precis-cli 0.1.3__py3-none-any.whl

app/cli/shell/commands/ai/generate.py

@@ -0,0 +1,285 @@
+# SPDX-License-Identifier: Apache-2.0
+#
+# Copyright 2026 Precis Team
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# backend/app/cli/shell/commands/ai/generate.py
+"""
+@fileoverview CLI AI 配置生成命令模块
+
+功能概述:
+- 提供 `ai generate` 子命令，从数据文件生成 Precis V2 配置
+- 支持 Agent 多轮优化模式（默认）和单次快速生成模式
+- 默认仅输出 YAML 预览，加 --apply 后写盘
+- 无文件参数时自动扫描项目 data/ 目录并交互式选择
+
+架构设计:
+- AIGenerateCommand 继承 Command 基类
+- 直接调用 ConfigGenerationService，不依赖后端 HTTP
+- 写盘与数据文件扫描逻辑委托 shared_services.generation_ops（CLI/TUI 同源）
+- _scan_data_files 保留为单参数向后兼容入口，等价 scan_data_files([], project_path)
+
+输入示例:
+    precis> ai generate data/users.xlsx
+    precis> ai generate data/*.xlsx --apply --max-iterations 3
+
+输出示例:
+    YAML 预览或写盘成功提示
+"""
+
+from __future__ import annotations
+
+import asyncio
+import glob
+import logging
+import os
+from pathlib import Path
+from typing import Any
+
+from app.cli.shared_services.generation_ops import (
+    SUPPORTED_EXTENSIONS,
+    apply_generated_config,
+    scan_data_files,
+)
+from app.cli.shell.commands.base import Command, CommandResult, ProjectContext
+from app.cli.shell.formatter import Formatter
+from app.shared.services.llm.generation import (
+    ConfigGenerationService,
+    GenerationOptions,
+    ProfilingOptions,
+)
+
+logger = logging.getLogger(__name__)
+
+
+def _scan_data_files(project_path: str) -> list[str]:
+    """扫描项目 data/ 目录下的支持文件（向后兼容入口）。
+
+    真正的实现已迁移到 shared_services.generation_ops.scan_data_files（CLI/TUI 同源）。
+    本函数保留单参数签名以兼容历史调用与现有测试，等价于 scan_data_files([], project_path)。
+
+    Args:
+        project_path: 项目根目录
+
+    Returns:
+        绝对路径列表
+    """
+    return scan_data_files([], project_path)
+
+
+class AIGenerateCommand(Command):
+    """AI 配置生成命令。
+
+    从数据文件分析并生成 Schema、Constraint、Regex 配置。
+    默认输出 YAML 预览，使用 --apply 写入项目目录。
+    """
+
+    def __init__(self) -> None:
+        super().__init__("generate", aliases=["gen"])
+
+    @property
+    def description(self) -> str:
+        return "从数据文件生成 Precis V2 配置"
+
+    @property
+    def usage(self) -> str:
+        return "ai generate [files...] [--apply] [--no-agent-mode] [--max-iterations N] [--sample-rows N]"
+
+    def execute(self, args: list[str], context: ProjectContext) -> CommandResult:
+        """执行配置生成命令。
+
+        Args:
+            args: 命令参数列表
+            context: 项目上下文
+
+        Returns:
+            命令执行结果
+        """
+        if not context.is_project_open:
+            return CommandResult.error("未打开项目，请先使用 'open <path>' 命令打开项目")
+
+        project_path = context.project_path
+        if not project_path:
+            return CommandResult.error("项目路径为空")
+
+        # 解析参数
+        file_patterns: list[str] = []
+        apply = False
+        agent_mode = True
+        max_iterations = 2
+        sample_rows = 100
+        sample_values = 100
+        generate_regex = False
+
+        i = 0
+        while i < len(args):
+            arg = args[i]
+            if arg == "--apply":
+                apply = True
+            elif arg == "--no-agent-mode":
+                agent_mode = False
+            elif arg == "--max-iterations":
+                i += 1
+                if i < len(args):
+                    try:
+                        max_iterations = max(1, min(5, int(args[i])))
+                    except ValueError:
+                        return CommandResult.error(f"--max-iterations 需要整数， got: {args[i]}")
+                else:
+                    return CommandResult.error("--max-iterations 需要参数")
