename)
+
+    viz.run()
+
+    outdir = Path("outdir")
+    files = list(outdir.rglob("system.xlsx"))
+
+    assert (
+        files
+    ), f"The file 'system.xlsx' does not exist in any subdirectory of '{outdir}'."