ltcai 3.6.0 → 4.0.1

package/latticeai/brain/discovery.py

@@ -0,0 +1,1455 @@
+from __future__ import annotations
+
+# ruff: noqa: F403,F405
+
+from ._kg_common import *  # noqa: F403,F401
+
+
+class KnowledgeGraphDiscoveryMixin:
+    def discover_local_roots(self) -> Dict[str, Any]:
+        """Return safe, cross-platform starting points for structure browsing."""
+        os_type = _current_os_type()
+        home = Path.home().expanduser()
+        roots: List[Dict[str, Any]] = []
+        seen: set = set()
+
+        def add(
+            label: str,
+            path: Path,
+            kind: str,
+            *,
+            recommended: bool = True,
+            warning: Optional[str] = None,
+        ) -> None:
+            try:
+                resolved = path.expanduser().resolve()
+            except OSError:
+                resolved = path.expanduser()
+            key = str(resolved)
+            if key in seen or not resolved.exists():
+                return
+            seen.add(key)
+            roots.append(
+                {
+                    "id": f"{kind}:{_path_fingerprint(resolved)}",
+                    "label": label,
+                    "path": key,
+                    "kind": kind,
+                    "recommended": recommended,
+                    "warning": warning or _root_warning(resolved, os_type),
+                }
+            )
+
+        add("홈", home, "home", warning=_root_warning(home, os_type))
+        for name, label in (
+            ("Documents", "문서"),
+            ("Desktop", "데스크탑"),
+            ("Downloads", "다운로드"),
+            ("Pictures", "사진"),
+            ("Projects", "프로젝트"),
+        ):
+            add(label, home / name, name.lower())
+
+        if os_type == "macos":
+            volumes = Path("/Volumes")
+            if volumes.exists():
+                try:
+                    for volume in sorted(
+                        volumes.iterdir(), key=lambda p: p.name.lower()
+                    ):
+                        add(volume.name, volume, "volume", recommended=False)
+                except OSError:
+                    pass
+        elif os_type == "windows":
+            for letter in "ABCDEFGHIJKLMNOPQRSTUVWXYZ":
+                drive = Path(f"{letter}:\\")
+                if drive.exists():
+                    add(
+                        f"{letter}: 드라이브",
+                        drive,
+                        "drive",
+                        recommended=(letter != "C"),
+                    )
+            for env_name, label in (
+                ("OneDrive", "OneDrive"),
+                ("OneDriveCommercial", "OneDrive"),
+            ):
+                raw = os.environ.get(env_name)
+                if raw:
+                    add(label, Path(raw), "cloud", recommended=False)
+        elif os_type == "linux":
+            for base in (Path("/mnt"), Path("/media")):
+                add(str(base), base, "mounts", recommended=False)
+                try:
+                    if base.exists():
+                        for mounted in sorted(
+                            base.iterdir(), key=lambda p: p.name.lower()
+                        ):
+                            add(mounted.name, mounted, "volume", recommended=False)
+                except OSError:
+                    pass
+
+        return {
+            "os_type": os_type,
+            "computer": platform.node() or "local",
+            "roots": roots,
+            "privacy_notice": "처음에는 드라이브와 폴더 구조만 확인하며, 파일 내용은 사용자가 동의한 뒤에만 읽습니다.",
+        }
+
+    def preview_local_tree(self, path: Path, *, max_items: int = 200) -> Dict[str, Any]:
+        """List one folder level using metadata only; file contents are not read."""
+        root = Path(path).expanduser().resolve()
+        if not root.exists():
+            raise ValueError(f"경로가 존재하지 않습니다: {path}")
+        if not root.is_dir():
+            raise ValueError(f"폴더가 아닙니다: {path}")
+
+        os_type = _current_os_type()
+        max_items = max(1, min(int(max_items or 200), 1000))
+        items: List[Dict[str, Any]] = []
+        inaccessible = 0
+        try:
+            children = sorted(
+                root.iterdir(), key=lambda p: (not p.is_dir(), p.name.lower())
+            )
+        except PermissionError as exc:
+            return {
+                "path": str(root),
+                "items": [],
+                "error": f"접근 권한 없음: {exc}",
+                "privacy_notice": "현재 단계에서는 파일 내용을 읽지 않고, 폴더와 파일의 이름/크기/수정일만 확인합니다.",
+            }
+
+        for child in children[:max_items]:
+            try:
+                is_dir = child.is_dir()
+                stat = child.stat()
+                reason = (
+                    _excluded_directory_reason(child, root=root, os_type=os_type)
+                    if is_dir
+                    else _sensitive_file_reason(child, root=root)
+                )
+                items.append(
+                    {
+                        "name": child.name,
+                        "path": str(child),
+                        "type": "directory" if is_dir else "file",
+                        "extension": "" if is_dir else child.suffix.lower(),
+                        "size_bytes": None if is_dir else stat.st_size,
+                        "modified_at": _safe_iso_from_stat_mtime(stat.st_mtime),
+                        "hidden": _is_hidden_path(child, root),
+                        "accessible": True,
+                        "excluded_reason": reason,
+                    }
+                )
+            except PermissionError:
+                inaccessible += 1
+                items.append(
+                    {
+                        "name": child.name,
+                        "path": str(child),
+                        "type": "unknown",
+                        "accessible": False,
+                        "excluded_reason": "permission_denied",
+                    }
+                )
+            except OSError as exc:
+                inaccessible += 1
+                items.append(
+                    {
+                        "name": child.name,
+                        "path": str(child),
+                        "type": "unknown",
+                        "accessible": False,
+                        "excluded_reason": str(exc),
+                    }
+                )
+
+        return {
+            "path": str(root),
+            "os_type": os_type,
+            "items": items,
+            "truncated": len(children) > max_items,
+            "inaccessible": inaccessible,
+            "warning": _root_warning(root, os_type),
+            "privacy_notice": "현재 단계에서는 파일 내용을 읽지 않고, 폴더와 파일의 이름/크기/수정일만 확인합니다.",
+        }
+
+    def _iter_local_scan_entries(
+        self, root: Path, *, max_files: int
+    ) -> Iterable[Dict[str, Any]]:
+        os_type = _current_os_type()
+        stack = [root]
+        files_seen = 0
+        while stack:
+            current = stack.pop()
+            try:
+                children = sorted(
+                    current.iterdir(), key=lambda p: (not p.is_dir(), p.name.lower())
+                )
+            except PermissionError as exc:
+                yield {
+                    "kind": "inaccessible_dir",
+                    "path": current,
