功能优化升级，增加数据模拟，备份定时任务。

CLAUDE.md

@@ -5,6 +5,14 @@
 # 总体要求
 - 每做一个任务就在工作记录中记录一次
 
+# 数据库建表 / 改表约束（2026-09-30 起）
+- **表和字段都必须写中文注释**，两处都要做：
+  1. ORM 模型（`web/backend/models/`）：表注释写在 `__table_args__ = (..., {"comment": "..."})`，字段注释写在 `Column(..., comment="...")`——新库 create_all 时直接带上；
+  2. 注释登记表 `web/backend/db/schema_comments.py` 的 `SCHEMA_COMMENTS`——后端启动时由 `apply_mysql_comments()` 自动给**已存在的库**补齐/修正表和字段注释（幂等）。
+- 新表必须注册进 `web/backend/models/__init__.py`，否则 create_all 会静默跳过建表。
+- 给**已有表加字段**时，create_all 不会生效，必须在 `web/backend/db/database.py` 写幂等迁移函数（inspect 判断列存在 → `ALTER TABLE ADD COLUMN`），并在 `init_db()` 里调用；新列同样要登记 SCHEMA_COMMENTS。
+- 发版会自动完成：重启后端 → create_all 建缺表 → 迁移函数补列 → 注释同步。正式库（192.168.50.10）无需手工动表。
+
 # 服务器发版本运行
 - `deploy/deploy_server.py`
 

web/backend/app.py

@@ -45,7 +45,9 @@ logger.info("DB: %s", engine.url.render_as_string(hide_password=True))
 @asynccontextmanager
 async def _lifespan(_: FastAPI):
     await asyncio.get_event_loop().run_in_executor(None, start_scheduler)
+    await asyncio.get_event_loop().run_in_executor(None, start_backup_scheduler)
     yield
+    stop_backup_scheduler()
     stop_scheduler()
 
 
@@ -81,6 +83,7 @@ from web.backend.routers.table_design import router as table_design_router  # no
 from web.backend.routers.llm_config import router as llm_config_router  # noqa: E402
 from web.backend.routers.table_prefs import router as table_prefs_router  # noqa: E402
 from web.backend.routers.analysis import router as analysis_router, start_scheduler, stop_scheduler  # noqa: E402
+from web.backend.routers.operations import start_backup_scheduler, stop_backup_scheduler  # noqa: E402
 app.include_router(task_groups_router, prefix="/api")
 app.include_router(tasks_router, prefix="/api")
 app.include_router(db_router, prefix="/api")

web/backend/core/ai_rule.py

@@ -40,14 +40,13 @@ _SYSTEM_PROMPT = "你是 Python 程序员，擅长写简短的数据校验函数
 
 _DESCRIPTION_SYSTEM_PROMPT = "你是数据质量规则专家，擅长用一句简洁的话说明数据校验规则。"
 
-_DESCRIPTION_TEMPLATE = """用一句中文生成 30-40 字规则说明，只输出说明本身。
+_DESCRIPTION_TEMPLATE = """用一句简洁、完整的中文说明下面的数据校验规则，只输出说明本身。
+不限定字数：以意思完整为准，关键格式 / 范围 / 排除项都要讲到，能省的字省掉即可。
 规则名称：{name}
 优先依据：{prompt}
 现有说明：{desc}
 规则类型：{rule_type}
-仅保留生成规则所需的关键格式或范围，不要补充额外校验细节。"""
-
-_MAX_AI_DESCRIPTION_LENGTH = 40
+不要补充规则未提及的校验细节。"""
 
 
 # ── User Prompt ─────────────────────────────────────────
@@ -232,9 +231,9 @@ def gen_rule_description(name: str, desc: str, rule_type: str, prompt: str = "")
     description = description.replace("\n", " ").strip().strip('"')
     if not description:
         return "", "AI 未生成规则说明"
-    # 兜底截断，避免模型未遵守长度要求时撑大规则列表。
-    description = description[:_MAX_AI_DESCRIPTION_LENGTH].rstrip("，、；：。 ")
-    return description, "已生成简短规则说明（30-40字）"
+    # 不截断字数（用户 2026-09-29：限制字数会把意思截断；DB 列是 Text 无长度约束，
+    # 长度交给提示词里的「以意思完整为准、能省则省」约束）。
+    return description, "已生成规则说明"
 
 
 # ── 主入口 ─────────────────────────────────────────────

web/backend/core/llm.py

@@ -238,7 +238,10 @@ class LLMClient:
                 {b for b in (initial_budget, max(initial_budget * 2, 2048), 8192) if b <= 8192}
             ))
             for attempt, max_tokens in enumerate(token_budgets, start=1):
-                kwargs = {**base_kwargs, "max_tokens": max_tokens}
+                # 思考型模型每一跳都可能超 30s（内部代理生成速度慢，实测 1024-2048
+                # 额度的思考+输出也要 30s+），统一给 90s 下限；配置了更大值则以配置为准。
+                attempt_timeout = max(self.cfg.timeout, 90)
+                kwargs = {**base_kwargs, "max_tokens": max_tokens, "timeout": attempt_timeout}
                 msg = self._provider.messages.create(**kwargs)
                 for block in msg.content:
                     text = getattr(block, "text", None)
@@ -325,10 +328,10 @@ def _format_api_error(exc: Exception) -> str:
             if hint:
                 return f"API Error: 请求被拒绝 ({status}) · {detail} ({hint})"
             return f"API Error: 请求失败 ({status}) · {detail}"
+        if isinstance(exc, APITimeoutError):
+            return f"API Error: 请求超时（模型响应时间过长，可调大 llm 配置 timeout）· {exc_msg}"
         if isinstance(exc, APIConnectionError):
             return f"API Error: 网络连接失败 · {exc_msg}"
-        if isinstance(exc, APITimeoutError):
-            return f"API Error: 请求超时 · {exc_msg}"
     except ImportError:
         pass
 

web/backend/core/operation_history.py

@@ -19,6 +19,10 @@ def record_operation(
     tables: Iterable[str],
     filename: str = "",
     execute_setting: str = "",
+    trigger_type: str = "manual",
+    backup_mode: str = "",
+    task_id: int | None = None,
+    task_name: str = "",
 ) -> OperationRecord:
     """保存一次已成功完成的操作；密码等敏感请求参数永不写入历史。
 
