[Python知识库]xls文件解析及数据入库

包括内容： 1、读取文件数据(open方法)

2、Python的迭代器使用? ?

3、文件的写入?

4、判断表头是否合理

    # xjy_create
    # 读取csv文件
    line_length = 0
    with open(filename, 'r')as read_csv:
        print("open filename as read_csv")
        # 读取5行数据
        lines = read_csv.readlines(5)
        flag = True  # 创建一个判断值：True or False

    # 创建一个迭代器
    for index in range(len(lines)):
        line = lines[index].split(',')  # 将每一行从 str -> list
        if index == 0:
            line_length = len(line)  # 每一行的元素个数
            print ("====line_length is:===")
            print line_length
        else:
            if line_length != len(line):
                flag = False
                break
    # 判断是一个合理的表格： 表头每一列个数都相同
    if flag:
        print ("if flag == True")
        # 将表头存入内存
        file_header = "/home/liuyy/dataImport/data/" + str(user_id) + " " + tbname + ".ctl"
        with open(filename, 'r')as read_csv:
            with open(file_header, 'w') as store_header:
                print ("===store table_header===")
                # 读取csv文件前五行  并写入内存
                test_line = 0
                for line in read_csv.readlines():
                    test_line = test_line + 1
                    # print line
                    if test_line <= 5:
                        store_header.write(line)
                    else:
                        break

        # 入库操作
        with open("/home/liuyy/dataImport/dataImport/loader_new.sh", 'r')as f_in:
            print ("===loader_new.sh===")
            with open(file4, 'w')as f_out:
                for line in f_in.readlines():
                    print ("****loader.sh : line***")
                    print line
                    if line.find("[ctl]") >= 0 or line.find("[log]") >= 0:
                        f_out.write(line.replace("[ctl]", file1).replace("[log]", log))
                    else:
                        f_out.write(line)

    # xjy_create_end