+                    "reason": f"permission_denied: {exc}",
+                }
+                continue
+            except OSError as exc:
+                yield {"kind": "inaccessible_dir", "path": current, "reason": str(exc)}
+                continue
+
+            for child in children:
+                if child.is_symlink():
+                    yield {"kind": "excluded", "path": child, "reason": "symlink"}
+                    continue
+                try:
+                    if child.is_dir():
+                        reason = _excluded_directory_reason(
+                            child, root=root, os_type=os_type
+                        )
+                        if reason:
+                            yield {
+                                "kind": "excluded_dir",
+                                "path": child,
+                                "reason": reason,
+                            }
+                        else:
+                            stack.append(child)
+                        continue
+                    if not child.is_file():
+                        yield {
+                            "kind": "excluded",
+                            "path": child,
+                            "reason": "not_regular_file",
+                        }
+                        continue
+                    stat = child.stat()
+                except PermissionError as exc:
+                    yield {
+                        "kind": "inaccessible_file",
+                        "path": child,
+                        "reason": f"permission_denied: {exc}",
+                    }
+                    continue
+                except OSError as exc:
+                    yield {
+                        "kind": "inaccessible_file",
+                        "path": child,
+                        "reason": str(exc),
+                    }
+                    continue
+
+                files_seen += 1
+                if files_seen > max_files:
+                    yield {
+                        "kind": "limit_reached",
+                        "path": child,
+                        "reason": "max_files",
+                    }
+                    return
+                yield {"kind": "file", "path": child, "stat": stat}
+
+    def _local_file_decision(
+        self, path: Path, root: Path, stat: os.stat_result
+    ) -> Dict[str, Any]:
+        ext = path.suffix.lower()
+        category = _file_category(ext)
+        parser_type = _parser_type_for_category(category, ext)
+        sensitive_reason = _sensitive_file_reason(path, root=root)
+        if sensitive_reason:
+            return {
+                "status": "sensitive_blocked",
+                "reason": sensitive_reason,
+                "category": category,
+                "parser_type": parser_type,
+                "indexable": False,
+            }
+        if category == "unsupported":
+            return {
+                "status": "unsupported",
+                "reason": "unsupported_extension",
+                "category": category,
+                "parser_type": parser_type,
+                "indexable": False,
+            }
+        limit = _size_limit_for_category(category)
+        if stat.st_size > limit:
+            return {
+                "status": "too_large",
+                "reason": f"size>{limit}",
+                "category": category,
+                "parser_type": parser_type,
+                "indexable": False,
+            }
+        return {
+            "status": "pending",
+            "reason": "",
+            "category": category,
+            "parser_type": parser_type,
+            "indexable": True,
+        }
+
+    def audit_local_folder(
+        self, path: Path, *, include_ocr: bool = False, max_files: int = 50_000
+    ) -> Dict[str, Any]:
+        """Safety-check a folder using metadata only; file bodies are not read."""
+        root = Path(path).expanduser().resolve()
+        if not root.exists():
+            raise ValueError(f"경로가 존재하지 않습니다: {path}")
+        if not root.is_dir():
+            raise ValueError(f"폴더가 아닙니다: {path}")
+
+        os_type = _current_os_type()
+        max_files = max(1, min(int(max_files or 50_000), 200_000))
+        status_counts: Counter = Counter()
+        category_counts: Counter = Counter()
+        extension_counts: Counter = Counter()
+        allowed_samples: List[Dict[str, Any]] = []
+        excluded_samples: List[Dict[str, Any]] = []
+        total_files = 0
+        readable_files = 0
+        inaccessible = 0
+        excluded_dirs = 0
+        limit_reached = False
+
+        for entry in self._iter_local_scan_entries(root, max_files=max_files):
+            kind = entry["kind"]
+            path_obj = entry["path"]
+            if kind == "limit_reached":
+                limit_reached = True
+                break
+            if kind == "excluded_dir":
+                excluded_dirs += 1
+                if len(excluded_samples) < 25:
+                    excluded_samples.append(
+                        _sample_file(
+                            path_obj, root, "excluded", entry.get("reason", "")
+                        )
+                    )
+                continue
+            if kind in {"inaccessible_dir", "inaccessible_file"}:
+                inaccessible += 1
+                status_counts["failed"] += 1
+                if len(excluded_samples) < 25:
+                    excluded_samples.append(
+                        _sample_file(path_obj, root, "failed", entry.get("reason", ""))
+                    )
+                continue
+            if kind == "excluded":
+                status_counts["excluded"] += 1
+                if len(excluded_samples) < 25:
+                    excluded_samples.append(
+                        _sample_file(
+                            path_obj, root, "excluded", entry.get("reason", "")
+                        )
+                    )
+                continue
+            if kind != "file":
+                continue
+
+            total_files += 1
+            stat = entry["stat"]
+            decision = self._local_file_decision(path_obj, root, stat)
+            status = decision["status"]
+            category = decision["category"]
+            ext = path_obj.suffix.lower() or "(none)"
+            category_counts[category] += 1
+            extension_counts[ext] += 1
+            if decision["indexable"]:
+                readable_files += 1
+                status_counts["readable"] += 1
+                if len(allowed_samples) < 25:
+                    allowed_samples.append(_sample_file(path_obj, root, "readable"))
+            else:
+                status_counts[status] += 1
+                if len(excluded_samples) < 25:
+                    excluded_samples.append(
+                        _sample_file(path_obj, root, status, decision["reason"])
+                    )
+
+        doc_weight = (
+            category_counts["pdf"] * 1.4
+            + category_counts["document"] * 0.9
+            + category_counts["slide_deck"] * 1.0
+        )
+        sheet_weight = category_counts["spreadsheet"] * 0.6
+        ocr_weight = category_counts["image"] * (1.8 if include_ocr else 0.1)