+            elif arg == "--sample-rows":
+                i += 1
+                if i < len(args):
+                    try:
+                        sample_rows = max(1, int(args[i]))
+                    except ValueError:
+                        return CommandResult.error(f"--sample-rows 需要整数， got: {args[i]}")
+                else:
+                    return CommandResult.error("--sample-rows 需要参数")
+            elif arg == "--sample-values":
+                i += 1
+                if i < len(args):
+                    try:
+                        sample_values = max(1, int(args[i]))
+                    except ValueError:
+                        return CommandResult.error(f"--sample-values 需要整数， got: {args[i]}")
+                else:
+                    return CommandResult.error("--sample-values 需要参数")
+            elif arg == "--generate-regex":
+                generate_regex = True
+            elif arg.startswith("--"):
+                return CommandResult.error(f"未知参数: {arg}\n用法: {self.usage}")
+            else:
+                file_patterns.append(arg)
+            i += 1
+
+        # 展开文件路径（支持通配符和相对路径）
+        file_paths: list[str] = []
+        for pattern in file_patterns:
+            if not os.path.isabs(pattern):
+                pattern = os.path.join(project_path, pattern)
+            matched = glob.glob(pattern)
+            if matched:
+                file_paths.extend(matched)
+            elif os.path.exists(pattern):
+                file_paths.append(pattern)
+
+        # 无文件参数时扫描 data/ 目录（委托 shared_services 双参数版本）
+        if not file_paths:
+            scanned = scan_data_files([], project_path)
+            if not scanned:
+                return CommandResult.error(
+                    "未找到数据文件。请在项目 data/ 目录放置 .xlsx/.csv/.json 文件，或显式指定文件路径。"
+                )
+            file_paths = scanned
+
+        # 过滤支持的文件类型
+        file_paths = [p for p in file_paths if p.lower().endswith(SUPPORTED_EXTENSIONS)]
+        if not file_paths:
+            return CommandResult.error("未找到支持的数据文件类型（.xlsx/.xls/.csv/.json/.jsonl）")
+
+        # 获取项目信息
+        config = context.project_config or {}
+        project_name = config.get("project", {}).get("name", Path(project_path).name)
+        project_id = config.get("project", {}).get("id", project_name)
+
+        # 输出待处理文件
+        print(Formatter.header("\nAI 配置生成"))
+        print(Formatter.info(f"项目: {project_name}"))
+        print(Formatter.info(f"Agent 模式: {'开启' if agent_mode else '关闭'}"))
+        print(Formatter.info(f"迭代次数: {max_iterations}"))
+        print(Formatter.info("数据文件:"))
+        for p in file_paths:
+            print(f"  - {os.path.relpath(p, project_path)}")
+
+        def progress_callback(stage: str, progress: float, extra: dict[str, Any] | None = None) -> None:
+            """终端进度回调。"""
+            msg = extra.get("message") if extra else None
+            prefix = f"[{stage}]" if not msg else f"[{stage}] {msg}"
+            print(Formatter.info(f"  {prefix} {progress:.0f}%"))
+
+        service = ConfigGenerationService()
+        profiling_opts = ProfilingOptions(
+            sample_rows=sample_rows,
+            sample_values_per_column=sample_values,
+        )
+        gen_opts = GenerationOptions(
+            generate_schemas=True,
+            generate_constraints=True,
+            generate_regex_nodes=generate_regex,
+            keep_existing=True,
+        )
+
+        # CLI 是同步环境：按模式构造一次协程，asyncio.run 创建一次性事件循环
+        # 并在结束时确保关闭（不再 get_running_loop/new_event_loop 泄漏循环）
+        if agent_mode:
