Python使用DOM和SAX解析XML的应用实例

<&＃63;xml version="1.0" encoding="UTF-8"&＃63;>


    id
  
    query
    varchar(200)
    false
    查询内容
  
    pv
    integer
    false
    页面访问量
  
    avg_money
    integer
    false
    平均金额
  

#!/usr/bin/python
# -*- coding:utf-8 -*-
# author: wklken
# desc: 用于读取数据库 XML 配置文件
# -----------------------
# 2012-02-18 创建
# ----------------------

import sys, os
from xml.dom import minidom, Node


def read_dbconfig_xml(xml_file_path):
    cOntent= {}

    root = minidom.parse(xml_file_path)
    table = root.getElementsByTagName("table")[0]

    # 读取数据库名和表名
    table_name = table.getAttribute("name")
    db_name = table.getAttribute("db_name")

    if len(table_name) > 0 and len(db_name) > 0:
        db_sql = f"CREATE DATABASE IF NOT EXISTS `{db_name}`; USE {db_name};"
        table_drop_sql = f"DROP TABLE IF EXISTS {table_name};"
        content.update({"db_sql": db_sql})
        content.update({"table_sql": table_drop_sql})
    else:
        print(f"Error: attribute not defined properly! db_name={db_name} ;table_name={table_name}")
        sys.exit(1)

    table_create_sql = f"CREATE TABLE {table_name} ("

    # 读取主键
    primary_key = table.getElementsByTagName("primary_key")[0]
    primary_key_name = primary_key.getElementsByTagName("name")[0].childNodes[0].nodeValue

    table_create_sql += f"{primary_key_name} INTEGER NOT NULL AUTO_INCREMENT PRIMARY KEY,"

    # 读取普通字段
    fields = table.getElementsByTagName("field")
    for field in fields:
        name = field.getElementsByTagName("name")[0].childNodes[0].nodeValue
        type = field.getElementsByTagName("type")[0].childNodes[0].nodeValue
        table_create_sql += f"{name} {type},"
        is_index = field.getElementsByTagName("is_index")[0].childNodes[0].nodeValue

    table_create_sql = table_create_sql.rstrip(',') + ");"
    content.update({"table_create_sql": table_create_sql})

    print(content)

if __name__ == "__main__":
    read_dbconfig_xml(sys.argv[1])

  
    100000
    男
    
北京，海淀区
    437
    
1989
    333
    242
    null
    大学
    
    1
    2月14日
  

from xml.sax import handler, parseString


class PersonHandler(handler.ContentHandler):
    def __init__(self, db_ops):
        self.db_ops = db_ops
        self.person = {}
        self.current_tag = ""
        self.in_quote = 0

    def startElement(self, name, attrs):
        if name == "person":
            self.person = {}
        self.current_tag = name
        self.in_quote = 1

    def endElement(self, name):
        if name == "person":
            in_fields = tuple([f'"{self.person.get(i, "")}"' for i in fields])
            print(in_sql % in_fields)
            self.db_ops.insert(in_sql % in_fields)
        self.in_quote = 0

    def characters(self, content):
        if self.in_quote:
            self.person[self.current_tag] = content


if __name__ == "__main__":
    f = open("./person.xml", 'r', encoding='gbk')
    db_ops = Db_Connect("127.0.0.1", "root", "root", "test")
    parseString(f.read(), PersonHandler(db_ops))
    f.close()
    db_ops.close()