+        estimated_seconds = round(
+            readable_files * 0.04 + doc_weight + sheet_weight + ocr_weight, 1
+        )
+
+        return {
+            "path": str(root),
+            "source_id": f"source:{_path_fingerprint(root)}",
+            "os_type": os_type,
+            "drive_id": _drive_id_for_path(root),
+            "warning": _root_warning(root, os_type),
+            "privacy_notice": "현재 단계에서는 파일 내용을 읽지 않고, 폴더와 파일의 이름/크기/수정일만 확인합니다.",
+            "include_ocr_requested": bool(include_ocr),
+            "summary": {
+                "total_files": total_files,
+                "readable_files": readable_files,
+                "excluded_files": int(
+                    status_counts["excluded"]
+                    + status_counts["sensitive_blocked"]
+                    + status_counts["too_large"]
+                    + status_counts["unsupported"]
+                ),
+                "sensitive_files": int(status_counts["sensitive_blocked"]),
+                "too_large_files": int(status_counts["too_large"]),
+                "unsupported_files": int(status_counts["unsupported"]),
+                "image_ocr_candidates": int(category_counts["image"]),
+                "inaccessible_items": inaccessible,
+                "excluded_dirs": excluded_dirs,
+                "estimated_seconds": estimated_seconds,
+                "storage_root": str(self.db_path.parent),
+                "limit_reached": limit_reached,
+            },
+            "by_status": dict(status_counts),
+            "by_category": dict(category_counts),
+            "by_extension": dict(extension_counts.most_common(40)),
+            "allowed_samples": allowed_samples,
+            "excluded_samples": excluded_samples,
+            "consent_required": {
+                "knowledge_source": True,
+                "image_ocr": bool(category_counts["image"]),
+                "watch": True,
+                "sensitive_files_default_excluded": True,
+            },
+        }
+
+    def local_sources(self) -> Dict[str, Any]:
+        with self._connect() as conn:
+            sources = [
+                {
+                    "id": row["id"],
+                    "root_path": row["root_path"],
+                    "os_type": row["os_type"],
+                    "drive_id": row["drive_id"],
+                    "label": row["label"],
+                    "status": row["status"],
+                    "include_ocr": bool(row["include_ocr"]),
+                    "watch_enabled": bool(row["watch_enabled"]),
+                    "consent": _safe_loads(row["consent_json"]),
+                    "created_at": row["created_at"],
+                    "updated_at": row["updated_at"],
+                    "last_scanned_at": row["last_scanned_at"],
+                }
+                for row in conn.execute(
+                    """
+                        SELECT id, root_path, os_type, drive_id, label, status, include_ocr,
+                               watch_enabled, consent_json, created_at, updated_at, last_scanned_at
+                        FROM knowledge_sources
+                        ORDER BY updated_at DESC, id ASC
+                        """
+                )
+            ]
+            status_rows = conn.execute(
+                "SELECT source_id, status, COUNT(*) AS count FROM local_file_index GROUP BY source_id, status"
+            ).fetchall()
+        counts: Dict[str, Dict[str, int]] = {}
+        for row in status_rows:
+            counts.setdefault(row["source_id"], {})[row["status"]] = row["count"]
+        for source in sources:
+            source["file_status"] = counts.get(source["id"], {})
+        return {"sources": sources}
+
+    def set_local_source_watch(self, source_id: str, enabled: bool) -> Dict[str, Any]:
+        source_id = str(source_id or "").strip()
+        if not source_id:
+            raise ValueError("source_id required")
+        with self._connect() as conn:
+            row = conn.execute(
+                "SELECT id FROM knowledge_sources WHERE id=?",
+                (source_id,),
+            ).fetchone()
+            if not row:
+                raise ValueError(f"knowledge source not found: {source_id}")
+            conn.execute(
+                "UPDATE knowledge_sources SET watch_enabled=?, updated_at=? WHERE id=?",
+                (1 if enabled else 0, _now(), source_id),
+            )
+        return {"source_id": source_id, "watch_enabled": bool(enabled)}
+
+    def remove_local_source(self, source_id: str) -> Dict[str, Any]:
+        """Remove one approved local source and its derived graph projection.
+
+        This is intentionally non-destructive for user files: only the LatticeAI
+        index rows, graph nodes, edges, and chunks derived from the source are
+        removed. The original folder and files are never touched.
+        """
+        source_id = str(source_id or "").strip()
+        if not source_id:
+            raise ValueError("source_id required")
+        with self._connect() as conn:
+            source = conn.execute(
+                "SELECT id, root_path FROM knowledge_sources WHERE id=?",
+                (source_id,),
+            ).fetchone()
+            if not source:
+                raise ValueError(f"knowledge source not found: {source_id}")
+            rows = conn.execute(
+                "SELECT graph_node_id FROM local_file_index WHERE source_id=? AND graph_node_id IS NOT NULL",
+                (source_id,),
+            ).fetchall()
+            graph_node_ids = [
+                row["graph_node_id"] for row in rows if row["graph_node_id"]
+            ]
+            for graph_node_id in graph_node_ids:
+                self._delete_local_file_graph(conn, graph_node_id)
+            conn.execute("DELETE FROM local_file_index WHERE source_id=?", (source_id,))
+            conn.execute("DELETE FROM knowledge_sources WHERE id=?", (source_id,))
+            self._cleanup_local_graph_orphans(conn, source_id)
+        return {
+            "source_id": source_id,
+            "root_path": source["root_path"],
+            "removed_graph_nodes": len(graph_node_ids),
+        }
+
+    def _extract_local_file_text(
+        self, path: Path, category: str, *, include_ocr: bool
+    ) -> Tuple[str, Dict[str, Any]]:
+        ext = path.suffix.lower()
+        meta: Dict[str, Any] = {"parser": _parser_type_for_category(category, ext)}
+        text = ""
+        if category in {"text", "code"} or ext == ".csv":
+            text = path.read_text(encoding="utf-8", errors="replace")
+        elif ext == ".pdf":
+            import pdfplumber
+
+            with pdfplumber.open(str(path)) as pdf:
+                meta["pages"] = len(pdf.pages)
+                text = "\n\n".join((page.extract_text() or "") for page in pdf.pages)
+        elif ext == ".docx":
+            from docx import Document
+
+            doc = Document(str(path))
+            paragraphs = [p.text for p in doc.paragraphs if p.text.strip()]
+            table_lines = []
+            for table in doc.tables:
+                for row in table.rows:
+                    cells = [_clean_text(cell.text) for cell in row.cells]
+                    if any(cells):
+                        table_lines.append("\t".join(cells))
