在用golang爬虫的时候
总会遇到 10天前 10分钟前 刚刚这种很影响我们爬取正常事件
所以我写了个方法 来格式化这种事件
golang 版本
package utils
import (
"fmt"
"regexp"
"strconv"
"strings"
"time"
)
func HoursBeforeDayBeforeFormat(arcTime string) string {
arcTime = strings.TrimSpace(arcTime)
timeList := [...]string{
"小时前",
"分钟前",
"刚刚",
}
t := time.Now() // 获取当前时间
currentTime := t.Format("2006-01-02 15:04:05") //获取当前格式的日期
for _, _value := range timeList {
isContains := strings.Contains(arcTime, _value)
if isContains {
nowHour := time.Now().Format("15")
pattern := regexp.MustCompile(`\d+`)
afterNumber := pattern.FindString(arcTime)
if afterNumber == "" {
if arcTime == "刚刚" {
return currentTime
} else {
return "请传入带有数字类型 比如：5小时前"
}
}
if afterNumber <= nowHour {
return currentTime
} else if strings.Contains(arcTime, "分钟前") {
return currentTime
} else {
beforeDay := t.AddDate(0, 0, -1)
return beforeDay.Format("2006-01-02 15:04:05")
}
}
}
if strings.Contains(arcTime, "天前") {
pattern := regexp.MustCompile(`\d+`)
afterNumber, err := strconv.Atoi(fmt.Sprint("-", pattern.FindString(arcTime)))
if err != nil {
return fmt.Sprintln(arcTime, "正则提取错误")
}
return fmt.Sprintln(t.AddDate(0, 0, afterNumber).Format("2006-01-02 15:04:05"))
} else {
return arcTime
}
}
Python版本
# -*- coding: utf-8 -*-
# @Time : 2022/8/19 11:31
# @Author : lzc
# @Email : hybpjx@163.com
# @File : __init__.py.py
# @Software: PyCharm
import operator
from datetime import datetime, timedelta
import re
def number_turn_to(input_data):
regex_list = [
{"name": "一", 'value': 1},
{"name": "二", 'value': 2},
{"name": "三", 'value': 3},
{"name": "四", 'value': 4},
{"name": "五", 'value': 5},
{"name": "六", 'value': 6},
{"name": "七", 'value': 7},
{"name": "八", 'value': 8},
{"name": "九", 'value': 9},
{"name": "十", 'value': 10},
]
for i1 in regex_list:
for i2 in i1['name']:
input_data = re.sub(i2, str(i1['value']), input_data)
return input_data
def hoursBefore_dayBefore_format(arc_time):
arc_time = number_turn_to(arc_time.strip())
time_list = [
"小时前",
"分钟前",
"刚刚",
]
now_time = datetime.strftime(datetime.today(), '%Y-%m-%d')
for t in time_list:
IsContains = operator.contains(arc_time, t)
if IsContains:
now_hour = datetime.strftime(datetime.today(), "%H")
pattern = re.search(r"(\d+)", arc_time)
if not pattern:
if arc_time == "刚刚":
return now_time
else:
return "请传入带有数字类型 小时前"
arc_hour = pattern.group(1)
if arc_hour <= now_hour:
# 24小时之前，一律替换成当天日期
return now_time
elif "分钟前" in arc_time:
return now_time
else:
arc_time = datetime.strftime(datetime.today() + timedelta(days=-1), '%Y-%m-%d') # 获取前1天日期
return arc_time
if "天前" in arc_time:
try:
d = [str(d) + '天前' for d in range(1, 15)]
arc_time = datetime.strftime(datetime.today() + timedelta(days=(d.index(arc_time) + 1) * -1),
'%Y-%m-%d')
return arc_time
except Exception:
return f"转换失败 当前时间{now_time},网站中的时间{arc_time}"
else:
return arc_time
if __name__ == '__main__':
print(hoursBefore_dayBefore_format("一天前"))
