四、元类及属性.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**难**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 29. 用纯属性取代get和set方法"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "我们很多时候会实现一些setter或getter方法："
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "class OldResistor:\n",
    "    def __init__(self, ohms):\n",
    "        self._ohms = ohms\n",
    "        \n",
    "    def get_ohms(self):\n",
    "        return self._ohms\n",
    "    \n",
    "    def set_ohms(self):\n",
    "        self._ohms = ohms"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "但这实在不像是python的代码，用起来会变得麻烦。但这些工作确实会让开发者在之后的工作中更加方便地封装功能、验证用法和限定取值范围。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "python中可以通过设置属性也能完成上面的工作，而且用起来比较方便，但我们也希望具有验证、限定等功能，怎么办？"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**一种合适的方式是使用@property修饰器和setter方法来进行**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "class VoltageResistance:\n",
    "    \"\"\" 一个用于记录电压的工作，同时在记录电压的同时，根据其电压来改变电流（current） \"\"\"\n",
    "    def __init__(self, ohms):\n",
    "        self.ohms = ohms\n",
    "        self._voltage = 0\n",
    "        self.current = 0\n",
    "    \n",
    "    @property\n",
    "    def voltage(self):\n",
    "        return self._voltage\n",
    "    \n",
    "    @voltage.setter\n",
    "    def voltage(self, voltage):\n",
    "        self._voltage = voltage\n",
    "        self.current = self._voltage / self.ohms  # 欧姆定理"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Before:     0 amps\n",
      "After:  0.01 amps\n"
     ]
    }
   ],
   "source": [
    "r2 = VoltageResistance(1e3)\n",
    "print(\"Before: %5r amps\" % r2.current)\n",
    "r2.voltage = 10\n",
    "print(\"After: %5r amps\" % r2.current)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "其能够完成数值和类型的验证："
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "class BoundedResistance(VoltageResistance):\n",
    "    \"\"\" 一个用于记录电压的工作，同时在记录电压的同时，根据其电压来改变电流（current） \"\"\"\n",
    "    def __init__(self, ohms):\n",
    "        super().__init__(ohms)\n",
    "    \n",
    "    @property\n",
    "    def ohms(self):\n",
    "        return self._ohms\n",
    "    \n",
    "    @ohms.setter\n",
    "    def ohms(self, ohms):\n",
    "        if ohms <= 0:\n",
    "            raise ValueError(\"%f ohms must be > 0\" % ohms)\n",
    "        self._ohms = ohms"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "ename": "ValueError",
     "evalue": "0.000000 ohms must be > 0",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mValueError\u001b[0m                                Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-12-4faf6805d103>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m\u001b[0m\n\u001b[0;32m      1\u001b[0m \u001b[0mr3\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mBoundedResistance\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;36m1e3\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m----> 2\u001b[1;33m \u001b[0mr3\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mohms\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;36m0\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[1;32m<ipython-input-11-33c88e6d8059>\u001b[0m in \u001b[0;36mohms\u001b[1;34m(self, ohms)\u001b[0m\n\u001b[0;32m     11\u001b[0m     \u001b[1;32mdef\u001b[0m \u001b[0mohms\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mohms\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     12\u001b[0m         \u001b[1;32mif\u001b[0m \u001b[0mohms\u001b[0m \u001b[1;33m<=\u001b[0m \u001b[1;36m0\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m---> 13\u001b[1;33m             \u001b[1;32mraise\u001b[0m \u001b[0mValueError\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;34m\"%f ohms must be > 0\"\u001b[0m \u001b[1;33m%\u001b[0m \u001b[0mohms\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m     14\u001b[0m         \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_ohms\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mohms\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;31mValueError\u001b[0m: 0.000000 ohms must be > 0"
     ]
    }
   ],
   "source": [
    "r3 = BoundedResistance(1e3)\n",
    "r3.ohms = 0"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "在构造的时候输入不正确的值也会出错（这是因为父类中我们有`self.ohms = ohms`的语句，这里会使得其调用setter，如果改成`self._ohms`就不会有这个效果了）："
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "ename": "ValueError",