+            meta["paragraphs"] = len(paragraphs)
+            meta["tables"] = len(doc.tables)
+            meta["table_rows"] = len(table_lines)
+            text = "\n\n".join([*paragraphs, *table_lines])
+        elif ext == ".xlsx":
+            from openpyxl import load_workbook
+
+            wb = load_workbook(str(path), read_only=True, data_only=True)
+            rows_all = []
+            non_empty_rows = 0
+            non_empty_cells = 0
+            char_count = 0
+            for ws in wb.worksheets:
+                sheet_rows = []
+                for row in ws.iter_rows(values_only=True):
+                    cells = [
+                        str(cell).strip() if cell is not None else "" for cell in row
+                    ]
+                    if not any(cells):
+                        continue
+                    line = "\t".join(cells)
+                    non_empty_rows += 1
+                    non_empty_cells += sum(1 for cell in cells if cell)
+                    sheet_rows.append(line)
+                    char_count += len(line) + 1
+                    if char_count > 200_000:
+                        break
+                if sheet_rows:
+                    rows_all.append(f"[Sheet: {ws.title}]")
+                    rows_all.extend(sheet_rows)
+                if char_count > 200_000:
+                    break
+            meta["sheets"] = len(wb.worksheets)
+            meta["rows"] = non_empty_rows
+            meta["cells"] = non_empty_cells
+            text = "\n".join(rows_all)
+        elif ext == ".pptx":
+            from pptx import Presentation
+
+            prs = Presentation(str(path))
+            slides_text = []
+            for index, slide in enumerate(prs.slides, 1):
+                parts = []
+                for shape in slide.shapes:
+                    if getattr(shape, "has_text_frame", False):
+                        slide_text = shape.text_frame.text.strip()
+                        if slide_text:
+                            parts.append(slide_text)
+                if parts:
+                    slides_text.append(f"[Slide {index}]\n" + "\n".join(parts))
+            meta["slides"] = len(prs.slides)
+            meta["text_slides"] = len(slides_text)
+            text = "\n\n".join(slides_text)
+        elif category == "image":
+            from PIL import Image
+
+            with Image.open(str(path)) as image:
+                meta.update(
+                    {
+                        "width": image.width,
+                        "height": image.height,
+                        "format": image.format,
+                        "mode": image.mode,
+                        "ocr_enabled": bool(include_ocr),
+                    }
+                )
+                if include_ocr:
+                    try:
+                        import pytesseract
+
+                        text = pytesseract.image_to_string(image)
+                        meta["ocr_chars"] = len(text)
+                    except (
+                        Exception
+                    ) as exc:  # pragma: no cover - depends on local OCR runtime
+                        meta["ocr_error"] = str(exc)
+                        text = ""
+        return text[:200_000], meta
+
+    def _ensure_local_hierarchy(
+        self,
+        conn: sqlite3.Connection,
+        *,
+        source_id: str,
+        root: Path,
+        file_path: Path,
+        os_type: str,
+        drive_id: str,
+    ) -> str:
+        computer_label = platform.node() or "내 컴퓨터"
+        computer_id = f"computer:{_slug(computer_label)}"
+        drive_node_id = f"drive:{_sha256_text(f'{os_type}:{drive_id}')[:24]}"
+        root_folder_id = f"folder:{_sha256_text(f'{source_id}:root')[:24]}"
+        self._upsert_node(
+            conn, computer_id, "Computer", computer_label, metadata={"os_type": os_type}
+        )
+        self._upsert_node(
+            conn,
+            drive_node_id,
+            "Drive",
+            drive_id,
+            metadata={"os_type": os_type, "drive_id": drive_id},
+        )
+        self._upsert_edge(
+            conn,
+            computer_id,
+            drive_node_id,
+            "포함함",
+            metadata={"source": "local_scan"},
+        )
+        self._upsert_node(
+            conn,
+            root_folder_id,
+            "Folder",
+            root.name or str(root),
+            summary=str(root),
+            metadata={"source_id": source_id, "path": str(root), "root": True},
+        )
+        self._upsert_edge(
+            conn,
+            drive_node_id,
+            root_folder_id,
+            "포함함",
+            metadata={"source": "local_scan"},
+        )
+
+        try:
+            relative_parent = file_path.parent.relative_to(root)
+        except ValueError:
+            relative_parent = Path()
+        parent_id = root_folder_id
+        current_path = root
+        for part in relative_parent.parts:
+            current_path = current_path / part
+            folder_id = (
+                f"folder:{_sha256_text(f'{source_id}:{current_path.as_posix()}')[:24]}"
+            )
+            self._upsert_node(
+                conn,
+                folder_id,
+                "Folder",
+                part,
+                summary=str(current_path),
+                metadata={
+                    "source_id": source_id,
+                    "path": str(current_path),
+                    "root": False,
+                },
+            )
+            self._upsert_edge(
+                conn, parent_id, folder_id, "포함함", metadata={"source": "local_scan"}
+            )
+            parent_id = folder_id
+        return parent_id
+
+    def _upsert_local_file_index(
+        self,
+        conn: sqlite3.Connection,
+        *,
+        source_id: str,
+        root: Path,
+        file_path: Path,
+        stat: Optional[os.stat_result],
+        os_type: str,
+        drive_id: str,
+        status: str,
+        parser_type: str,
+        sha256: Optional[str] = None,
+        graph_node_id: Optional[str] = None,
+        error_message: Optional[str] = None,
+        metadata: Optional[Dict[str, Any]] = None,
+    ) -> str:
+        try:
+            relative_path = file_path.relative_to(root).as_posix()
+        except ValueError:
+            relative_path = file_path.name
+        index_id = f"local-index:{_sha256_text(f'{source_id}:{relative_path}')[:24]}"
+        now = _now()
+        size = stat.st_size if stat else None
+        modified_at = _safe_iso_from_stat_mtime(stat.st_mtime) if stat else ""
+        conn.execute(
+            """
+                INSERT INTO local_file_index(
+                  id, source_id, os_type, drive_id, root_path, file_path, relative_path,
+                  file_name, extension, size_bytes, modified_at, sha256, last_scanned_at,
+                  last_indexed_at, parser_type, status, error_message, graph_node_id,
+                  deleted, metadata_json
+                )
+                VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
+                ON CONFLICT(source_id, relative_path) DO UPDATE SET
+                  os_type=excluded.os_type,
+                  drive_id=excluded.drive_id,
+                  root_path=excluded.root_path,
