RO-crate content appear in the nf-core lint warning if there is a TODO

CHANGELOG.md

@@ -19,6 +19,8 @@
 - Fix: linting with comments after the input directive ([#3458](https://github.com/nf-core/tools/pull/3458))
 - EDAM ontology fixes ([#3460](https://github.com/nf-core/tools/pull/3460))
 - Fix default linting of nf-core components when `nf-core pipelines lint` is ran ([#3480](https://github.com/nf-core/tools/pull/3480))
+- Fix the unexpected warning and sychronize the README.md and RO-crate-metadata.json
+  ([#3493](https://github.com/nf-core/tools/pull/3493))
 - Adapt the linter to the new notation used to include the centralized nf-core configs ([#3491](https://github.com/nf-core/tools/pull/3491))
 
 ### Modules

docs/api/_src/pipeline_lint_tests/index.md

@@ -21,6 +21,7 @@
     - [pipeline_todos](./pipeline_todos/)
     - [plugin_includes](./plugin_includes/)
     - [readme](./readme/)
+    - [rocrate_readme_sync](./rocrate_readme_sync/)
     - [schema_description](./schema_description/)
     - [schema_lint](./schema_lint/)
     - [schema_params](./schema_params/)

docs/api/_src/pipeline_lint_tests/rocrate_readme_sync.md

@@ -0,0 +1,5 @@
+# rocrate_readme_sync
+
+    ```{eval-rst}
+    .. automethod:: nf_core.pipelines.lint.PipelineLint.rocrate_readme_sync
+    ```

nf_core/pipelines/lint/__init__.py

@@ -50,6 +50,7 @@
 from .pipeline_todos import pipeline_todos
 from .plugin_includes import plugin_includes
 from .readme import readme
+from .rocrate_readme_sync import rocrate_readme_sync
 from .schema_description import schema_description
 from .schema_lint import schema_lint
 from .schema_params import schema_params
@@ -104,6 +105,8 @@ class PipelineLint(nf_core.utils.Pipeline):
     schema_lint = schema_lint
     schema_params = schema_params
     system_exit = system_exit
+    rocrate_readme_sync = rocrate_readme_sync
+
     template_strings = template_strings
     version_consistency = version_consistency
     included_configs = included_configs
@@ -160,6 +163,7 @@ def _get_all_lint_tests(release_mode):
             "base_config",
             "modules_config",
             "nfcore_yml",
+            "rocrate_readme_sync",
         ] + (["version_consistency", "included_configs"] if release_mode else [])
 
     def _load(self) -> bool:

nf_core/pipelines/lint/pipeline_todos.py

@@ -39,7 +39,8 @@ def pipeline_todos(self, root_dir=None):
     if root_dir is None:
         root_dir = self.wf_path
 
-    ignore = [".git"]
+    # Ignore ro-crate-metadata.json to avoid warnings when TODOs are not deleted.
+    ignore = [".git", "ro-crate-metadata.json"]
     if Path(root_dir, ".gitignore").is_file():
         with open(Path(root_dir, ".gitignore"), encoding="latin1") as fh:
             for line in fh:

nf_core/pipelines/lint/rocrate_readme_sync.py

@@ -0,0 +1,70 @@
+import json
+import logging
+from pathlib import Path
+
+log = logging.getLogger(__name__)
+
+
+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.
+    """
+
+    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")
+    readme_file = Path(self.wf_path, "README.md")
+
+    # Only proceed if both files exist
+    if not (metadata_file.exists() and readme_file.exists()):
+        if not metadata_file.exists():
+            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}
+
+    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}
+
+    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]:
+            ignored.append(
+                "No description found in ro-crate-metadata.json, add a description to the RO-Crate metadata."
+            )
+            graph[0]["description"] = " "
+    rc_description_graph = metadata_dict.get("@graph", [{}])[0].get("description")
+
+    readme_content = readme_file.read_text(encoding="utf-8")
+
+    # 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
+            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 lint --fix` to update."
+            )
+            could_fix = True
+    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}

tests/pipelines/lint/test_rocrate_readme_sync.py

@@ -0,0 +1,26 @@
+from pathlib import Path
+
+from ..test_lint import TestLint
+
+
+class TestLintROcrateReadmeSync(TestLint):
+    def test_rocrate_readme_sync_pass(self):
+        self.lint_obj._load()
+        results = self.lint_obj.rocrate_readme_sync()
+        assert len(results.get("warned", [])) == 0
+        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, "w") as f:
+            f.write("{}")
+
+        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 lint --fix` to update."
+            in results.get("failed", [])
+        )