fix(diffctx): hub dampening + noise filter fixes — 98→64 failures

Author: nikolay-e

src/treemapper/diffctx/filtering.py

@@ -13,9 +13,9 @@
 _PROXIMITY_FLOOR_MAX = 0.01
 _PROXIMITY_HALF_DECAY = 50
 _DEFINITION_PROXIMITY_HALF_DECAY = 5
-_HUB_REVERSE_THRESHOLD = 3
+_HUB_REVERSE_THRESHOLD = 2
 _MAX_CONTEXT_FRAGMENTS_PER_FILE = 10
-_LOW_RELEVANCE_THRESHOLD = 0.005
+_LOW_RELEVANCE_THRESHOLD = 0.02
 _SIZE_PENALTY_BASE_TOKENS = 100
 _SIZE_PENALTY_EXPONENT = 0.5
 
@@ -96,10 +96,10 @@ def _find_hub_noise_paths(
 
     noise_counts: dict[Path, int] = {}
     for deps in reverse_deps.values():
-        if len(deps) > _HUB_REVERSE_THRESHOLD:
+        if len(deps) >= _HUB_REVERSE_THRESHOLD:
             for dep in deps:
                 noise_counts[dep] = noise_counts.get(dep, 0) + 1
-    return {p for p, count in noise_counts.items() if count >= 3 and p not in direct_edge_paths}
+    return {p for p, count in noise_counts.items() if count >= 1 and p not in direct_edge_paths}
 
 
 def _find_config_generic_code_files(

src/treemapper/diffctx/graph.py

@@ -84,6 +84,9 @@ def build_graph(fragments: list[Fragment], repo_root: Path | None = None) -> Gra
 _SUPPRESSION_EXEMPT = frozenset({"semantic", "structural", "test_edge"})
 
 
+_HUB_OUT_DEGREE_THRESHOLD = 3
+
+
 def _apply_hub_suppression(
     edges: dict[tuple[FragmentId, FragmentId], float],
     edge_categories: dict[tuple[FragmentId, FragmentId], str],
@@ -110,4 +113,14 @@ def _apply_hub_suppression(
             weight = weight / max(1.0, math.log(1 + deg))
         suppressed[(src, dst)] = weight
 
+    sem_out_degree: dict[FragmentId, int] = {}
+    for (src, _dst), cat in edge_categories.items():
+        if cat == "semantic":
+            sem_out_degree[src] = sem_out_degree.get(src, 0) + 1
+
+    for src, dst in list(suppressed.keys()):
+        out_deg = sem_out_degree.get(src, 0)
+        if out_deg >= _HUB_OUT_DEGREE_THRESHOLD and edge_categories.get((src, dst)) == "semantic":
+            suppressed[(src, dst)] = suppressed[(src, dst)] / math.sqrt(out_deg)
+
     return suppressed

tests/cases/diff/julia_003_project_toml.yaml

@@ -275,7 +275,3 @@ oracle:
   - no_global_3
   - no_global_4
   - no_global_5
-xfail:
-  category: null
-  reason: TOML fragmentation — version bump line not included in selected [deps] fragment
-  issue: null

tests/cases/diff/regression_003_large_config_fragmentation.yaml

@@ -158,7 +158,3 @@ oracle:
   allowed: []
   forbidden:
   - no_global_0
-xfail:
-  category: null
-  reason: package.json fragmentation — config files need better merge strategy
-  issue: null

tests/cases/diff/regression_007_migration_enum_tracing.yaml

@@ -342,7 +342,3 @@ oracle:
   - no_global_13
   - no_global_14
   - no_global_15
-xfail:
-  category: null
-  reason: Enum fragment truncated under budget pressure — QUARTER value missing
-  issue: null