@@ -36,6 +40,10 @@ def record_operation(
         operation=operation,
         source_id=source.id if source else None,
         source_name=source.name if source else "",
+        trigger_type=trigger_type or "manual",
+        backup_mode=backup_mode or "",
+        task_id=task_id,
+        task_name=(task_name or "")[:120],
         database_name=database or "",
         scope=scope or "",
         tables_json=tables_json,
@@ -46,3 +54,15 @@ def record_operation(
     db.commit()
     db.refresh(row)
     return row
+
+
+def update_operation(db: Session, record_id: int, *, status: str, execute_setting: str = "", error_message: str = ""):
+    row = db.get(OperationRecord, record_id)
+    if row is None:
+        return None
+    row.status = status
+    row.execute_setting = (execute_setting or "")[:400]
+    row.error_message = (error_message or "")[:1000]
+    db.commit()
+    db.refresh(row)
+    return row

web/backend/db/database.py

@@ -5,7 +5,7 @@ from contextlib import contextmanager
 import json
 from pathlib import Path
 
-from sqlalchemy import create_engine, event
+from sqlalchemy import create_engine, event, inspect
 from sqlalchemy.orm import sessionmaker, Session, declarative_base
 
 from web.backend.config import APP_ENV, DATA_DIR, DATABASE_CONNECT_ARGS, DATABASE_URL
@@ -66,6 +66,8 @@ def init_db() -> None:
     """初始化当前环境的 MySQL 表，并保留 SQLite 旧库升级兼容。"""
     from web.backend.db import seed
     Base.metadata.create_all(engine)
+    _migrate_operation_record_error()
+    _migrate_operation_record_context()
     if IS_SQLITE:
         # 一次性迁移：补齐 connection_preset 的 UNIQUE(db_type, name) 约束
         # 历史背景：connection_preset 最初是手动跑 schema.sql 片段建的（仅 7 张主表里没有它），
@@ -100,14 +102,48 @@ def init_db() -> None:
         if changed:
             logger.info("[init_db] 已更新 %s 项 MySQL 中文注释", changed)
     # 灌种子（每个 seed 函数内部判重，可重复调用）
+    # 2026-09-29：不再灌示例数据源 —— 测试/正式环境每次发版都会多出 3 条
+    # 127.0.0.1 占位数据源（seed_data_sources 保留函数但不再调用）
     seed.seed_task_groups()
     seed.seed_connection_presets()
-    seed.seed_data_sources()
     _migrate_llm_config_from_file()
     _migrate_backup_config_to_database()
     print(f"[init_db] 初始化完成：{engine.url.render_as_string(hide_password=True)}")
 
 
+def _migrate_operation_record_error() -> None:
+    """Add async operation error text to existing SQLite installations."""
+    inspector = inspect(engine)
+    if "operation_record" not in inspector.get_table_names():
+        return
+    cols = {column["name"] for column in inspector.get_columns("operation_record")}
+    if "error_message" in cols:
+        return
+    with engine.begin() as conn:
+        if IS_SQLITE:
+            conn.exec_driver_sql("ALTER TABLE operation_record ADD COLUMN error_message TEXT NULL")
+        else:
+            conn.exec_driver_sql("ALTER TABLE operation_record ADD COLUMN error_message TEXT NULL")
+
+
+def _migrate_operation_record_context() -> None:
+    """补齐操作记录的触发方式、备份模式和任务关联字段。"""
+    inspector = inspect(engine)
+    if "operation_record" not in inspector.get_table_names():
+        return
+    existing = {column["name"] for column in inspector.get_columns("operation_record")}
+    columns = (
+        ("trigger_type", "VARCHAR(20) NOT NULL DEFAULT 'manual'"),
+        ("backup_mode", "VARCHAR(20) NOT NULL DEFAULT ''"),
+        ("task_id", "INTEGER NULL"),
+        ("task_name", "VARCHAR(120) NOT NULL DEFAULT ''"),
+    )
+    with engine.begin() as conn:
+        for name, definition in columns:
+            if name not in existing:
+                conn.exec_driver_sql(f"ALTER TABLE operation_record ADD COLUMN {name} {definition}")
+
+
 def _migrate_connection_preset_unique() -> None:
     """确保 connection_preset 表有显式 UNIQUE INDEX uq_connection_preset_db_type_name。
 

web/backend/db/schema.sql

@@ -198,6 +198,10 @@ CREATE TABLE operation_record (
   id              INTEGER PRIMARY KEY AUTOINCREMENT,
   operation       TEXT NOT NULL,
   source_id       INTEGER REFERENCES data_source(id) ON DELETE SET NULL,
+  trigger_type    TEXT NOT NULL DEFAULT 'manual',
+  backup_mode     TEXT NOT NULL DEFAULT '',
+  task_id         INTEGER,
+  task_name       TEXT NOT NULL DEFAULT '',
   source_name     TEXT NOT NULL DEFAULT '',
   database_name   TEXT NOT NULL DEFAULT '',
   scope           TEXT NOT NULL DEFAULT '',
@@ -205,10 +209,29 @@ CREATE TABLE operation_record (
   filename        TEXT NOT NULL DEFAULT '',
   execute_setting TEXT NOT NULL DEFAULT '',
   status          TEXT NOT NULL DEFAULT '已完成',
+  error_message   TEXT,
   created_at      TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP
 );
 CREATE INDEX idx_operation_record_operation_created ON operation_record(operation, created_at DESC);
 