+            coro = service.generate_with_agent(
+                file_paths=file_paths,
+                project_name=project_name,
+                project_id=project_id,
+                config_path=project_path,
+                profiling_options=profiling_opts,
+                generation_options=gen_opts,
+                max_iterations=max_iterations,
+                progress_callback=progress_callback,
+            )
+        else:
+            coro = service.generate(
+                file_paths=file_paths,
+                project_name=project_name,
+                project_id=project_id,
+                config_path=project_path,
+                profiling_options=profiling_opts,
+                generation_options=gen_opts,
+                progress_callback=lambda stage, progress: progress_callback(stage, progress, None),
+            )
+        try:
+            result = asyncio.run(coro)
+        except Exception as e:
+            logger.error(f"配置生成失败: {e}", exc_info=True)
+            return CommandResult.error(f"配置生成失败: {e}")
+
+        if not result.get("success"):
+            error = result.get("error") or "配置生成失败"
+            return CommandResult.error(error)
+
+        yaml_preview = result.get("yaml_preview", "")
+        warnings = result.get("warnings", [])
+
+        if warnings:
+            print(Formatter.warning("\n生成警告:"))
+            for w in warnings:
+                print(f"  - {w}")
+
+        if apply:
+            try:
+                apply_generated_config(result, project_path)
+            except Exception as e:
+                logger.error(f"写盘失败: {e}", exc_info=True)
+                return CommandResult.error(f"配置已生成，但写盘失败: {e}")
+
+            schema_count = len(result.get("schemas", {}))
+            constraint_count = len(result.get("constraints", {}))
+            regex_count = len(result.get("regex_nodes", {}))
+            return CommandResult.ok(
+                f"配置已生成并写入项目：{schema_count} 个 schema，"
+                f"{constraint_count} 个 constraint，{regex_count} 个 regex。"
+            )
+
+        # 默认仅输出预览
+        print(Formatter.header("\nYAML 预览（未写盘，加 --apply 可应用）"))
+        print(yaml_preview)
+        return CommandResult.ok("配置生成完成，以上为 YAML 预览。")
+
+
+__all__ = ["AIGenerateCommand"]

app/cli/shell/commands/ai/interaction.py

@@ -0,0 +1,262 @@
+# SPDX-License-Identifier: Apache-2.0
+#
+# Copyright 2026 Precis Team
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# backend/app/cli/shell/commands/ai/interaction.py
+"""
+@fileoverview AI 交互辅助模块
+
+功能概述:
+- 在执行修改操作前向用户展示操作列表并确认
+- 从项目 schema 文件中提取表格和列信息作为上下文，供 AI 理解项目结构
+- 提供 Spinner 动画辅助函数（创建与停止）
+
+架构设计:
+- confirm_actions(): 展示 AI 计划执行的操作列表，等待用户输入 y/yes 确认
+- build_context_data(): 读取项目 schema 目录下的 YAML 文件，构建 AI 上下文
+- create_spinner()/stop_spinner(): 简单的终端加载动画线程控制
+
+输入示例:
+    confirm_actions(actions, "AI 回复说明")
+    build_context_data("用户消息", context)
+
+输出示例:
+    True/False（用户是否确认执行）
+    {"message": "...", "context": {...}, "projectOverview": {...}}
+"""
+
+from __future__ import annotations
+
+import logging
+import threading
+import time
+from pathlib import Path
+from typing import Any
+
+from app.cli.shell.formatter import Colors, Formatter
+from app.shared.services.ai.utils import get_project_overview
+
+logger = logging.getLogger(__name__)
+
+
+def confirm_actions(actions: list[dict[str, Any]], reply: str) -> bool:
+    """显示操作确认提示，等待用户确认。
+
+    在执行修改操作前，向用户展示 AI 计划执行的操作列表，
+    等待用户输入 y/yes 确认后才继续。
+
+    Args:
+        actions: 动作列表
+        reply: AI 的回复说明
+
+    Returns:
+        用户是否确认执行
+    """
