• 数据库：Hive转Presto（三）

    继续上节代码。

    1. import re
    2. import os
    3. import tkinter.filedialog
    4. from tkinter import *
    5.  
    6.  
    7. class Hive2Presto:
    8.     def __int__(self):
    9.         self.t_funcs = ['substr', 'nvl', 'substring', 'unix_timestamp'] + \
    10.                        ['to_date', 'concat', 'sum', 'avg', 'abs', 'year', 'month', 'ceiling', 'floor']
    11.         self.time_funcs = ['date_add', 'datediff', 'add_months']
    12.         self.funcs = self.t_funcs + self.time_funcs
    13.         self.current_path = os.path.abspath(__file__)
    14.         self.dir = os.path.dirname(self.current_path)
    15.         self.result = []
    16.         self.error = []
    17.         self.filename = ''
    18.  
    19.     def main(self):
    20.         self.root = Tk()
    21.         self.root.config(bg='#ff741d')  # 背景颜色设置为公司主题色^_^
    22.         self.root.title('Hive转Presto')
    23.         self.win_width = 550
    24.         self.win_height = 500
    25.         self.screen_width = self.root.winfo_screenwidth()
    26.         self.screen_height = self.root.winfo_screenheight()
    27.         self.x = (self.screen_width - self.win_width) // 2
    28.         self.y = (self.screen_height - self.win_height) // 2
    29.         self.root.geometry(f'{self.win_width}x{self.win_height}+{self.x}+{self.y}')
    30.  
    31.         font = ('楷体', 11)
    32.  
    33.         self.button = Button(self.root, text='转换', command=self.trans, bg='#ffcc8c', font=font, anchor='e')
    34.         self.button.grid(row=0, column=0, padx=100, pady=10, sticky=W)
    35.  
    36.         self.file_button = Button(self.root, text='选择文件', command=self.choose_file, bg='#ffcc8c', font=font,
    37.                                   anchor='e')
    38.         self.file_button.grid(row=0, column=1, padx=0, pady=10, sticky=W)
    39.  
    40.         self.entry = Entry(self.root, width=65, font=font)
    41.         self.entry.insert(0, '输入Hive代码')
    42.         self.entry.grid(row=1, column=0, padx=10, pady=10, columnspan=2)
    43.         self.entry.bind('', self.delete_text)
    44.  
    45.         self.text = Text(self.root, width=75, height=20)
    46.         self.text.grid(row=2, column=0, padx=10, pady=10, columnspan=2)
    47.  
    48.         self.des_label = Label(self.root, text='可以复制结果，也有生成的文件，与选取的文件同文件夹', bg='#ffcc8c',
    49.                                font=('楷体', 10))
    50.         self.des_label.grid(row=3, column=0, padx=10, pady=10, columnspan=2)
    51.  
    52.         s = ''
    53.         for i in range(0, (n := len(self.funcs)), 4):
    54.             if i + 4 <= n:
    55.                 s += ','.join(self.funcs[i:i + 4]) + '\n'
    56.             else:
    57.                 s += ','.join(self.funcs[i:]) + '\n'
    58.         s = s[:-1]
    59.         self.des_label1 = Label(self.root, text=s, bg='#ffcc8c',
    60.                                 font=('楷体', 10))
    61.         self.des_label1.grid(row=4, column=0, padx=10, pady=10, columnspan=2)
    62.  
    63.         self.root.columnconfigure(0, minsize=10)
    64.         self.root.columnconfigure(1, minsize=10)
    65.         self.root.columnconfigure(0, pad=5)
    66.  
    67.         self.root.mainloop()
    68.  
