sourcecode 1.32.0__py3-none-any.whl → 1.32.2__py3-none-any.whl

sourcecode/__init__.py

@@ -1,3 +1,3 @@
 """sourcecode — Deterministic codebase context maps for AI coding agents."""
 
-__version__ = "1.32.0"
+__version__ = "1.32.2"

sourcecode/architecture_analyzer.py

@@ -152,14 +152,15 @@ LAYER_PATTERNS: dict[str, dict[str, list[str]]] = {
 
 # Higher value = wins when score ties
 _PATTERN_PRIORITY: dict[str, int] = {
-    "cqrs":       8,
-    "clean":      7,
-    "onion":      6,
-    "hexagonal":  5,
-    "monorepo":   4,
-    "mvc":        3,
-    "layered":    2,
-    "fullstack":  1,
+    "cqrs":               8,
+    "clean":              7,
+    "onion":              6,
+    "hexagonal":          5,
+    "monorepo":           4,
+    "spring_mvc_layered": 3,
+    "mvc":                3,
+    "layered":            2,
+    "fullstack":          1,
 }
 
 
@@ -248,9 +249,9 @@ class ArchitectureAnalyzer:
         pattern, layers = self._detect_layers(filtered)
         if pattern in (None, "flat", "unknown"):
             if pattern == "flat":
-                limitations.append("Patron de capas no detectado: proyecto con estructura plana")
+                limitations.append("Layer pattern not detected: project has a flat directory structure")
             elif pattern == "unknown":
-                limitations.append("Patron de capas no reconocido: estructura de directorios sin senales claras")
+                limitations.append("Unrecognized layer pattern: directory structure has no clear architectural signals")
 
         # Step 3b: monorepo override — workspace config is hard evidence.
         # Overrides all weak inferred patterns; only truly specialised patterns
@@ -269,7 +270,7 @@ class ArchitectureAnalyzer:
                 pattern = "monorepo"
                 layers = mono_layers
                 limitations.append(
-                    "Workspace config detectado — arquitectura refleja topologia de paquetes"
+                    "Workspace config detected — architecture reflects package topology"
                 )
                 ws_files = [p for p in sm.file_paths if p.split("/")[-1] in _WORKSPACE_CONFIG_FILES]
                 evidence.append({
@@ -570,28 +571,76 @@ class ArchitectureAnalyzer:
                 ))
             return best_pattern, layers
 
-        # 2. Microservices structural detection (before file-naming heuristics)
+        # 2. Spring domain-module detection (petclinic-style: deep common prefix + feature dirs)
+        spring_result = self._detect_spring_domain_modules(source_paths)
+        if spring_result is not None:
+            return spring_result
+
+        # 3. Microservices structural detection (before file-naming heuristics)
         microservices_result = self._detect_microservices(source_paths)
         if microservices_result is not None:
             return microservices_result
 
-        # 3. Functional file-naming heuristic: *_analyzer.py, cli.py, schema.py, …
+        # 5. Functional file-naming heuristic: *_analyzer.py, cli.py, schema.py, …
         func_result = self._detect_layered_functional(source_paths)
         if func_result is not None:
             return func_result
 
-        # 4. Modular sub-package heuristic: ≥2 distinct named sub-packages
+        # 6. Modular sub-package heuristic: ≥2 distinct named sub-packages
         modular_result = self._detect_modular(source_paths)
         if modular_result is not None:
             return modular_result
 
-        # 5. Fallback: flat (shallow) vs truly unknown (deep but unrecognised)
+        # 7. Fallback: flat (shallow) vs truly unknown (deep but unrecognised)
         max_depth = max(
             (len(p.replace("\\", "/").split("/")) - 1 for p in source_paths),
             default=0,
         )
         return ("flat" if max_depth <= 2 else "unknown"), []
 
+    def _detect_spring_domain_modules(
+        self, paths: list[str]
+    ) -> Optional[tuple[str, list[ArchitectureLayer]]]:
+        """Detect Spring Boot domain-organized packages (petclinic-style).
+
+        When all source paths share a deep common package prefix
+        (e.g. src/main/java/org/springframework/samples/petclinic/),
+        strips that prefix and detects feature/domain modules in the remainder.
+        Requires ≥3 distinct domain directories to avoid false positives.
+        """
+        if len(paths) < 6:
+            return None
+
+        parts_list = [p.replace("\\", "/").split("/") for p in paths]
+        min_depth = min(len(p) for p in parts_list)
+        common_depth = 0
+        for i in range(min_depth - 1):
+            seg = parts_list[0][i]
+            if all(pl[i] == seg for pl in parts_list):
+                common_depth = i + 1
+            else:
+                break
+
+        if common_depth < 3:
+            return None
+
+        module_files: dict[str, list[str]] = {}
+        for orig, pl in zip(paths, parts_list):
+            remaining = pl[common_depth:]
+            if remaining and remaining[0] not in _GENERIC_NAMES:
+                module_files.setdefault(remaining[0], []).append(orig)
+
+        meaningful = {k: v for k, v in module_files.items() if len(v) >= 2}
+        if len(meaningful) < 3:
+            return None
+
+        return "spring_mvc_layered", [
+            ArchitectureLayer(
+                name=k, pattern="spring_mvc_layered", files=v, confidence="medium"
+            )
+            for k, v in meaningful.items()
+        ]
+
     def _detect_microservices(
         self, paths: list[str]
     ) -> Optional[tuple[str, list[ArchitectureLayer]]]:

sourcecode/classifier.py

@@ -118,6 +118,10 @@ class TypeClassifier:
         ):
             return "webapp"
 