+                  file_path=excluded.file_path,
+                  file_name=excluded.file_name,
+                  extension=excluded.extension,
+                  size_bytes=excluded.size_bytes,
+                  modified_at=excluded.modified_at,
+                  sha256=excluded.sha256,
+                  last_scanned_at=excluded.last_scanned_at,
+                  last_indexed_at=excluded.last_indexed_at,
+                  parser_type=excluded.parser_type,
+                  status=excluded.status,
+                  error_message=excluded.error_message,
+                  graph_node_id=excluded.graph_node_id,
+                  deleted=excluded.deleted,
+                  metadata_json=excluded.metadata_json
+                """,
+            (
+                index_id,
+                source_id,
+                os_type,
+                drive_id,
+                str(root),
+                str(file_path),
+                relative_path,
+                file_path.name,
+                file_path.suffix.lower(),
+                size,
+                modified_at,
+                sha256,
+                now,
+                now if status == "indexed" else None,
+                parser_type,
+                status,
+                error_message,
+                graph_node_id,
+                0 if status != "deleted" else 1,
+                _json(metadata),
+            ),
+        )
+        return index_id
+
+    def _upsert_local_file_node(
+        self,
+        conn: sqlite3.Connection,
+        *,
+        source_id: str,
+        root: Path,
+        file_path: Path,
+        stat: os.stat_result,
+        os_type: str,
+        drive_id: str,
+        sha256: str,
+        category: str,
+        parser_type: str,
+        text: str,
+        parser_meta: Dict[str, Any],
+    ) -> str:
+        text = _clean_text(text)
+        if not text:
+            raise ValueError("텍스트 추출 결과가 비어 있습니다.")
+        try:
+            relative_path = file_path.relative_to(root).as_posix()
+        except ValueError:
+            relative_path = file_path.name
+        file_node_id = f"local-file:{_sha256_text(f'{source_id}:{relative_path}')[:24]}"
+        parent_folder_id = self._ensure_local_hierarchy(
+            conn,
+            source_id=source_id,
+            root=root,
+            file_path=file_path,
+            os_type=os_type,
+            drive_id=drive_id,
+        )
+        child_rows = conn.execute(
+            """
+                SELECT e.to_node AS id
+                FROM edges e
+                JOIN nodes n ON n.id=e.to_node
+                WHERE e.from_node=? AND n.type IN ('Chunk', 'ImageText', 'Section')
+                """,
+            (file_node_id,),
+        ).fetchall()
+        child_ids = [row["id"] for row in child_rows]
+        conn.execute("DELETE FROM chunks WHERE source_node=?", (file_node_id,))
+        if child_ids:
+            placeholders = ",".join("?" * len(child_ids))
+            conn.execute(f"DELETE FROM nodes WHERE id IN ({placeholders})", child_ids)
+            self._v2_delete_nodes(conn, child_ids)
+        conn.execute("DELETE FROM edges WHERE from_node=?", (file_node_id,))
+        self._v2_delete_edges_from(conn, file_node_id)
+
+        metadata = {
+            "source": "local_folder",
+            "source_id": source_id,
+            "root_path": str(root),
+            "file_path": str(file_path),
+            "relative_path": relative_path,
+            "filename": file_path.name,
+            "ext": file_path.suffix.lower(),
+            "category": category,
+            "parser_type": parser_type,
+            "bytes": stat.st_size,
+            "modified_at": _safe_iso_from_stat_mtime(stat.st_mtime),
+            "sha256": sha256,
+            "parser": parser_meta,
+        }
+        self._upsert_node(
+            conn,
+            file_node_id,
+            _node_type_for_category(category),
+            file_path.name,
+            summary=text[:700],
+            metadata=metadata,
+            raw=metadata,
+        )
+        self._upsert_edge(
+            conn,
+            parent_folder_id,
+            file_node_id,
+            "포함함",
+            weight=1.0,
+            metadata={"source": "local_scan"},
+        )
+
+        target_for_concepts = text
+        if category == "image" and text:
+            image_text_id = f"imagetext:{_sha256_text(f'{file_node_id}:ocr')[:24]}"
+            self._upsert_node(
+                conn,
+                image_text_id,
+                "ImageText",
+                f"{file_path.name} OCR",
+                summary=_clean_text(text)[:700],
+                metadata={
+                    "source_node": file_node_id,
+                    "source_id": source_id,
+                    "chars": len(text),
+                },
+            )
+            self._upsert_edge(
+                conn,
+                file_node_id,
+                image_text_id,
+                "포함함",
+                weight=0.8,
+                metadata={"source": "ocr"},
+            )
+
+        for index, chunk in enumerate(_chunks(text)):
+            chunk_id = f"chunk:{_sha256_text(f'{file_node_id}:{index}:{chunk}')[:24]}"
+            self._upsert_node(
+                conn,
+                chunk_id,
+                "Chunk",
+                f"{file_path.name} chunk {index + 1}",
+                summary=chunk[:500],
+                metadata={
+                    "index": index,
+                    "source_node": file_node_id,
+                    "source_id": source_id,
+                },
+            )
+            self._upsert_chunk(
+                conn,
+                chunk_id=chunk_id,
+                source_node=file_node_id,
+                text=chunk,
+                metadata={
+                    "index": index,
+                    "source_node": file_node_id,
+                    "source_id": source_id,
+                },
+            )
+            self._upsert_edge(
+                conn,
+                file_node_id,
+                chunk_id,
+                "포함함",
+                weight=0.7,
+                metadata={"source": "local_scan"},
+            )
+
+        concepts = _extract_concepts(target_for_concepts, limit=18)
+        concept_ids: Dict[str, str] = {}
+        for concept in concepts:
+            node_t = _classify_node_type(concept, target_for_concepts)
+            concept_id = f"{node_t.lower()}:{_slug(concept)}"
+            concept_ids[concept.lower()] = concept_id
+            self._upsert_node(
+                conn,
+                concept_id,
+                node_t,
+                concept,
+                metadata={
+                    "auto_extracted": True,
+                    "source": "local_folder",
+                    "source_id": source_id,
+                },
+            )
+            self._upsert_edge(
+                conn,
+                file_node_id,
+                concept_id,
+                "언급함",
+                weight=0.75,
+                metadata={"source": "local_scan"},
+            )
+
+        for triple in _extract_triples(target_for_concepts, concepts, limit=20):
+            subj_id = concept_ids.get(triple["subject"].lower())
