refactor: 数据存储重构,分离不同类型的存储实现

This commit is contained in:
Relakkes
2024-01-14 22:06:31 +08:00
parent e31aebbdfb
commit 894dabcf63
37 changed files with 1427 additions and 864 deletions

View File

@@ -0,0 +1,81 @@
# -*- coding: utf-8 -*-
# @Author : relakkes@gmail.com
# @Time : 2024/1/14 19:34
# @Desc :
from typing import List
import config
from .bilibili_store_db_types import *
from .bilibili_store_impl import *
class BiliStoreFactory:
STORES = {
"csv": BiliCsvStoreImplement,
"db": BiliDbStoreImplement,
"json": BiliJsonStoreImplement
}
@staticmethod
def create_store() -> AbstractStore:
store_class = BiliStoreFactory.STORES.get(config.SAVE_DATA_OPTION)
if not store_class:
raise ValueError(
"[BiliStoreFactory.create_store] Invalid save option only supported csv or db or json ...")
return store_class()
async def update_bilibili_video(video_item: Dict):
video_item_view: Dict = video_item.get("View")
video_user_info: Dict = video_item_view.get("owner")
video_item_stat: Dict = video_item_view.get("stat")
video_id = str(video_item_view.get("aid"))
save_content_item = {
"video_id": video_id,
"video_type": "video",
"title": video_item_view.get("title", "")[:500],
"desc": video_item_view.get("desc", "")[:500],
"create_time": video_item_view.get("pubdate"),
"user_id": str(video_user_info.get("mid")),
"nickname": video_user_info.get("name"),
"avatar": video_user_info.get("face", ""),
"liked_count": str(video_item_stat.get("like", "")),
"video_play_count": str(video_item_stat.get("view", "")),
"video_danmaku": str(video_item_stat.get("danmaku", "")),
"video_comment": str(video_item_stat.get("reply", "")),
"last_modify_ts": utils.get_current_timestamp(),
"video_url": f"https://www.bilibili.com/video/av{video_id}",
"video_cover_url": video_item_view.get("pic", ""),
}
utils.logger.info(
f"[store.bilibili.update_bilibili_video] bilibili video id:{video_id}, title:{save_content_item.get('title')}")
await BiliStoreFactory.create_store().store_content(content_item=save_content_item)
async def batch_update_bilibili_video_comments(video_id: str, comments: List[Dict]):
if not comments:
return
for comment_item in comments:
await update_bilibili_video_comment(video_id, comment_item)
async def update_bilibili_video_comment(video_id: str, comment_item: Dict):
comment_id = str(comment_item.get("rpid"))
content: Dict = comment_item.get("content")
user_info: Dict = comment_item.get("member")
save_comment_item = {
"comment_id": comment_id,
"create_time": comment_item.get("ctime"),
"video_id": str(video_id),
"content": content.get("message"),
"user_id": user_info.get("mid"),
"nickname": user_info.get("uname"),
"avatar": user_info.get("avatar"),
"sub_comment_count": str(comment_item.get("rcount", 0)),
"last_modify_ts": utils.get_current_timestamp(),
}
utils.logger.info(
f"[store.bilibili.update_bilibili_video_comment] Bilibili video comment: {comment_id}, content: {save_comment_item.get('content')}")
await BiliStoreFactory.create_store().store_comment(comment_item=save_comment_item)

View File

@@ -0,0 +1,55 @@
# -*- coding: utf-8 -*-
# @Author : relakkes@gmail.com
# @Time : 2024/1/14 19:34
# @Desc : B站存储到DB的模型类集合
from tortoise import fields
from tortoise.models import Model
class BilibiliBaseModel(Model):
id = fields.IntField(pk=True, autoincrement=True, description="自增ID")
user_id = fields.CharField(null=True, max_length=64, description="用户ID")
nickname = fields.CharField(null=True, max_length=64, description="用户昵称")
avatar = fields.CharField(null=True, max_length=255, description="用户头像地址")
add_ts = fields.BigIntField(description="记录添加时间戳")
last_modify_ts = fields.BigIntField(description="记录最后修改时间戳")
class Meta:
abstract = True
class BilibiliVideo(BilibiliBaseModel):
video_id = fields.CharField(max_length=64, index=True, description="视频ID")
video_type = fields.CharField(max_length=16, description="视频类型")
title = fields.CharField(null=True, max_length=500, description="视频标题")
desc = fields.TextField(null=True, description="视频描述")
create_time = fields.BigIntField(description="视频发布时间戳", index=True)
liked_count = fields.CharField(null=True, max_length=16, description="视频点赞数")
video_play_count = fields.CharField(null=True, max_length=16, description="视频播放数量")
video_danmaku = fields.CharField(null=True, max_length=16, description="视频弹幕数量")
video_comment = fields.CharField(null=True, max_length=16, description="视频评论数量")
video_url = fields.CharField(null=True, max_length=512, description="视频详情URL")
video_cover_url = fields.CharField(null=True, max_length=512, description="视频封面图 URL")
class Meta:
table = "bilibili_video"
table_description = "B站视频"
def __str__(self):
return f"{self.video_id} - {self.title}"
class BilibiliComment(BilibiliBaseModel):
comment_id = fields.CharField(max_length=64, index=True, description="评论ID")
video_id = fields.CharField(max_length=64, index=True, description="视频ID")
content = fields.TextField(null=True, description="评论内容")
create_time = fields.BigIntField(description="评论时间戳")
sub_comment_count = fields.CharField(max_length=16, description="评论回复数")
class Meta:
table = "bilibili_video_comment"
table_description = "B 站视频评论"
def __str__(self):
return f"{self.comment_id} - {self.content}"