+        _SERVERSIDE_TEMPLATE_FRAMEWORKS = frozenset({"Thymeleaf", "FreeMarker"})
+        if framework_names & _SERVERSIDE_TEMPLATE_FRAMEWORKS:
+            return "web_mvc"
+
         if framework_names & _API_FRAMEWORKS:
             return "api"
 

sourcecode/cli.py

@@ -140,7 +140,21 @@ def _check_pipeline_coherence(sm: "SourceMap") -> list[str]:  # type: ignore[nam
 
     return issues
 
-_HELP = """\
+def _build_help_text() -> str:
+    """Build --help text dynamically based on current license state."""
+    try:
+        from sourcecode.license import is_pro as _is_pro
+    except Exception:
+        _is_pro = False
+
+    if _is_pro:
+        plan_badge = "[bold green]● Pro[/bold green]"
+    else:
+        plan_badge = "[yellow]Free[/yellow]  ·  [dim]sourcecode activate <key>[/dim] to unlock Pro"
+
+    text = f"""\
+[bold]sourcecode[/bold]  {plan_badge}
+
 Deterministic codebase context for AI coding agents.
 
 [bold]Primary usage:[/bold]
@@ -164,6 +178,26 @@ Deterministic codebase context for AI coding agents.
   version
 """
 
+    if not _is_pro:
+        text += """\
+
+[dim bold]Locked (Pro):[/dim bold]
+  [dim]impact                              blast radius before any change[/dim]
+  [dim]modernize (full)                    dead zones, tangles, full coupling[/dim]
+  [dim]fix-bug (full)                      complete risk-ranked file list[/dim]
+  [dim]review-pr (expanded)               CI-grade PR review[/dim]
+  [dim]prepare-context delta               incremental context for CI/CD[/dim]
+  [dim]prepare-context generate-tests      test gap analysis[/dim]
+  [dim]--full                              removes all truncation limits[/dim]
+
+  [dim cyan]→ sourcecode activate <key>[/dim cyan]
+"""
+
+    return text
+
+
+_HELP = _build_help_text()
+
 # Known subcommand names — tokens matching these are routed as subcommands,
 # not consumed as a repository path.
 _SUBCOMMANDS: frozenset[str] = frozenset(
@@ -344,6 +378,10 @@ def _get_command_with_preprocessing(typer_instance: Any) -> Any:
     cmd = _orig_get_command(typer_instance)
     if typer_instance is not app:
         return cmd  # only wrap the root app, not telemetry_app etc.
+
+    # Refresh help text at invocation time so it reflects current license state.
+    cmd.help = _build_help_text()
+
     _orig_cmd_main = cmd.main
 
     def _cmd_main(args: Optional[list[str]] = None, **kwargs: Any) -> Any:
@@ -2011,6 +2049,15 @@ def main(
         except Exception:
             pass  # non-fatal: cache write failure
 
+    # Update RIS with aggregated snapshot data (non-fatal side-effect).
+    if not no_cache and not _pipeline_error and _core_key:
+        try:
+            from sourcecode.serializer import core_view as _ris_core_view
+            from sourcecode.ris import maybe_update_ris as _ris_update
+            _ris_update(target, _ris_core_view(sm), _git_sha)
+        except Exception:
+            pass
+
     if _pipeline_error:
         raise typer.Exit(code=2)
 
@@ -3147,6 +3194,14 @@ def endpoints_cmd(
         raise typer.Exit(code=1)
 
     data = _extract_java_endpoints(target)
+
+    # Update RIS api_surface section (non-fatal side-effect).
+    try:
+        from sourcecode.ris import update_ris_api_surface as _ris_ep
+        _ris_ep(target, data)
+    except Exception:
+        pass
+
     output = _serialize_dict(data, format)
 
     if output_path is not None:
@@ -3824,7 +3879,7 @@ def mcp_init(
         raise typer.Exit(code=1)
 
     typer.echo("MCP integration active.")
-    typer.echo("  Note:    repo_path debe usar forward slashes: C:/Users/... o /ruta/unix")
+    typer.echo("  Note:    repo_path must use forward slashes: C:/Users/... or /unix/path")
     typer.echo("")
 
     # Post-write: validate config and warn if client not running
@@ -3976,7 +4031,7 @@ def mcp_status() -> None:
     typer.echo(sep)
     typer.echo("  Note:    'configured' and 'running' are checked independently.")
     typer.echo("           A running app still needs restart after first-time config.")
-    typer.echo("  Path:    repo_path debe usar forward slashes: C:/Users/... o /ruta/unix")
+    typer.echo("  Path:    repo_path must use forward slashes: C:/Users/... or /unix/path")
     typer.echo("  Setup:   sourcecode mcp init")
     typer.echo("  Remove:  sourcecode mcp remove")
 

sourcecode/context_summarizer.py

@@ -28,6 +28,7 @@ _STACK_LABELS: dict[str, str] = {
 
 _TYPE_LABELS: dict[str, str] = {
     "api": "REST API",
+    "web_mvc": "Spring MVC web app",
     "webapp": "Web app",
     "fullstack": "Full-stack app",
     "cli": "CLI tool",

sourcecode/dependency_analyzer.py

@@ -1118,7 +1118,7 @@ class DependencyAnalyzer:
         try:
             tree = ET.parse(pom)
         except (ET.ParseError, OSError):
-            return [], ["java: error al parsear pom.xml"]
+            return [], ["java: error parsing pom.xml"]
 
         root_elem = tree.getroot()
         ns_match = re.match(r"\{[^}]+\}", root_elem.tag)
@@ -1217,10 +1217,10 @@ class DependencyAnalyzer:
                 try:
                     content = gradle_file.read_text(encoding="utf-8", errors="replace")
                 except OSError:
-                    return [], [f"gradle: error al leer {filename}"]
+                    return [], [f"gradle: error reading {filename}"]
                 props = self._parse_gradle_properties(root, content)
                 records = self._parse_gradle_dependencies(content, props, filename)
-                return records, ["gradle: sin lockfile compatible; dependencias transitivas no disponibles"]
+                return records, ["gradle: no compatible lockfile found; transitive dependencies unavailable"]
         return [], []
 
