Files
MediaCrawler/store/douyin/_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

198 lines
5.8 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 : 抖音存储实现类
import asyncio
import json
import os
import pathlib
from typing import Dict
from sqlalchemy import select
import config
from base.base_crawler import AbstractStore
from database.db_session import get_session
from database.models import DouyinAweme, DouyinAwemeComment, DyCreator
from tools import utils, words
from tools.async_file_writer import AsyncFileWriter
from var import crawler_type_var
class DouyinCsvStoreImplement(AbstractStore):
def __init__(self):
self.file_writer = AsyncFileWriter(
crawler_type=crawler_type_var.get(),
platform="douyin"
)
async def store_content(self, content_item: Dict):
"""
Douyin content CSV storage implementation
Args:
content_item: note item dict
Returns:
"""
await self.file_writer.write_to_csv(
item=content_item,
item_type="contents"
)
async def store_comment(self, comment_item: Dict):
"""
Douyin comment CSV storage implementation
Args:
comment_item: comment item dict
Returns:
"""
await self.file_writer.write_to_csv(
item=comment_item,
item_type="comments"
)
async def store_creator(self, creator: Dict):
"""
Douyin creator CSV storage implementation
Args:
creator: creator item dict
Returns:
"""
await self.file_writer.write_to_csv(
item=creator,
item_type="creators"
)
class DouyinDbStoreImplement(AbstractStore):
async def store_content(self, content_item: Dict):
"""
Douyin content DB storage implementation
Args:
content_item: content item dict
"""
aweme_id = content_item.get("aweme_id")
async with get_session() as session:
result = await session.execute(select(DouyinAweme).where(DouyinAweme.aweme_id == aweme_id))
aweme_detail = result.scalar_one_or_none()
if not aweme_detail:
content_item["add_ts"] = utils.get_current_timestamp()
if content_item.get("title"):
new_content = DouyinAweme(**content_item)
session.add(new_content)
else:
for key, value in content_item.items():
setattr(aweme_detail, key, value)
await session.commit()
async def store_comment(self, comment_item: Dict):
"""
Douyin comment DB storage implementation
Args:
comment_item: comment item dict
"""
comment_id = comment_item.get("comment_id")
async with get_session() as session:
result = await session.execute(select(DouyinAwemeComment).where(DouyinAwemeComment.comment_id == comment_id))
comment_detail = result.scalar_one_or_none()
if not comment_detail:
comment_item["add_ts"] = utils.get_current_timestamp()
new_comment = DouyinAwemeComment(**comment_item)
session.add(new_comment)
else:
for key, value in comment_item.items():
setattr(comment_detail, key, value)
await session.commit()
async def store_creator(self, creator: Dict):
"""
Douyin creator DB storage implementation
Args:
creator: creator dict
"""
user_id = creator.get("user_id")
async with get_session() as session:
result = await session.execute(select(DyCreator).where(DyCreator.user_id == user_id))
user_detail = result.scalar_one_or_none()
if not user_detail:
creator["add_ts"] = utils.get_current_timestamp()
new_creator = DyCreator(**creator)
session.add(new_creator)
else:
for key, value in creator.items():
setattr(user_detail, key, value)
await session.commit()
class DouyinJsonStoreImplement(AbstractStore):
def __init__(self):
self.file_writer = AsyncFileWriter(
crawler_type=crawler_type_var.get(),
platform="douyin"
)
async def store_content(self, content_item: Dict):
"""
content JSON storage implementation
Args:
content_item:
Returns:
"""
await self.file_writer.write_single_item_to_json(
item=content_item,
item_type="contents"
)
async def store_comment(self, comment_item: Dict):
"""
comment JSON storage implementation
Args:
comment_item:
Returns:
"""
await self.file_writer.write_single_item_to_json(
item=comment_item,
item_type="comments"
)
async def store_creator(self, creator: Dict):
"""
creator JSON storage implementation
Args:
creator:
Returns:
"""
await self.file_writer.write_single_item_to_json(
item=creator,
item_type="creators"
)
class DouyinSqliteStoreImplement(DouyinDbStoreImplement):
pass