+-- 9.1) backup_task · 定时备份任务配置
+CREATE TABLE backup_task (
+  id INTEGER PRIMARY KEY AUTOINCREMENT,
+  name TEXT NOT NULL,
+  source_id INTEGER NOT NULL REFERENCES data_source(id) ON DELETE CASCADE,
+  database_name TEXT NOT NULL DEFAULT '',
+  schema_name TEXT,
+  scope TEXT NOT NULL DEFAULT 'database',
+  tables_json TEXT NOT NULL DEFAULT '[]',
+  backup_mode TEXT NOT NULL DEFAULT 'full',
+  include_schema INTEGER NOT NULL DEFAULT 1,
+  include_data INTEGER NOT NULL DEFAULT 1,
+  cron_expression TEXT NOT NULL,
+  enabled INTEGER NOT NULL DEFAULT 1,
+  created_at TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+  updated_at TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP
+);
+
 -- 10) table_design_config · 表设计校验配置与最近扫描结果
 CREATE TABLE table_design_config (
   id                   INTEGER PRIMARY KEY AUTOINCREMENT,

web/backend/db/schema_comments.py

@@ -44,7 +44,7 @@ SCHEMA_COMMENTS: Dict[str, dict] = {
     },
     "operation_record": {
         "table": "数据库操作记录",
-        "columns": {"id": "主键", "operation": "操作类型", "source_id": "数据源ID", "source_name": "数据源名称", "database_name": "数据库名称", "scope": "操作范围", "tables_json": "数据表列表JSON", "filename": "备份文件名", "execute_setting": "执行设置", "status": "执行状态", "created_at": "创建时间"},
+        "columns": {"id": "主键", "operation": "操作类型", "source_id": "数据源ID", "source_name": "数据源名称", "database_name": "数据库名称", "scope": "操作范围", "tables_json": "数据表列表JSON", "filename": "备份文件名", "execute_setting": "执行设置", "status": "执行状态", "trigger_type": "触发方式：manual 手动 / scheduled 定时", "backup_mode": "备份方式：full 全量 / incremental 增量", "task_id": "关联备份任务ID（无则手动操作）", "task_name": "关联备份任务名称", "error_message": "失败原因", "created_at": "创建时间"},
     },
     "table_design_config": {
         "table": "表设计校验配置",
@@ -83,4 +83,26 @@ SCHEMA_COMMENTS: Dict[str, dict] = {
         "table": "系统参数配置",
         "columns": {"setting_key": "配置参数键", "setting_value": "配置参数值", "updated_at": "更新时间"},
     },
+    "backup_task": {
+        "table": "定时备份任务配置",
+        "columns": {
+            "id": "主键", "name": "任务名称", "source_id": "数据源ID", "database_name": "数据库/Schema名称",
+            "schema_name": "Schema名称", "scope": "备份范围：database 整库 / tables 指定表",
+            "tables_json": "指定数据表清单JSON", "backup_mode": "备份方式：full 全量 / incremental 增量",
+            "include_schema": "是否备份表结构：1 是 0 否", "include_data": "是否备份表数据：1 是 0 否",
+            "cron_expression": "Cron执行表达式（标准5段）", "enabled": "是否启用：1 启用 0 停用",
+            "created_at": "创建时间", "updated_at": "更新时间",
+        },
+    },
+    "table_pref": {
+        "table": "列表页表格偏好（表设置：列宽/列显隐/列对齐/列顺序/行高/线框粗细）",
+        "columns": {
+            "id": "主键", "page_key": "页面标识", "col_key": "列标识（表头文字；__page__ 为页面级偏好行）",
+            "width": "列宽 px；NULL=未自定义", "hidden": "是否隐藏：0 显示 1 隐藏",
+            "align": "列对齐：left/center/right；NULL=默认", "ord": "列顺序（0 起）；NULL=原始位置",
+            "row_height": "行高 px（36/40/48）；仅页面级行使用",
+            "border_width": "线框粗细档位（0=细 1=中 2=粗）；仅页面级行使用",
+            "updated_at": "最近更新时间",
+        },
+    },
 }

web/backend/models/__init__.py

@@ -18,5 +18,6 @@ from .table_design_config import TableDesignConfig
 from .llm_config import LLMConfigRecord
 from .system_config import SystemConfig
 from .analysis import AnalysisTask, AnalysisRun, AnalysisTableSnapshot
+from .backup_task import BackupTask
 
-__all__ = ["ConnectionPreset", "Field", "Rule", "RuleLibrary", "FieldRule", "Task", "TaskGroup", "DataSource", "OperationRecord", "TableDesignConfig", "LLMConfigRecord", "SystemConfig", "TablePref", "AnalysisTask", "AnalysisRun", "AnalysisTableSnapshot"]
+__all__ = ["ConnectionPreset", "Field", "Rule", "RuleLibrary", "FieldRule", "Task", "TaskGroup", "DataSource", "OperationRecord", "TableDesignConfig", "LLMConfigRecord", "SystemConfig", "TablePref", "AnalysisTask", "AnalysisRun", "AnalysisTableSnapshot", "BackupTask"]

web/backend/models/backup_task.py

+"""Persistent scheduled backup configurations."""
+from sqlalchemy import Column, DateTime, ForeignKey, Integer, String, Text
+from sqlalchemy.sql import func
+from web.backend.db.database import Base
+
+
+class BackupTask(Base):
+    __tablename__ = "backup_task"
+    __table_args__ = (
+        # 表注释（新库建表时写入 MySQL COMMENT；已存在的表由 apply_mysql_comments 启动时同步）
+        {"comment": "定时备份任务配置"},
+    )
+
+    id = Column(Integer, primary_key=True, autoincrement=True, comment="主键")
+    name = Column(String(120), nullable=False, comment="任务名称")
+    source_id = Column(Integer, ForeignKey("data_source.id", ondelete="CASCADE"), nullable=False, comment="数据源ID")
+    database_name = Column(String(255), nullable=False, default="", comment="数据库/Schema名称")
+    schema_name = Column(String(255), nullable=True, comment="Schema名称")
+    scope = Column(String(20), nullable=False, default="database", comment="备份范围：database 整库 / tables 指定表")
+    tables_json = Column(Text, nullable=False, default="[]", comment="指定数据表清单JSON")
+    backup_mode = Column(String(20), nullable=False, default="full", comment="备份方式：full 全量 / incremental 增量")
+    include_schema = Column(Integer, nullable=False, default=1, comment="是否备份表结构：1 是 0 否")
+    include_data = Column(Integer, nullable=False, default=1, comment="是否备份表数据：1 是 0 否")
+    cron_expression = Column(String(100), nullable=False, comment="Cron执行表达式（标准5段）")
+    enabled = Column(Integer, nullable=False, default=1, comment="是否启用：1 启用 0 停用")
+    created_at = Column(DateTime, nullable=False, server_default=func.current_timestamp(), comment="创建时间")
+    updated_at = Column(DateTime, nullable=False, server_default=func.current_timestamp(), onupdate=func.current_timestamp(), comment="更新时间")
+
+    def to_dict(self, next_run_at=None):
+        try:
+            import json
+            tables = json.loads(self.tables_json or "[]")
+        except Exception:
+            tables = []
+        return {"id": self.id, "name": self.name, "source_id": self.source_id,
+                "database": self.database_name, "schema": self.schema_name,
+                "scope": self.scope, "tables": tables, "backup_mode": self.backup_mode,
+                "include_schema": bool(self.include_schema), "include_data": bool(self.include_data),
+                "cron_expression": self.cron_expression, "enabled": bool(self.enabled),
+                "created_at": self.created_at, "next_run_at": next_run_at}

