Compare commits

...

2 Commits

Author SHA1 Message Date
1da3f446a8 http request. 2023-11-22 14:34:03 +08:00
49fc606c33 11 2023-11-08 18:02:51 +08:00
2 changed files with 98 additions and 23 deletions

View File

@ -1,38 +1,93 @@
import aiohttp import asyncio
import re
import time
import httpx
import pymysql
base_url = "https://dm.video.qq.com/barrage/base/" base_url = "https://dm.video.qq.com/barrage/base/"
dm_url = "https://dm.video.qq.com/barrage/segment/p0047eb164c/t/v1/0/30000" dm_url = "https://dm.video.qq.com/barrage/segment/p0047eb164c/t/v1/0/30000"
async def getTxVideoData(url, video_id): async def getTxVideoData(url, video_id):
url = "https://example.com/api/data.json" # 替换成你要请求的JSON数据的URL # 使用正则表达式获取视频ID p0047eb164c https://v.qq.com/x/cover/mzc00200pppnnhl/p0047eb164c.html
global connection
global cursor
vid = re.findall(r'/([a-zA-Z0-9]+)\.html', url)[0]
print(vid)
url = f"https://dm.video.qq.com/barrage/base/{vid}" # 替换成你要请求的JSON数据的URL
json_data = await fetch_json(url) json_data = await fetch_json(url)
if json_data: if json_data:
# 在这里处理解析后的JSON数据 # 在这里处理解析后的JSON数据
print(json_data) span = json_data['segment_span']
print(span)
items = json_data["segment_index"].items()
page = len(items)
# 最大数据索引
max_data_index = (page - 1) * 30000
# 页数
print(page)
# 获取的数据量
data_num = 5000
# 每页获取的数据量
number_per_page = int(data_num / page)
print(number_per_page)
total_num = 0
# 连接数据库
try:
# 连接数据库
connection = pymysql.connect(host='127.0.0.1', port=3306, user='root', password='root', db='dm')
cursor = connection.cursor()
except pymysql.Error as e:
# 异常 连接失败
print("数据库连接失败")
# finally:
# # 关闭数据库连接
# if connection:
# connection.close()
for i in range(0, page):
data_url_suffix = json_data["segment_index"][str(i * 30000)]["segment_name"]
data_url = f"https://dm.video.qq.com/barrage/segment/{vid}/{data_url_suffix}"
print(data_url)
data_json = await fetch_json(data_url)
if data_json:
barrage_list = data_json["barrage_list"]
data_len = len(barrage_list)
timestamp = int(time.time())
# 遍历barrage_list
sql = "INSERT INTO `danmaku_list` (`id`, `type`, `text`, `color`, `size`, `videotime`, `ip`, `time`) VALUES "
data_index = 0
for barrage in barrage_list:
content = barrage["content"].replace("\\", "").replace("'", "\\'")
time_offset = float(barrage["time_offset"]) / 1000
if data_index == 0:
sql += f"('{video_id}', 'right', '{content}','#FFFFFF', '27.5px', {time_offset}, '127.0.0.1', {timestamp})"
else:
sql += f",('{video_id}', 'right', '{content}','#FFFFFF', '27.5px', {time_offset}, '127.0.0.1', {timestamp})"
data_index += 1
total_num += data_len
# print(sql)
# cursor.execute(sql)
# connection.commit()
print(f"总数据量:{total_num}")
async def fetch_json(url): async def fetch_json(url):
async with aiohttp.ClientSession() as session: async with httpx.AsyncClient() as client:
async with session.get(url) as response: response = await client.get(url)
if response.status == 200: return response.json()
data = await response.json()
return data
else:
print(f"Failed to fetch data. Status code: {response.status}")
return None
if __name__ == "__main__": if __name__ == "__main__":
url = "https://v.qq.com/x/cover/mzc00200pppnnhl/j00468dc2og.html"
url = "https://v.qq.com/x/cover/mzc00200pppnnhl/p0047eb164c.html"
if "v.qq.com" in url: if "v.qq.com" in url:
# 调用 v.qq.com 对应的接口 # 调用 v.qq.com 对应的接口
await getTxVideoData(url, "p0047eb164c") asyncio.run(getTxVideoData(url, "j00468dc2og"))
# 检查是否存在 "rainss.cn" # 检查是否存在 "rainss.cn"
elif "rainss.cn" in url: elif "rainss.cn" in url:
# 调用 rainss.cn 对应的接口 # 调用 rainss.cn 对应的接口
print("调用 rainss.cn 的接口") print("调用 rainss.cn 的接口")
# 检查是否存在 "allms.cn" # 检查是否存在 "allms.cn"
elif "allms.cn" in url: elif "allms.cn" in url:
# 调用 allms.cn 对应的接口 # 调用 allms.cn 对应的接口
@ -40,4 +95,4 @@ if __name__ == "__main__":
# 如果都不匹配 # 如果都不匹配
else: else:
print("未匹配到任何子字符串") print("未匹配到任何子字符串")