     "evalue": "-5.000000 ohms must be > 0",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mValueError\u001b[0m                                Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-13-9fc5cf3426e1>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m\u001b[0m\n\u001b[1;32m----> 1\u001b[1;33m \u001b[0mBoundedResistance\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m-\u001b[0m\u001b[1;36m5\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[1;32m<ipython-input-11-33c88e6d8059>\u001b[0m in \u001b[0;36m__init__\u001b[1;34m(self, ohms)\u001b[0m\n\u001b[0;32m      2\u001b[0m     \u001b[1;34m\"\"\" 一个用于记录电压的工作，同时在记录电压的同时，根据其电压来改变电流（current） \"\"\"\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m      3\u001b[0m     \u001b[1;32mdef\u001b[0m \u001b[0m__init__\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mohms\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m----> 4\u001b[1;33m         \u001b[0msuper\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m__init__\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mohms\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m      5\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m      6\u001b[0m     \u001b[1;33m@\u001b[0m\u001b[0mproperty\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m<ipython-input-10-0aa74ee86231>\u001b[0m in \u001b[0;36m__init__\u001b[1;34m(self, ohms)\u001b[0m\n\u001b[0;32m      2\u001b[0m     \u001b[1;34m\"\"\" 一个用于记录电压的工作，同时在记录电压的同时，根据其电压来改变电流（current） \"\"\"\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m      3\u001b[0m     \u001b[1;32mdef\u001b[0m \u001b[0m__init__\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mohms\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m----> 4\u001b[1;33m         \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mohms\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mohms\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m      5\u001b[0m         \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_voltage\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;36m0\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m      6\u001b[0m         \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mcurrent\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;36m0\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m<ipython-input-11-33c88e6d8059>\u001b[0m in \u001b[0;36mohms\u001b[1;34m(self, ohms)\u001b[0m\n\u001b[0;32m     11\u001b[0m     \u001b[1;32mdef\u001b[0m \u001b[0mohms\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mohms\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     12\u001b[0m         \u001b[1;32mif\u001b[0m \u001b[0mohms\u001b[0m \u001b[1;33m<=\u001b[0m \u001b[1;36m0\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m---> 13\u001b[1;33m             \u001b[1;32mraise\u001b[0m \u001b[0mValueError\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;34m\"%f ohms must be > 0\"\u001b[0m \u001b[1;33m%\u001b[0m \u001b[0mohms\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m     14\u001b[0m         \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_ohms\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mohms\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;31mValueError\u001b[0m: -5.000000 ohms must be > 0"
     ]
    }
   ],
   "source": [
    "BoundedResistance(-5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "另一个功能，防止父类的属性遭到篡改："
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "class FixedResistance(VoltageResistance):\n",
    "    \"\"\" 一个用于记录电压的工作，同时在记录电压的同时，根据其电压来改变电流（current） \"\"\"\n",
    "    def __init__(self, ohms):\n",
    "        super().__init__(ohms)\n",
    "    \n",
    "    @property\n",
    "    def ohms(self):\n",
    "        return self._ohms\n",
    "    \n",
    "    @ohms.setter\n",
    "    def ohms(self, ohms):\n",
    "        if hasattr(self, \"_ohms\"):\n",
    "            raise AttributeError(\"Can't set attribute\")\n",
    "        self._ohms = ohms"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "ename": "AttributeError",
     "evalue": "Can't set attribute",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mAttributeError\u001b[0m                            Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-15-7fbedb722fb3>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m\u001b[0m\n\u001b[0;32m      1\u001b[0m \u001b[0mr4\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mFixedResistance\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;36m1e3\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m----> 2\u001b[1;33m \u001b[0mr4\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mohms\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;36m2e3\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[1;32m<ipython-input-14-17d49dd64843>\u001b[0m in \u001b[0;36mohms\u001b[1;34m(self, ohms)\u001b[0m\n\u001b[0;32m     11\u001b[0m     \u001b[1;32mdef\u001b[0m \u001b[0mohms\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mohms\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     12\u001b[0m         \u001b[1;32mif\u001b[0m \u001b[0mhasattr\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;34m\"_ohms\"\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m---> 13\u001b[1;33m             \u001b[1;32mraise\u001b[0m \u001b[0mAttributeError\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;34m\"Can't set attribute\"\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m     14\u001b[0m         \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_ohms\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mohms\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;31mAttributeError\u001b[0m: Can't set attribute"
     ]
    }
   ],
   "source": [
    "r4 = FixedResistance(1e3)\n",
    "r4.ohms = 2e3"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**`@property`的最大缺点是无法在类似的类中进行共享，因为`@ohms.setter`需要属性名。**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "还有，**实现`getter`的时候，不要改变其他属性的值。**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 30. 考虑用@property来代替属性重构"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "我们的例子是漏桶算法，其作用是将网络流量进行整流，得到稳定速率的数据流。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "from datetime import datetime, timedelta\n",
    "\n",
    "class Bucket:\n",
    "    def __init__(self, period):\n",
    "        self.period_delta = timedelta(seconds=period)\n",
    "        self.reset_time = datetime.now()\n",
    "        self.quota = 0\n",
    "    \n",
    "    def __repr__(self):\n",
    "        return \"Bucket(quota=%d)\" % self.quota"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "def fill(bucket, amount):\n",
    "    \"\"\" 填充桶，如果时间超过了周期，则重置桶的状态，不然，将量填充到配额中 \"\"\"\n",
    "    now = datetime.now()\n",
    "    if now - bucket.reset_time > bucket.period_delta:\n",
    "        bucket.quota = 0\n",
    "        bucket.reset_time = now\n",
    "    bucket.quota += amount"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "def deduct(bucket, amount):\n",
    "    \"\"\" 扣除，如果超过了周期或扣除的数量超过了桶中的配额，则返回False，不然扣除配额，返回True \"\"\"\n",
    "    now = datetime.now()\n",
    "    if now - bucket.reset_time > bucket.period_delta:\n",
    "        return False\n",
    "    if bucket.quota - amount < 0:\n",
    "        return False\n",
    "    bucket.quota -= amount\n",
    "    return True"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Bucket(quota=100)\n",
      "Had 99 quota\n",
      "Bucket(quota=1)\n",
      "Not enough for 3 quota\n",
      "Bucket(quota=1)\n"
     ]
    }
   ],
   "source": [
    "bucket = Bucket(60)                      # 创建一个漏桶\n",
    "fill(bucket, 100)                        # 往漏桶中填充100的份额\n",
    "print(bucket)\n",
    "\n",
    "if deduct(bucket, 99):                   # 提取99的份额\n",
    "    print(\"Had 99 quota\")\n",
    "else:\n",
    "    print(\"Not enough for 99 quota\")\n",
    "print(bucket)\n",
    "\n",
    "if deduct(bucket, 3):                    # 如果剩下的份额不够，则会阻止这次提取\n",
    "    print(\"Had 3 quota\")\n",
    "else:\n",
    "    print(\"Not enough for 3 quota\")\n",
    "print(bucket)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "但以上的实现，无法得知提取被阻止的时候到底是因为配额不够，还是因为一开始就没有配额。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Bucket:\n",
    "    \"\"\"\n",
    "    实际上储存的不是quota，而是max_quota和quota_consumed，然后使用这两者实时计算quota\n",
    "    \n",
    "    max_quota是一开始加入的配额数量，而quota_consumed是消耗的综合（一个周期内）\n",
    "    \"\"\"\n",
    "    def __init__(self, period):\n",
    "        self.period_delta = timedelta(seconds=period)\n",
    "        self.reset_time = datetime.now()\n",
    "        self.max_quota = 0\n",
    "        self.quota_consumed = 0\n",
    "        \n",
    "    def __repr__(self):\n",
    "        return (\"Bucket(max_quota=%d, quota_consumed=%d)\" %\n",
    "               (self.max_quota, self.quota_consumed))\n",
    "    \n",
    "    @property\n",
    "    def quota(self):\n",
    "        return self.max_quota - self.quota_consumed\n",
    "    \n",
    "    @quota.setter\n",
    "    def quota(self, amount):\n",
    "        delta = self.max_quota - amount\n",
    "        if amount == 0:\n",
    "            # 如果quota被设置为0，则被看做是重置状态\n",
    "            self.quota_consumed = 0\n",
    "            self.max_quota = 0\n",
    "        elif delta < 0:\n",
    "            # 如果本次配额的消耗会导致所存有配额的消失，则\n",
    "            # 1. 如果这是一次消耗造成的，则将最大存有配额改成此次消耗的量\n",
    "            # 2. 不然，则阻止\n",
    "            assert self.quota_consumed == 0\n",
    "            self.max_quota = amount\n",
    "        else:\n",
    "            assert self.max_quota >= self.quota_consumed\n",
    "            self.quota_consumed += delta"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Bucket(max_quota=100, quota_consumed=0)\n",
      "Had 99 quota\n",
      "Bucket(max_quota=100, quota_consumed=99)\n",
      "Not enough for 3 quota\n",
      "Bucket(max_quota=100, quota_consumed=99)\n"
     ]
    }
   ],
   "source": [
    "bucket = Bucket(60)                      # 创建一个漏桶\n",
    "fill(bucket, 100)                        # 往漏桶中填充100的份额\n",
    "print(bucket)\n",
    "\n",
    "if deduct(bucket, 99):                   # 提取99的份额\n",
    "    print(\"Had 99 quota\")\n",
    "else:\n",
    "    print(\"Not enough for 99 quota\")\n",
    "print(bucket)\n",
    "\n",
    "if deduct(bucket, 3):                    # 如果剩下的份额不够，则会阻止这次提取\n",
    "    print(\"Had 3 quota\")\n",
    "else:\n",
    "    print(\"Not enough for 3 quota\")\n",
    "print(bucket)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**`@property`适合逐步完善数据模型，但如果用的过于频繁，则需要考虑彻底重构该类并修改相关的调用代码。**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 31. 用描述符来改写需要复用的`@property`方法"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`@property`方法有一个明显的缺点，是不便于复用。就是如果我们要实现多个相同逻辑的属性，则需要手写多个`@property`方法。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "一种更好的实现上述功能的方法是**Python的描述符（descriptor）**。因为python会对访问操作进行一定的转义，类似`len(foo)`被转义成了`foo.__len__()`，**访问操作被转义成了`__get__`和`__set__`方法**。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Grade:\n",
    "    def __get__(*args, **kwargs):\n",
    "        pass\n",
    "    \n",
    "    def __set__(*args, **kwargs):\n",
    "        pass\n",
    "     \n",
    "class Exam:\n",
    "    pass\n",
    "    math_grade = Grade()\n",
    "    writing_grade = Grade()\n",
    "    science_grade = Grade()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "exam = Exam()\n",
    "exam.writing_grade = 40   # <==> Exam.__dict__[\"writing_grade\"].__set__(exam, 40)\n",
    "exam.writing_grade # <==> Exam.__dict__[\"writing_grade\"].__get__(exam, Exam)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "只所以有上面的转义，关键在于`object`类的`__getattribute__`方法。**如果Exam实例中没有`writing_grade`属性，则python会转向Exam类，并在该类中查找同名的类属性（必须是类属性，如果是实例属性，赋值会直接把这个`Grade`实例替换掉）。这个类属性如果实现了`__get__`和`__set__`，则Python认为此对象遵从描述符协议。**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Grade:\n",
    "    def __init__(self):\n",
    "        self._value = 0\n",
    "        \n",
    "    def __get__(self, instance, instance_type):\n",
    "        return self._value\n",
    "    \n",
    "    def __set__(self, instance, value):\n",
    "        if not (0 <= value <= 100):\n",
    "            raise ValueError(\"Grade must be between 0 and 100\")\n",
    "        self._value = value"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Exam:\n",
    "    math_grade = Grade()\n",
    "    writing_grade = Grade()\n",
    "    science_grade = Grade()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Writing 82\n",
      "Science 99\n"
     ]
    }
   ],
   "source": [
    "first_exam = Exam()\n",
    "first_exam.writing_grade = 82\n",
    "first_exam.science_grade = 99\n",
    "print(\"Writing\", first_exam.writing_grade)\n",
    "print(\"Science\", first_exam.science_grade)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "但如果出现多个Exam实例，就导致了错误的结果："
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Second 75 is right\n",
      "First 75 is wrong\n"
     ]
    }
   ],
   "source": [
    "second_exam = Exam()\n",
    "second_exam.writing_grade = 75\n",
    "print(\"Second\", second_exam.writing_grade, \"is right\")\n",
    "print(\"First\", first_exam.writing_grade, \"is wrong\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**这源于我们使用的是类属性的缘故。**一种解决的方式是在Grade属性中使用一个字典，对不同的实例进行记录。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Grade:\n",
    "    \"\"\" 描述符 \"\"\"\n",
    "    def __init__(self):\n",
    "        self._values = {}\n",
    "    \n",
    "    def __get__(self, instance, instance_type):  # 从这里可以看出，instance指的是Grade作为类属性的类的实例\n",
    "        if instance is None:\n",
    "            return self\n",
    "        return self._values.get(instance, 0)\n",
    "    \n",
    "    def __set__(self, instance, value):\n",
    "        if not (0 <= value <= 100):\n",
    "            raise ValueError(\"Grade must be between 0 and 100\")\n",
    "        self._values[instance] = value"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "但以上的实现有个问题：`_values`字典保存了每个instance的引用，所以这些instance不会被垃圾收集器收集，导致**泄露内存**。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "我们可以使用`WeekKeyDictionary`来替代普通字典，其特殊之处在于：如果发现字典中所持有的的引用是整个程序中的最后引用，则系统自动将这个实例从字典的键中移除。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "from weakref import WeakKeyDictionary\n",
