sourcecode 1.12.0__tar.gz → 1.13.0__tar.gz

{sourcecode-1.12.0 → sourcecode-1.13.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: sourcecode
-Version: 1.12.0
+Version: 1.13.0
 Summary: Deterministic codebase context for AI coding agents
 License:                                  Apache License
                                    Version 2.0, January 2004

{sourcecode-1.12.0 → sourcecode-1.13.0}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "sourcecode"
-version = "1.12.0"
+version = "1.13.0"
 description = "Deterministic codebase context for AI coding agents"
 readme = "README.md"
 requires-python = ">=3.9"

{sourcecode-1.12.0 → sourcecode-1.13.0}/src/sourcecode/__init__.py

@@ -1,3 +1,3 @@
 """sourcecode — Deterministic codebase context maps for AI coding agents."""
 
-__version__ = "1.12.0"
+__version__ = "1.13.0"

{sourcecode-1.12.0 → sourcecode-1.13.0}/src/sourcecode/architecture_summary.py

@@ -173,7 +173,7 @@ class ArchitectureSummarizer:
             fw_names = list(dict.fromkeys(fw_names))[:3]  # dedup, preserve order
         else:
             # Filter out frameworks from auxiliary or tooling stacks (docs/, benchmarks/, etc.)
-            fw_names = [
+            fw_names = list(dict.fromkeys(
                 f.name
                 for s in sm.stacks
                 if not self._is_tooling_path(s.root or "")
@@ -181,7 +181,7 @@ class ArchitectureSummarizer:
                 and not self._is_auxiliary_path(s.root or "")
                 and not self._is_auxiliary_path(s.workspace or "")
                 for f in s.frameworks[:2]
-            ][:3]
+            ))[:3]
 
         fw_str = f" using {', '.join(fw_names)}" if fw_names else ""
         if runtime:
@@ -288,7 +288,7 @@ class ArchitectureSummarizer:
 
     def _summarize_java_entry(self, path: str, content: str, stacks: list[StackDetection]) -> list[str]:
         lines: list[str] = []
-        frameworks = [f.name for stack in stacks for f in stack.frameworks]
+        frameworks = list(dict.fromkeys(f.name for stack in stacks for f in stack.frameworks))
         if frameworks:
             lines.append(f"Frameworks detectados: {', '.join(frameworks)}.")
         annotations = re.findall(r"@(SpringBootApplication|QuarkusMain|MicronautApplication|Application)\b", content)

{sourcecode-1.12.0 → sourcecode-1.13.0}/src/sourcecode/classifier.py

@@ -150,9 +150,18 @@ class TypeClassifier:
             score = {"low": 0, "medium": 1, "high": 2}.get(stack.confidence, 0)
             manifest_weight = 1 if stack.manifests else 0
             priority = 0
-            if project_type in {"webapp", "fullstack"} and stack.stack in {"nodejs", "elixir"} or project_type == "api" and stack.stack in _API_STACKS or project_type == "cli" and any(
-                framework.name in _CLI_FRAMEWORKS for framework in stack.frameworks
-            ) or project_type == "cli" and stack.stack in {"cpp", "dotnet"}:
+            # Backend server-side stacks with manifest evidence outrank frontend stacks
+            # in fullstack projects (e.g. Java+Spring wins over nodejs+Angular).
+            if (project_type in {"fullstack", "api"}
+                    and stack.stack in _API_STACKS
+                    and stack.manifests):
+                priority = 4
+            elif (project_type in {"webapp", "fullstack"} and stack.stack in {"nodejs", "elixir"}
+                  or project_type == "api" and stack.stack in _API_STACKS
+                  or project_type == "cli" and any(
+                      framework.name in _CLI_FRAMEWORKS for framework in stack.frameworks
+                  )
+                  or project_type == "cli" and stack.stack in {"cpp", "dotnet"}):
                 priority = 3
             return (priority, score, manifest_weight)
 

{sourcecode-1.12.0 → sourcecode-1.13.0}/src/sourcecode/cli.py

@@ -705,6 +705,11 @@ def main(
     elif mode not in _CONTRACT_MODES and mode != "raw":
         mode = "contract"   # unknown → safe default
 
