feat(diffctx): hub suppression categories, edge weights, lambda tuning

Author: nikolay-e

src/treemapper/diffctx/config/weights.py

@@ -21,8 +21,8 @@ class EdgeWeightConfig:
     "doc_structure": EdgeWeightConfig(0.30, 0.83),
     "anchor_link": EdgeWeightConfig(0.55, 0.64),
     "citation": EdgeWeightConfig(0.25, 1.0),
-    "python_call": EdgeWeightConfig(0.55, 0.70),
-    "python_symbol": EdgeWeightConfig(0.60, 0.70),
+    "python_call": EdgeWeightConfig(0.65, 0.70),
+    "python_symbol": EdgeWeightConfig(0.70, 0.70),
     "python_type": EdgeWeightConfig(0.50, 0.70),
     "javascript_call": EdgeWeightConfig(0.70, 0.50),
     "javascript_symbol": EdgeWeightConfig(0.75, 0.50),
@@ -80,7 +80,7 @@ class LangWeights:
 
 
 LANG_WEIGHTS: dict[str, LangWeights] = {
-    "python": LangWeights(0.55, 0.60, 0.50, 0.20, 0.35),
+    "python": LangWeights(0.65, 0.70, 0.50, 0.20, 0.35),
     "javascript": LangWeights(0.50, 0.55, 0.45, 0.25, 0.35),
     "jsx": LangWeights(0.50, 0.55, 0.45, 0.25, 0.35),
     "typescript": LangWeights(0.70, 0.75, 0.65, 0.15, 0.25),

src/treemapper/diffctx/edges/structural/sibling.py

@@ -11,6 +11,7 @@
 class SiblingEdgeBuilder(EdgeBuilder):
     weight = EDGE_WEIGHTS["sibling"].forward
     reverse_weight_factor = EDGE_WEIGHTS["sibling"].reverse_factor
+    category = "sibling"
 
     def build(self, fragments: list[Fragment], repo_root: Path | None = None) -> EdgeDict:
         by_dir = self._group_files_by_dir(fragments)

src/treemapper/diffctx/edges/structural/test.py

@@ -92,6 +92,7 @@ class TestEdgeBuilder(EdgeBuilder):
     weight_direct = EDGE_WEIGHTS["test_direct"].forward
     weight_naming = EDGE_WEIGHTS["test_naming"].forward
     reverse_weight_factor = EDGE_WEIGHTS["test_reverse"].forward / EDGE_WEIGHTS["test_direct"].forward
+    category = "test_edge"
 
     def discover_related_files(
         self,

src/treemapper/diffctx/graph.py

@@ -79,7 +79,7 @@ def build_graph(fragments: list[Fragment], repo_root: Path | None = None) -> Gra
     return graph
 
 
-_SUPPRESSION_EXEMPT = frozenset({"semantic", "structural", "config", "document"})
+_SUPPRESSION_EXEMPT = frozenset({"semantic", "structural"})
 
 
 def _apply_hub_suppression(

src/treemapper/diffctx/ppr.py

@@ -106,7 +106,7 @@ def personalized_pagerank(
     seeds: set[FragmentId],
     alpha: float = 0.60,
     tol: float = 1e-4,
-    lam: float = 0.5,
+    lam: float = 0.4,
     seed_weights: dict[FragmentId, float] | None = None,
 ) -> dict[FragmentId, float]:
     if not graph.nodes: