Added verbose command for detailed report, tweaked W003 regex and updated codemeta.json

Author: Anas-Elhounsri

codemeta.json

@@ -1,34 +1,29 @@
 {
-    "@context": "https://w3id.org/codemeta/3.0",
-    "type": "SoftwareSourceCode",
-    "author": [
-        {
-            "id": "https://orcid.org/0009-0005-6868-4511",
-            "type": "Person",
-            "affiliation": {
-                "type": "Organization",
-                "name": "Ontology Engineering Group, Universidad Politecnica de Madrid"
-            },
-            "email": "a.elhounsri@upm.es",
-            "familyName": "El Hounsri",
-            "givenName": "Anas"
-        }
-    ],
-    "codeRepository": "git+https://github.com/SoftwareUnderstanding/RsMetaCheck.git",
-    "dateModified": "2025-09-19",
-    "description": "Automated tool to detect metadata quality pitfalls in software repositories (Python, Java, C++, etc.). Analyzes SoMEF output files for version mismatches, license issues, broken URLs, and more. ",
-    "downloadUrl": "https://github.com/SoftwareUnderstanding/RsMetaCheck/archive/refs/heads/main.tar.gz",
-    "keywords": [
-        "codemeta",
-        "pitfalls",
-        "metadata",
-        "analysis"
-    ],
-    "license": "https://spdx.org/licenses/MIT",
-    "name": "RsMetaCheck",
-    "operatingSystem": "Linux",
-    "programmingLanguage": "Python",
-    "version": "0.1.1",
-    "developmentStatus": "active",
-    "issueTracker": "https://github.com/SoftwareUnderstanding/RsMetaCheck/issues"
-}
+  "@context": "https://w3id.org/codemeta/3.0",
+  "type": "SoftwareSourceCode",
+  "author": [
+    {
+      "id": "https://orcid.org/0009-0005-6868-4511",
+      "type": "Person",
+      "affiliation": {
+        "type": "Organization",
+        "name": "Ontology Engineering Group, Universidad Politecnica de Madrid"
+      },
+      "email": "a.elhounsri@upm.es",
+      "familyName": "El Hounsri",
+      "givenName": "Anas"
+    }
+  ],
+  "codeRepository": "git+https://github.com/SoftwareUnderstanding/RsMetaCheck.git",
+  "dateModified": "2025-09-19",
+  "description": "Automated tool to detect metadata quality pitfalls in software repositories (Python, Java, C++, etc.). Analyzes SoMEF output files for version mismatches, license issues, broken URLs, and more. ",
+  "downloadUrl": "https://github.com/SoftwareUnderstanding/RsMetaCheck/archive/refs/heads/main.tar.gz",
+  "keywords": ["codemeta", "pitfalls", "metadata", "analysis"],
+  "license": "https://spdx.org/licenses/MIT",
+  "name": "RsMetaCheck",
+  "operatingSystem": "Linux",
+  "programmingLanguage": "Python",
+  "version": "0.2.0",
+  "developmentStatus": "active",
+  "issueTracker": "https://github.com/SoftwareUnderstanding/RsMetaCheck/issues"
+}

poetry.lock

@@ -35,6 +35,12 @@ def cli():
         help="SoMEF confidence threshold (default: 0.8). Only used when running SoMEF."
     )
 
+    parser.add_argument(
+        "--verbose",
+        action="store_true",
+        help="Include both detected AND undetected pitfalls in the output JSON-LD."
+    )
+
     args = parser.parse_args()
 
     if args.skip_somef:
@@ -52,7 +58,7 @@ def cli():
             return
 
         print(f"Analyzing {len(somef_json_paths)} SoMEF output files...")
-        run_analysis(somef_json_paths, args.pitfalls_output, args.analysis_output)
+        run_analysis(somef_json_paths, args.pitfalls_output, args.analysis_output, verbose=args.verbose)
 
     else:
         threshold = args.threshold
@@ -71,7 +77,7 @@ def cli():
                 print(f"Warning: Skipping invalid input (not a URL or existing file): {input_item}")
 
         print(f"\nRunning analysis on outputs in {somef_output_dir}...")
-        run_analysis(somef_output_dir, args.pitfalls_output, args.analysis_output)
+        run_analysis(somef_output_dir, args.pitfalls_output, args.analysis_output, verbose=args.verbose)
 
 
 if __name__ == "__main__":

src/metacheck/cli.py

@@ -38,7 +38,7 @@
 from metacheck.scripts.warnings.w010 import detect_git_remote_shorthand_pitfall
 
