feat(extract): add opt-in --cargo crate dependency extractor

README.md

@@ -574,6 +574,7 @@ graphify extract ./docs --backend claude-cli   # route through Claude Code CLI -
 graphify extract ./docs --backend azure        # Azure OpenAI (set AZURE_OPENAI_API_KEY + AZURE_OPENAI_ENDPOINT)
 graphify extract ./docs --max-workers 16       # AST parallelism (also GRAPHIFY_MAX_WORKERS)
 graphify extract --postgres "postgresql://user:pass@host/db"   # introspect live PostgreSQL schema directly
+graphify extract ./my-workspace --cargo        # introspect Rust Cargo workspace dependencies directly
 graphify extract ./docs --token-budget 30000   # smaller semantic chunks for local/small models
 graphify extract ./docs --max-concurrency 2    # fewer parallel LLM calls (useful for local inference)
 graphify extract ./docs --api-timeout 900      # longer HTTP timeout for slow local models (default 600s)

graphify/__main__.py

@@ -2182,6 +2182,7 @@ def main() -> None:
         print("    --postgres DSN          extract schema from a live PostgreSQL database")
         print("                            maps tables, views, functions + FK relationships;")
         print("                            column-level detail is not represented in the graph")
+        print("    --cargo                 extract crate→crate deps from Cargo.toml")
         print("    --global                also merge the resulting graph into the global graph")
         print("    --as <tag>              repo tag for --global (default: target directory name)")
         print("  global add <graph.json>  add/update a project graph in the global graph (~/.graphify/global-graph.json)")
@@ -3904,7 +3905,7 @@ def _load_graph(p: str):
                 "Usage: graphify extract <path> [--backend gemini|kimi|claude|openai|deepseek|ollama] "
                 "[--model M] [--mode deep] [--out DIR] [--google-workspace] [--no-cluster] "
                 "[--max-workers N] [--token-budget N] [--max-concurrency N] "
-                "[--api-timeout S] [--postgres DSN]",
+                "[--api-timeout S] [--postgres DSN] [--cargo]",
                 file=sys.stderr,
             )
             sys.exit(1)
@@ -3924,6 +3925,7 @@ def _load_graph(p: str):
         extract_mode: str | None = None
         out_dir: Path | None = None
         cli_postgres_dsn: str | None = None
+        cli_cargo: bool = False
         no_cluster = False
         dedup_llm = False
         google_workspace = False
@@ -4023,6 +4025,9 @@ def _parse_float(name: str, raw: str) -> float:
                 cli_postgres_dsn = args[i + 1]; i += 2
             elif a.startswith("--postgres="):
                 cli_postgres_dsn = a.split("=", 1)[1]; i += 1
+            elif a == "--cargo":
+                cli_cargo = True
+                i += 1
             else:
                 i += 1
 
@@ -4322,13 +4327,25 @@ def _progress(idx: int, total: int, _result: dict) -> None:
             print(f"[graphify extract] PostgreSQL: {len(pg_result['nodes'])} nodes, "
                   f"{len(pg_result['edges'])} edges")
 
-        # Merge AST + semantic + pg_result. Order matters for deduplication: passing AST
+        cargo_result: dict = {"nodes": [], "edges": []}
+        if cli_cargo:
+            from graphify.cargo_introspect import introspect_cargo
+            print("[graphify extract] introspecting Cargo workspace...")
+            try:
+                cargo_result = introspect_cargo(target)
+            except (ConnectionError, ImportError) as exc:
+                print(f"error: {exc}", file=sys.stderr)
+                sys.exit(1)
+            print(f"[graphify extract] Cargo: {len(cargo_result['nodes'])} nodes, "
+                  f"{len(cargo_result['edges'])} edges")
+
+        # Merge AST + semantic + pg_result + cargo_result. Order matters for deduplication: passing AST
         # first means semantic node attributes win on collision (richer labels
         # for symbols also referenced in docs). Hyperedges only come from the
         # semantic side.
         merged: dict = {
-            "nodes": list(ast_result.get("nodes", [])) + list(sem_result.get("nodes", [])) + list(pg_result.get("nodes", [])),
-            "edges": list(ast_result.get("edges", [])) + list(sem_result.get("edges", [])) + list(pg_result.get("edges", [])),
+            "nodes": list(ast_result.get("nodes", [])) + list(sem_result.get("nodes", [])) + list(pg_result.get("nodes", [])) + list(cargo_result.get("nodes", [])),
+            "edges": list(ast_result.get("edges", [])) + list(sem_result.get("edges", [])) + list(pg_result.get("edges", [])) + list(cargo_result.get("edges", [])),
             "hyperedges": list(sem_result.get("hyperedges", [])),
             "input_tokens": ast_result.get("input_tokens", 0) + sem_result.get("input_tokens", 0),
             "output_tokens": ast_result.get("output_tokens", 0) + sem_result.get("output_tokens", 0),

