python 代码解读

# -*- coding: utf-8 -*-
# date: 2017-01-23
# author: starking
# module function: extract log structure from log files
import os
import time
import json


#这个类的对象代表了单条日志
class Log:
    #定义在类内，但是不是类的方法内的变量叫做类的静态变量
    __SOH = '\x01'  # parameter separator in log text
    __NULL = '\N'  # indicate the log parameter is null
    __params = ['row_key', 'user_id', 'app_key', 'user_ip', 'question',
                'answer', 'process_record', 'gps', 'ts', 'result_code',
                'error', 'total_time', 'voiceId', 'request_id', 'server_code',
                'screen_type_code', 'platform_type_code', 'history', 'method',
                'city', 'time', 'scenario', 'scenario_param', 'wakeup_word',
                'service_code', 'oneshotkey_property', 'create_date']

    def __init__(self):
        self.text = None  # 初始化赋值为None
        for param in Log.__params:
            setattr(self, param, None)#obj,属性名，属性值这里赋值

    def __init__(self, log_text):
        self.parse(log_text)

    # parse a log string
    def parse(self, log_text):
        self.text = log_text
        for i, item in enumerate(log_text.split(Log.__SOH)):
            item = None if item == Log.__NULL or item.strip() == '' else item
            setattr(self, Log.__params[i], item)
        self.__parse_param_values()

    # extract parameter value when the param has non-string structure
    def __parse_param_values(self):
        self.total_time_value = int(self.total_time) if self.total_time is not None else None

        if self.answer is None:
            self.answer_json = None
        else:
            self.answer_json = json.loads(self.answer)
            if type(self.answer_json) is list:
                self.answer_json = self.answer_json[0]
        self.process_record_json = json.loads(self.process_record) if self.process_record is not None else None
        self.create_date_value = time.strptime(self.create_date, '%Y-%m-%d') if self.create_date is not None else None
        self.ts_value = time.strptime(self.ts, '%Y-%m-%d %H:%M:%S.0') if self.ts is not None else None
        if self.time is None:
            self.time_value = None
        else:
            try:
                self.time_value = time.strptime(self.time, '%Y-%m-%d %H:%M:%S')
            except:
                self.time_value = None
            try:
                self.time_value = time.strptime(self.time, '%Y-%m-%d%H:%M:%S')
            except:
                self.time_value = None

    # the print info
    def __str__(self):
        return self.display()

    def display(self):
        return '{' + os.linesep.join(
                ['(%s: %s)' % (param, getattr(self, param)) for param in Log.__params]) + '}'

    # return the log input
    def to_text(self):
        return self.text


# 代表了多条日志，就是一个日志文件，这里覆盖了 __iter__方法所以是一个iterable，同时覆盖了__next__ 所以是一个iterator;
#python 需要覆盖的放法一般 __ 开头同时__ 结尾
class LogStream:
    def __init__(self, filepaths):
        self.files = [open(fp, mode='r') for fp in filepaths]
        self.f_iters = [iter(f) for f in self.files]
        self.f_n = len(self.files)
        self.f_i = 0

    # LogStream is iterable
    def __iter__(self):
        return self

    # LogStream is an iterator
    def __next__(self):
        while self.f_i < self.f_n:
            try:
                val = self.f_iters[self.f_i].next()
                if val is None or len(val.strip()) == 0:
                    continue
                return Log(val.strip())
            except StopIteration:
                self.f_i += 1
        else:
            raise StopIteration()

    # close all file sources
    def __del__(self):
        for f in self.files:
            f.close()

    # reset stream
    def reset(self):
        self.f_i = 0
        for f in self.files:
            f.seek(0)

if __name__ == '__main__':
    log_path = r'*.log'
    stream = LogStream([log_path])
    # just print the first log for test
    for log in stream:
        print log
        break#因为break所以只打印了一条日志