+            obj_id = concept_ids.get(triple["object"].lower())
+            if subj_id and obj_id and subj_id != obj_id:
+                self._upsert_edge(
+                    conn,
+                    subj_id,
+                    obj_id,
+                    triple["relation"],
+                    weight=0.9,
+                    metadata={
+                        "context": triple.get("context", "")[:240],
+                        "source_id": source_id,
+                    },
+                )
+
+        for item in _semantic_items(target_for_concepts):
+            sem_type = item["type"]
+            sem_title = item["title"]
+            sem_id = f"{sem_type.lower()}:{_sha256_text(f'{file_node_id}:{sem_type}:{sem_title}')[:24]}"
+            self._upsert_node(
+                conn,
+                sem_id,
+                sem_type,
+                sem_title,
+                summary=item["summary"],
+                metadata={
+                    "auto_extracted": True,
+                    "source_node": file_node_id,
+                    "filename": file_path.name,
+                },
+                raw=item,
+            )
+            self._upsert_edge(conn, file_node_id, sem_id, "포함함", weight=0.9)
+
+        return file_node_id
+
+    def _delete_local_file_graph(
+        self, conn: sqlite3.Connection, file_node_id: Optional[str]
+    ) -> None:
+        if not file_node_id:
+            return
+
+        file_row = conn.execute(
+            "SELECT metadata_json FROM nodes WHERE id=?",
+            (file_node_id,),
+        ).fetchone()
+        source_id = None
+        if file_row:
+            source_id = _safe_loads(file_row["metadata_json"]).get("source_id")
+
+        linked_rows = conn.execute(
+            """
+                SELECT n.id, n.type, n.metadata_json
+                FROM edges e
+                JOIN nodes n ON n.id=e.to_node
+                WHERE e.from_node=?
+                """,
+            (file_node_id,),
+        ).fetchall()
+        owned_ids: set = set()
+        auto_candidate_ids: set = set()
+        for row in linked_rows:
+            metadata = _safe_loads(row["metadata_json"])
+            if (
+                row["type"] in {"Chunk", "ImageText", "Section"}
+                or metadata.get("source_node") == file_node_id
+            ):
+                owned_ids.add(row["id"])
+            elif (
+                metadata.get("auto_extracted")
+                and metadata.get("source") == "local_folder"
+            ):
+                auto_candidate_ids.add(row["id"])
+
+        conn.execute("DELETE FROM chunks WHERE source_node=?", (file_node_id,))
+        conn.execute(
+            "DELETE FROM edges WHERE from_node=? OR to_node=?",
+            (file_node_id, file_node_id),
+        )
+        conn.execute("DELETE FROM nodes WHERE id=?", (file_node_id,))
+        self._v2_delete_nodes(conn, [file_node_id])
+
+        def delete_nodes(node_ids: set) -> None:
+            if not node_ids:
+                return
+            placeholders = ",".join("?" * len(node_ids))
+            params = list(node_ids)
+            conn.execute(
+                f"DELETE FROM chunks WHERE source_node IN ({placeholders})", params
+            )
+            conn.execute(
+                f"DELETE FROM edges WHERE from_node IN ({placeholders}) OR to_node IN ({placeholders})",
+                params * 2,
+            )
+            conn.execute(f"DELETE FROM nodes WHERE id IN ({placeholders})", params)
+            self._v2_delete_nodes(conn, params)
+
+        delete_nodes(owned_ids)
+
+        removable_auto_ids: set = set()
+        for node_id in auto_candidate_ids:
+            remaining_edges = conn.execute(
+                "SELECT from_node, to_node FROM edges WHERE from_node=? OR to_node=?",
+                (node_id, node_id),
+            ).fetchall()
+            if all(
+                (
+                    row["from_node"] in auto_candidate_ids
+                    and row["to_node"] in auto_candidate_ids
+                )
+                for row in remaining_edges
+            ):
+                removable_auto_ids.add(node_id)
+        delete_nodes(removable_auto_ids)
+        if source_id:
+            self._cleanup_local_graph_orphans(conn, str(source_id))
+
+    def _cleanup_local_graph_orphans(
+        self, conn: sqlite3.Connection, source_id: str
+    ) -> None:
+        while True:
+            folder_rows = conn.execute(
+                "SELECT id, metadata_json FROM nodes WHERE type='Folder'"
+            ).fetchall()
+            leaf_ids = []
+            for row in folder_rows:
+                metadata = _safe_loads(row["metadata_json"])
+                if metadata.get("source_id") != source_id:
+                    continue
+                has_children = conn.execute(
+                    "SELECT 1 FROM edges WHERE from_node=? LIMIT 1",
+                    (row["id"],),
+                ).fetchone()
+                if not has_children:
+                    leaf_ids.append(row["id"])
+            if not leaf_ids:
+                break
+            placeholders = ",".join("?" * len(leaf_ids))
+            conn.execute(
+                f"DELETE FROM edges WHERE from_node IN ({placeholders}) OR to_node IN ({placeholders})",
+                leaf_ids * 2,
+            )
+            conn.execute(f"DELETE FROM nodes WHERE id IN ({placeholders})", leaf_ids)
+            self._v2_delete_nodes(conn, leaf_ids)
+
+        for node_type in ("Drive", "Computer"):
+            rows = conn.execute(
+                "SELECT id FROM nodes WHERE type=?", (node_type,)
+            ).fetchall()
+            removable = []
+            for row in rows:
+                has_children = conn.execute(
+                    "SELECT 1 FROM edges WHERE from_node=? LIMIT 1",
+                    (row["id"],),
+                ).fetchone()
+                if not has_children:
+                    removable.append(row["id"])
+            if removable:
+                placeholders = ",".join("?" * len(removable))
+                conn.execute(
+                    f"DELETE FROM edges WHERE from_node IN ({placeholders}) OR to_node IN ({placeholders})",
+                    removable * 2,
+                )
+                conn.execute(
+                    f"DELETE FROM nodes WHERE id IN ({placeholders})", removable
+                )
+                self._v2_delete_nodes(conn, removable)
+
+    def _local_file_index_has_extracted_text(self, row: sqlite3.Row) -> bool:
+        metadata = _safe_loads(row["metadata_json"])
+        parser = metadata.get("parser") if isinstance(metadata, dict) else {}
+        if not isinstance(parser, dict):
+            return False
+        try:
+            return int(parser.get("extracted_chars") or 0) > 0
+        except (TypeError, ValueError):
+            return False
+
+    def index_local_folder(
+        self,
+        path: Path,
+        *,
+        include_ocr: bool = False,
+        watch_enabled: bool = False,
+        user_email: Optional[str] = None,
+        consent: Optional[Dict[str, Any]] = None,
+        max_files: int = 5_000,
+    ) -> Dict[str, Any]:
+        """Read approved files from a local folder and connect them to Graph RAG."""