     def _parse_gradle_properties(self, root: Path, content: str) -> dict[str, str]:

sourcecode/detectors/java.py

@@ -23,6 +23,7 @@ _MAX_ANNOTATION_ENTRY_POINTS = 1000
 _REST_CONTROLLER_RE = re.compile(r'@RestController\b')
 _MVC_CONTROLLER_RE = re.compile(r'@Controller\b')
 _REQUEST_MAPPING_RE = re.compile(r'@RequestMapping\b')
+_ANY_MAPPING_RE = re.compile(r'@(?:Request|Get|Post|Put|Delete|Patch)Mapping\b')
 _CONTROLLER_ADVICE_RE = re.compile(r'@ControllerAdvice\b')
 _WEB_FILTER_RE = re.compile(r'@WebFilter\b')
 _FILTER_BEAN_RE = re.compile(r'FilterRegistrationBean\b')
@@ -298,6 +299,10 @@ class JavaDetector(AbstractDetector):
             frameworks.append(FrameworkDetection(name="Jakarta EE", source=source))
         if "mybatis" in text:
             frameworks.append(FrameworkDetection(name="MyBatis", source=source))
+        if "thymeleaf" in text:
+            frameworks.append(FrameworkDetection(name="Thymeleaf", source=source))
+        if "freemarker" in text:
+            frameworks.append(FrameworkDetection(name="FreeMarker", source=source))
         if "spring-boot-starter-security" in text or "spring-security-core" in text:
             frameworks.append(FrameworkDetection(name="Spring Security", source=source))
         if "spring-boot-starter-data-jpa" in text or "spring-data-jpa" in text:
@@ -470,7 +475,7 @@ class JavaDetector(AbstractDetector):
                 path=rel_path, stack="java", kind="exception_handler",
                 source="annotation", confidence="medium",
             )]
-        if _MVC_CONTROLLER_RE.search(content) and _REQUEST_MAPPING_RE.search(content):
+        if _MVC_CONTROLLER_RE.search(content) and _ANY_MAPPING_RE.search(content):
             http_path_match = _HTTP_PATH_RE.search(content)
             http_path = http_path_match.group(1) if http_path_match else None
             verb_match = _REQUEST_METHOD_VERB_RE.search(content)

sourcecode/mcp/server.py

@@ -109,6 +109,42 @@ def _check_repo_path(path: str) -> "CallToolResult | None":
     return None
 
 
+@mcp.tool()
+def get_cold_start_context(repo_path: str = ".") -> dict:
+    """Instant session bootstrap from persisted Repository Intelligence Snapshot (RIS).
+
+    CALL THIS FIRST at the start of every MCP session.  Returns cached structural
+    context built from prior analysis runs — zero re-analysis cost.
+
+    status values:
+      "cold_start_ready"  — RIS exists and matches the current git HEAD.
+      "cold_start_stale"  — RIS exists but HEAD has changed since last analysis.
+                            Data is still useful; run get_compact_context to refresh.
+      "no_ris"            — No RIS yet for this repo; run get_compact_context first.
+
+    Returns: status, repo_id, git_head, stale (bool), last_updated_at,
+             summary (compact snapshot), entrypoints, endpoints, hotspots.
+
+    repo_path: absolute path to the repository (default: current working directory).
+    """
+    _raw = repo_path
+    try:
+        if not isinstance(repo_path, str):
+            return _err("repo_path must be a string", "INVALID_ARGUMENT")
+        repo_path = _normalize_repo_path(repo_path)
+        _path_err = _check_repo_path(repo_path)
+        if _path_err is not None:
+            return _path_err
+        from pathlib import Path as _Path
+        from sourcecode.ris import get_cold_start_context as _gcs
+        return _ok(_gcs(_Path(repo_path)))
+    except Exception as exc:
+        return _err(
+            f"Internal error: {type(exc).__name__}: {exc} — repo_path: {_raw}",
+            "INTERNAL_ERROR",
+        )
+
+
 @mcp.tool()
 def get_compact_context(repo_path: str = ".", git_context: bool = False) -> dict:
     """Compact human/LLM summary of a repository (~1000-3000 tokens). USE THIS FIRST.

sourcecode/prepare_context.py

@@ -981,12 +981,18 @@ class TaskContextBuilder:
 
             dep_records, dep_summary = DependencyAnalyzer().analyze(self.root)
             primary_eco = stacks[0].stack if stacks else ""
-            direct = [
+            _direct_raw = [
                 d for d in dep_records
                 if d.scope != "transitive" and d.source in {"manifest", "lockfile"}
                 and (d.role or "unknown") in {"runtime", "parsing", "serialization", "observability", "infra"}
                 and d.scope not in {"dev"}
             ]
+            _seen_dep: set[str] = set()
+            direct = []
+            for _d in _direct_raw:
+                if _d.name not in _seen_dep:
+                    _seen_dep.add(_d.name)
+                    direct.append(_d)
             # Rank by framework centrality: core infra (ORM, Spring) > serialization > other.
             # Penalise vendored tooling (closure-compiler, shaded utilities) so that
             # Hibernate/JPA/Solr appear before minor build-time dependencies.