+    print(Formatter.info("\n即将执行以下操作:"))
+    print("-" * 40)
+
+    # 显示 AI 的说明
+    if reply:
+        print(Formatter.info(f"说明: {reply}"))
+        print("-" * 40)
+
+    # 显示每个动作详情
+    for i, action in enumerate(actions, 1):
+        action_type = action.get("actionType", "UNKNOWN")
+
+        # 处理 VALIDATE_PROJECT 特殊显示
+        if action_type == "VALIDATE_PROJECT":
+            spec = action.get("constraintSpec", {})
+            table_name = spec.get("tableName", spec.get("targetNodeId", "所有表"))
+            tables = spec.get("tables") or spec.get("tableIds")
+
+            if tables:
+                if len(tables) == 1:
+                    display = f"校验表: {tables[0]}"
+                else:
+                    display = f"校验 {len(tables)} 张表: {', '.join(tables)}"
+            elif table_name and table_name != "所有表":
+                display = f"校验表: {table_name}"
+            else:
+                display = "校验所有表"
+
+            print(f"  {i}. {display}")
+            continue
+
+        # 通用动作描述映射
+        action_desc = {
+            "ADD_TO_CANVAS": "显示到画布（只读）",
+            "ADD_CONSTRAINT_NODE": "添加约束",
+            "UPDATE_CONSTRAINT_NODE": "更新约束",
+            "DELETE_CONSTRAINT_NODE": "删除约束",
+            "ADD_SCHEMA": "创建表",
+            "UPDATE_SCHEMA": "修改表结构",
+            "DELETE_SCHEMA": "删除表",
+            "ADD_REGEX": "创建正则校验",
+            "UPDATE_REGEX": "更新正则校验",
+            "DELETE_REGEX": "删除正则校验",
+            "ADD_TRANSFORM": "创建数据转换",
+            "UPDATE_TRANSFORM": "更新数据转换",
+            "DELETE_TRANSFORM": "删除数据转换",
+            "UPDATE_SETTINGS": "修改项目设置",
+        }.get(action_type, action_type)
+
+        # 根据动作类型提取展示信息
+        if action_type == "ADD_TO_CANVAS":
+            spec = action.get("canvasSpec", {})
+            resource_kind = spec.get("resourceKind", "未知")
+            resource_name = spec.get("name", spec.get("resourceId", "未知"))
+            print(f"  {i}. {action_desc}: {resource_kind} / {resource_name}")
+        elif action_type in ("ADD_CONSTRAINT_NODE", "UPDATE_CONSTRAINT_NODE", "DELETE_CONSTRAINT_NODE"):
+            spec = action.get("constraintSpec", {})
+            constraint_type = spec.get("type", "Unknown")
+            table_name = spec.get("tableName", spec.get("targetNodeId", "未知"))
+            column_name = spec.get("targetColumn", spec.get("targetColumnId", "未知"))
+            print(f"  {i}. {action_desc}")
+            print(f"     表: {table_name}, 字段: {column_name}, 类型: {constraint_type}")
+        elif action_type in ("ADD_SCHEMA", "UPDATE_SCHEMA", "DELETE_SCHEMA"):
+            spec = action.get("schemaSpec", {})
+            name = spec.get("name", spec.get("schemaId", "未知"))
+            columns = spec.get("columns", [])
+            print(f"  {i}. {action_desc}: {name}")
+            if columns:
+                col_names = [c.get("name", "?") for c in columns]
+                print(f"     列: {', '.join(col_names)}")
+        elif action_type in ("ADD_REGEX", "UPDATE_REGEX", "DELETE_REGEX"):
+            spec = action.get("regexSpec", {})
+            name = spec.get("name", spec.get("regexId", "未知"))
+            print(f"  {i}. {action_desc}: {name}")
+        elif action_type in ("ADD_TRANSFORM", "UPDATE_TRANSFORM", "DELETE_TRANSFORM"):
+            spec = action.get("transformSpec", {})
+            t_type = spec.get("type", "未知")
+            print(f"  {i}. {action_desc}: {t_type}")
+        elif action_type == "UPDATE_SETTINGS":
+            spec = action.get("settingsSpec", {})
+            category = spec.get("category", "未知")