+        root = Path(path).expanduser().resolve()
+        if not root.exists():
+            raise ValueError(f"경로가 존재하지 않습니다: {path}")
+        if not root.is_dir():
+            raise ValueError(f"폴더가 아닙니다: {path}")
+
+        os_type = _current_os_type()
+        drive_id = _drive_id_for_path(root)
+        source_id = f"source:{_path_fingerprint(root)}"
+        now = _now()
+        max_files = max(1, min(int(max_files or 5_000), 50_000))
+        consent_payload = {
+            "approved_at": now,
+            "approved_by": user_email,
+            "knowledge_source": True,
+            "include_ocr": bool(include_ocr),
+            "watch_enabled": bool(watch_enabled),
+            "sensitive_files_default_excluded": True,
+            **(consent or {}),
+        }
+        counts: Counter = Counter()
+        seen_relative_paths: set = set()
+        indexed_nodes: List[str] = []
+        errors: List[Dict[str, str]] = []
+        limit_reached = False
+
+        with self._connect() as conn:
+            conn.execute(
+                """
+                    INSERT INTO knowledge_sources(
+                      id, root_path, os_type, drive_id, label, status, include_ocr,
+                      watch_enabled, consent_json, created_at, updated_at, last_scanned_at
+                    )
+                    VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
+                    ON CONFLICT(id) DO UPDATE SET
+                      root_path=excluded.root_path,
+                      os_type=excluded.os_type,
+                      drive_id=excluded.drive_id,
+                      label=excluded.label,
+                      status=excluded.status,
+                      include_ocr=excluded.include_ocr,
+                      watch_enabled=excluded.watch_enabled,
+                      consent_json=excluded.consent_json,
+                      updated_at=excluded.updated_at,
+                      last_scanned_at=excluded.last_scanned_at
+                    """,
+                (
+                    source_id,
+                    str(root),
+                    os_type,
+                    drive_id,
+                    root.name or str(root),
+                    "scanning",
+                    1 if include_ocr else 0,
+                    1 if watch_enabled else 0,
+                    _json(consent_payload),
+                    now,
+                    now,
+                    now,
+                ),
+            )
+
+            for entry in self._iter_local_scan_entries(root, max_files=max_files):
+                kind = entry["kind"]
+                file_path = entry["path"]
+                if kind == "limit_reached":
+                    counts["limit_reached"] += 1
+                    limit_reached = True
+                    break
+                if kind in {"excluded_dir", "excluded"}:
+                    counts["excluded"] += 1
+                    continue
+                if kind in {"inaccessible_dir", "inaccessible_file"}:
+                    counts["failed"] += 1
+                    errors.append(
+                        {
+                            "path": str(file_path),
+                            "error": entry.get("reason", "inaccessible"),
+                        }
+                    )
+                    continue
+                if kind != "file":
+                    continue
+
+                stat = entry["stat"]
+                try:
+                    relative_path = file_path.relative_to(root).as_posix()
+                except ValueError:
+                    relative_path = file_path.name
+                seen_relative_paths.add(relative_path)
+                modified_at = _safe_iso_from_stat_mtime(stat.st_mtime)
+                existing = conn.execute(
+                    """
+                        SELECT size_bytes, modified_at, sha256, graph_node_id, status, metadata_json
+                        FROM local_file_index
+                        WHERE source_id=? AND relative_path=?
