Files
MediaCrawler/store/tieba/_store_impl.py
persist-1 be306c6f54 refactor(database): 重构数据库存储实现,使用SQLAlchemy ORM替代原始SQL操作
- 删除旧的async_db.py和async_sqlite_db.py实现
- 新增SQLAlchemy ORM模型和数据库会话管理
- 统一各平台存储实现到_store_impl.py文件
- 添加数据库初始化功能支持
- 更新.gitignore和pyproject.toml依赖配置
- 优化文件存储路径和命名规范
2025-09-06 04:10:20 +08:00

192 lines
5.9 KiB
Python

# 声明:本代码仅供学习和研究目的使用。使用者应遵守以下原则:
# 1. 不得用于任何商业用途。
# 2. 使用时应遵守目标平台的使用条款和robots.txt规则。
# 3. 不得进行大规模爬取或对平台造成运营干扰。
# 4. 应合理控制请求频率,避免给目标平台带来不必要的负担。
# 5. 不得用于任何非法或不当的用途。
#
# 详细许可条款请参阅项目根目录下的LICENSE文件。
# 使用本代码即表示您同意遵守上述原则和LICENSE中的所有条款。
# -*- coding: utf-8 -*-
# @Author : persist1@126.com
# @Time : 2025/9/5 19:34
# @Desc : 贴吧存储实现类
import asyncio
import csv
import json
import os
import pathlib
from typing import Dict
import aiofiles
from sqlalchemy import select
from sqlalchemy.ext.asyncio import AsyncSession
import config
from base.base_crawler import AbstractStore
from database.models import TiebaNote, TiebaComment, TiebaCreator
from tools import utils, words
from database.db_session import get_session
from var import crawler_type_var
def calculate_number_of_files(file_store_path: str) -> int:
"""计算数据保存文件的前部分排序数字,支持每次运行代码不写到同一个文件中
Args:
file_store_path;
Returns:
file nums
"""
if not os.path.exists(file_store_path):
return 1
try:
return max([int(file_name.split("_")[0]) for file_name in os.listdir(file_store_path)]) + 1
except ValueError:
return 1
class TieBaCsvStoreImplement(AbstractStore):
def __init__(self, **kwargs):
super().__init__(**kwargs)
self.writer = AsyncFileWriter(platform="tieba", crawler_type=kwargs.get("crawler_type"))
async def store_content(self, content_item: Dict):
"""
tieba content CSV storage implementation
Args:
content_item: note item dict
Returns:
"""
await self.writer.write_to_csv(item_type="contents", item=content_item)
async def store_comment(self, comment_item: Dict):
"""
tieba comment CSV storage implementation
Args:
comment_item: comment item dict
Returns:
"""
await self.writer.write_to_csv(item_type="comments", item=comment_item)
async def store_creator(self, creator: Dict):
"""
tieba content CSV storage implementation
Args:
creator: creator dict
Returns:
"""
await self.writer.write_to_csv(item_type="creators", item=creator)
class TieBaDbStoreImplement(AbstractStore):
async def store_content(self, content_item: Dict):
"""
tieba content DB storage implementation
Args:
content_item: content item dict
"""
note_id = content_item.get("note_id")
async with get_session() as session:
stmt = select(TiebaNote).where(TiebaNote.note_id == note_id)
res = await session.execute(stmt)
db_note = res.scalar_one_or_none()
if db_note:
for key, value in content_item.items():
setattr(db_note, key, value)
else:
db_note = TiebaNote(**content_item)
session.add(db_note)
await session.commit()
async def store_comment(self, comment_item: Dict):
"""
tieba content DB storage implementation
Args:
comment_item: comment item dict
"""
comment_id = comment_item.get("comment_id")
async with get_session() as session:
stmt = select(TiebaComment).where(TiebaComment.comment_id == comment_id)
res = await session.execute(stmt)
db_comment = res.scalar_one_or_none()
if db_comment:
for key, value in comment_item.items():
setattr(db_comment, key, value)
else:
db_comment = TiebaComment(**comment_item)
session.add(db_comment)
await session.commit()
async def store_creator(self, creator: Dict):
"""
tieba content DB storage implementation
Args:
creator: creator dict
"""
user_id = creator.get("user_id")
async with get_session() as session:
stmt = select(TiebaCreator).where(TiebaCreator.user_id == user_id)
res = await session.execute(stmt)
db_creator = res.scalar_one_or_none()
if db_creator:
for key, value in creator.items():
setattr(db_creator, key, value)
else:
db_creator = TiebaCreator(**creator)
session.add(db_creator)
await session.commit()
class TieBaJsonStoreImplement(AbstractStore):
def __init__(self, **kwargs):
super().__init__(**kwargs)
self.writer = AsyncFileWriter(platform="tieba", crawler_type=kwargs.get("crawler_type"))
async def store_content(self, content_item: Dict):
"""
tieba content JSON storage implementation
Args:
content_item: note item dict
Returns:
"""
await self.writer.write_single_item_to_json(item_type="contents", item=content_item)
async def store_comment(self, comment_item: Dict):
"""
tieba comment JSON storage implementation
Args:
comment_item: comment item dict
Returns:
"""
await self.writer.write_single_item_to_json(item_type="comments", item=comment_item)
async def store_creator(self, creator: Dict):
"""
tieba content JSON storage implementation
Args:
creator: creator dict
Returns:
"""
await self.writer.write_single_item_to_json(item_type="creators", item=creator)
class TieBaSqliteStoreImplement(TieBaDbStoreImplement):
"""
Tieba sqlite store implement
"""
pass