    "\n",
    "class Grade:\n",
    "    def __init__(self):\n",
    "        self._values = WeakKeyDictionary()\n",
    "    \n",
    "    def __get__(self, instance, instance_type):\n",
    "        if instance is None:\n",
    "            return self\n",
    "        return self._values.get(instance, 0)\n",
    "    \n",
    "    def __set__(self, instance, value):\n",
    "        if not (0 <= value <= 100):\n",
    "            raise ValueError(\"Grade must be between 0 and 100\")\n",
    "        self._values[instance] = value\n",
    "class Exam:\n",
    "    math_grade = Grade()\n",
    "    writing_grade = Grade()\n",
    "    science_grade = Grade()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Writing 82\n",
      "Science 99\n",
      "Second 75 is right\n",
      "First 82 is right\n"
     ]
    }
   ],
   "source": [
    "first_exam = Exam()\n",
    "first_exam.writing_grade = 82\n",
    "first_exam.science_grade = 99\n",
    "print(\"Writing\", first_exam.writing_grade)\n",
    "print(\"Science\", first_exam.science_grade)\n",
    "second_exam = Exam()\n",
    "second_exam.writing_grade = 75\n",
    "print(\"Second\", second_exam.writing_grade, \"is right\")\n",
    "print(\"First\", first_exam.writing_grade, \"is right\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 32. 使用`__getattr__`、`__getattribute__`和`__setattr__`实现按需生成的属性"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "如果在`__dict__`找不到想要的属性，则会调用`__getattr__`方法："
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "class LazyDB:\n",
    "    def __init__(self):\n",
    "        self.exists = 5\n",
    "        \n",
    "    def __getattr__(self, name):\n",
    "        value = \"Value for %s\" % name\n",
    "        setattr(self, name, value)  # 把这个补充到__dict__中\n",
    "        return value"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Before: {'exists': 5}\n",
      "foo:    Value for foo\n",
      "After:  {'exists': 5, 'foo': 'Value for foo'}\n"
     ]
    }
   ],
   "source": [
    "data = LazyDB()\n",
    "print(\"Before:\", data.__dict__)\n",
    "print(\"foo:   \", data.foo)\n",
    "print(\"After: \", data.__dict__)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "exists: 5\n",
      "Called __getattr__(foo)\n",
      "foo:    Value for foo\n",
      "foo:    Value for foo\n"
     ]
    }
   ],
   "source": [
    "class LoggingLazyDB(LazyDB):\n",
    "    def __getattr__(self, name):\n",
    "        print(\"Called __getattr__(%s)\" % name)\n",
    "        return super().__getattr__(name)\n",
    "\n",
    "data = LoggingLazyDB()\n",
    "print(\"exists:\", data.exists)\n",
    "print(\"foo:   \", data.foo)\n",
    "print(\"foo:   \", data.foo)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "我们看到，`__getattr__`方法在存在指定属性的时候是不调用的，但如果我们希望不管这个属性存在还是不存在都调用一些方法，则需要使用`__getattribute__`方法。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [],
   "source": [
    "class ValidatingDB:\n",
    "    def __init__(self):\n",
    "        self.exists = 5\n",
    "        \n",
    "    def __getattribute__(self, name):\n",
    "        print(\"Called __getattribute__(%s)\" % name)\n",
    "        try:\n",
    "            return super().__getattribute__(name)  # 这里实际上调用的是object.__getattribute__，实际上就是正常的读取属性的过程，如果没有这个属性会返回一个AttributeError\n",
    "        except AttributeError:\n",
    "            value = \"Value for %s\" % name\n",
    "            setattr(self, name, value)\n",
    "            return value"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Called __getattribute__(exists)\n",
      "exists: 5\n",
      "Called __getattribute__(foo)\n",
      "foo:    Value for foo\n",
      "Called __getattribute__(foo)\n",
      "foo:    Value for foo\n"
     ]
    }
   ],
   "source": [
    "data = ValidatingDB()\n",
    "print(\"exists:\", data.exists)\n",
    "print(\"foo:   \", data.foo)\n",
    "print(\"foo:   \", data.foo)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "上面我们用到一个函数`setattr`，实际上调用的是`__setattr__`方法。而且不止这样会调用，即便使用`self.attr = 5`这样的方式也是会调用这个方法的。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**在使用这些方法的时候，一定有注意不要陷入反复递归的陷阱。**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "class BrokenDictionaryDB:\n",
    "    def __init__(self, data):\n",
    "        self._data = data\n",
    "        \n",
    "    def __getattribute__(self, name):\n",
    "        print(\"Called __getattribute__(%s)\" % name)\n",
    "        return self._data[name]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# data = BrokenDictionaryDB({\"foo\": 3})\n",
