Fix edam comments

CHANGELOG.md

@@ -9,6 +9,10 @@
 - switch changelog bot trigger only on comments ([#4241](https://github.com/nf-core/tools/pull/4241))
 - fix indentation in generated api docs ([#4245](https://github.com/nf-core/tools/pull/4245))
 
+### Linting
+
+- Store `EDAM.tsv` in `NFCORE_CACHE_DIR` and fix yaml comment loss ([#4242](https://github.com/nf-core/tools/pull/4242))
+
 ### Modules
 
 - Allow task.ext.prefix2 in modules linting ([#4234](https://github.com/nf-core/tools/pull/4234))

nf_core/modules/lint/__init__.py

@@ -611,13 +611,21 @@ def _add_edam_ontologies(section, edam_formats, desc):
                 section["ontologies"] = []
             log.debug(f"expected ontologies for {desc}: {expected_ontologies}")
             log.debug(f"current ontologies for {desc}: {current_ontologies}")
-            for ontology, ext in expected_ontologies:
-                if ontology not in current_ontologies:
+            for ontology_url, ext in expected_ontologies:
+                comment_text = edam_formats[ext][1]
+                if ontology_url not in current_ontologies:
                     try:
-                        section["ontologies"].append(ruamel.yaml.comments.CommentedMap({"edam": ontology}))
-                        section["ontologies"][-1].yaml_add_eol_comment(f"{edam_formats[ext][1]}", "edam")
+                        cm = ruamel.yaml.comments.CommentedMap()
+                        cm["edam"] = ontology_url
+                        cm.yaml_add_eol_comment(comment_text, key="edam")
+                        section["ontologies"].append(cm)
                     except KeyError:
                         log.warning(f"Could not add ontologies in {desc}")
+                else:
+                    for item in section["ontologies"]:
+                        if isinstance(item, ruamel.yaml.comments.CommentedMap) and item.get("edam") == ontology_url:
+                            item.yaml_add_eol_comment(comment_text, key="edam")
+                            break
 
         # EDAM ontologies
         edam_formats = nf_core.modules.modules_utils.load_edam()
@@ -681,7 +689,14 @@ def _add_edam_ontologies(section, edam_formats, desc):
 
         def _ensure_string_keys(obj):
             """Recursively ensure all dict keys are strings (e.g., convert 1.2 -> "1.2")"""
-            if isinstance(obj, dict):
+            # This first block is needed to keep the comments in the yml
+            if isinstance(obj, ruamel.yaml.comments.CommentedMap):
+                for key in list(obj.keys()):
+                    value = obj.pop(key)
+                    new_key = str(key) if not isinstance(key, str) else key
+                    obj[new_key] = _ensure_string_keys(value)
+                return obj
+            elif isinstance(obj, dict):
                 return {str(k) if not isinstance(k, str) else k: _ensure_string_keys(v) for k, v in obj.items()}
             elif isinstance(obj, list):
                 return [_ensure_string_keys(item) for item in obj]

nf_core/modules/modules_utils.py

@@ -1,13 +1,19 @@
 import logging
+import time
 from pathlib import Path
 from urllib.parse import urlparse
 
 import requests
 
+from nf_core.utils import NFCORE_CACHE_DIR
+
 from ..components.nfcore_component import NFCoreComponent
 
 log = logging.getLogger(__name__)
 
+EDAM_TSV_URL = "https://edamontology.org/EDAM.tsv"
+EDAM_CACHE_TTL = 7 * 24 * 60 * 60  # one week
+
 
 class ModuleExceptionError(Exception):
     """Exception raised when there was an error with module commands"""
@@ -96,15 +102,41 @@ def get_installed_modules(directory: Path, repo_type="modules") -> tuple[list[st
     return local_modules, nfcore_modules
 
 
+def cache_is_expired(path: Path) -> bool:
+    """Return True if the cache file is older than the configured TTL."""
+    age = time.time() - path.stat().st_mtime
+    return age > EDAM_CACHE_TTL
+
+
 def load_edam():
     """Load the EDAM ontology from the nf-core repository"""
     edam_formats = {}
-    try:
-        response = requests.get("https://edamontology.org/EDAM.tsv")
-    except requests.exceptions.RequestException as e:
-        log.warning(f"Failed to load EDAM ontology: {e}")
-        return edam_formats
-    for line in response.content.splitlines():
+    cache_path = Path(NFCORE_CACHE_DIR) / "EDAM.tsv"
+
+    # Remove stale cache file
+    if cache_path.exists() and cache_is_expired(cache_path):
+        log.debug("Cached EDAM ontology expired; removing old cache file")
+        cache_path.unlink(missing_ok=True)
+
+    if not cache_path.exists():
+        log.debug("EDAM.tsv file not found in NFCORE_CACHE_DIR; downloading")
+        try:
+            response = requests.get(EDAM_TSV_URL, timeout=15)
+            response.raise_for_status()
+            data_bytes = response.content
+            cache_path.write_bytes(data_bytes)
+        except requests.exceptions.RequestException as e:
+            log.warning(f"Failed to download EDAM ontology: {e}")
+            return edam_formats
+    else:
+        log.debug("Using EDAM.tsv file found in NFCORE_CACHE_DIR")
+        try:
+            data_bytes = cache_path.read_bytes()
+        except OSError as e:
+            log.warning(f"Failed to load EDAM ontology: {e}")
+            return edam_formats
+
+    for line in data_bytes.splitlines():
         fields = line.decode("utf-8").split("\t")
         if fields[0].split("/")[-1].startswith("format") and fields[14]:  # We choose an already provided extension
             extensions = fields[14].split("|")

tests/modules/test_modules_utils.py

@@ -1,3 +1,6 @@
+from pathlib import Path
+from unittest.mock import patch
+
 import nf_core.modules.modules_utils
 
 from ..test_modules import TestModules
@@ -82,3 +85,37 @@ def test_filter_modules_by_name_empty_list(self):
 
         filtered = nf_core.modules.modules_utils.filter_modules_by_name(modules, "fastqc")
         assert len(filtered) == 0
+
+    @patch("nf_core.modules.modules_utils.NFCORE_CACHE_DIR", new="test_cache")
+    def test_load_edam(self):
+        """Test EDAM ontology loading"""
+
+        cache_dir = Path("test_cache")
+        cache_path = cache_dir / "EDAM.tsv"
+
+        # Ensure clean state
+        if cache_dir.exists():
+            for f in cache_dir.iterdir():
+                f.unlink()
+            cache_dir.rmdir()
+
+        cache_dir.mkdir()
+
+        # Cache should not exist before loading
+        assert not cache_path.exists()
+
+        edam_formats = nf_core.modules.modules_utils.load_edam()
+
+        # Cache file should now exist
+        assert cache_path.exists()
+
+        first_key, first_value = next(iter(edam_formats.items()))
+
+        assert isinstance(first_key, str)
+        assert isinstance(first_value, tuple)
+        assert len(first_value) == 2
+
+        # Cleanup (important since we're not using tmp_path)
+        for f in cache_dir.iterdir():
+            f.unlink()
+        cache_dir.rmdir()