View File

@@ -0,0 +1,129 @@
# -*- coding: utf-8 -*-
# @Author : relakkes@gmail.com
# @Time : 2024/1/14 19:34
# @Desc : B站存储实现类
import csv
import pathlib
from typing import Dict
import aiofiles
from tortoise.contrib.pydantic import pydantic_model_creator
from base.base_crawler import AbstractStore
from tools import utils
from var import crawler_type_var
class BiliCsvStoreImplement(AbstractStore):
csv_store_path: str = "data/bilibili"
def make_save_file_name(self, store_type: str) -> str:
"""
make save file name by store type
Args:
store_type: contents or comments
Returns: eg: data/bilibili/search_comments_20240114.csv ...
"""
return f"{self.csv_store_path}/{crawler_type_var.get()}_{store_type}_{utils.get_current_date()}.csv"
async def save_data_to_csv(self, save_item: Dict, store_type: str):
"""
Below is a simple way to save it in CSV format.
Args:
save_item: save content dict info
store_type: Save type contains content and commentscontents | comments
Returns: no returns
"""
pathlib.Path(self.csv_store_path).mkdir(parents=True, exist_ok=True)
save_file_name = self.make_save_file_name(store_type=store_type)
async with aiofiles.open(save_file_name, mode='a+', encoding="utf-8-sig", newline="") as f:
writer = csv.writer(f)
if await f.tell() == 0:
await writer.writerow(save_item.keys())
await writer.writerow(save_item.values())
async def store_content(self, content_item: Dict):
"""
Bilibili content CSV storage implementation
Args:
content_item: note item dict
Returns:
"""
await self.save_data_to_csv(save_item=content_item, store_type="contents")
async def store_comment(self, comment_item: Dict):
"""
Bilibili comment CSV storage implementation
Args:
comment_item: comment item dict
Returns:
"""
await self.save_data_to_csv(save_item=comment_item, store_type="comments")
class BiliDbStoreImplement(AbstractStore):
async def store_content(self, content_item: Dict):
"""
Bilibili content DB storage implementation
Args:
content_item: content item dict
Returns:
"""
from .bilibili_store_db_types import BilibiliVideo
video_id = content_item.get("video_id")
if not await BilibiliVideo.filter(video_id=video_id).exists():
content_item["add_ts"] = utils.get_current_timestamp()
bilibili_video_pydantic = pydantic_model_creator(BilibiliVideo, name='BilibiliVideoCreate', exclude=('id',))
bilibili_data = bilibili_video_pydantic(**content_item)
bilibili_video_pydantic.model_validate(bilibili_data)
await BilibiliVideo.create(**bilibili_data.model_dump())
else:
bilibili_video_pydantic = pydantic_model_creator(BilibiliVideo, name='BilibiliVideoUpdate',
exclude=('id', 'add_ts'))
bilibili_data = bilibili_video_pydantic(**content_item)
bilibili_video_pydantic.model_validate(bilibili_data)
await BilibiliVideo.filter(video_id=video_id).update(**bilibili_data.model_dump())
async def store_comment(self, comment_item: Dict):
"""
Bilibili content DB storage implementation
Args:
comment_item: comment item dict
Returns:
"""
from .bilibili_store_db_types import BilibiliComment
comment_id = comment_item.get("comment_id")
if not await BilibiliComment.filter(comment_id=comment_id).exists():
comment_item["add_ts"] = utils.get_current_timestamp()
comment_pydantic = pydantic_model_creator(BilibiliComment, name='BilibiliVideoCommentCreate',
exclude=('id',))
comment_data = comment_pydantic(**comment_item)
comment_pydantic.model_validate(comment_data)
await BilibiliComment.create(**comment_data.model_dump())
else:
comment_pydantic = pydantic_model_creator(BilibiliComment, name='BilibiliVideoCommentUpdate',
exclude=('id', 'add_ts'))
comment_data = comment_pydantic(**comment_item)
comment_pydantic.model_validate(comment_data)
await BilibiliComment.filter(comment_id=comment_id).update(**comment_data.model_dump())
class BiliJsonStoreImplement(AbstractStore):
async def store_content(self, content_item: Dict):
pass
async def store_comment(self, comment_item: Dict):
pass