web/backend/models/operation_record.py

@@ -16,6 +16,11 @@ class OperationRecord(Base):
     id = Column(Integer, primary_key=True, autoincrement=True)
     operation = Column(String(20), nullable=False)
     source_id = Column(Integer, ForeignKey("data_source.id", ondelete="SET NULL"), nullable=True)
+    # 记录来源和备份模式，保证历史列表可以区分手动/自动、全量/增量。
+    trigger_type = Column(String(20), nullable=False, default="manual", server_default="manual")
+    backup_mode = Column(String(20), nullable=False, default="", server_default="")
+    task_id = Column(Integer, nullable=True)
+    task_name = Column(String(120), nullable=False, default="", server_default="")
     source_name = Column(String(100), nullable=False, default="")
     database_name = Column(String(255), nullable=False, default="")
     scope = Column(String(20), nullable=False, default="")
@@ -23,6 +28,7 @@ class OperationRecord(Base):
     filename = Column(String(255), nullable=False, default="")
     execute_setting = Column(Text, nullable=False, default="")
     status = Column(String(20), nullable=False, default="已完成")
+    error_message = Column(Text, nullable=True, default="")
     created_at = Column(DateTime, nullable=False, server_default=func.current_timestamp())
 
     def to_dict(self) -> dict:
@@ -35,11 +41,16 @@ class OperationRecord(Base):
             "operation": self.operation,
             "source_id": self.source_id,
             "source_name": self.source_name,
+            "trigger_type": self.trigger_type or "manual",
+            "backup_mode": self.backup_mode or "",
+            "task_id": self.task_id,
+            "task_name": self.task_name or "",
             "database": self.database_name,
             "scope": self.scope,
             "tables": tables if isinstance(tables, list) else [],
             "filename": self.filename,
             "execute_setting": self.execute_setting,
             "status": self.status,
+            "error_message": self.error_message or "",
             "created_at": self.created_at.isoformat() if isinstance(self.created_at, datetime) else self.created_at,
         }

web/backend/routers/analysis.py

@@ -271,6 +271,32 @@ def _snapshots(db: Session, run_id: int) -> dict[str, AnalysisTableSnapshot]:
             .filter(AnalysisTableSnapshot.run_id == run_id).all()}
 
 
+def _column_changes(old_cols: dict[str, dict], current_cols: dict[str, dict]) -> list[dict]:
+    """Compare column properties; a shifted position after add/drop is not a modification."""
+    common = old_cols.keys() & current_cols.keys()
+    def relative_order(cols: dict[str, dict]) -> dict[str, int]:
+        names = sorted(common, key=lambda name: (int(cols[name].get("ordinal_position") or 0), name))
+        return {name: index for index, name in enumerate(names)}
+
+    old_order, current_order = relative_order(old_cols), relative_order(current_cols)
+    changes = []
+    for col_name in sorted(old_cols.keys() | current_cols.keys()):
+        left, right = old_cols.get(col_name), current_cols.get(col_name)
+        if left is None:
+            changes.append({"column": col_name, "kind": "added", "before": None, "after": right})
+        elif right is None:
+            changes.append({"column": col_name, "kind": "removed", "before": left, "after": None})
+        else:
+            fields = [key for key in _COLUMNS if key not in ("column_name", "ordinal_position")
+                      and left.get(key) != right.get(key)]
+            if old_order[col_name] != current_order[col_name]:
+                fields.append("ordinal_position")
+            if fields:
+                changes.append({"column": col_name, "kind": "changed", "before": left,
+                                "after": right, "changed_fields": fields})
+    return changes
+
+
 def _report(db: Session, run: AnalysisRun) -> dict:
     previous = (db.query(AnalysisRun).filter(AnalysisRun.task_id == run.task_id,
                                            AnalysisRun.status == "success", AnalysisRun.id < run.id)
@@ -282,15 +308,7 @@ def _report(db: Session, run: AnalysisRun) -> dict:
         now, before = current.get(name), old.get(name)
         current_cols = {c["column_name"]: c for c in json.loads(now.columns_json)} if now else {}
         old_cols = {c["column_name"]: c for c in json.loads(before.columns_json)} if before else {}
-        changes = []
-        for col_name in (sorted(set(current_cols) | set(old_cols)) if previous else []):
-            left, right = old_cols.get(col_name), current_cols.get(col_name)
-            if left is None:
-                changes.append({"column": col_name, "kind": "added", "before": None, "after": right})
-            elif right is None:
-                changes.append({"column": col_name, "kind": "removed", "before": left, "after": None})
-            elif left != right:
-                changes.append({"column": col_name, "kind": "changed", "before": left, "after": right})
+        changes = _column_changes(old_cols, current_cols) if previous else []
         tables.append({"table_name": name, "change": "baseline" if previous is None else "added" if before is None else "removed" if now is None else
                        "changed" if changes or now.table_comment != before.table_comment else "unchanged",
                        "table_comment_before": before.table_comment if before else None,
@@ -360,7 +378,7 @@ def _column_def(col: dict, dialect: str) -> str:
     """列定义片段：类型 + NOT NULL + 默认值（注释由调用方按方言拼）。"""
     ctype = str(col.get("column_type") or col.get("data_type") or "VARCHAR(255)")
     definition = f"{quote_ident(str(col.get('column_name') or ''), dialect)} {ctype}"
-    if str(col.get("is_nullable") or "YES").upper() == "NO":
+    if str(col.get("is_nullable") or "YES").upper() in ("NO", "N"):
         definition += " NOT NULL"
     default = col.get("column_default")
     if default is not None and str(default).strip() != "":
@@ -460,7 +478,16 @@ def analyze_table(run_id: int, table_name: str, db: Session = Depends(get_sessio
     change_lines = []
     for change in (row.get("column_changes") or [])[:60]:
         if change["kind"] == "changed" and change.get("before") and change.get("after"):
-            detail = f"（{change['before'].get('column_type')} → {change['after'].get('column_type')}）"
+            before, after = change["before"], change["after"]
+            labels = {"column_type": "类型", "data_type": "数据类型", "char_max_length": "字符长度",
+                      "numeric_precision": "数值精度", "numeric_scale": "小数位数",
+                      "is_nullable": "可空", "column_default": "默认值", "column_comment": "注释",
+                      "extra": "附加属性", "ordinal_position": "顺序"}
+            fields = [key for key in change.get("changed_fields", [])
+                      if key == "column_type" or "column_type" not in change["changed_fields"]
+                      or key not in ("data_type", "char_max_length", "numeric_precision", "numeric_scale")]
+            detail = "（" + "；".join(
+                f"{labels[key]}：{before.get(key)} → {after.get(key)}" for key in fields) + "）"
         else:
             detail = ""
         label = {"added": "新增", "removed": "删除", "changed": "修改"}.get(change["kind"], change["kind"])