    "# data.foo"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**这是因为`self._data`实际上会再次调用`__getattribute__`方法，从而陷入无限递归。解决方案是使用`super().__getattribute__(\"attrname\")`方法，其因为是`object`的方法，即标准的提取属性方法，其效果就是从属性字典中把值提取出来。**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "class DictionaryDB:\n",
    "    def __init__(self, data):\n",
    "        self._data = data\n",
    "        \n",
    "    def __getattribute__(self, name):\n",
    "        data_dict = super().__getattribute__(\"_data\")\n",
    "        return data_dict[name]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data = DictionaryDB({\"foo\": 3})\n",
    "data.foo"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**同理，如果要在`__setattr__`中完成修改属性，则需要使用`super().__setattr__`完成。**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 33. 用元类来验证子类"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "元类最简单的任务是去检查保证类的风格协调一致，确保一些方法得到了覆写，确保类属性之间具备某些严格的关系。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "定义元类的时候，需要从`type`开始继承，对于使用元类的其他类来说，python会将这些类的`class`语句中的相关信息都发送给元类的`__new__`方法，我们可以通过以下的方法是来看到这些信息："
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Meta(type):\n",
    "    def __new__(meta, name, bases, class_dict):\n",
    "        print((meta, name, bases, class_dict))\n",
    "        return type.__new__(meta, name, bases, class_dict)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(<class '__main__.Meta'>, 'MyClass', (), {'__module__': '__main__', '__qualname__': 'MyClass', 'stuff': 123, 'foo': <function MyClass.foo at 0x00000214803501E0>})\n"
     ]
    }
   ],
   "source": [
    "class MyClass(metaclass=Meta):\n",
    "    stuff = 123\n",
    "    \n",
    "    def foo(self):\n",
    "        pass"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**因此，我们可以将相关的逻辑代码放在`Meta.__new__`方法中。**比如，我们希望创建一个表示多边形的类，所以我们定义一个特殊的验证类，使得多边形体系的基类将其作为自己的元类。注意，元类中编写的逻辑，是针对的该基类的子类（要做到这一点，需要检验`bases`）。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "class ValidatePolygon(type):\n",
    "    def __new__(meta, name, bases, class_dict):\n",
    "        if bases != (object,) and bases != ():\n",
    "            if class_dict[\"sides\"] < 3:\n",
    "                raise ValueError(\"Polygons need 3+ sides\")\n",
    "        return type.__new__(meta, name, bases, class_dict)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Polygon(metaclass=ValidatePolygon):\n",
    "    sides = None\n",
    "    \n",
    "    @classmethod\n",
    "    def interior_angles(cls):\n",
    "        return (cls.sides - 2) * 180"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Triangle(Polygon):\n",
    "    sides = 3"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Before class\n",
      "Before sides\n",
      "After sides\n",
      "After class\n"
     ]
    },
    {
     "ename": "ValueError",
     "evalue": "Polygons need 3+ sides",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mValueError\u001b[0m                                Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-11-6271bba46ccf>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m\u001b[0m\n\u001b[0;32m      1\u001b[0m \u001b[0mprint\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;34m\"Before class\"\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m----> 2\u001b[1;33m \u001b[1;32mclass\u001b[0m \u001b[0mLine\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mPolygon\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m      3\u001b[0m     \u001b[0mprint\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;34m\"Before sides\"\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m      4\u001b[0m     \u001b[0msides\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;36m1\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m      5\u001b[0m     \u001b[0mprint\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;34m\"After sides\"\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m<ipython-input-8-4d68d3e30a17>\u001b[0m in \u001b[0;36m__new__\u001b[1;34m(meta, name, bases, class_dict)\u001b[0m\n\u001b[0;32m      3\u001b[0m         \u001b[1;32mif\u001b[0m \u001b[0mbases\u001b[0m \u001b[1;33m!=\u001b[0m \u001b[1;33m(\u001b[0m\u001b[0mobject\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;32mand\u001b[0m \u001b[0mbases\u001b[0m \u001b[1;33m!=\u001b[0m \u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m      4\u001b[0m             \u001b[1;32mif\u001b[0m \u001b[0mclass_dict\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;34m\"sides\"\u001b[0m\u001b[1;33m]\u001b[0m \u001b[1;33m<\u001b[0m \u001b[1;36m3\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m----> 5\u001b[1;33m                 \u001b[1;32mraise\u001b[0m \u001b[0mValueError\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;34m\"Polygons need 3+ sides\"\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m      6\u001b[0m         \u001b[1;32mreturn\u001b[0m \u001b[0mtype\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m__new__\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mmeta\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mname\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mbases\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mclass_dict\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;31mValueError\u001b[0m: Polygons need 3+ sides"