+    # --changed-only forces compact output to bound size; full contract/raw scan
+    # of a large repo produces 100KB+ even with only 2 changed files.
+    if changed_only and not compact and not agent:
+        compact = True
+
     # Legacy flags imply raw mode unless --mode was explicitly overridden.
     # --format yaml and --graph-modules are now compatible with contract_view:
     #   yaml is a serialization format (not an output-section flag)
@@ -1498,33 +1503,6 @@ def main(
             content = json.dumps(data, indent=2, ensure_ascii=False)
     elif agent:
         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)
-            _jvm_stacks_agent = {"java", "kotlin", "scala", "groovy"}
-            _is_jvm_agent = any(s.stack in _jvm_stacks_agent for s in sm.stacks)
-            _MAX_AGENT_CONTRACTS = 50 if _is_jvm_agent else 10
-            _all_contracts = sm.file_contracts
-            _sorted = sorted(_all_contracts, key=lambda c: getattr(c, "relevance_score", 0.0), reverse=True)
-            _sampled = _sorted[:_MAX_AGENT_CONTRACTS]
-            _total_contracts = len(_all_contracts)
-            data["contracts"] = [_serialize_contract_minimal(c) for c in _sampled]
-            data["contracts_meta"] = {
-                "total": _total_contracts,
-                "shown": len(_sampled),
-                "truncated": _total_contracts > _MAX_AGENT_CONTRACTS,
-            }
-            if sm.contract_summary is not None:
-                cs = sm.contract_summary
-                data["contract_summary"] = {
-                    "files": cs.extracted_files,
-                    "total": cs.total_files,
-                }
-                if cs.method_breakdown:
-                    data["contract_summary"]["methods"] = cs.method_breakdown
         if not no_redact:
             data = redact_dict(data)
         content = json.dumps(data, indent=2, ensure_ascii=False)
@@ -1701,7 +1679,10 @@ def prepare_context_cmd(
         "project_summary": output.project_summary,
         "architecture_summary": output.architecture_summary,
         "confidence": output.confidence,
-        "relevant_files": [asdict(f) for f in output.relevant_files],
+        "relevant_files": [
+            {k: v for k, v in asdict(f).items() if v != ""}
+            for f in output.relevant_files
+        ],
         "why_these_files": output.why_these_files,
         "key_dependencies": output.key_dependencies,
     }

{sourcecode-1.12.0 → sourcecode-1.13.0}/src/sourcecode/detectors/java.py

@@ -37,8 +37,9 @@ _REQUEST_METHOD_VERB_RE = re.compile(
 # @M3FiltroSeguridad custom security annotation
 _M3_FILTRO_RE = re.compile(r'@M3FiltroSeguridad\b')
 _M3_FILTRO_PARAMS_RE = re.compile(
-    r'@M3FiltroSeguridad\s*\(\s*(?:nombreRecurso\s*=\s*"([^"]*)")?'
-    r'(?:[^)]*nivelRequerido\s*=\s*(\d+))?'
+    r'@M3FiltroSeguridad\s*\(\s*'
+    r'(?:nombreRecurso\s*=\s*(?:"([^"]*)"|([\w.]+)))?'  # group 1: string literal, group 2: constant ref
+    r'(?:[^)]*nivelRequerido\s*=\s*(\d+))?'  # group 3: nivel
 )
 
 # Security config detection
@@ -391,8 +392,8 @@ class JavaDetector(AbstractDetector):
             security_evidence = None
             m3_match = _M3_FILTRO_PARAMS_RE.search(content)
             if m3_match:
-                nombre = m3_match.group(1) or ""
-                nivel = m3_match.group(2) or ""
+                nombre = m3_match.group(1) or m3_match.group(2) or ""
+                nivel = m3_match.group(3) or ""
                 security_evidence = f"@M3FiltroSeguridad(nombreRecurso={nombre!r}, nivelRequerido={nivel})"
             return [EntryPoint(
                 path=rel_path, stack="java", kind="rest_controller",
@@ -416,8 +417,8 @@ class JavaDetector(AbstractDetector):
             security_evidence = None
             m3_match = _M3_FILTRO_PARAMS_RE.search(content)
             if m3_match:
-                nombre = m3_match.group(1) or ""
-                nivel = m3_match.group(2) or ""
+                nombre = m3_match.group(1) or m3_match.group(2) or ""
+                nivel = m3_match.group(3) or ""
                 security_evidence = f"@M3FiltroSeguridad(nombreRecurso={nombre!r}, nivelRequerido={nivel})"
             return [EntryPoint(
                 path=rel_path, stack="java", kind="mvc_controller",