graphify/cargo_introspect.py

@@ -0,0 +1,98 @@
+"""Cargo manifest introspection for workspace-internal crate dependencies."""
+
+from __future__ import annotations
+
+from pathlib import Path
+from typing import Any
+
+
+_CONFIDENCE_EXTRACTED = "EXTRACTED"
+
+
+def _load_toml(path: Path) -> dict[str, Any]:
+    try:
+        import tomllib  # type: ignore[import-not-found]
+    except ModuleNotFoundError:
+        try:
+            import tomli as tomllib  # type: ignore[import-not-found,no-redef]
+        except ModuleNotFoundError:
+            raise ImportError(
+                "--cargo on Python 3.10 needs tomli. Install with: pip install tomli"
+            ) from None
+
+    with path.open("rb") as manifest:
+        return tomllib.load(manifest)
+
+
+def _member_manifest_paths(root: Path, root_data: dict[str, Any]) -> list[Path]:
+    paths: list[Path] = []
+    if isinstance(root_data.get("package"), dict):
+        paths.append(root / "Cargo.toml")
+
+    workspace = root_data.get("workspace")
+    members = workspace.get("members", []) if isinstance(workspace, dict) else []
+    if not isinstance(members, list):
+        return paths
+
+    for pattern in members:
+        if not isinstance(pattern, str):
+            continue
+        for member in sorted(root.glob(pattern)):
+            manifest = member / "Cargo.toml"
+            if manifest.is_file() and manifest not in paths:
+                paths.append(manifest)
+    return paths
+
+
+def introspect_cargo(root: str | Path) -> dict[str, Any]:
+    """Return crate nodes and internal dependency edges from Cargo manifests."""
+    root_path = Path(root).resolve()
+    root_manifest = root_path / "Cargo.toml"
+    root_data = _load_toml(root_manifest)
+
+    manifests = _member_manifest_paths(root_path, root_data)
+    crates: dict[str, tuple[str, Path, dict[str, Any]]] = {}
+
+    for manifest in manifests:
+        data = root_data if manifest == root_manifest else _load_toml(manifest)
+        package = data.get("package")
+        if not isinstance(package, dict):
+            continue
+        name = package.get("name")
+        if isinstance(name, str):
+            crates[name] = (f"crate:{name}", manifest, data)
+
+    nodes = [
+        {
+            "id": crate_id,
+            "label": name,
+            "source_file": manifest.relative_to(root_path).as_posix(),
+            "source_location": "L1",
+        }
+        for name, (crate_id, manifest, _data) in sorted(crates.items())
+    ]
+
+    edges: list[dict[str, Any]] = []
+    for source_name, (source_id, manifest, data) in sorted(crates.items()):
+        dependencies = data.get("dependencies", {})
+        if not isinstance(dependencies, dict):
+            continue
+        source_file = manifest.relative_to(root_path).as_posix()
+        for dependency_name in sorted(dependencies):
+            target = crates.get(dependency_name)
+            if target is None:
+                continue
+            edges.append(
+                {
+                    "source": source_id,
+                    "target": target[0],
+                    "relation": "crate_depends_on",
+                    "context": "cargo_dependency",
+                    "weight": 1.0,
+                    "confidence": _CONFIDENCE_EXTRACTED,
+                    "source_file": source_file,
+                    "source_location": "L1",
+                }
+            )
+
+    return {"nodes": nodes, "edges": edges}

graphify/watch.py

@@ -838,7 +838,7 @@ def on_any_event(self, event):
             nonlocal last_trigger, pending
             if event.is_directory:
                 return
-            path = Path(event.src_path)
+            path = Path(os.fsdecode(event.src_path))
             # Check .graphifyignore BEFORE the extension/dotfile/out filters so
             # the cheapest short-circuit for users with broad ignore patterns
             # (node_modules/, .venv/, build/, …) fires first. _is_ignored
@@ -848,9 +848,13 @@ def on_any_event(self, event):
                 return
             if path.suffix.lower() not in _WATCHED_EXTENSIONS:
                 return
-            if any(part.startswith(".") for part in path.parts):
+            try:
+                filter_parts = path.relative_to(watch_root_for_ignore).parts
+            except ValueError:
+                filter_parts = path.parts
+            if any(part.startswith(".") for part in filter_parts):
                 return
-            if _GRAPHIFY_OUT in path.parts:
+            if _GRAPHIFY_OUT in filter_parts:
                 return
             last_trigger = time.monotonic()
             pending = True