per-language base channel configs (closes #2)

- Dfg.base_cfg_for ext: .py adds string_values+field_names on top of the
  global base (N8 sweep: ALL MRR 0.955->0.969, evolved 0.862->0.929,
  samefile FPR halved; those channels cost recall on Rust)
- Signature.extract ?ext selects the per-language cfg; threaded through
  every extraction site (units/bench/corpus/lsp/mcp/cli); diff_functions
  and graph build per-side/per-file cfgs
- sigdb: format_version 2; cache tag carries rust+python channel bits
- --with still composes on the global base for ablations
- acceptance: python benchmark reproduces combo numbers with no flags;
  16 sanity + 4 regression + protocol smokes green

Author: thebabush

nonna/cli/bench.ml

@@ -376,7 +376,7 @@ let rank (pairs_file : string) (corpus_paths : string list) =
   let sig_of : (int, Signature.t) Hashtbl.t = Hashtbl.create 8192 in
   Units.units_of_paths corpus_paths
   |> List.iter (fun (u : Units.unit_info) ->
-         let sg = Signature.extract u.Units.ucfg in
+         let sg = Signature.extract ~ext:(Filename.extension u.Units.ufile) u.Units.ucfg in
          if Signature.size sg >= Units.min_features then (
            let fid =
              Engine.add eng
@@ -486,7 +486,7 @@ let eval ?(scores_out : string option) (pairs_file : string) =
              Units.units_of_file file
              |> List.iter (fun (u : Units.unit_info) ->
                     Hashtbl.replace t u.Units.uline_start
-                      (Signature.extract u.Units.ucfg))
+                      (Signature.extract ~ext:(Filename.extension u.Units.ufile) u.Units.ucfg))
            with _ -> ());
           Hashtbl.replace cache file t;
           t

nonna/cli/corpus.ml

@@ -101,7 +101,7 @@ let cache_dir () : string =
 let index_dir (dir : string) : Sigdb.entry list =
   Units.units_of_paths [ dir ]
   |> List.filter_map (fun (u : Units.unit_info) ->
-         let sg = Signature.extract u.Units.ucfg in
+         let sg = Signature.extract ~ext:(Filename.extension u.Units.ufile) u.Units.ucfg in
          if Signature.size sg < Units.min_features then None
          else
            Some

nonna/cli/lsp_server.ml

@@ -103,7 +103,7 @@ let index_workspace_async (root : string) : unit =
            let eng = Engine.create () in
            Units.units_of_paths [ root ]
            |> List.iter (fun (u : Units.unit_info) ->
-                  let sg = Signature.extract u.Units.ucfg in
+                  let sg = Signature.extract ~ext:(Filename.extension u.Units.ufile) u.Units.ucfg in
                   if Signature.size sg >= Units.min_features then
                     ignore
                       (Engine.add eng
@@ -131,7 +131,7 @@ let line_range (line0 : int) : J.t =
 let diagnostics_for (path : string) : J.t list =
   Units.units_of_file path
   |> List.filter_map (fun (u : Units.unit_info) ->
-         let sg = Signature.extract u.Units.ucfg in
+         let sg = Signature.extract ~ext:(Filename.extension u.Units.ufile) u.Units.ucfg in
          if Signature.size sg < Units.min_features then None
          else
            Engine.query !engine sg ~threshold:report_threshold ~max_results:5
@@ -218,7 +218,7 @@ let find_similar (uri : string) (line0 : int) : J.t =
   match Units.unit_at path (line0 + 1) with
   | None -> `Assoc [ ("query", `Null); ("hits", `List []) ]
   | Some u ->
-      let sg = Signature.extract u.Units.ucfg in
+      let sg = Signature.extract ~ext:(Filename.extension u.Units.ufile) u.Units.ucfg in
       let hits =
         Engine.query !engine sg ~threshold:0.2 ~max_results:15
         |> List.filter (fun (h : Engine.hit) ->

nonna/cli/main.ml

@@ -20,7 +20,7 @@ let cmd_features (file : string) =
   Units.units_of_file file
   |> List.iter (fun (u : Units.unit_info) ->
          let feats = Dfg.extract u.Units.ucfg in
-         let sg = Signature.extract u.Units.ucfg in
+         let sg = Signature.extract ~ext:(Filename.extension u.Units.ufile) u.Units.ucfg in
          Printf.printf "=== %s (%s) — %d dfg features, %d total\n"
            u.Units.uname (Units.loc_str u) (List.length feats)
            (Signature.size sg);
@@ -50,7 +50,7 @@ let cmd_query (corpus : string list) (draft : string) (threshold : float)
   Printf.printf "corpus: %d units indexed\n" (List.length kept);
   Units.units_of_file draft
   |> List.iter (fun (u : Units.unit_info) ->
-         let sg = Signature.extract u.Units.ucfg in
+         let sg = Signature.extract ~ext:(Filename.extension u.Units.ufile) u.Units.ucfg in
          Printf.printf "\n── %s (%s) — %d features\n" u.Units.uname
            (Units.loc_str u) (Signature.size sg);
          if Signature.size sg < Units.min_features then
@@ -103,7 +103,10 @@ let cmd_graph (file : string) (fn_filter : string option) (outdir : string) =
   in
   selected
   |> List.iter (fun (u : Units.unit_info) ->
-         let g = Dfg.graph_of u.Units.ucfg in
+         let g =
+           Dfg.graph_of ~fc:(Dfg.base_cfg_for (Filename.extension file))
+             u.Units.ucfg
+         in
          let source =
            if u.Units.uline_start > 0 && file_lines <> [] then
              Some