web/backend/routers/queries.py

@@ -444,9 +444,19 @@ def start_query(req: StartQueryRequest, db: Session = Depends(get_session)):
             count_row = dbc.fetchone(count_sql)
             total_rows = int((count_row or {}).get("n") or 0)
 
-            if schema:
+            # 表设置需要能选择未配置校验规则的字段。规则字段仍排在前面，
+            # 其余字段补入 field_list，并加入显式 SELECT，避免前端只能看到问题字段。
+            # Oracle 的 conn.db 是 service name，不能拿来当 OWNER；未填写
+            # schema 时使用连接用户读取当前用户表的字段元数据。
+            meta_schema = schema
+            if not meta_schema:
+                if cfg.db_type == "oracle":
+                    meta_schema = ((conn or {}).get("user") or "").strip()
+                else:
+                    meta_schema = ((conn or {}).get("db") or "").strip()
+            if meta_schema:
                 try:
-                    col_meta = dbc.list_columns(schema, task.source_table) or []
+                    col_meta = dbc.list_columns(meta_schema, task.source_table) or []
                     comments_by_key: dict[str, str] = {
                         (m.get("column_name") or "").lower(): (m.get("column_comment") or "")
                         for m in col_meta
@@ -456,6 +466,22 @@ def start_query(req: StartQueryRequest, db: Session = Depends(get_session)):
                         c = comments_by_key.get(f_out["field_key"].lower(), "")
                         if c:
                             f_out["field_comment"] = c
+                    existing = {f["field_key"].lower() for f in field_list_out}
+                    for meta in col_meta:
+                        field_key = (meta.get("column_name") or "").strip()
+                        if not field_key or field_key.lower() in existing:
+                            continue
+                        field_list_out.append({
+                            "id": None,
+                            "field_key": field_key,
+                            "field_comment": meta.get("column_comment") or "",
+                            "show_default": False,
+                            "ord": len(field_list_out),
+                            "rules": 0,
+                            "rule_list": [],
+                        })
+                        col_names.append(field_key)
+                        existing.add(field_key.lower())
                 except Exception as e:
                     logger.warning(f"[queries] 获取字段注释失败，跳过：{type(e).__name__}: {e}")
     except Exception as e:

web/backend/tests/test_analysis_report.py

+import unittest
+
+from web.backend.routers.analysis import _column_changes, _generate_sql
+
+
+def column(name, position, column_type="VARCHAR2(1)", default=None):
+    return {"column_name": name, "ordinal_position": position,
+            "column_type": column_type, "data_type": "VARCHAR2",
+            "char_max_length": 1, "numeric_precision": None, "numeric_scale": None,
+            "is_nullable": "N", "column_default": default,
+            "column_comment": None, "extra": None}
+
+
+class AnalysisReportDiffTests(unittest.TestCase):
+    def test_removal_does_not_modify_shifted_columns_or_generate_modify_sql(self):
+        old = {c["column_name"]: c for c in
+               [column("FIRST", 1), column("REMOVED", 2), column("ISHJTS", 3)]}
+        current = {c["column_name"]: c for c in
+                   [column("FIRST", 1), column("ISHJTS", 2)]}
+
+        changes = _column_changes(old, current)
+
+        self.assertEqual([(item["column"], item["kind"]) for item in changes],
+                         [("REMOVED", "removed")])
+        sql = _generate_sql({"table_name": "GD_GRDK", "columns": current,
+                             "column_changes": changes}, "oracle")["diff_sql"]
+        self.assertIn('DROP COLUMN "REMOVED"', sql)
+        self.assertNotIn("MODIFY", sql)
+
+    def test_real_property_change_is_reported(self):
+        old = {"ISHJTS": column("ISHJTS", 1, default="'0'")}
+        current = {"ISHJTS": column("ISHJTS", 1, default="'1'")}
+
+        changes = _column_changes(old, current)
+
+        self.assertEqual(changes[0]["kind"], "changed")
+        self.assertEqual(changes[0]["changed_fields"], ["column_default"])
+
+    def test_reorder_of_existing_columns_is_reported(self):
+        old = {"A": column("A", 1), "B": column("B", 2)}
+        current = {"A": column("A", 2), "B": column("B", 1)}
+
+        changes = _column_changes(old, current)
+
+        self.assertEqual([item["changed_fields"] for item in changes],
+                         [["ordinal_position"], ["ordinal_position"]])

web/src/api/backup.js

@@ -7,6 +7,11 @@ export const listOperationRecords = (operation) => http.get('/backup/records', {
 export const importOperationRecords = (operation, records) => http.post(`/backup/records/import?operation=${encodeURIComponent(operation)}`, { records })
 export const createBackup = (sourceId, payload) => http.post(`/data-sources/${sourceId}/backup`, payload)
 export const restoreBackup = (payload) => http.post('/backup/restore', payload)
+export const listBackupTasks = () => http.get('/backup/tasks')
+export const createBackupTask = (payload) => http.post('/backup/tasks', payload)
+export const updateBackupTask = (id, payload) => http.put(`/backup/tasks/${id}`, payload)
+export const deleteBackupTask = (id) => http.delete(`/backup/tasks/${id}`)
+export const runBackupTask = (id) => http.post(`/backup/tasks/${id}/run`)
 export const encryptBackup = (payload) => http.post('/backup/encrypt', payload)
 export const decryptBackup = (payload) => http.post('/backup/decrypt', payload)
 export const encryptData = (payload) => http.post('/backup/data-encrypt', payload)