 
-def detect_all_pitfalls(json_files: Iterable[Path], pitfalls_output_dir: Union[str, Path], output_file: Union[str, Path]):
+def detect_all_pitfalls(json_files: Iterable[Path], pitfalls_output_dir: Union[str, Path], output_file: Union[str, Path], verbose: bool = False):
     """
     Detect all software repository pitfalls in SoMEF output files using modular detectors.
     Now also generates individual JSON-LD files for each repository.
@@ -363,8 +363,8 @@ def detect_all_pitfalls(json_files: Iterable[Path], pitfalls_output_dir: Union[s
                     for result in repo_pitfall_results
                 )
 
-                if has_any_issue:
-                    jsonld_data = create_pitfall_jsonld(somef_data, repo_pitfall_results, json_file.name)
+                if has_any_issue or verbose:
+                    jsonld_data = create_pitfall_jsonld(somef_data, repo_pitfall_results, json_file.name, verbose=verbose)
                     saved_file = save_individual_pitfall_jsonld(jsonld_data, pitfalls_output_dir, json_file.name)
 
                     if saved_file:
@@ -412,7 +412,7 @@ def detect_all_pitfalls(json_files: Iterable[Path], pitfalls_output_dir: Union[s
         print(f"Error writing output file: {e}")
 
 
-def main(input_dir=None, somef_json_paths=None, pitfalls_dir=None, analysis_output=None):
+def main(input_dir=None, somef_json_paths=None, pitfalls_dir=None, analysis_output=None, verbose=False):
     """
     Main function to run all pitfall detections.
 
@@ -421,6 +421,7 @@ def main(input_dir=None, somef_json_paths=None, pitfalls_dir=None, analysis_outp
         somef_json_paths (Iterable[Path], optional): Explicit list of SoMEF output JSON files.
         pitfalls_dir (str|Path, optional): Directory to save pitfall JSON-LD files.
         analysis_output (str|Path, optional): Path to save summary results JSON.
+        verbose (bool, optional): Include both detected AND undetected pitfalls in JSON-LD.
 
     Note: Provide either input_dir OR somef_json_paths, not both.
           If both are provided, somef_json_paths takes precedence.
@@ -448,7 +449,7 @@ def main(input_dir=None, somef_json_paths=None, pitfalls_dir=None, analysis_outp
         print("No JSON files found for analysis.")
         return
 
-    detect_all_pitfalls(json_files, pitfalls_directory, output_file)
+    detect_all_pitfalls(json_files, pitfalls_directory, output_file, verbose)
 
 if __name__ == "__main__":
     main()

src/metacheck/detect_pitfalls_main.py

@@ -3,7 +3,7 @@
 from metacheck.detect_pitfalls_main import main
 
 
-def run_analysis(somef_input: Union[str, Path, Iterable[Path]], pitfalls_dir: Union[str, Path], analysis_file: Union[str, Path]):
+def run_analysis(somef_input: Union[str, Path, Iterable[Path]], pitfalls_dir: Union[str, Path], analysis_file: Union[str, Path], verbose: bool = False):
     """
     Run metadata analysis using existing code.
 
@@ -12,17 +12,18 @@ def run_analysis(somef_input: Union[str, Path, Iterable[Path]], pitfalls_dir: Un
                      or an iterable of Path objects pointing to specific SoMEF JSON files
         pitfalls_dir: Directory to save pitfall JSON-LD files
         analysis_file: Path to save summary results JSON
+        verbose: bool indicating if both detected and undetected checks should be logged.
     """
     print(f"\nRunning analysis...")
 
     if isinstance(somef_input, (str, Path)):
         somef_path = Path(somef_input)
         if somef_path.is_dir():
             print(f"Using directory: {somef_input}")
-            main(input_dir=somef_input, pitfalls_dir=pitfalls_dir, analysis_output=analysis_file)
+            main(input_dir=somef_input, pitfalls_dir=pitfalls_dir, analysis_output=analysis_file, verbose=verbose)
         else:
             print(f"Error: {somef_input} is not a valid directory")
     else:
         json_files = list(somef_input)
         print(f"Using {len(json_files)} specified JSON files")
-        main(somef_json_paths=json_files, pitfalls_dir=pitfalls_dir, analysis_output=analysis_file)
+        main(somef_json_paths=json_files, pitfalls_dir=pitfalls_dir, analysis_output=analysis_file, verbose=verbose)

