Automatically sync the Rocrate and README content during pipelines lint (#3688)

* format

* format

* update

* update MultiQC version

Author: ningyuxin1999

nf_core/pipelines/lint/rocrate_readme_sync.py

@@ -8,14 +8,12 @@
 def rocrate_readme_sync(self):
     """
     Check if the RO-Crate description in ro-crate-metadata.json matches the README.md content.
-    If the --fix is set, the RO-Crate description will be updated to match the README.md content.
+    If not, the RO-Crate description will be automatically updated to match the README.md content during linting.
     """
 
     passed = []
-    failed = []
     ignored = []
     fixed = []
-    could_fix: bool = False
 
     # Check if the file exists before trying to load it
     metadata_file = Path(self.wf_path, "ro-crate-metadata.json")
@@ -27,46 +25,35 @@ def rocrate_readme_sync(self):
             ignored.append("`ro-crate-metadata.json` not found")
         if not readme_file.exists():
             ignored.append("`README.md` not found")
-        return {"passed": passed, "failed": failed, "ignored": ignored}
+        return {"passed": passed, "fixed": fixed, "ignored": ignored}
 
     try:
         metadata_content = metadata_file.read_text(encoding="utf-8")
         metadata_dict = json.loads(metadata_content)
     except json.JSONDecodeError as e:
         log.error("Failed to decode JSON from `ro-crate-metadata.json`: %s", e)
         ignored.append("Invalid JSON in `ro-crate-metadata.json`")
-        return {"passed": passed, "failed": failed, "ignored": ignored}
+        return {"passed": passed, "fixed": fixed, "ignored": ignored}
     readme_content = readme_file.read_text(encoding="utf-8")
     graph = metadata_dict.get("@graph")
+
     if not graph or not isinstance(graph, list) or not graph[0] or not isinstance(graph[0], dict):
         ignored.append("Invalid RO-Crate metadata structure.")
     else:
         # Check if the 'description' key is present
         if "description" not in graph[0]:
-            if "rocrate_readme_sync" in self.fix:
-                metadata_dict.get("@graph")[0]["description"] = readme_content
-                fixed.append("Fixed: add the same description from `README.md` to the RO-Crate metadata.")
-            else:
-                ignored.append("No description found in `ro-crate-metadata.json`.")
-                return {"passed": passed, "failed": failed, "ignored": ignored}
+            metadata_dict.get("@graph")[0]["description"] = readme_content
+            fixed.append("Fixed: add the same description from `README.md` to the RO-Crate metadata.")
 
     rc_description_graph = metadata_dict.get("@graph", [{}])[0].get("description")
 
     # Compare the two strings and add a linting error if they don't match
     if readme_content != rc_description_graph:
-        # If the --fix flag is set, you could overwrite the RO-Crate description with the README content:
-        if "rocrate_readme_sync" in self.fix:
-            metadata_dict.get("@graph")[0]["description"] = readme_content
-            fixed.append("Fixed: add the same description from `README.md` to the RO-Crate metadata.")
-            with metadata_file.open("w", encoding="utf-8") as f:
-                json.dump(metadata_dict, f, indent=4)
-            passed.append("RO-Crate description matches the `README.md`.")
-            fixed.append("Mismatch fixed: RO-Crate description updated from `README.md`.")
-        else:
-            failed.append(
-                "The RO-Crate descriptions do not match the README.md content. Use `nf-core pipelines lint --fix rocrate_readme_sync` to update."
-            )
-            could_fix = True
+        metadata_dict.get("@graph")[0]["description"] = readme_content
+        with metadata_file.open("w", encoding="utf-8") as f:
+            json.dump(metadata_dict, f, indent=4)
+        passed.append("RO-Crate description matches the `README.md`.")
+        fixed.append("Mismatch fixed: RO-Crate description updated from `README.md`.")
     else:
         passed.append("RO-Crate descriptions are in sync with `README.md`.")
-    return {"passed": passed, "failed": failed, "ignored": ignored, "fixed": fixed, "could_fix": could_fix}
+    return {"passed": passed, "fixed": fixed, "ignored": ignored}

tests/pipelines/lint/test_rocrate_readme_sync.py

@@ -12,25 +12,6 @@ def test_rocrate_readme_sync_pass(self):
         assert len(results.get("failed", [])) == 0
         assert len(results.get("passed", [])) > 0
 
-    def test_rocrate_readme_sync_fail(self):
-        self.lint_obj._load()
-
-        json_path = Path(self.lint_obj.wf_path, "ro-crate-metadata.json")
-        with open(json_path) as f:
-            try:
-                rocrate = json.load(f)
-            except json.JSONDecodeError as e:
-                raise UserWarning(f"Unable to load JSON file '{json_path}' due to error {e}")
-        rocrate["@graph"][0]["description"] = "This is a test script"
-        with open(json_path, "w") as f:
-            json.dump(rocrate, f, indent=4)
-        results = self.lint_obj.rocrate_readme_sync()
-        assert len(results.get("failed", [])) == 1
-        assert (
-            "The RO-Crate descriptions do not match the README.md content. Use `nf-core pipelines lint --fix rocrate_readme_sync` to update."
-            in results.get("failed", [])
-        )
-
     def test_rocrate_readme_sync_fixed(self):
         self.lint_obj._load()
         json_path = Path(self.lint_obj.wf_path, "ro-crate-metadata.json")
@@ -43,12 +24,6 @@ def test_rocrate_readme_sync_fixed(self):
         with open(json_path, "w") as f:
             json.dump(rocrate, f, indent=4)
 
-        results = self.lint_obj.rocrate_readme_sync()
-        assert len(results.get("failed", [])) == 1
-
-        # Fix the issue
-        assert "rocrate_readme_sync" in self.lint_obj.lint_tests
-        self.lint_obj.fix = ["rocrate_readme_sync"]
-        self.lint_obj._load()
         results = self.lint_obj.rocrate_readme_sync()
         assert len(results.get("failed", [])) == 0
+        assert len(results.get("fixed", [])) == 1