sourcecode 0.48.0__tar.gz → 0.49.0__tar.gz

{sourcecode-0.48.0 → sourcecode-0.49.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: sourcecode
-Version: 0.48.0
+Version: 0.49.0
 Summary: Deterministic codebase context for AI coding agents
 License:                                  Apache License
                                    Version 2.0, January 2004

{sourcecode-0.48.0 → sourcecode-0.49.0}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "sourcecode"
-version = "0.48.0"
+version = "0.49.0"
 description = "Deterministic codebase context for AI coding agents"
 readme = "README.md"
 requires-python = ">=3.9"

{sourcecode-0.48.0 → sourcecode-0.49.0}/src/sourcecode/__init__.py

@@ -1,3 +1,3 @@
 """sourcecode — Deterministic codebase context maps for AI coding agents."""
 
-__version__ = "0.48.0"
+__version__ = "0.49.0"

{sourcecode-0.48.0 → sourcecode-0.49.0}/src/sourcecode/cli.py

@@ -1445,8 +1445,10 @@ def main(
         data = agent_view(sm)
         # When contract pipeline ran (mode=contract, no legacy flags), include
         # per-file contracts in agent output so agents get structural context.
+        # Remove file_relevance — contracts cover this signal with more detail.
         if _is_contract_mode and sm.file_contracts:
             from sourcecode.serializer import _serialize_contract_minimal
+            data.pop("file_relevance", None)
             data["contracts"] = [_serialize_contract_minimal(c) for c in sm.file_contracts]
             if sm.contract_summary is not None:
                 cs = sm.contract_summary

{sourcecode-0.48.0 → sourcecode-0.49.0}/src/sourcecode/contract_pipeline.py

@@ -27,8 +27,8 @@ from sourcecode.schema import EntryPoint, MonorepoPackageInfo
 # ---------------------------------------------------------------------------
 
 _MAX_FILES = 500          # hard cap on files extracted per run
-_MAX_CONTRACTS = 25       # default top-N output cap — omit rather than flood
-_MIN_CONTRACT_SCORE = 0.10  # drop contracts below this relevance threshold
+_MAX_CONTRACTS = 15       # default top-N output cap — omit rather than flood
+_MIN_CONTRACT_SCORE = 0.15  # drop contracts below this relevance threshold
 _SRC_EXTENSIONS: frozenset[str] = frozenset(_LANGUAGE_MAP.keys())
 
 

{sourcecode-0.48.0 → sourcecode-0.49.0}/src/sourcecode/serializer.py

@@ -1130,9 +1130,6 @@ def _contract_view_minimal(
         "project": project,
     }
 
-    if sm.metadata.traversal_topology:
-        result["traversal"] = sm.metadata.traversal_topology
-
     # Per-file contracts
     if contracts:
         serialized: list[dict[str, Any]] = []
@@ -1310,6 +1307,10 @@ def _compress_sig(name: str, sig: str, max_len: int = 100) -> str:
     return full
 
 
+_MAX_FN_PER_CONTRACT = 5   # max function signatures per contract (token budget)
+_MAX_SIG_LEN = 60          # max chars per compressed signature
+
+
 def _serialize_contract_minimal(c: Any) -> dict[str, Any]:
     """Serialize one FileContract to minimal format."""
     item: dict[str, Any] = {"path": c.path, "role": c.role}
@@ -1317,38 +1318,47 @@ def _serialize_contract_minimal(c: Any) -> dict[str, Any]:
     if c.is_changed:
         item["changed"] = True
 
-    # Exports: flat string for functions/unknown, {name,k} for others
-    # When all exports are same non-function kind, group them
-    if c.exports:
-        exs: list[Any] = []
-        kinds = {e.kind for e in c.exports}
-        if len(kinds) == 1 and "function" not in kinds and "unknown" not in kinds:
-            # All same non-function kind — compact: {"k": "class", "names": [...]}
-            only_kind = next(iter(kinds))
-            exs = [{"k": only_kind, "names": sorted(e.name for e in c.exports)}]
-        else:
-            for e in sorted(c.exports, key=lambda e: e.name):
-                if e.kind in ("function", "unknown"):
-                    exs.append(e.name)
-                else:
-                    exs.append({"name": e.name, "k": e.kind})
-        item["exports"] = exs
-
-    # External deps (non-stdlib already filtered in extractor)
-    if c.dependencies:
-        item["deps"] = sorted(c.dependencies)
-
-    # Exported function signatures — compressed
+    # Exported function signatures — compressed, capped
     exported_names = {e.name for e in c.exports}
+    fn_names_in_sigs: set[str] = set()
     if c.functions:
         fns = []
         for f in sorted(c.functions, key=lambda f: f.name):
             if not (f.exported or f.name in exported_names):
                 continue
-            fns.append(_compress_sig(f.name, f.signature))
+            fns.append(_compress_sig(f.name, f.signature, max_len=_MAX_SIG_LEN))
+            fn_names_in_sigs.add(f.name)
+            if len(fns) >= _MAX_FN_PER_CONTRACT:
+                break
         if fns:
             item["fn"] = fns
 
+    # Exports: omit function names already shown in fn; keep non-function exports
+    if c.exports:
+        exs: list[Any] = []
+        non_fn_exports = [e for e in c.exports if e.kind not in ("function", "unknown")]
+        fn_exports_not_in_sig = [
+            e for e in c.exports
+            if e.kind in ("function", "unknown") and e.name not in fn_names_in_sigs
+        ]
+        remaining = non_fn_exports + fn_exports_not_in_sig
+        if remaining:
+            kinds = {e.kind for e in remaining}
+            if len(kinds) == 1 and "function" not in kinds and "unknown" not in kinds:
+                only_kind = next(iter(kinds))
+                exs = [{"k": only_kind, "names": sorted(e.name for e in remaining)}]
+            else:
+                for e in sorted(remaining, key=lambda e: e.name):
+                    if e.kind in ("function", "unknown"):
+                        exs.append(e.name)
+                    else:
+                        exs.append({"name": e.name, "k": e.kind})
+            item["exports"] = exs
+
+    # External deps (non-stdlib already filtered in extractor)
+    if c.dependencies:
+        item["deps"] = sorted(c.dependencies)
+
     # Types: skip if fully covered by exports (avoids duplication in model files)
     if c.types:
         export_names_set = {e.name for e in c.exports}
@@ -1363,12 +1373,6 @@ def _serialize_contract_minimal(c: Any) -> dict[str, Any]:
     if c.hooks_used:
         item["hooks"] = c.hooks_used
 
-    # Ranking signals: why this file was ranked here
-    if getattr(c, "ranking_reasons", None):
-        non_trivial = [r for r in c.ranking_reasons if r not in ("source file", "noise")]
-        if non_trivial:
-            item["why"] = non_trivial
-
     return item