src/metacheck/run_analyzer.py

@@ -26,12 +26,13 @@ def detect_dual_license_missing_codemeta_pitfall(somef_data: Dict, file_name: st
         r"dually[\s-]?licen[cs]ed?",
         r"multiple[\s-]?licen[cs]es?",
         r"(?:is|are)\s+licen[cs]ed?\s+under.*(?:and|or).*licen[cs]e", 
-        r"choose.*(?:between|from).*licen[cs]e",  
+        r"choose.*(?:between|from|your).*licen[cs]e",  
         r"either.*or.*licen[cs]e",  
         r"\d+\..*licen[cs]e.*\n.*\d+\..*licen[cs]e",
         r"licen[cs]e.*options?",
-        r"available\s+under.*(?:two|multiple|either).*licen[cs]es?" 
-]
+        r"available\s+under.*(?:two|multiple|either).*licen[cs]es?",
+        r"licen[cs]ed? under.*(?:and|or)"
+    ]
 
     has_dual_license_indicator = False
     dual_license_source = None

src/metacheck/scripts/warnings/w003.py

@@ -31,7 +31,7 @@ def detect_development_status_url_pitfall(somef_data: Dict, file_name: str) -> D
     Detect when codemeta.json developmentStatus is a URL instead of a string.
     """
     result = {
-        "has_pitfall": False,
+        "has_warning": False,
         "file_name": file_name,
         "development_status": None,
         "source": None,
@@ -54,7 +54,7 @@ def detect_development_status_url_pitfall(somef_data: Dict, file_name: str) -> D
                 dev_status = entry["result"]["value"]
 
                 if is_url(dev_status):
-                    result["has_pitfall"] = True
+                    result["has_warning"] = True
                     result["development_status"] = dev_status
                     result["source"] = source
                     result["is_url"] = True

src/metacheck/scripts/warnings/w009.py

@@ -32,7 +32,7 @@ def detect_git_remote_shorthand_pitfall(somef_data: Dict, file_name: str) -> Dic
     Detect when metadata files use Git remote-style shorthand in codeRepository.
     """
     result = {
-        "has_pitfall": False,
+        "has_warning": False,
         "file_name": file_name,
         "repository_url": None,
         "source": None,
@@ -64,7 +64,7 @@ def detect_git_remote_shorthand_pitfall(somef_data: Dict, file_name: str) -> Dic
                 repo_url = entry["result"]["value"]
 
                 if is_git_remote_shorthand(repo_url):
-                    result["has_pitfall"] = True
+                    result["has_warning"] = True
                     result["repository_url"] = repo_url
                     result["source"] = source if source else f"technique: {technique}"
                     result["metadata_source_file"] = extract_metadata_source_filename(source)

src/metacheck/scripts/warnings/w010.py

@@ -444,10 +444,11 @@ def convert_sets_to_lists(obj):
         return obj
 
 
-def create_pitfall_jsonld(somef_data: Dict, pitfall_results: List[Dict], file_name: str) -> Dict:
+def create_pitfall_jsonld(somef_data: Dict, pitfall_results: List[Dict], file_name: str, verbose: bool = False) -> Dict:
     """
     Create a JSON-LD structure for detected pitfalls following the sample format.
     """
+    import hashlib
     software_info = extract_software_info_from_somef(somef_data)
     description_info = extract_description_info(somef_data)
 
@@ -468,9 +469,17 @@ def create_pitfall_jsonld(somef_data: Dict, pitfall_results: List[Dict], file_na
     }
 
     for pitfall_result in pitfall_results:
-        if pitfall_result.get("has_pitfall", False) or pitfall_result.get("has_warning", False):
+        has_pitfall = pitfall_result.get("has_pitfall", False)
+        has_warning = pitfall_result.get("has_warning", False)
+        has_issue = has_pitfall or has_warning
+        
+        if has_issue or verbose:
             pitfall_code = pitfall_result.get("pitfall_code", "Unknown")
             category = get_pitfall_category(pitfall_code)
+            
+            output_val = "true" if has_issue else "false"
+            evidence_val = format_evidence_text(pitfall_code, pitfall_result) if has_issue else f"{pitfall_code} not detected:"
+            suggestion_val = get_suggestion_text(pitfall_code) if has_issue else ""
 
             check_result = {
                 "@type": "CheckResult",
@@ -483,10 +492,13 @@ def create_pitfall_jsonld(somef_data: Dict, pitfall_results: List[Dict], file_na
                 },
                 "process": get_pitfall_description(pitfall_code),
                 "status": {"@id": "schema:CompletedActionStatus"},
-                "checkId": pitfall_code,
-                "evidence": format_evidence_text(pitfall_code, pitfall_result),
-                "suggestion": get_suggestion_text(pitfall_code)
+                "output": output_val,
+                "evidence": evidence_val,
+                "suggestion": suggestion_val
             }
+            
+            check_hash = hashlib.sha256(json.dumps(check_result, sort_keys=True).encode("utf-8")).hexdigest()
+            check_result["checkId"] = check_hash
 
             jsonld_output["checks"].append(check_result)