bex/tag_preprocessor/analyze.py

@@ -280,7 +280,7 @@ def analyze_clusters(file_paths, extension, project_root="", min_coverage=DEFAUL
         return [("(all methods)", None, len(sequences), meta)]
     unique_seqs = len(set(tuple(s) for s in symbol_seqs))
     unique_ratio = unique_seqs / n_methods
-    if unique_ratio > 0.9:
+    if unique_ratio > 0.95:
         meta["skip_reason"] = "too_diverse"
         return [("(all methods)", None, len(sequences), meta)]
 
@@ -437,7 +437,7 @@ def _infer_group(label, group_seqs, group_files, project_root, min_coverage, pre
     unique_seqs = len(set(tuple(s) for s in symbol_seqs))
     unique_ratio = unique_seqs / n_methods
     # Skip diversity check when decomposing (decomposition creates diverse fragments)
-    if unique_ratio > 0.9 and not decompose:
+    if unique_ratio > 0.95 and not decompose:
         meta = {"files": group_files, "imports": imports, "arg_patterns": arg_patterns, "packages": packages, "skip_reason": "too_diverse"}
         return (label, None, len(filtered), meta)
 
@@ -902,7 +902,7 @@ def _build_json_output(results):
     return json.dumps(output, indent=2)
 
 
-def _build_yaml_output(results, dir_path, max_mdl=200.0, min_structure=0.0):
+def _build_yaml_output(results, dir_path, max_mdl=500.0, min_structure=0.0):
     """Build YAML output grouped by top-level module, sorted by MDL.
 
     Filters out (other), no-grammar groups, groups above max_mdl,
@@ -1038,8 +1038,8 @@ def _parse_args(argv=None):
         help="Scoring method: langsize (default, Bex et al.) or mdl (fallback)",
     )
     parser.add_argument(
-        "--min-methods", type=int, default=3,
-        help="Minimum methods per group to infer grammar (default: 3). Groups with fewer are skipped.",
+        "--min-methods", type=int, default=2,
+        help="Minimum methods per group to infer grammar (default: 2). Groups with fewer are skipped.",
     )
     parser.add_argument(
         "--crx-method", choices=["standard", "refined"], default="standard",