web/src/api/dataSources.js

@@ -5,3 +5,4 @@ export const createDataSource = (payload) => http.post('/data-sources', payload)
 export const updateDataSource = (id, payload) => http.put(`/data-sources/${id}`, payload)
 export const deleteDataSource = (id) => http.delete(`/data-sources/${id}`)
 export const testDataSource = (id) => http.post(`/data-sources/${id}/test`)
+export const insertSimulatedData = (id, payload) => http.post(`/data-sources/${id}/simulate/insert`, payload)

web/src/layouts/MainLayout.vue

@@ -59,6 +59,7 @@
         </el-menu-item>
         <el-sub-menu index="backup">
           <template #title><el-icon><Download /></el-icon><span>备份</span></template>
+          <el-menu-item index="/backup/tasks"><template #title>任务</template></el-menu-item>
           <el-menu-item index="/backup"><template #title>备份</template></el-menu-item>
           <el-menu-item index="/restore"><template #title>还原</template></el-menu-item>
           <el-menu-item index="/clear-tables"><template #title>清表</template></el-menu-item>
@@ -74,13 +75,17 @@
           <el-menu-item index="/validation"><template #title>数据</template></el-menu-item>
         </el-sub-menu>
         <el-sub-menu index="analysis">
-          <template #title><el-icon><TrendCharts /></el-icon><span>分析</span></template>
+          <template #title><el-icon><TrendCharts /></el-icon><span>监测</span></template>
           <el-menu-item index="/analysis/tasks"><template #title>任务</template></el-menu-item>
           <el-menu-item index="/analysis/reports"><template #title>报告</template></el-menu-item>
         </el-sub-menu>
+        <el-sub-menu index="data">
+          <template #title><el-icon><DataAnalysis /></el-icon><span>数据</span></template>
+          <el-menu-item index="/data/simulation"><template #title>模拟</template></el-menu-item>
+          <el-menu-item index="/config/data-sources"><template #title>数据源</template></el-menu-item>
+        </el-sub-menu>
         <el-sub-menu index="config">
           <template #title><el-icon><Setting /></el-icon><span>配置</span></template>
-          <el-menu-item index="/config/data-sources"><template #title>数据源</template></el-menu-item>
           <el-menu-item index="/config/rule-config"><template #title>规则</template></el-menu-item>
           <el-menu-item index="/config/system"><template #title>配置</template></el-menu-item>
           <el-menu-item index="/config/model"><template #title>模型</template></el-menu-item>
@@ -118,11 +123,12 @@ import { listTaskGroups } from '@/api/taskGroups'
 import { fontScaleOptions, useUiPrefs } from '@/composables/useUiPrefs'
 import brandLogoDark from '@/assets/brand-logo-dark.png'
 import brandLogoLight from '@/assets/brand-logo-light.png'
+import { DataAnalysis } from '@element-plus/icons-vue'
 
 const route = useRoute()
 const router = useRouter()
 const sidebarCollapsed = ref(false)
-const openGroups = ref(['backup', 'crypto', 'config', 'validation', 'analysis'])
+const openGroups = ref(['backup', 'crypto', 'config', 'validation', 'analysis', 'data'])
 
 // ── UI 偏好（主题 / 字号，见 composables/useUiPrefs.js）────
 const { theme, fontScale, currentScaleLabel, themeTip, themeIcon, setFontScale, toggleTheme } = useUiPrefs()

web/src/router/index.js

@@ -5,6 +5,7 @@ import DataQualityView from '@/views/DataQualityView.vue'
 import TaskConfigView from '@/views/TaskConfigView.vue'
 import DataSourceView from '@/views/DataSourceView.vue'
 import OperationView from '@/views/OperationView.vue'
+import BackupTasksView from '@/views/BackupTasksView.vue'
 import ClearTablesView from '@/views/ClearTablesView.vue'
 import SystemConfigView from '@/views/SystemConfigView.vue'
 import ModelConfigView from '@/views/ModelConfigView.vue'
@@ -15,6 +16,7 @@ import HomeView from '@/views/HomeView.vue'
 import AnalysisTasksView from '@/views/AnalysisTasksView.vue'
 import AnalysisReportsView from '@/views/AnalysisReportsView.vue'
 import AnalysisReportDetailView from '@/views/AnalysisReportDetailView.vue'
+import DataSimulationView from '@/views/DataSimulationView.vue'
 
 const routes = [
   {
@@ -47,7 +49,9 @@ const routes = [
       { path: 'config/data-sources', name: 'DataSources', component: DataSourceView, meta: { title: '数据源' } },
       { path: 'config/system', name: 'SystemConfig', component: SystemConfigView, meta: { title: '系统配置' } },
       { path: 'config/model', name: 'ModelConfig', component: ModelConfigView, meta: { title: '模型配置' } },
+      { path: 'data/simulation', name: 'DataSimulation', component: DataSimulationView, meta: { title: '数据模拟' } },
       { path: 'backup', name: 'Backup', component: OperationView, props: { type: 'backup' }, meta: { title: '备份' } },
+      { path: 'backup/tasks', name: 'BackupTasks', component: BackupTasksView, meta: { title: '任务' } },
       { path: 'restore', name: 'Restore', component: OperationView, props: { type: 'restore' }, meta: { title: '还原' } },
       { path: 'clear-tables', name: 'ClearTables', component: ClearTablesView, meta: { title: '清表' } },
       { path: 'encrypt', name: 'Encrypt', component: OperationView, props: { type: 'encrypt' }, meta: { title: '加密' } },

web/src/styles/page-shell.css

@@ -57,6 +57,9 @@
   font-size: var(--font-size-base);
   font-weight: 400;
 }
+.list-shell-page .filter-form .table-settings-item {
+  margin-left: auto;
+}
 