+            settings = spec.get("settings", {})
+            print(f"  {i}. {action_desc}: {category}")
+            print(f"     设置: {settings}")
+        else:
+            print(f"  {i}. {action_desc}")
+
+    print("-" * 40)
+
+    # 等待用户确认
+    try:
+        confirm = input(Formatter.warning("确认执行? (y/N): ")).strip().lower()
+        return confirm in ("y", "yes")
+    except (KeyboardInterrupt, EOFError):
+        print()
+        return False
+
+
+def build_context_data(message: str, context: Any) -> dict[str, Any]:
+    """构建上下文数据。
+
+    从项目配置中提取表格和列信息作为上下文，供 AI 理解项目结构。
+
+    Args:
+        message: 用户消息
+        context: 命令上下文
+
+    Returns:
+        上下文数据字典，包含 message、context 和 projectOverview
+    """
+    project_path = context.project_path
+    selected_nodes = []
+
+    # 尝试从 schema 文件中提取表格信息
+    if project_path:
+        schemas_dir = Path(project_path) / "schemas"
+        if schemas_dir.exists():
+            for schema_file in schemas_dir.glob("*.yaml"):
+                try:
+                    import yaml
+
+                    with open(schema_file, encoding="utf-8") as f:
+                        schema_data = yaml.safe_load(f) or {}
+
+                    # Schema 文件直接包含 columns 列表
+                    table_name = schema_data.get("name", "")
+                    table_id = schema_data.get("id", table_name)
+
+                    columns = schema_data.get("columns", [])
+                    column_info = []
+                    for c in columns:
+                        col_name = c.get("name", "")
+                        col_id = c.get("id", col_name)
+                        column_info.append({"id": col_id, "name": col_name})
+
+                    if table_name or table_id:
+                        # 优先使用 schema 文件中定义的 id
+                        node_id = table_id
+                        selected_nodes.append(
+                            {
+                                "id": node_id,
+                                "type": "schema",
+                                "data": {
+                                    "label": table_name or table_id,
+                                    "columns": column_info,
+                                },
+                            }
+                        )
+                except Exception as e:
+                    logger.debug(f"读取 schema 文件失败 {schema_file}: {e}")
+
+    # 获取项目概览（表结构和约束）
+    project_overview = get_project_overview(project_path)
+
+    return {
+        "message": message,
+        "context": {
+            "hasContext": len(selected_nodes) > 0,
+            "selectedNodes": selected_nodes,
+        },
+        "projectOverview": project_overview,
+    }
+
+
+def create_spinner(message: str = "AI> ") -> tuple[threading.Event, threading.Thread]:
+    """创建并启动一个 spinner 线程。
+
+    Args:
+        message: spinner 前显示的消息前缀
+
+    Returns:
+        tuple: (stop_event, spinner_thread)
+    """
+    stop_event = threading.Event()
+
+    def spinner() -> None:
+        chars = ["⠋", "⠙", "⠹", "⠸", "⠼", "⠴", "⠦", "⠧", "⠇", "⠏"]
+        i = 0
+        while not stop_event.is_set():
+            print(f"\r{Formatter.colorize(message, Colors.CYAN)}{chars[i % len(chars)]}", end="", flush=True)
+            time.sleep(0.1)
+            i += 1
+
+    thread = threading.Thread(target=spinner)
+    thread.start()
+    return stop_event, thread
+
+
+def stop_spinner(stop_event: threading.Event, thread: threading.Thread) -> None:
+    """停止 spinner 线程。
+
+    Args:
+        stop_event: 停止事件
+        thread: spinner 线程
+    """
+    stop_event.set()
+    thread.join()