vmysql-cli 0.1.0__py3-none-any.whl

vmysql_cli-0.1.0.dist-info/METADATA

@@ -0,0 +1,24 @@
+Metadata-Version: 2.4
+Name: vmysql_cli
+Version: 0.1.0
+Summary: 自然语言操作MySQL——输入中文即可查询、插入、修改数据库
+Author: vvv
+License: MIT
+Project-URL: Homepage, https://github.com/你的用户名/vmysql-cli
+Project-URL: Issues, https://github.com/你的用户名/vmysql-cli/issues
+Keywords: mysql,nl2sql,cli,ai,deepseek
+Classifier: Development Status :: 3 - Alpha
+Classifier: Intended Audience :: Developers
+Classifier: License :: OSI Approved :: MIT License
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.9
+Classifier: Programming Language :: Python :: 3.10
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Topic :: Database
+Requires-Python: >=3.9
+Description-Content-Type: text/markdown
+Requires-Dist: sqlalchemy>=2.0
+Requires-Dist: pandas>=2.0
+Requires-Dist: openai>=1.0
+Requires-Dist: pymysql>=1.0

vmysql_cli-0.1.0.dist-info/RECORD

@@ -0,0 +1,6 @@
+vmysql_cli.py,sha256=KSPqSmHFc9niTA_-Da9vSApuSKLqe5VxwIMG9GfpQZc,8268
+vmysql_cli-0.1.0.dist-info/METADATA,sha256=rDI0KlbETJdTaUrlPPptygeoR0SKElPuN9sac0kZ_j4,971
+vmysql_cli-0.1.0.dist-info/WHEEL,sha256=aeYiig01lYGDzBgS8HxWXOg3uV61G9ijOsup-k9o1sk,91
+vmysql_cli-0.1.0.dist-info/entry_points.txt,sha256=X3xiEh4KENcYcCntuYaRYN-RsiX98OGy4GZvK_2L-jk,47
+vmysql_cli-0.1.0.dist-info/top_level.txt,sha256=NQdlomBkpsyh2BtlpDVv84Fj2S0uNQNBtUoTR6M_B-Y,11
+vmysql_cli-0.1.0.dist-info/RECORD,,

vmysql_cli-0.1.0.dist-info/WHEEL

@@ -0,0 +1,5 @@
+Wheel-Version: 1.0
+Generator: setuptools (82.0.1)
+Root-Is-Purelib: true
+Tag: py3-none-any
+