 /* ---- 主操作组（重置 / + 新建 / 危险操作）：按钮长度对齐参考项目（min-width 72，内容自适应） ---- */
 .filter-main-actions .el-form-item__content {

web/src/views/AnalysisReportDetailView.vue

@@ -26,8 +26,7 @@
         <div v-for="change in row.column_changes" :key="change.column" class="change-line">
           <el-tag :type="changeTag(change.kind)" size="small">{{ changeLabel(change.kind) }}</el-tag>
           <b>{{ change.column }}</b>
-          <span v-if="change.kind === 'changed'">{{ change.before?.column_type }} → {{ change.after?.column_type }}
-            <template v-if="commentDiff(change)">，注释：{{ change.before?.column_comment || '（空）' }} → {{ change.after?.column_comment || '（空）' }}</template></span>
+          <span v-if="change.kind === 'changed'">{{ changeDetails(change) }}</span>
           <span v-else-if="change.kind === 'added'">{{ change.after?.column_type }}{{ change.after?.column_comment ? `，注释：${change.after.column_comment}` : '' }}</span>
         </div>
       </div>
@@ -41,7 +40,7 @@
           <tbody>
             <tr v-for="col in columnList" :key="col.column_name">
               <td>{{ col.column_name }}</td><td>{{ col.column_type }}</td>
-              <td>{{ col.is_nullable === 'NO' ? '否' : '是' }}</td>
+              <td>{{ isNotNull(col.is_nullable) ? '否' : '是' }}</td>
               <td>{{ col.column_default ?? '—' }}</td>
               <td>{{ col.column_comment || '—' }}</td>
             </tr>
@@ -97,7 +96,26 @@ const number = (v) => v === null || v === undefined ? '—' : Number(v).toLocale
 const delta = (v) => v === null || v === undefined ? '基线' : `${v > 0 ? '+' : ''}${Number(v).toLocaleString()}`
 const changeLabel = (k) => ({ added: '新增', removed: '删除', changed: '修改' }[k] || k)
 const changeTag = (k) => ({ added: 'success', removed: 'danger', changed: 'warning' }[k] || 'info')
-const commentDiff = (c) => (c.before?.column_comment || '') !== (c.after?.column_comment || '')
+const isNotNull = (value) => ['NO', 'N'].includes(String(value || '').toUpperCase())
+function changeDetails(change) {
+  const before = change.before || {}; const after = change.after || {}
+  const fields = change.changed_fields || []
+  const parts = []
+  const show = (label, key) => parts.push(`${label}：${before[key] ?? '（空）'} → ${after[key] ?? '（空）'}`)
+  if (fields.includes('column_type')) show('类型', 'column_type')
+  else {
+    for (const [key, label] of [['data_type', '数据类型'], ['char_max_length', '字符长度'],
+      ['numeric_precision', '数值精度'], ['numeric_scale', '小数位数']]) {
+      if (fields.includes(key)) show(label, key)
+    }
+  }
+  if (fields.includes('is_nullable')) parts.push(`可空：${isNotNull(before.is_nullable) ? '否' : '是'} → ${isNotNull(after.is_nullable) ? '否' : '是'}`)
+  for (const [key, label] of [['column_default', '默认值'], ['column_comment', '注释'],
+    ['extra', '附加属性'], ['ordinal_position', '顺序']]) {
+    if (fields.includes(key)) show(label, key)
+  }
+  return parts.join('，')
+}
 
 async function loadDetail() {
   if (!Number.isFinite(runId) || !tableName) { error.value = '缺少报告参数'; loading.value = false; return }

web/src/views/DataQualityView.vue

@@ -32,6 +32,13 @@
             @click="onCancel"
           >{{ cancelling ? '中断中…' : '中断' }}</el-button>
         </el-form-item>
+        <el-form-item v-if="resultFields.length" class="table-settings-item">
+          <TableSettings
+            :state="tableSettingsState"
+            @visible="onFieldVisible"
+            @reorder="onFieldReorder"
+          />
+        </el-form-item>
       </el-form>
     </section>
 
@@ -67,6 +74,7 @@
         :rows="rows"
         :append-mode="true"
         :row-key="rowKeyFn"
+        :visible-fields="visibleFieldKeys"
         @explain-issue="onExplainIssue"
       />
     </article>
@@ -87,6 +95,7 @@ import { listTasks } from '@/api/tasks'
 import { startQuery, pullQuery, cancelQuery } from '@/api/queries'
 import { exportToExcel } from '@/utils/excel'
 import ResultTable from '@/components/ResultTable.vue'
+import TableSettings from '@/components/TableSettings.vue'
 import ExplainIssueDialog from '@/components/ExplainIssueDialog.vue'
 
 const route = useRoute()
@@ -149,6 +158,8 @@ function takeSnapshot(typeKey) {
     taskOptions: taskOptions.value,
     rows: rows.value,
     resultFields: resultFields.value,
+    visibleFieldKeys: visibleFieldKeys.value,
+    tableSettingsState: tableSettingsState.value,
     lastResult: lastResult.value || partial,
     // 续跑所需（2026-09-24 流式版）：新 session 会全表重扫，已收到的行靠
     // __row_index > lastRowIndex 去重（旧分页版按 receivedPages 页号去重已删）
@@ -169,6 +180,8 @@ function applySnapshot(snap) {
     taskOptions.value = []
     rows.value = []
     resultFields.value = []
+    visibleFieldKeys.value = []
+    tableSettingsState.value = { columns: [], rowHeight: 40, borderWidth: 1, colPrefs: {} }
     lastResult.value = null
     scannedCount.value = 0
     totalRows.value = 0
@@ -179,6 +192,15 @@ function applySnapshot(snap) {
   taskOptions.value = snap.taskOptions
   rows.value = snap.rows
   resultFields.value = snap.resultFields
+  visibleFieldKeys.value = snap.visibleFieldKeys?.length
+    ? snap.visibleFieldKeys
+    : resultFields.value.filter((f) => (f.rules || 0) > 0 || f.showDefault !== false).map((f) => f.key)
+  tableSettingsState.value = snap.tableSettingsState || {
+    columns: resultFields.value.map((f) => ({ key: f.key, label: f.cn })),
+    rowHeight: 40,
+    borderWidth: 1,
+    colPrefs: Object.fromEntries(resultFields.value.map((f) => [f.key, { visible: visibleFieldKeys.value.includes(f.key) }])),
+  }
   lastResult.value = snap.lastResult
   scannedCount.value = snap.lastResult?.scanned || 0
   // 恢复 in-flight 进度（fire-and-forget 启动续跑）
@@ -237,6 +259,8 @@ const lastResult = ref(null)                            // 最近一次查询结
 const queryError = ref('')
 const rows = ref([])                                    // 喂给 ResultTable 的行
 const resultFields = ref([])
+const visibleFieldKeys = ref([])
+const tableSettingsState = ref({ columns: [], rowHeight: 40, borderWidth: 1, colPrefs: {} })
 
