import csv
import re
from collections import Counter



# 正则表达式匹配时间
time_pattern = re.compile(r'\d{4}-\d{2}')

# 初始化计数器
date_counts = Counter()

# 打开并读取CSV文件
with open('congtent.csv', mode='r', encoding='utf-8') as file:
    csv_reader = csv.reader(file)
    # print(csv_reader)
    # header = next(csv_reader)  # 跳过标题行

    for row in csv_reader:
        # print(row)
        time = row[14]
        # print("time", time)
        # 搜索时间值
        time_match = time_pattern.search(time)
        if time_match:
            # print(f"Found time: {time_match.group()}")
            date = time_match.group()
            date_counts[date] += 1

# 打印每一天的时间出现次数
for date, count in date_counts.items():
    print(f"{date}: {count} times")

# 可视化展示
dates = list(date_counts.keys())
counts = list(date_counts.values())

print(dates)
print(counts)

from pyecharts.charts import Bar
from pyecharts import options as opts



# 创建柱状图
bar = (
    Bar()
    .add_xaxis(dates)
    .add_yaxis("数量", counts)
    .set_series_opts(itemstyle_opts={'color': 'blue'})
    .set_global_opts(title_opts=opts.TitleOpts(title="日期对应数量柱状图"))
)

# 生成html文件（可选）
bar.render("bar_chart.html")