{sourcecode-1.12.0 → sourcecode-1.13.0}/src/sourcecode/prepare_context.py

@@ -563,6 +563,15 @@ class TaskContextBuilder:
         except Exception:
             pass
 
+        # ── 5c. Delta: resolve git-changed files BEFORE ranking ───────────────
+        # For delta task, relevant_files must rank only files changed in the
+        # specified git range, not the full repo by generic entrypoint scoring.
+        _delta_files: Optional[set[str]] = None
+        if task_name == "delta":
+            _delta_raw = self._get_git_changed_files(since=since)
+            if _delta_raw:
+                _delta_files = set(_delta_raw)
+
         # ── 5c. review-pr suspected_areas (needs git uncommitted_files) ──────
         if task_name == "review-pr" and spec.enable_code_notes:
             pr_areas: dict[str, int] = {}
@@ -593,6 +602,7 @@ class TaskContextBuilder:
             git_hotspots=git_hotspots,
             uncommitted_files=uncommitted_files,
             code_notes=cn_notes_for_ranking if cn_notes_for_ranking else None,
+            delta_files=_delta_files,
         )
 
         # ── 7. Test gaps (generate-tests only) ────────────────────────────
@@ -640,11 +650,11 @@ class TaskContextBuilder:
             rf.path: rf.reason for rf in relevant_files
         }
 
-        # ── 10. Delta: git changed files ──────────────────────────────────────
+        # ── 10. Delta: git changed files (reuse pre-computed set from step 5c) ──
         changed_files: list[str] = []
         affected_entry_points: list[str] = []
         if task_name == "delta":
-            changed_files = self._get_git_changed_files(since=since)
+            changed_files = sorted(_delta_files) if _delta_files else self._get_git_changed_files(since=since)
             ep_set = {ep.path for ep in entry_points}
             affected_entry_points = [f for f in changed_files if f in ep_set]
 
@@ -746,6 +756,7 @@ class TaskContextBuilder:
         git_hotspots: Optional[dict[str, int]] = None,
         uncommitted_files: Optional[set[str]] = None,
         code_notes: Optional[list] = None,
+        delta_files: Optional[set[str]] = None,
     ) -> list[RelevantFile]:
         from sourcecode.ranking_engine import RankingEngine
         from sourcecode.file_classifier import FileClassifier
@@ -791,7 +802,10 @@ class TaskContextBuilder:
 
         scored: list[tuple[float, str, RelevantFile]] = []
 
-        for path in all_paths:
+        # For delta task, score only files changed in the specified git range.
+        paths_to_score = [p for p in all_paths if p in delta_files] if delta_files else all_paths
+
+        for path in paths_to_score:
             if Path(path).suffix.lower() not in _ALL_EXTENSIONS:
                 continue
             if any(pen in path for pen in spec.ranking_penalties):

{sourcecode-1.12.0 → sourcecode-1.13.0}/src/sourcecode/serializer.py

@@ -700,17 +700,26 @@ def compact_view(sm: SourceMap, *, no_tree: bool = False) -> dict[str, Any]:
             for ep in ep_groups["production"][:_EP_PRODUCTION_CAP]
         ]
 
-    # Stacks — name + method + confidence + frameworks (names only)
+    # Stacks — deduplicated: for same stack name, prefer manifest over heuristic
+    _stack_best: dict[str, Any] = {}
+    for _s in sm.stacks:
+        _existing = _stack_best.get(_s.stack)
+        if _existing is None:
+            _stack_best[_s.stack] = _s
+        elif _s.detection_method != "heuristic" and _existing.detection_method == "heuristic":
+            _stack_best[_s.stack] = _s
+        elif _s.primary and not _existing.primary and _s.detection_method == _existing.detection_method:
+            _stack_best[_s.stack] = _s
     stacks_compact = [
         {
             "stack": s.stack,
             "detection_method": s.detection_method,
             "confidence": s.confidence,
             **({"primary": True} if s.primary else {}),
-            **({"frameworks": [f.name for f in s.frameworks]} if s.frameworks else {}),
+            **({"frameworks": list(dict.fromkeys(f.name for f in s.frameworks))} if s.frameworks else {}),
             **({"package_manager": s.package_manager} if s.package_manager else {}),
         }
-        for s in sm.stacks
+        for s in _stack_best.values()
     ]
 
     # Confidence — overall only + anomalies