+                        """,
+                    (source_id, relative_path),
+                ).fetchone()
+                decision = self._local_file_decision(file_path, root, stat)
+                parser_type = decision["parser_type"]
+                if not decision["indexable"]:
+                    counts[decision["status"]] += 1
+                    if existing and existing["graph_node_id"]:
+                        self._delete_local_file_graph(conn, existing["graph_node_id"])
+                    self._upsert_local_file_index(
+                        conn,
+                        source_id=source_id,
+                        root=root,
+                        file_path=file_path,
+                        stat=stat,
+                        os_type=os_type,
+                        drive_id=drive_id,
+                        status=decision["status"],
+                        parser_type=parser_type,
+                        metadata={
+                            "reason": decision["reason"],
+                            "category": decision["category"],
+                        },
+                    )
+                    continue
+
+                if (
+                    existing
+                    and existing["status"] == "indexed"
+                    and existing["graph_node_id"]
+                    and self._local_file_index_has_extracted_text(existing)
+                    and existing["size_bytes"] == stat.st_size
+                    and existing["modified_at"] == modified_at
+                ):
+                    counts["skipped_unchanged"] += 1
+                    self._upsert_local_file_index(
+                        conn,
+                        source_id=source_id,
+                        root=root,
+                        file_path=file_path,
+                        stat=stat,
+                        os_type=os_type,
+                        drive_id=drive_id,
+                        status="indexed",
+                        parser_type=parser_type,
+                        sha256=existing["sha256"],
+                        graph_node_id=existing["graph_node_id"],
+                        metadata={
+                            **_safe_loads(existing["metadata_json"]),
+                            "category": decision["category"],
+                            "unchanged": True,
+                        },
+                    )
+                    continue
+
+                try:
+                    data = file_path.read_bytes()
+                    digest = _sha256_bytes(data)
+                except Exception as exc:
+                    counts["failed"] += 1
+                    errors.append({"path": str(file_path), "error": str(exc)})
+                    if existing and existing["graph_node_id"]:
+                        self._delete_local_file_graph(conn, existing["graph_node_id"])
+                    self._upsert_local_file_index(
+                        conn,
+                        source_id=source_id,
+                        root=root,
+                        file_path=file_path,
+                        stat=stat,
+                        os_type=os_type,
+                        drive_id=drive_id,
+                        status="failed",
+                        parser_type=parser_type,
+                        error_message=str(exc),
+                        metadata={"category": decision["category"]},
+                    )
+                    continue
+
+                if (
+                    existing
+                    and existing["sha256"] == digest
+                    and existing["graph_node_id"]
+                    and self._local_file_index_has_extracted_text(existing)
+                ):
+                    counts["skipped_unchanged"] += 1
+                    self._upsert_local_file_index(
+                        conn,
+                        source_id=source_id,
+                        root=root,
+                        file_path=file_path,
+                        stat=stat,
+                        os_type=os_type,
+                        drive_id=drive_id,
+                        status="indexed",
+                        parser_type=parser_type,
+                        sha256=digest,
+                        graph_node_id=existing["graph_node_id"],
+                        metadata={
+                            **_safe_loads(existing["metadata_json"]),
+                            "category": decision["category"],
+                            "sha256_unchanged": True,
+                        },
+                    )
+                    continue
+
+                try:
+                    text, parser_meta = self._extract_local_file_text(
+                        file_path,
+                        decision["category"],
+                        include_ocr=include_ocr,
+                    )
+                    text = _clean_text(text)
+                    parser_meta = {**parser_meta, "extracted_chars": len(text)}
+                    if not text:
+                        counts["skipped_empty_text"] += 1
+                        if existing and existing["graph_node_id"]:
+                            self._delete_local_file_graph(
+                                conn, existing["graph_node_id"]
+                            )
+                        self._upsert_local_file_index(
+                            conn,
+                            source_id=source_id,
+                            root=root,
+                            file_path=file_path,
+                            stat=stat,
+                            os_type=os_type,
+                            drive_id=drive_id,
+                            status="skipped_empty_text",
+                            parser_type=parser_type,
+                            sha256=digest,
+                            error_message="텍스트 추출 결과가 비어 있습니다.",
+                            metadata={
+                                "category": decision["category"],
+                                "parser": parser_meta,
+                            },
+                        )
+                        continue
+                    graph_node_id = self._upsert_local_file_node(
+                        conn,
+                        source_id=source_id,
+                        root=root,
+                        file_path=file_path,
+                        stat=stat,
+                        os_type=os_type,
+                        drive_id=drive_id,
+                        sha256=digest,
+                        category=decision["category"],
+                        parser_type=parser_type,
+                        text=text,
+                        parser_meta=parser_meta,
+                    )
+                    self._upsert_local_file_index(
+                        conn,
+                        source_id=source_id,
+                        root=root,
+                        file_path=file_path,
+                        stat=stat,
+                        os_type=os_type,
+                        drive_id=drive_id,
+                        status="indexed",
+                        parser_type=parser_type,
+                        sha256=digest,
+                        graph_node_id=graph_node_id,
+                        metadata={
+                            "category": decision["category"],
+                            "parser": parser_meta,
+                        },
+                    )
+                    counts["indexed"] += 1
+                    indexed_nodes.append(graph_node_id)
+                except Exception as exc:
+                    counts["failed"] += 1
+                    errors.append({"path": str(file_path), "error": str(exc)})
+                    if existing and existing["graph_node_id"]:
+                        self._delete_local_file_graph(conn, existing["graph_node_id"])
+                    self._upsert_local_file_index(
+                        conn,
+                        source_id=source_id,
+                        root=root,
+                        file_path=file_path,
+                        stat=stat,
+                        os_type=os_type,
+                        drive_id=drive_id,
+                        status="failed",
+                        parser_type=parser_type,
+                        sha256=digest,
+                        error_message=str(exc),
+                        metadata={"category": decision["category"]},
+                    )
+
+            if not limit_reached:
+                existing_rows = {
+                    row["relative_path"]: row["graph_node_id"]
+                    for row in conn.execute(
+                        "SELECT relative_path, graph_node_id FROM local_file_index WHERE source_id=?",
+                        (source_id,),
+                    )
+                }
+                deleted_paths = set(existing_rows) - seen_relative_paths
+                for relative_path in deleted_paths:
+                    self._delete_local_file_graph(
+                        conn, existing_rows.get(relative_path)
+                    )
+                    conn.execute(
+                        """
+                            UPDATE local_file_index
+                            SET status='deleted', deleted=1, last_scanned_at=?, error_message=NULL, graph_node_id=NULL
+                            WHERE source_id=? AND relative_path=?
+                            """,
+                        (_now(), source_id, relative_path),
+                    )
+                counts["deleted"] = len(deleted_paths)
+            conn.execute(
+                """
+                    UPDATE knowledge_sources
+                    SET status='active', updated_at=?, last_scanned_at=?
+                    WHERE id=?
+                    """,
+                (_now(), _now(), source_id),
+            )
+
+        return {
+            "status": "ok",
+            "source": {
+                "id": source_id,
+                "root_path": str(root),
+                "os_type": os_type,
+                "drive_id": drive_id,
+                "include_ocr": bool(include_ocr),
+                "watch_enabled": bool(watch_enabled),
+            },
+            "counts": dict(counts),
+            "indexed_nodes": indexed_nodes[:100],
+            "errors": errors[:50],
+            "notice": "Lattice AI는 사용자가 선택한 폴더만 AI 지식으로 변환합니다.",
+        }