sourcecode/repository_ir.py

@@ -1697,6 +1697,9 @@ def _canonical_subsystem_pkg(fqn: str) -> str:
     (even if uppercase) to force at least 2-segment grouping.
     """
     _TOP_LEVEL = {"com", "org", "net", "io", "java", "javax"}
+    # Well-known framework namespaces that are not application boundaries;
+    # go one level deeper (depth 5) so callers get the actual app module.
+    _FRAMEWORK_NS = {"springframework", "apache", "eclipse", "google", "jetbrains"}
     parts: list[str] = []
     for segment in fqn.split("."):
         if "#" in segment or (segment and segment[0].isupper()):
@@ -1705,6 +1708,8 @@ def _canonical_subsystem_pkg(fqn: str) -> str:
     if not parts:
         return fqn.rsplit(".", 1)[0] if "." in fqn else fqn
     if parts[0] in _TOP_LEVEL and len(parts) >= 3:
+        if len(parts) >= 5 and len(parts) > 3 and parts[1] in _FRAMEWORK_NS:
+            return ".".join(parts[:5])
         return ".".join(parts[:3])
     # Prevent bare TLD collapse: "org" or "com" alone as subsystem key is meaningless
     # and groups ALL classes under that TLD into a single giant component.
@@ -3338,7 +3343,7 @@ def compute_blast_radius(
             _seen_mapper_fqns.add(fqn)
             _mapper_entry: dict = {
                 "fqn": fqn,
-                "role": role or "mapper",
+                "role": role or ("mapper" if symbol_kind == "mapper_interface" else "repository"),
                 "source_file": node_dict.get("source_file") or "",
             }
             if canonical != fqn:

sourcecode/ris.py

@@ -0,0 +1,372 @@
+"""Repository Intelligence Snapshot (RIS) — persistent repo-level semantic state.
+
+RIS aggregates outputs from per-commit L1/L2 snapshots into a single artifact
+that survives across git commits.  It enables cold-start bootstrapping for MCP
+sessions: on first tool call, agents receive structural context immediately
+without triggering re-analysis.
+
+Storage layout:
+    ~/.sourcecode/cache/<repo_id>/ris.json.gz
+
+Build lifecycle:
+    1. After every successful analysis run, ``maybe_update_ris`` is called with
+       the L1 core_dict (already in scope in cli.py after ``write_core``).
+    2. If the git HEAD matches the stored RIS → only compact/agent/git sections
+       are refreshed (the API surface from ``endpoints`` is preserved).
+    3. If the git HEAD changed → all sections are rebuilt from the new core_dict
+       (api_surface preserved if the caller does not provide updated endpoints).
+    4. ``get_cold_start_context`` reads the RIS and returns a lightweight
+       bootstrap object.  It is safe to call from MCP on every session init.
+
+RIS is NOT a command cache.  It stores a semantic model of the repository
+derived from existing snapshot outputs, not raw command results.
+"""
+from __future__ import annotations
+
+import gzip
+import json
+import subprocess
+from dataclasses import asdict, dataclass
+from datetime import datetime, timezone
+from pathlib import Path
+from typing import Optional
+
+RIS_SCHEMA_VERSION: str = "1.0"
+_RIS_FILENAME: str = "ris.json.gz"
+
+
+# ---------------------------------------------------------------------------
+# Schema
+# ---------------------------------------------------------------------------
+
+@dataclass
+class RepositoryIntelligenceSnapshot:
+    repo_id: str
+    created_at: str
+    last_updated_at: str
+    git_head: str
+    version: str
+    structural_map: dict
+    api_surface: dict
+    dependency_graph: dict
+    compact_summary: dict
+    agent_index: dict
+    git_context_snapshot: dict
+    metadata: dict
+
+
+# ---------------------------------------------------------------------------
+# Storage helpers
+# ---------------------------------------------------------------------------
+
+def _ris_path(repo_root: Path) -> Path:
+    from sourcecode.cache import cache_dir as _cache_dir
+    return _cache_dir(repo_root) / _RIS_FILENAME
+
+
+def load_ris(repo_root: Path) -> Optional[RepositoryIntelligenceSnapshot]:
+    """Load RIS from disk.  Returns None on any error (missing, corrupt, stale schema)."""
+    try:
+        p = _ris_path(repo_root)
+        if not p.exists():
+            return None
+        raw = gzip.decompress(p.read_bytes())
+        d = json.loads(raw)
+        if not isinstance(d, dict) or d.get("version") != RIS_SCHEMA_VERSION:
+            return None
+        return RepositoryIntelligenceSnapshot(
+            repo_id=d.get("repo_id", ""),
+            created_at=d.get("created_at", ""),
+            last_updated_at=d.get("last_updated_at", ""),
+            git_head=d.get("git_head", ""),
+            version=d.get("version", ""),
+            structural_map=d.get("structural_map", {}),
+            api_surface=d.get("api_surface", {}),
+            dependency_graph=d.get("dependency_graph", {}),
+            compact_summary=d.get("compact_summary", {}),
+            agent_index=d.get("agent_index", {}),
+            git_context_snapshot=d.get("git_context_snapshot", {}),
+            metadata=d.get("metadata", {}),
+        )
+    except Exception:
+        return None
+
+
+def save_ris(repo_root: Path, ris: RepositoryIntelligenceSnapshot) -> None:
+    """Write RIS to disk as gzip-compressed JSON.  Never raises."""
+    try:
+        p = _ris_path(repo_root)
+        p.parent.mkdir(parents=True, exist_ok=True)
+        raw = json.dumps(asdict(ris), ensure_ascii=False, separators=(",", ":"))
+        p.write_bytes(gzip.compress(raw.encode("utf-8"), compresslevel=6))
+    except Exception:
+        pass
+
+
+# ---------------------------------------------------------------------------
+# Data extraction from L1 core_dict
+# ---------------------------------------------------------------------------
+
+def _extract_structural_map(agent_data: dict) -> dict:
+    """Extract structural map from agent_view output."""
+    entry_points = agent_data.get("entry_points", [])
+
+    layers = []
+    domains = []
+    arch = agent_data.get("architecture")
+    if isinstance(arch, dict):
+        layers = arch.get("layers") or []
+        domains = arch.get("domains") or []
+
+    controllers: list = []
+    services: list = []
+    for layer in layers:
+        if not isinstance(layer, dict):
+            continue
+        name = (layer.get("name") or "").lower()
+        files = layer.get("files") or []
+        if any(kw in name for kw in ("controller", "rest", "api", "endpoint")):
+            controllers.extend(files)
+        elif any(kw in name for kw in ("service", "business", "domain", "use_case")):
+            services.extend(files)
+
+    return {
+        "entrypoints": entry_points,
+        "controllers": controllers,
+        "services": services,
+        "modules": domains,
+    }
+
+
+def _extract_api_surface(agent_data: dict) -> dict:
+    """Extract API surface from agent_view output (Java security surface if present)."""
+    endpoints: list = []
+    controllers_index: list = []
+
+    signals = agent_data.get("signals") or {}
+    sec = signals.get("security_surface") or {}
+    if isinstance(sec, dict):
+        raw_endpoints = sec.get("endpoints") or []
+        if isinstance(raw_endpoints, list):
+            endpoints = raw_endpoints
+        controllers_index = sec.get("controllers") or []
+
+    return {
+        "endpoints": endpoints,
+        "controllers_index": controllers_index,
+    }
+
+
+def _extract_dependency_graph(compact_data: dict, agent_data: dict) -> dict:
+    """Extract dependency graph summary from view outputs."""
+    nodes = agent_data.get("key_dependencies") or compact_data.get("key_dependencies") or []
+    summary = compact_data.get("dependency_summary") or {}
+
+    return {
+        "nodes": nodes,
+        "edges": [],
+        "summary": summary,
+    }
+
+
+def _extract_git_context_snapshot(agent_data: dict) -> dict:
+    """Extract git context from agent_view output."""
+    gc = agent_data.get("git_context") or {}
+    if not isinstance(gc, dict):
+        return {"last_commits": [], "hotspots": []}
+    return {
+        "last_commits": gc.get("recent_commits") or [],
+        "hotspots": gc.get("top_hotspots") or [],
+    }
+
+
+# ---------------------------------------------------------------------------
+# Build / update
+# ---------------------------------------------------------------------------
+
+def _now_iso() -> str:
+    return datetime.now(timezone.utc).isoformat()
+
+
+def _build_from_core(
+    repo_root: Path,
+    core_dict: dict,
+    git_head: str,
+    existing_api_surface: Optional[dict] = None,
+) -> RepositoryIntelligenceSnapshot:
+    from sourcecode.cache import repo_id as _repo_id_fn
+
+    compact_data = core_dict.get("_compact") or {}