     ]
    }
   ],
   "source": [
    "print(\"Before class\")\n",
    "class Line(Polygon):\n",
    "    print(\"Before sides\")\n",
    "    sides = 1\n",
    "    print(\"After sides\")\n",
    "    print(\"After class\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**可以看到，验证是在`class`语句体整个完成后进行的。**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 34. 用元类去注册子类"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "import json\n",
    "\n",
    "class Serializable:\n",
    "    def __init__(self, *args):\n",
    "        self.args = args\n",
    "        \n",
    "    def serialize(self):\n",
    "        return json.dumps({\"args\": self.args})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Point2D(Serializable):\n",
    "    def __init__(self, x, y):\n",
    "        super().__init__(x, y)\n",
    "        self.x = x\n",
    "        self.y = y\n",
    "    def __repr__(self):\n",
    "        return \"Point2D(%d, %d)\" % (self.x, self.y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Object:     Point2D(5, 3)\n",
      "Serialized: {\"args\": [5, 3]}\n"
     ]
    }
   ],
   "source": [
    "point = Point2D(5, 3)\n",
    "print(\"Object:    \", point)\n",
    "print(\"Serialized:\", point.serialize())  # Point2D对象序列化"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Deserializable(Serializable):\n",
    "    \"\"\" 反序列化，子类可以通过类方法将json_data还原成其对应的类实例 \"\"\"\n",
    "    @classmethod\n",
    "    def deserialize(cls, json_data):\n",
    "        params = json.loads(json_data)\n",
    "        return cls(*params[\"args\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "class BetterPoint2D(Deserializable):\n",
    "    def __init__(self, x, y):\n",
    "        super().__init__(x, y)\n",
    "        self.x = x\n",
    "        self.y = y\n",
    "    def __repr__(self):\n",
    "        return \"Point2D(%d, %d)\" % (self.x, self.y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Before: Point2D(5, 3)\n",
      "Serialized: {\"args\": [5, 3]}\n",
      "After: Point2D(5, 3)\n"
     ]
    }
   ],
   "source": [
    "point = BetterPoint2D(5, 3)\n",
    "print(\"Before:\", point)\n",
    "data = point.serialize()\n",
    "print(\"Serialized:\", data)\n",
    "after = BetterPoint2D.deserialize(data)\n",
    "print(\"After:\", after)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**这种方法的缺点是：我们需要知道现在的json数据是储存的是`BetterPoint2D`的数据，所以才能使用`BetterPoint2D`的类方法来反序列化。但理想的方案应该是更加通用的，不需要知道子类就可以进行。**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "一种可行的方式是将类名记录到JSON数据中："
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "class BetterSerializable:\n",
    "    def __init__(self, *args):\n",
    "        self.args = args\n",
    "        \n",
    "    def serialize(self):\n",
    "        return json.dumps({\n",
    "            \"class\": self.__class__.__name__,\n",
    "            \"args\": self.args\n",
    "        })"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**然后，将类名和类实例构造器之间的映射关系保存到一份字典中，然后在反序列化的时候取出来用。**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "registry = {}\n",
    "\n",
    "def register_class(target_class):\n",
    "    registry[target_class.__name__] = target_class\n",
    "    \n",
    "def deserialize(data):\n",
    "    params = json.loads(data)\n",
    "    name = params[\"class\"]\n",
    "    target_class = registry[name]\n",
    "    return target_class(*params[\"args\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "class EvenBetterPoint2D(BetterSerializable):\n",
    "    def __init__(self, x, y):\n",
    "        super().__init__(x, y)\n",
    "        self.x = x\n",
    "        self.y = y\n",
    "        \n",
    "    def __repr__(self):\n",
    "        return \"%s(%d, %d)\" % (self.__class__.__name__, self.x, self.y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "register_class(EvenBetterPoint2D)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Before: EvenBetterPoint2D(5, 3)\n",
      "Serialized: {\"class\": \"EvenBetterPoint2D\", \"args\": [5, 3]}\n",