    69.     def replace_func(self, s, res):
    70.         """
    71.         把搜索到函数整体取出来，处理括号中的参数
    72.         :param s:
    73.         :param res:
    74.         :return:
    75.         """
    76.         for f in res:
    77.             f1 = f.replace('\n', '').strip()
    78.             f1 = re.sub(r'(\s*)', '(', f1)
    79.             # 搜索括号里的字符串
    80.             if re.findall(r'(\w*)\(', f1):
    81.                 func_name = re.findall(r'(\w*)\(', f1)[0].strip()
    82.             else:
    83.                 continue
    84.             try:
    85.                 if 'date_add' == func_name.lower():
    86.                     date, date_num = self.extact_func(f1, func_name)
    87.                     s_n = f"date_add('day',{date_num},cast(substr(cast{date} as varchar,1,10) as date))"
    88.                     s = s.replace(f, s_n)
    89.                 elif 'datediff' == func_name.lower():
    90.                     date1, date2 = self.extact_func(f1, func_name)
    91.                     s_n = f"date_add('day',{date2},cast(substr(cast{date} as varchar,1,10) as date),cast(substr(cast{date1} as varchar),1,10) as date))"
    92.                     s = s.replace(f, s_n)
    93.                 elif 'nvl' == func_name.lower():
    94.                     s1, s2 = self.extact_func(f1, func_name)
    95.                     s_n = f"coalesce({s1},{s2})"
    96.                     s = s.replace(f, s_n)
    97.                 elif 'substr' == func_name.lower():
    98.                     date, start, end = self.extact_func(f1, func_name)
    99.                     s_n = f"substr(cast({date} as varchar),{start},{end}"
    100.                     s = s.replace(f, s_n)
    101.                 elif 'substring' == func_name.lower():
    102.                     date, start, end = self.extact_func(f1, func_name)
    103.                     s_n = f"substring(cast({date} as varchar),{start},{end}"
    104.                     s = s.replace(f, s_n)
    105.                 elif 'unit_timestamp' == func_name.lower():
    106.                     date = self.extact_func(f1, func_name)[0]
    107.                     s_n = f"to_unixtime(cast({date} as timestanp))"
    108.                     s = s.replace(f, s_n)
    109.                 elif 'to_date' == func_name.lower():
    110.                     date = self.extact_func(f1, func_name)[0]
    111.                     s_n = f"cast({date} as date)"
    112.                     s = s.replace(f, s_n)
    113.                 elif 'concat' == func_name.lower():
    114.                     res = self.extact_func(f1, func_name)[0]
    115.                     s_n = f'concat('
    116.                     for r in res:
    117.                         r = r.strip().replace('\n', '')
    118.                         s_n += f"cast({r} as varchar),"
    119.                     s_n = s_n[:-1] + ')'
    120.                     s = s.replace(f, s_n)
    121.                 elif 'sum' == func_name.lower():
    122.                     if 'unix_timestamp' in f1 or 'to_unixtime' in f1:
    123.                         continue
    124.                     ss = self.extact_func(f1, func_name)[0]
    125.                     if 'if(' in ss.replace(' ', ''):
    126.                         continue
    127.                     s = self.func_trans(f, f1, func_name, ss, s)
    128.                 elif 'avg' == func_name.lower():
    129.                     if 'unix_timestamp' in f1 or 'to_unixtime' in f1:
    130.                         continue
    131.                     ss = self.extact_func(f1, func_name)[0]
    132.                     if 'if(' in ss.replace(' ', ''):
    133.                         continue
    134.                     s = self.func_trans(f, f1, func_name, ss, s)
    135.                 elif 'abs' == func_name.lower():
    136.                     if 'unix_timestamp' in f1 or 'to_unixtime' in f1:
    137.                         continue
    138.                     ss = self.extact_func(f1, func_name)[0]
    139.                     if 'if(' in ss.replace(' ', ''):
    140.                         continue
    141.                     s = self.func_trans(f, f1, func_name, ss, s)
    142.                 elif 'ceiling' == func_name.lower():
    143.                     if 'unix_timestamp' in f1 or 'to_unixtime' in f1:
    144.                         continue
    145.                     ss = self.extact_func(f1, func_name)[0]
    146.                     if 'if(' in ss.replace(' ', ''):
    147.                         continue
    148.                     s = self.func_trans(f, f1, func_name, ss, s)
    149.                 elif 'floor' == func_name.lower():
    150.                     if 'unix_timestamp' in f1 or 'to_unixtime' in f1:
    151.                         continue
    152.                     ss = self.extact_func(f1, func_name)[0]
    153.                     if 'if(' in ss.replace(' ', ''):
    154.                         continue
    155.                     s = self.func_trans(f, f1, func_name, ss, s)
    156.                 elif 'year' == func_name.lower():
    157.                     date = self.extact_func(f1, func_name)[0]
    158.                     s_n = f"year(cast(substr(cast({date} as varchar,1,10) as date))"
    159.                     s = s.replace(f, s_n)
    160.                 elif 'month' == func_name.lower():
    161.                     date = self.extact_func(f1, func_name)[0]
    162.                     s_n = f"month(cast(substr(cast({date} as varchar,1,10) as date))"
    163.                     s = s.replace(f, s_n)
    164.             except:
    165.                 self.error.append(f"源代码中{func_name}函数参数输入可能有错误，具体为：{f1}")
    166.                 continue
    167.         if self.error:
    168.             self.entry.delete(0, END)
    169.             self.text.delete("1.0", END)
    170.             self.text.insert("end", f"{s}")
    171.             self.error.insert(0, '转换失败，有部分没有转成功\n')
    172.  