+    agent_data = core_dict.get("_agent") or {}
+
+    # Strip file_relevance — large and not needed for cold-start
+    agent_index = {k: v for k, v in agent_data.items() if k != "file_relevance"}
+
+    now = _now_iso()
+    return RepositoryIntelligenceSnapshot(
+        repo_id=_repo_id_fn(repo_root),
+        created_at=now,
+        last_updated_at=now,
+        git_head=git_head,
+        version=RIS_SCHEMA_VERSION,
+        structural_map=_extract_structural_map(agent_data),
+        api_surface=existing_api_surface or _extract_api_surface(agent_data),
+        dependency_graph=_extract_dependency_graph(compact_data, agent_data),
+        compact_summary=compact_data,
+        agent_index=agent_index,
+        git_context_snapshot=_extract_git_context_snapshot(agent_data),
+        metadata={
+            "snapshot_source": "existing_snapshot_system",
+            "confidence": float(
+                1.0 if agent_data else (0.5 if compact_data else 0.0)
+            ),
+            "partial": not bool(agent_data),
+        },
+    )
+
+
+def maybe_update_ris(repo_root: Path, core_dict: dict, git_head: str) -> None:
+    """Update (or create) the RIS artifact from an L1 core_dict.
+
+    Called from cli.py after every successful write_core.  Never raises.
+    """
+    try:
+        if not isinstance(core_dict, dict):
+            return
+
+        existing = load_ris(repo_root)
+
+        if existing is not None and existing.git_head == git_head and git_head:
+            # Same commit — refresh analysis sections, preserve api_surface
+            compact_data = core_dict.get("_compact") or {}
+            agent_data = core_dict.get("_agent") or {}
+            agent_index = {k: v for k, v in agent_data.items() if k != "file_relevance"}
+
+            updated = RepositoryIntelligenceSnapshot(
+                repo_id=existing.repo_id,
+                created_at=existing.created_at,
+                last_updated_at=_now_iso(),
+                git_head=git_head,
+                version=RIS_SCHEMA_VERSION,
+                structural_map=_extract_structural_map(agent_data) if agent_data else existing.structural_map,
+                api_surface=existing.api_surface,  # preserve explicit endpoint data
+                dependency_graph=_extract_dependency_graph(compact_data, agent_data) if compact_data else existing.dependency_graph,
+                compact_summary=compact_data if compact_data else existing.compact_summary,
+                agent_index=agent_index if agent_index else existing.agent_index,
+                git_context_snapshot=_extract_git_context_snapshot(agent_data) if agent_data else existing.git_context_snapshot,
+                metadata={
+                    "snapshot_source": "existing_snapshot_system",
+                    "confidence": float(1.0 if agent_data else (0.5 if compact_data else 0.0)),
+                    "partial": not bool(agent_data),
+                },
+            )
+            save_ris(repo_root, updated)
+        else:
+            # New commit or first build — rebuild all sections (preserve api_surface if available)
+            existing_api = existing.api_surface if existing is not None else None
+            ris = _build_from_core(repo_root, core_dict, git_head, existing_api_surface=existing_api)
+            if existing is not None:
+                # Preserve creation timestamp
+                ris = RepositoryIntelligenceSnapshot(
+                    **{**asdict(ris), "created_at": existing.created_at}
+                )
+            save_ris(repo_root, ris)
+    except Exception:
+        pass
+
+
+def update_ris_api_surface(repo_root: Path, endpoints_data: dict) -> None:
+    """Update the api_surface section from an ``endpoints`` command output.
+
+    Called from endpoints_cmd after _extract_java_endpoints().  Never raises.
+    """
+    try:
+        if not isinstance(endpoints_data, dict):
+            return
+        endpoints = endpoints_data.get("endpoints") or []
+        existing = load_ris(repo_root)
+        if existing is None:
+            # No RIS yet — create a minimal stub so api_surface is persisted
+            from sourcecode.cache import repo_id as _repo_id_fn
+            now = _now_iso()
+            existing = RepositoryIntelligenceSnapshot(
+                repo_id=_repo_id_fn(repo_root),
+                created_at=now,
+                last_updated_at=now,
+                git_head="",
+                version=RIS_SCHEMA_VERSION,
+                structural_map={},
+                api_surface={},
+                dependency_graph={},
+                compact_summary={},
+                agent_index={},
+                git_context_snapshot={},
+                metadata={"snapshot_source": "existing_snapshot_system", "confidence": 0.0, "partial": True},
+            )
+
+        # Build controllers_index from endpoint controller fields
+        ctrl_set: set[str] = set()
+        for ep in endpoints:
+            if isinstance(ep, dict):
+                ctrl = ep.get("controller") or ep.get("controller_class") or ""
+                if ctrl:
+                    ctrl_set.add(ctrl)
+
+        updated_api = {
+            "endpoints": endpoints,
+            "controllers_index": sorted(ctrl_set),
+        }
+        updated = RepositoryIntelligenceSnapshot(
+            **{**asdict(existing), "api_surface": updated_api, "last_updated_at": _now_iso()}
+        )
+        save_ris(repo_root, updated)
+    except Exception:
+        pass
+
+
+# ---------------------------------------------------------------------------
+# Cold-start context
+# ---------------------------------------------------------------------------
+
+def _current_git_head(repo_root: Path) -> str:
+    """Return current HEAD SHA.  Returns '' on any error or non-git directory."""
+    try:
+        result = subprocess.run(
+            ["git", "-C", str(repo_root), "rev-parse", "HEAD"],
+            capture_output=True,
+            text=True,
+            timeout=2,
+        )
+        if result.returncode == 0:
+            return result.stdout.strip()
+    except Exception:
+        pass
+    return ""
+
+
+def get_cold_start_context(repo_root: Path) -> dict:
+    """Return a lightweight bootstrap object from the persisted RIS.
+
+    Never raises.  Returns ``{"status": "no_ris"}`` when no RIS exists.
+    """
+    try:
+        ris = load_ris(repo_root)
+        if ris is None:
+            return {"status": "no_ris"}
+
+        current_head = _current_git_head(repo_root)
+        stale = bool(current_head and ris.git_head and current_head != ris.git_head)
+
+        return {
+            "status": "cold_start_stale" if stale else "cold_start_ready",
+            "repo_id": ris.repo_id,
+            "git_head": ris.git_head,
+            "stale": stale,
+            "last_updated_at": ris.last_updated_at,
+            "summary": ris.compact_summary,
+            "entrypoints": ris.structural_map.get("entrypoints", []),
+            "endpoints": ris.api_surface.get("endpoints", []),
+            "hotspots": ris.git_context_snapshot.get("hotspots", []),
+        }
+    except Exception:
+        return {"status": "no_ris"}