nonna/cli/mcp_server.ml

@@ -72,7 +72,7 @@ let index_async (root : string) : unit =
             let eng = Engine.create () in
             Units.units_of_paths [ root ]
             |> List.iter (fun (u : Units.unit_info) ->
-                   let sg = Signature.extract u.Units.ucfg in
+                   let sg = Signature.extract ~ext:(Filename.extension u.Units.ufile) u.Units.ucfg in
                    if Signature.size sg >= Units.min_features then
                      ignore
                        (Engine.add eng
@@ -188,7 +188,7 @@ let hit_block (h : Engine.hit) : string =
 
 let query_unit (u : Units.unit_info) ~(threshold : float) ~(top_k : int) :
     Engine.hit list =
-  let sg = Signature.extract u.Units.ucfg in
+  let sg = Signature.extract ~ext:(Filename.extension u.Units.ufile) u.Units.ucfg in
   if Signature.size sg < Units.min_features then []
   else
     Engine.query !engine sg ~threshold ~max_results:(top_k + 1)
@@ -262,10 +262,18 @@ let diff_functions (args : J.t) : J.t =
   match (resolve_side args "a", resolve_side args "b") with
   | Error e, _ | _, Error e -> tool_text ~is_error:true e
   | Ok (la, ua), Ok (lb, ub) ->
-      let sa = Signature.extract ua.Units.ucfg in
-      let sb = Signature.extract ub.Units.ucfg in
-      let ga = Dfg.graph_of ua.Units.ucfg in
-      let gb = Dfg.graph_of ub.Units.ucfg in
+      let sa = Signature.extract ~ext:(Filename.extension ua.Units.ufile) ua.Units.ucfg in
+      let sb = Signature.extract ~ext:(Filename.extension ub.Units.ufile) ub.Units.ucfg in
+      let ga =
+        Dfg.graph_of
+          ~fc:(Dfg.base_cfg_for (Filename.extension ua.Units.ufile))
+          ua.Units.ucfg
+      in
+      let gb =
+        Dfg.graph_of
+          ~fc:(Dfg.base_cfg_for (Filename.extension ub.Units.ufile))
+          ub.Units.ucfg
+      in
       let text =
         String.concat "\n\n"
           [

nonna/cli/sigdb.ml

@@ -8,17 +8,18 @@ module Engine = Nonna_index.Engine
 module Signature = Nonna_features.Signature
 
 (* bump on any change to hashing/features/weights *)
-let format_version = 1
+let format_version = 2
 
 type entry = { meta : Engine.meta; sg : Signature.t }
 
 let profile_tag () : string =
   let module Dfg = Nonna_features.Dfg in
-  Printf.sprintf "%s-i%d-b%x"
+  Printf.sprintf "%s-i%d-b%x-p%x"
     (if !Signature.default_profile = Signature.full_profile then "full"
      else "structural")
     !Dfg.iterations
-    (Dfg.cfg_bits !Dfg.base_cfg)
+    (Dfg.cfg_bits (Dfg.base_cfg_for ".rs"))
+    (Dfg.cfg_bits (Dfg.base_cfg_for ".py"))
 
 let save (path : string) (entries : entry list) : unit =
   let tmp = path ^ ".tmp" in

nonna/cli/units.ml

@@ -171,7 +171,7 @@ let index_units (units : unit_info list) :
   let kept =
     units
     |> List.filter_map (fun u ->
-           let sg = Signature.extract u.ucfg in
+           let sg = Signature.extract ~ext:(Filename.extension u.ufile) u.ucfg in
            if Signature.size sg < min_features then None
            else (
              ignore

nonna/features/dfg.ml

@@ -141,6 +141,18 @@ let keep_value (fc : cfg) (kind : string) : bool =
 let base_cfg : cfg ref =
   ref { structural_cfg with call_names = true; int_values = true }
 
+(* Per-language base (N8 sweeps, issue #2): channel optima differ by
+   language. Python additionally wants string_values + field_names
+   (stringly-typed: literals and attribute names are API identity there;
+   ALL MRR 0.955→0.969, evolved 0.862→0.929, samefile FPR halved) — the
+   same channels COST recall on Rust. Composed on top of the global
+   base_cfg so --with ablations still apply everywhere. *)
+let base_cfg_for (ext : string) : cfg =
+  let b = !base_cfg in
+  match ext with
+  | ".py" -> { b with string_values = true; field_names = true }
+  | _ -> b
+
 (* compact tag for cache keys / reports *)
 let cfg_bits (c : cfg) : int =
   (if c.call_names then 1 else 0)

nonna/features/signature.ml

@@ -43,14 +43,16 @@ let full_profile =
 (* Set once at CLI startup (--profile). *)
 let default_profile : profile ref = ref structural_profile
 
-let extract ?(profile : profile option) (fcfg : IL.fun_cfg) : t =
+(* [ext] selects the per-language base channels (Dfg.base_cfg_for). *)
+let extract ?(profile : profile option) ?(ext = "") (fcfg : IL.fun_cfg) : t =
   let p = match profile with Some p -> p | None -> !default_profile in
+  let fc = Dfg.base_cfg_for ext in
   let sem = Semantic.extract_parts fcfg in
   let all : (Fhash.t * float) list =
-    (Dfg.extract fcfg
+    (Dfg.extract ~fc fcfg
     |> List.map (fun (f : Dfg.feature) -> (f.Dfg.hash, dfg_weight f.Dfg.tag)))
     @ (if p.w_delta_sem > 0. then
-         Dfg.extract_delta ~rich:Dfg.semantic_cfg fcfg
+         Dfg.extract_delta ~base:fc ~rich:Dfg.semantic_cfg fcfg
          |> List.map (fun (f : Dfg.feature) ->
                 (f.Dfg.hash, p.w_delta_sem *. dfg_weight f.Dfg.tag))
        else [])