    173.             root_ex = Tk()
    174.             root_ex.title('错误')
    175.             win_width = 600
    176.             win_height = 200
    177.             screen_width = root_ex.winfo_screenwidth()
    178.             screen_height = root_ex.winfo_screenheight()
    179.             x = (screen_width - win_width) // 2
    180.             y = (screen_height - win_height) // 2
    181.             root_ex.geometry(f'{win_width}x{win_height}+{x}+{y}')
    182.             label_ex = Label(root_ex, text="\n".join(self.error), font=("楷体", 10))
    183.             label_ex.pack()
    184.             root_ex.mainloop()
    185.  
    186.             return s
    187.  
    188.     def func_trans(self, f, f1, func_name, ss, s):
    189.         if not ('+' in ss or '-' in ss or '*' in ss or '/' in ss):
    190.             date = self.extact_func(f1, func_name)[0]
    191.             s_n = f'{func_name}(cast{date} as double))'
    192.             s = s.replace(f, s_n)
    193.         else:
    194.             res1 = self.mysplit(f1)
    195.             s_n = f
    196.             n = len(s_n)
    197.             for item in res1:
    198.                 if any(c.isalpha() for c in item.replace(' ', '')):
    199.                     idxs = s_n.find(item)
    200.                     idxs = [idxs] if type(idxs) != list else idxs
    201.                     for idx in idxs:
    202.                         if idx + len(item) + 3 <= n:
    203.                             if not 'as' in s_n[idx:idx + len(item) + 4]:
    204.                                 s_n = re.sub(rf'\b{item}\b', f'cast({item} as double)', s_n)
    205.                         else:
    206.                             s_n = re.sub(rf'\b{item}\b', f'cast({item} as double)', s_n)
    207.                     s = s.replace(f, s_n)
    208.         return s
    209.  
    210.     def choose_file(self):
    211.         """
    212.         如果代码太多，从text中输入会很卡，直接选择代码文件输入会很快
    213.         :return:
    214.         """
    215.         self.filename = tkinter.filedialog.askopenfilename()
    216.         if '/' in self.filename:
    217.             self.filename = self.filename.replace('/', '\\')
    218.         self.entry.delete(0, END)
    219.         self.entry.insert(0, self.filename)
    220.  
    221.     def findvar(self, ss):
    222.         """
    223.         搜索与计算有关的字段
    224.         :param ss:
    225.         :return:
    226.         """
    227.         b = ['+', '-', '*', '/', '=', '!=', '>', '<', '<=', '>=', '<>']
    228.         result1 = []
    229.         result2 = []
    230.         result1_n = []
    231.         result1_n = []
    232.         res_ops = []
    233.         res1_ops = []
    234.         res_adj = []
    235.         res1_adj = []
    236.         
    237.     def mysplit(self, s):
    238.         """
    239.         分割字段
    240.         :param s:
    241.         :return:
    242.         """
    243.         pass
    244.  
    245.     def extact_func(self, s, func_name):
    246.         pass
    247.  
    248.     def delete_text(self, event):
    249.         pass
    250.  
    251.     def trans(self):
    252.         pass
    253.  
    254.  
    255. if __name__ == '__main__':
    256.     pro = Hive2Presto()
    257.     pro.__int__()
    258.     pro.main()

  • 相关阅读:
    SpringBoot 测试实践 - 1：常用的工具
    HarmonyOS UI 开发
    1.UEFI环境搭建
    java计算机毕业设计在线影院系统源码+系统+mysql数据库+lw文档+部署
    Flutter 高级教程之如何开发iOS Widget小组件展示SQLite本地数据库数据（教程含完整源码）
    查看Visual Studio软件_MSC_VER值（MSVC编译器版本）的方法
    Ubuntu18保姆级教程及其jdk和hadoop安装含资源
    怎么制作一个网站？怎样搭建一个高质量的网站？
    halcon之区域：多种区域（Region）特征（6）
    【JQuery】JQuery入门——模拟用户分组以及页面换肤
  • 原文地址：https://blog.csdn.net/Sanfenpai6/article/details/133637214