sourcecode/summarizer.py

@@ -279,6 +279,7 @@ class ProjectSummarizer:
     _TYPE_LABELS: dict[str, str] = {
         "cli": "CLI",
         "api": "API",
+        "web_mvc": "Spring MVC web app",
         "webapp": "Web application",
         "library": "Library",
         "monorepo": "Monorepo",

{sourcecode-1.32.0.dist-info → sourcecode-1.32.2.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: sourcecode
-Version: 1.32.0
+Version: 1.32.2
 Summary: Deterministic codebase context for AI coding agents
 License:                                  Apache License
                                    Version 2.0, January 2004

{sourcecode-1.32.0.dist-info → sourcecode-1.32.2.dist-info}/RECORD

@@ -1,20 +1,20 @@
-sourcecode/__init__.py,sha256=1pcmq6UuzqBpI1Q4E_5ukKd_IJ8s8CN4xrW1_EyV0Gw,103
+sourcecode/__init__.py,sha256=tfPDIqlBjWLMfih5Bz8aKlWeo9EJQJnFkI8wjOakF7w,103
 sourcecode/adaptive_scanner.py,sha256=XffluXKzJUXrMtjEiAOnSNPZnztdIcts17T9ouHeID0,10521
-sourcecode/architecture_analyzer.py,sha256=Ry3aYT9dc7XuLmWLT5IZ93RkCf_P14Qtew0nGPvUl_8,42184
+sourcecode/architecture_analyzer.py,sha256=qh749a7ykPtGmQI1MR9y6j8TtL_jBdVYFx9YRsLqOMw,44121
 sourcecode/architecture_summary.py,sha256=z34_6v7cSwy98cof2UVciGho7SCrZ93tiqMmq5WNzRQ,20405
 sourcecode/ast_extractor.py,sha256=_btmeOJIe3t-NicF94D5ZAesa2YIJ0_QNExGnbHxGFE,50578
 sourcecode/cache.py,sha256=pBrPdpPrOgpXHHQO670U3aUfVf5N3A3obsTKgiZtN4I,23030
 sourcecode/canonical_ir.py,sha256=_HM3AUmKSdna9u4dCoU6rpgSA6HdF8gzOKZykIUCNGY,23277
-sourcecode/classifier.py,sha256=yWeq6agTjkFa3zuNa-gdVIHtjoBoPoVlJnX-b7tdVJs,7851
-sourcecode/cli.py,sha256=7GfBpmy2_6lUbuNz8zft4Vof-WOeGnzpeSNwxzDQndM,163863
+sourcecode/classifier.py,sha256=2lYoSH3vOTkXZYPU7Go2WIet1-IuNzTWVhc-ULnXtgw,8024
+sourcecode/cli.py,sha256=l3PJ9pK6JQ_JUxYSD3xnpvCIvR9kvOvbEEvHvqfij9A,165743
 sourcecode/code_notes_analyzer.py,sha256=EJemNCNc9Dn-1RZYu-aNbK0ELzmsyC4s6FdHi3XyNEI,9392
 sourcecode/confidence_analyzer.py,sha256=_jckZSxksV-OU38vbkxfVNBnWCtlCq8Vwfg23x1uspA,19054
 sourcecode/context_scorer.py,sha256=QpChSpsmaAYz91rXA4Ue5xzQmNz_ZboZN09YOHScq1U,14679
-sourcecode/context_summarizer.py,sha256=CiQrfBEzun949bWvmLabWoj2HhPn6Lw62ofqnsy0FlQ,6503
+sourcecode/context_summarizer.py,sha256=zlbm8ytdvJToohU108-dwBmEl52xl0gXpf6PZBOW_2A,6540
 sourcecode/contract_model.py,sha256=nRxJKPMs1VHwFTa8AVXhGmaLjti3Lr2sjHDpWgv1bfE,3917
 sourcecode/contract_pipeline.py,sha256=gvTdDniedm_mjq4vaHqnBY2UkQ0s00gtXqzTLILNXHc,28719
 sourcecode/coverage_parser.py,sha256=q0LeZJaX1bnntLu-ImksdBsMlpsVmk_iUfSaB4eaJGo,19702
-sourcecode/dependency_analyzer.py,sha256=Po7GKJnClCkXty0np1B4F1zo_bPeKAtgbehazhXuaBM,56493
+sourcecode/dependency_analyzer.py,sha256=wE7IgXOepfagGdRJX4-u5mhps_kfR2njz2RFgPikFFY,56491
 sourcecode/doc_analyzer.py,sha256=05bjTUbDbmnbajD_cgRnACzS8T7xxBKVX4CjkJlhZg8,24411
 sourcecode/entrypoint_classifier.py,sha256=MTa7yqbeuJ9XPbGCPuvtR9IqY-SN3hoXXyVtb3iXDhs,4316
 sourcecode/env_analyzer.py,sha256=9_q_U_Q1tjiY5FSaeEImSDgToP2uHGrjBAFF7ihJn2I,22204
@@ -28,19 +28,20 @@ sourcecode/metrics_analyzer.py,sha256=m0ENgtqKeBL17kUIK3fmGkgo7UfXBNHxCMj0H_Y5K7
 sourcecode/output_budget.py,sha256=43307mJEyUPU3MI-QEQoVxrcAvNyUzdzF_SAPgisBQE,6603
 sourcecode/path_filters.py,sha256=ROFRQ8eSLBEMiixK9f45-RO7um4VEEcjoD5AA4I427I,3739
 sourcecode/pr_comment_renderer.py,sha256=smHslxiG14lrytCkq5nFrFu-qTHgA-t-LFYfdrfjz2o,14423
-sourcecode/prepare_context.py,sha256=RM7ka0rduJy8kwGHzLU9if6q7D9ST7tGjOf5LnsdTuw,201451
+sourcecode/prepare_context.py,sha256=TFg8GDJ8kotJtR4rOTAvwQGVNSj0x-Usamy-mtjxeqs,201681
 sourcecode/progress.py,sha256=qn30sWaHOkjTgXsSBmiPkz7Rsbwc5oSlIe6JNEMYp_k,3149
 sourcecode/ranking_engine.py,sha256=ZAucq_YX2KkWUuAZf4P0lhtQ_38vEFnUhuGtSZd1S0E,12970
 sourcecode/redactor.py,sha256=SB4hwIvg8h-hvcqKcDWaZvA-aSyn-at-BIRwa0tUv5E,3227
 sourcecode/relevance_scorer.py,sha256=MYF4FFkveAQps9SmTeTlh6ODiBz2F--_hWNeHMLtUHQ,8405
 sourcecode/repo_classifier.py,sha256=FG1vaWKdWXsWdl-S8hjVMiTqcwgaRXkDyvK4rPcOGtQ,22681