      "After: EvenBetterPoint2D(5, 3)\n"
     ]
    }
   ],
   "source": [
    "point = EvenBetterPoint2D(5, 3)\n",
    "print(\"Before:\", point)\n",
    "data = point.serialize()\n",
    "print(\"Serialized:\", data)\n",
    "after = deserialize(data)\n",
    "print(\"After:\", after)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "以上的整个操作过程都是成功的，但有个问题：**开发者很有可能忘记调用`register_class`函数**。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**所以我们现在应该想个办法，在继承`BetterSerializable`的时候，程序自动调用`register_class`函数。这个功能可以通过元类来实现**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Meta(type):\n",
    "    def __new__(meta, name, bases, class_dict):\n",
    "        cls = type.__new__(meta, name, bases, class_dict)  # 这里可以看到，`type.__new__`函数返回的是类本身，和实例中的`__new__`方法不同\n",
    "        register_class(cls)\n",
    "        return cls\n",
    "    \n",
    "class RegisteredSerializable(BetterSerializable, metaclass=Meta):\n",
    "    pass"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Vector3D(RegisteredSerializable):\n",
    "    def __init__(self, x, y, z):\n",
    "        super().__init__(x, y, z)\n",
    "        self.x, self.y, self.z = x, y, z"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Before: <__main__.Vector3D object at 0x000002148000B550>\n",
      "Serialized: {\"class\": \"Vector3D\", \"args\": [10, -7, 3]}\n",
      "After: <__main__.Vector3D object at 0x000002148000B668>\n"
     ]
    }
   ],
   "source": [
    "v3 = Vector3D(10, -7, 3)\n",
    "print(\"Before:\", v3)\n",
    "data = v3.serialize()\n",
    "print(\"Serialized:\", data)\n",
    "print(\"After:\", deserialize(data))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 35. 用元类来注解类的属性"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**元类的一个更加有用的功能，是可以在某个类刚定义好但是尚未使用的时候，提前修改或注解该类的属性。**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Field:\n",
    "    def __init__(self, name):\n",
    "        self.name = name\n",
    "        self.internal_name = \"_\" + self.name\n",
    "    \n",
    "    def __get__(self, instance, instance_type):\n",
    "        if instance is None:\n",
    "            return self\n",
    "        return getattr(instance, self.internal_name, \"\")\n",
    "    \n",
    "    def __set__(self, instance, value):\n",
    "        setattr(instance, self.internal_name, value)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Customer:\n",
    "    # Class attributes\n",
    "    first_name = Field(\"first_name\")  # 这里有个小问题，即我们必须将first_name重复写两次\n",
    "    last_name = Field(\"last_name\")\n",
    "    prefix = Field(\"prefix\")\n",
    "    suffix = Field(\"suffix\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Before: '' {}\n",
      "After: 'Euclid' {'_first_name': 'Euclid'}\n"
     ]
    }
   ],
   "source": [
    "foo = Customer()\n",
    "print(\"Before:\", repr(foo.first_name), foo.__dict__)\n",
    "foo.first_name = \"Euclid\"\n",
    "print(\"After:\", repr(foo.first_name), foo.__dict__)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "解决办法是使用元类："
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Meta(type):\n",
    "    def __new__(meta, name, bases, class_dict):\n",
    "        for key, value in class_dict.items():\n",
    "            if isinstance(value, Field):\n",
    "                value.name = key\n",
    "                value.internal_name = \"_\" + key\n",
    "        cls = type.__new__(meta, name, bases, class_dict)\n",
    "        return cls"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "class DatabaseRow(metaclass=Meta):\n",
    "    pass"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Field:\n",
    "    def __init__(self):\n",
    "        self.name = None\n",
    "        self.internal_name = None\n",
    "    \n",
    "    def __get__(self, instance, instance_type):\n",
    "        if instance is None:\n",
    "            return self\n",
    "        return getattr(instance, self.internal_name, \"\")\n",
    "    \n",
    "    def __set__(self, instance, value):\n",
    "        setattr(instance, self.internal_name, value)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "class BetterCustomer(DatabaseRow):\n",
    "    first_name = Field()\n",
    "    last_name = Field()\n",
    "    prefix = Field()\n",
    "    suffix = Field()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Before: '' {}\n",
      "After: 'Euler' {'_first_name': 'Euler'}\n"
     ]
    }
   ],
   "source": [
    "foo = BetterCustomer()\n",
    "print(\"Before:\", repr(foo.first_name), foo.__dict__)\n",
    "foo.first_name = \"Euler\"\n",
    "print(\"After:\", repr(foo.first_name), foo.__dict__)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}