vmysql_cli-0.1.0.dist-info/entry_points.txt

@@ -0,0 +1,2 @@
+[console_scripts]
+vmysql_cli = vmysql_cli:main

vmysql_cli-0.1.0.dist-info/top_level.txt

@@ -0,0 +1 @@
+vmysql_cli

vmysql_cli.py

@@ -0,0 +1,214 @@
+import time
+import json
+from pathlib import Path
+import sqlalchemy as a
+import pandas as b
+from openai import OpenAI
+
+ds_client = None
+DB_URL = None
+CONFIG_FILE = Path.home() / ".vmysql_cli_config.json"
+
+def _load_config():
+    """加载配置：环境变量 > 本地配置文件 > 交互式输入"""
+    global ds_client, DB_URL
+    config = {}
+
+    # 1. 尝试从配置文件加载
+    if CONFIG_FILE.exists():
+        try:
+            config = json.loads(CONFIG_FILE.read_text(encoding="utf-8"))
+        except Exception:
+            pass
+
+    # 2. 缺失的配置交互式询问
+    prompted = False
+    if "api_key" not in config:
+        config["api_key"] = input("请输入 DeepSeek API Key: ").strip()
+        prompted = True
+    if "base_url" not in config:
+        print("API 地址示例：")
+        print("  DeepSeek → https://api.deepseek.com/v1")
+        print("  Ollama   → http://localhost:11434/v1")
+        config["base_url"] = input("请输入 API Base URL: ").strip()
+        prompted = True
+    # 3. 缺失的数据库配置交互式询问
+    if "db_url" not in config:
+        host = input("数据库 IP 地址 (默认 127.0.0.1): ").strip() or "127.0.0.1"
+        port = input("端口 (默认 3306): ").strip() or "3306"
+        user = input("用户名 (默认 root): ").strip() or "root"
+        pwd = input("密码: ").strip()
+        config["db_url"] = f"mysql+pymysql://{user}:{pwd}@{host}:{port}/db0?charset=utf8"
+        prompted = True
+
+    # 3. 保存配置文件（仅在首次交互输入时）
+    if prompted:
+        try:
+            CONFIG_FILE.write_text(json.dumps(config, ensure_ascii=False, indent=2), encoding="utf-8")
+            print(f"配置已保存到 {CONFIG_FILE}")
+        except Exception:
+            pass
+        pass
+
+    ds_client = OpenAI(api_key=config["api_key"], base_url=config["base_url"])
+    DB_URL = config["db_url"]
+
+_engine = None
+_schema_cache = None
+_preview_cache = None
+_sql_cache = {}
+
+def _get_engine():
+    global _engine
+    if _engine is None:
+        _engine = a.create_engine(DB_URL)
+    return _engine
+
+def get_database_schema():
+    global _schema_cache
+    if _schema_cache is not None:
+        return _schema_cache
+    engine = _get_engine()
+    
+    # 2. 查询数据库所有表
+    tables_df = b.read_sql("SHOW TABLES", engine)
+    table_names = tables_df.iloc[:, 0].tolist()  # 提取所有表名
+    
+    # 3. 遍历每张表，查询字段结构，并拼接成文本
+    lines = ["【当前数据库完整结构】\n"]
+    for table in table_names:
+        lines.append(f"\n表名：{table}\n")
+        struct_df = b.read_sql(f"SHOW FULL COLUMNS FROM `{table}`", engine)
+        for _, row in struct_df.iterrows():
+            extras = []
+            if row['Null'] == 'NO':
+                extras.append('NOT NULL')
+            if row['Key'] == 'PRI':
+                extras.append('PRIMARY KEY')
+            if 'auto_increment' in str(row['Extra']).lower():
+                extras.append('AUTO_INCREMENT')
+            if row['Default'] is not None:
+                extras.append(f'DEFAULT={row["Default"]}')
+            extra_str = ' [' + ' '.join(extras) + ']' if extras else ''
+            lines.append(f"- {row['Field']} ({row['Type']}){extra_str}\n")
+    _schema_cache = ''.join(lines)
+    return _schema_cache
+def get_data_preview():
+    """查每张表的前20条数据，让AI看到实际内容（有缓存）"""
+    global _preview_cache
+    if _preview_cache is not None:
+        return _preview_cache
+    try:
+        tables = list(b.read_sql("SHOW TABLES", _get_engine()).iloc[:, 0])
+    except Exception:
+        return ""
+    preview = "\n【当前各表数据预览（最多20行）】\n"
+    for t in tables:
+        try:
+            df = b.read_sql(f"SELECT * FROM `{t}` LIMIT 20", _get_engine())
+            preview += f"\n表 {t} ({len(df)} 行):\n{df.to_string(index=False)}\n"
+        except Exception:
+            pass
+    _preview_cache = preview
+    return preview
+
+_prompt_prefix_cache = None
+
+def _get_prompt_prefix():
+    """缓存 prompt 前缀，避免每次重建大片文本"""
+    global _prompt_prefix_cache
+    if _prompt_prefix_cache is not None:
+        return _prompt_prefix_cache
+    _prompt_prefix_cache = f'''
+我现在的数据库真实结构如下：
+{get_database_schema()}
+{get_data_preview()}
+你是MySQL数据库，数据库名为db0。上面列出的就是全部可用的表和字段，没有其他表。
+请结合数据预览理解用户意图（如"无意义的数据""异常数据"等需根据实际数据判断）。
+
+请严格遵守规则：
+1. 只输出长度为2的Python列表，无多余文字
+2. 第1项：SQL语句，必须使用:xxx命名占位符
+3. 第2项：Python字典，作为查询参数，不是字符串
+4. 允许SELECT/INSERT/UPDATE/DELETE/DROP/CREATE，DELETE和UPDATE必须带WHERE
+5. 操作已有表时只能使用上面列出的表名和字段名，CREATE TABLE可以定义新字段
+6. 禁止使用information_schema、sys、mysql等系统库
+7. INSERT时永远不要包含自增主键列(如id)，让数据库自动生成
+
+示例1：["select * from student where id = :id", {{"id": 1}}]
+示例2：["select * from student", {{}}]
+问"有哪些表"示例3：["show tables", {{}}]
+插入示例：["insert into student(name,age) values(:n,:a)", {{"n":"张三","a":20}}]
+批量插入：["insert into student(name,age) values(:n1,:a1),(:n2,:a2)", {{"n1":"张三","a1":20,"n2":"李四","a2":22}}]
+    '''.strip()
+    return _prompt_prefix_cache
+
+def get_nl_to_sql(user_input):
+    if user_input in _sql_cache:
+        return _sql_cache[user_input], 0.0, 0.0
+    t0 = time.perf_counter()
+    prompt = _get_prompt_prefix()
+    t1 = time.perf_counter()
+
+    res = ds_client.chat.completions.create(
+        model="deepseek-chat",
+        messages=[{"role":"system","content":prompt},{"role":"user","content":user_input}],
+        temperature=0
+    )
+    raw = res.choices[0].message.content.strip()
+    t2 = time.perf_counter()
+    result = eval(raw)
+    _sql_cache[user_input] = result
+    return result, t1 - t0, t2 - t1
+def _fill_sql(sql, params):
+    """把 :xxx 占位符替换为实际参数值，按key长度降序避免 :n 错误替换 :n1"""
+    result = sql
+    for k in sorted(params.keys(), key=len, reverse=True):
+        v = params[k]
+        if v is None:
+            result = result.replace(f":{k}", "NULL")
+        elif isinstance(v, (int, float)):
+            result = result.replace(f":{k}", str(v))
+        else:
+            result = result.replace(f":{k}", f"'{v}'")
+    return result
+
+def sql(c):
+    c = c.strip()
+    # 本地处理表结构查询，不走AI（仅精确匹配）
+    if c in ('表结构', '有哪些表', '有什么表', '表有哪些', '表', '库', '结构', 'show tables'):
+        print("[耗时] Schema: 0.00s | AI: 0.00s | 执行: 0.00s (本地处理)")
+        if c == '表':
+            tables = b.read_sql("SHOW TABLES", _get_engine()).iloc[:, 0].tolist()
+            return '\n'.join(tables)
+        schema = _schema_cache or get_database_schema()
+        return schema
+    d, t_schema, t_ai = get_nl_to_sql(c)
+    t1 = time.perf_counter()
+    sql_upper = d[0].strip().upper()
+    if sql_upper.startswith(('INSERT', 'UPDATE', 'DELETE', 'DROP', 'CREATE', 'ALTER')):
+        with _get_engine().connect() as conn:
+            r = conn.execute(a.text(d[0]), d[1])
+            conn.commit()
+        t_exec = time.perf_counter() - t1
+        print(f"[耗时] Schema: {t_schema:.2f}s | AI: {t_ai:.2f}s | 执行: {t_exec:.2f}s")
+        print(f"[SQL] {_fill_sql(d[0], d[1])}")
+        return f"影响了 {r.rowcount} 行"
+    result = b.read_sql(
+        sql=a.text(d[0]),
+        con=_get_engine(),
+        params=d[1]
+    )
+    t_exec = time.perf_counter() - t1
+    print(f"[耗时] Schema: {t_schema:.2f}s | AI: {t_ai:.2f}s | 执行: {t_exec:.2f}s")
+    print(f"[SQL] {_fill_sql(d[0], d[1])}")
+    return result
+def main():
+    _load_config()
+    while 1:
+        result = sql(input(""))
+        print(result)
+
+if __name__ == "__main__":
+    main()
+