-sourcecode/repository_ir.py,sha256=rga0I9L0K-3kSMTF8R8slBH4Fkwwrh-ut4Y3h6GxLvY,152757
+sourcecode/repository_ir.py,sha256=-NjBQUT7zyya4ng8Hq0-ChoiHZkUif9lr-Q878gmj8M,153163
+sourcecode/ris.py,sha256=iy1O04SiQ3cFIK5Qn8iHt3fZTiXIgbXvWkjbBG0fTBo,14000
 sourcecode/runtime_classifier.py,sha256=uTAD6BDCiBLUZEDRfqk718kM4RTT_vAbfkcOI2_Xx58,18432
 sourcecode/scanner.py,sha256=WdOQ78mMzjR1NjmKTlbxdgwinnCTfAhxCVLBEFQiFHU,8899
 sourcecode/schema.py,sha256=aHNXDf8LGyUC8ZDE_VS9kiskC2-Oswhi_WnpdGy6HDw,24897
 sourcecode/semantic_analyzer.py,sha256=TDuC3wzZR2DPm1mgrAg1YSLk2QzJoueS3TZAmyGGpCU,89417
 sourcecode/serializer.py,sha256=3JBvsMDj5pt1RXpi1zJpk5DGWUKbeb2Jl622-kmYWD4,123312
-sourcecode/summarizer.py,sha256=BMHJA0Do4rBnabc1_BxHoETTNb5ew0VqCX_eY3_PdCg,20706
+sourcecode/summarizer.py,sha256=YspHEVeYJVmltq0FMtGZF8kIP3qiR2KLcanGL6Y7uTI,20747
 sourcecode/tree_utils.py,sha256=8GAkIfQAsvtEudIeW1l4ooH_oRtrWR8cpJQJsEa_Pfw,2093
 sourcecode/workspace.py,sha256=X_6NmNnitvT3_38V-JDChydo_sR68s249hLFlrQskU0,8271
 sourcecode/detectors/__init__.py,sha256=A0AACJFF6HWf_RgatNtWu3PUzstcKtIGM9f1PoFcJug,1987
@@ -52,7 +53,7 @@ sourcecode/detectors/elixir.py,sha256=jCpvt5Yi6jvplc80ovRtWh17q-11ZGo9qX7o8b57TJ
 sourcecode/detectors/go.py,sha256=2r66uRQfeTWsqxr4HDhT6vExZErby0t46QXLHVBRv9w,2782
 sourcecode/detectors/heuristic.py,sha256=7cRxrip4yIaggYzZJB6ef8yHKh-gHgiH_pXMFcjlyFU,3723
 sourcecode/detectors/hybrid.py,sha256=IGFRUVsAZ1ooRlFdznCeJAV6vy1yVDx-VyghvLtddXc,9101
-sourcecode/detectors/java.py,sha256=YmmbRChVt_E4yTLfz3jzdNtbCtyQGS4RD7Xt95_rtdg,28979
+sourcecode/detectors/java.py,sha256=BMdDLBe6vzMa8dqNc8cZ6An0OpQmW5oWcJ02ULzHC_Q,29288
 sourcecode/detectors/jvm_ext.py,sha256=EgHJ5W8EE-ZTN9V607mVzohyKgZE8Mc2jCi-DF8RAZU,2616
 sourcecode/detectors/nodejs.py,sha256=Hg3Gmr7yIMJFiLoDwOTk2wtu00wxIs6kZf-oQujTFUA,13187
 sourcecode/detectors/parsers.py,sha256=ugPg8yNUf0Ai1gA7Fnn6wAkYGFjTxRodSP3IeViYJJ4,2290
@@ -66,7 +67,7 @@ sourcecode/detectors/terraform.py,sha256=cxORPR_zVLOJpHlh4e9JnFpkQsn_UnqMMom5yG6
 sourcecode/detectors/tooling.py,sha256=8CKbtxwQoABP-WyBRNmdAmHDOvAH57AR1cF4UKuWEdQ,2074
 sourcecode/mcp/__init__.py,sha256=XU4HfRGbdid8wdUA0x_4f7uKZD1z3mv_XUY_WU_T9Mw,179
 sourcecode/mcp/runner.py,sha256=YSw2DXEICau6mCBr3Gfia3D_tKxMbRvIIXEh4cHC1SY,1390
-sourcecode/mcp/server.py,sha256=zbc4G5j2VAlVnluAksoSk6zV_wwgDevaH5Q4oVsl2kA,24018
+sourcecode/mcp/server.py,sha256=jUCDNjJKUIhse52M8-Lp63CDxlm21xOYUOVlv3z6_uk,25556
 sourcecode/mcp/onboarding/__init__.py,sha256=sj2PWqEBmMc4zBNkomg89WtL0M6S7A9yb7_wAuSWNP4,66
 sourcecode/mcp/onboarding/applier.py,sha256=B9CneieWTpaDSDIyW3S5nrlRlBpvfqUcgi93-mm_ApQ,2135
 sourcecode/mcp/onboarding/backup.py,sha256=ihqGOR8QTX8HASRSEDyfFyXr5bkXrygPHamv4p9KTmk,1452
@@ -78,8 +79,8 @@ sourcecode/telemetry/consent.py,sha256=wLMvGNJeSSyZoNkQXpoUioY6mMv4Qdvuw7S9jAEWn
 sourcecode/telemetry/events.py,sha256=oEvvulfsv5GIDWG2174gSS6tNB95w38AIYiYeifGKlE,2294
 sourcecode/telemetry/filters.py,sha256=Asa71oRl7q3Wt_FMwuufIZJFzSYdgRNKS8LHCIyFeYE,4805
 sourcecode/telemetry/transport.py,sha256=KJeIPCPWMdmbCP3ySGs2iUlia34U6vWne2dZsUezesw,1560
-sourcecode-1.32.0.dist-info/METADATA,sha256=FHXYjifmhRxjdL38qMdsKuEz9_H2nJIrodG2c7R2LHM,31100
-sourcecode-1.32.0.dist-info/WHEEL,sha256=QccIxa26bgl1E6uMy58deGWi-0aeIkkangHcxk2kWfw,87
-sourcecode-1.32.0.dist-info/entry_points.txt,sha256=ex3F9rmbXeyDIoFQHtkEqTsKSaJow8F0LrVu8XfIktQ,57
-sourcecode-1.32.0.dist-info/licenses/LICENSE,sha256=7DdHrU9Z_3e7dSvq4ISijZNjnuHo5NIHNiHDouMQ9JU,10491
-sourcecode-1.32.0.dist-info/RECORD,,
+sourcecode-1.32.2.dist-info/METADATA,sha256=PwMFojaNy2nH80htoS7Q_-p-AQ2GSwiTr49Mv8EGXJE,31100
+sourcecode-1.32.2.dist-info/WHEEL,sha256=QccIxa26bgl1E6uMy58deGWi-0aeIkkangHcxk2kWfw,87
+sourcecode-1.32.2.dist-info/entry_points.txt,sha256=ex3F9rmbXeyDIoFQHtkEqTsKSaJow8F0LrVu8XfIktQ,57
+sourcecode-1.32.2.dist-info/licenses/LICENSE,sha256=7DdHrU9Z_3e7dSvq4ISijZNjnuHo5NIHNiHDouMQ9JU,10491
+sourcecode-1.32.2.dist-info/RECORD,,