 // 本端刚发过 cancel（onCancel / abortInFlight / unmount）→ 之后 /pull 的 404 静默退出
 // （非本端取消的 404 = 会话被别人取消 / TTL 过期，要给用户提示）
@@ -318,6 +342,10 @@ async function pollLoop(ac, { lastRowIndex = -1 } = {}) {
         queryError.value = `扫描失败：${res.error}`
         try { await cancelQuery(sessionId.value) } catch (_) { /* best-effort 释放 */ }
       } else {
+        // done 只表示「扫描线程结束」，不表示缓冲区已取干：扫描比轮询快时缓冲区
+        // 还有积压（2026-09-29 线上 87187 vs 69478 就是这么丢的）。满额包 → 立即
+        // 续拉；尾包（<PULL_CHUNK_MAX）才说明取干，此时定稿。
+        if (res.bad_rows.length >= PULL_CHUNK_MAX) continue
         // 定稿：bad_rows 用 bad_total（含截断丢弃部分），rows 是实际展示的
         lastResult.value = { scanned: res.scanned, bad_rows: res.bad_total, rows: rows.value }
         if (res.truncated) {
@@ -408,8 +436,21 @@ async function onQuery() {
       comment: f.field_comment || '',
       rules: f.rules || 0,
       ruleList: f.rule_list || [],
-      showDefault: f.show_default !== false,
+      // 有规则的字段始终默认展示，避免历史配置 show_default=false 时把问题字段隐藏；
+      // 未配置规则的源表字段仍按 show_default（后端补入字段默认 false）。
+      showDefault: (f.rules || 0) > 0 || f.show_default !== false,
     }))
+    visibleFieldKeys.value = resultFields.value
+      .filter((f) => (f.rules || 0) > 0 || f.showDefault !== false)
+      .map((f) => f.key)
+    tableSettingsState.value = {
+      columns: resultFields.value.map((f) => ({ key: f.key, label: f.cn })),
+      rowHeight: 40,
+      borderWidth: 1,
+      colPrefs: Object.fromEntries(
+        resultFields.value.map((f) => [f.key, { visible: visibleFieldKeys.value.includes(f.key) }]),
+      ),
+    }
     if (start.total_rows === 0) {
       // COUNT=0：表里没数据
       lastResult.value = { scanned: 0, bad_rows: 0, rows: [] }
@@ -454,6 +495,23 @@ async function onCancel() {
 // 直接塞进 explainContext → 打开 ExplainIssueDialog 自动调 LLM
 const explainDialogOpen = ref(false)
 const explainContext = ref(null)
+function onFieldVisible({ key, visible }) {
+  const next = visibleFieldKeys.value.filter((k) => k !== key)
+  if (visible) next.push(key)
+  // 按表设置当前列顺序重建可见列，勾选字段时不重置用户已调整的顺序。
+  const orderedKeys = tableSettingsState.value.columns.map((c) => c.key)
+  visibleFieldKeys.value = orderedKeys.filter((k) => next.includes(k))
+  tableSettingsState.value.colPrefs[key] = { ...(tableSettingsState.value.colPrefs[key] || {}), visible }
+}
+function onFieldReorder({ from, to }) {
+  const keys = tableSettingsState.value.columns.map((c) => c.key)
+  const a = keys.indexOf(from), b = keys.indexOf(to)
+  if (a < 0 || b < 0) return
+  keys.splice(a, 1); keys.splice(b, 0, from)
+  tableSettingsState.value.columns = keys.map((k) => resultFields.value.find((f) => f.key === k)).filter(Boolean).map((f) => ({ key: f.key, label: f.cn }))
+  visibleFieldKeys.value = keys.filter((k) => visibleFieldKeys.value.includes(k))
+}
+
 function onExplainIssue(payload) {
   explainContext.value = payload
   explainDialogOpen.value = true
@@ -469,7 +527,7 @@ async function onExport() {
     return
   }
   const exportCols = [
-    ...resultFields.value.filter((c) => c.key !== '__reason' && c.showDefault !== false),
+    ...resultFields.value.filter((c) => c.key !== '__reason' && visibleFieldKeys.value.includes(c.key)),
     { key: '__reason', cn: '备注' },
   ]
   if (exportCols.length <= 1) {

web/src/views/RuleLibraryView.vue

@@ -47,7 +47,7 @@
         <el-form-item label="规则名称" required><el-input v-model="form.name" placeholder="如：身份证格式校验" /></el-form-item>
         <el-form-item label="规则类型"><el-radio-group v-model="form.rule_type"><el-radio-button label="regex">正则</el-radio-button><el-radio-button label="number">数值</el-radio-button><el-radio-button label="date">日期</el-radio-button><el-radio-button label="string">字符串</el-radio-button></el-radio-group></el-form-item>
         <el-form-item label="AI生成提示词"><el-input v-model="form.ai_prompt" type="textarea" :rows="3" maxlength="2000" show-word-limit placeholder="如：校验 18 位身份证号" /></el-form-item>
-        <el-form-item label="规则说明" required><el-input v-model="form.description" type="textarea" :rows="4" maxlength="500" show-word-limit placeholder="30-40 字说明，可手动修改" /></el-form-item>
+        <el-form-item label="规则说明" required><el-input v-model="form.description" type="textarea" :rows="4" maxlength="500" show-word-limit placeholder="简洁且意思完整的说明，可手动修改" /></el-form-item>
         <el-form-item :label="form.rule_type === 'regex' ? '正则表达式' : '校验代码'" required>
           <el-input v-model="ruleContent" type="textarea" :rows="form.rule_type === 'regex' ? 2 : 8" :placeholder="form.rule_type === 'regex' ? '^.+$' : 'def check(value):\n    return True'" />
           <div class="expression-actions">