32
dm.py
View File

@ -54,18 +54,24 @@ class TableWidgetExample(QMainWindow):
self.connection = None self.connection = None
self.cursor = None self.cursor = None
self.is_connect = False self.is_connect = False
self.logger_area = None
self.initLogger() self.initLogger()
self.initUI() self.initUI()
def initUI(self): def initUI(self):
self.setWindowTitle("采集") self.setWindowTitle("采集")
self.setGeometry(100, 100, 800, 600) self.setGeometry(100, 100, 1000, 600)
central_widget = QWidget() central_widget = QWidget()
center_layout = QVBoxLayout(central_widget) central_layout = QHBoxLayout(central_widget)
# central_layout.setSpacing(0)
self.setCentralWidget(central_widget) self.setCentralWidget(central_widget)
left_widget = QWidget()
left_layout = QVBoxLayout(left_widget)
left_layout.setContentsMargins(0, 0, 0, 0)
central_layout.addWidget(left_widget, 3)
top_widget = QWidget() top_widget = QWidget()
top_layout = QVBoxLayout(top_widget) top_layout = QVBoxLayout(top_widget)
@ -79,7 +85,7 @@ class TableWidgetExample(QMainWindow):
header.setSectionResizeMode(QtWidgets.QHeaderView.Stretch) header.setSectionResizeMode(QtWidgets.QHeaderView.Stretch)
top_layout.addWidget(self.table) top_layout.addWidget(self.table)
center_layout.addWidget(top_widget) left_layout.addWidget(top_widget)
# 下部分布局 # 下部分布局
bottom_widget = QWidget() bottom_widget = QWidget()
@ -126,7 +132,21 @@ class TableWidgetExample(QMainWindow):
action_layout.addWidget(button_execute) action_layout.addWidget(button_execute)
bottom_layout.addLayout(action_layout) bottom_layout.addLayout(action_layout)
center_layout.addWidget(bottom_widget) left_layout.addWidget(bottom_widget)
right_widget = QWidget()
right_layout = QVBoxLayout(right_widget)
right_layout.setContentsMargins(0, 0, 0, 0)
central_layout.addWidget(right_widget, 1)
logger_widget = QWidget()
# logger_widget.setFixedHeight(200)
logger_layout = QVBoxLayout(logger_widget)
self.logger_area = QTextEdit()
self.logger_area.setPlaceholderText("日志...")
self.logger_area.setReadOnly(True)
logger_layout.addWidget(self.logger_area)
right_layout.addWidget(logger_widget)
def initLogger(self): def initLogger(self):
logging.basicConfig(level=logging.DEBUG) logging.basicConfig(level=logging.DEBUG)
@ -278,11 +298,11 @@ class TableWidgetExample(QMainWindow):
else: else:
print("未匹配到任何子字符串") print("未匹配到任何子字符串")
def generate_insert_sql(self, data): def generate_insert_sql(self, data):
sql = "INSERT INTO `danmaku_list` (`id`, `cid`, `type`, `text`, `color`, `size`, `videotime`, `ip`, `time`) VALUES ('de9d5e3f5b7f9ea2197d35f383530643', NULL, 'right', '不错哦', '#FFFFFF', '27.5px', '48.275', '58.62.32.191', '1666319003')" sql = "INSERT INTO `danmaku_list` (`id`, `cid`, `type`, `text`, `color`, `size`, `videotime`, `ip`, `time`) VALUES ('de9d5e3f5b7f9ea2197d35f383530643', NULL, 'right', '不错哦', '#FFFFFF', '27.5px', '48.275', '58.62.32.191', '1666319003')"
return sql return sql
if __name__ == "__main__": if __name__ == "__main__":
app = QApplication(sys.argv) app = QApplication(sys.argv)
window = TableWidgetExample() window = TableWidgetExample()