mirror of
https://github.com/NanmiCoder/MediaCrawler.git
synced 2026-06-09 19:37:25 +08:00
refactor: 数据存储重构，分离不同类型的存储实现
This commit is contained in:
Relakkes
95
store/douyin/__init__.py
Normal file
95
store/douyin/__init__.py
Normal file
@@ -0,0 +1,95 @@
|
||||
# -*- coding: utf-8 -*-
|
||||
# @Author : relakkes@gmail.com
|
||||
# @Time : 2024/1/14 18:46
|
||||
# @Desc :
|
||||
from typing import List
|
||||
|
||||
import config
|
||||
|
||||
from .douyin_store_db_types import *
|
||||
from .douyin_store_impl import *
|
||||
|
||||
|
||||
class DouyinStoreFactory:
|
||||
STORES = {
|
||||
"csv": DouyinCsvStoreImplement,
|
||||
"db": DouyinDbStoreImplement,
|
||||
"json": DouyinJsonStoreImplement
|
||||
}
|
||||
|
||||
@staticmethod
|
||||
def create_store() -> AbstractStore:
|
||||
store_class = DouyinStoreFactory.STORES.get(config.SAVE_DATA_OPTION)
|
||||
if not store_class:
|
||||
raise ValueError(
|
||||
"[DouyinStoreFactory.create_store] Invalid save option only supported csv or db or json ...")
|
||||
return store_class()
|
||||
|
||||
|
||||
async def update_douyin_aweme(aweme_item: Dict):
|
||||
aweme_id = aweme_item.get("aweme_id")
|
||||
user_info = aweme_item.get("author", {})
|
||||
interact_info = aweme_item.get("statistics", {})
|
||||
save_content_item = {
|
||||
"aweme_id": aweme_id,
|
||||
"aweme_type": str(aweme_item.get("aweme_type")),
|
||||
"title": aweme_item.get("desc", ""),
|
||||
"desc": aweme_item.get("desc", ""),
|
||||
"create_time": aweme_item.get("create_time"),
|
||||
"user_id": user_info.get("uid"),
|
||||
"sec_uid": user_info.get("sec_uid"),
|
||||
"short_user_id": user_info.get("short_id"),
|
||||
"user_unique_id": user_info.get("unique_id"),
|
||||
"user_signature": user_info.get("signature"),
|
||||
"nickname": user_info.get("nickname"),
|
||||
"avatar": user_info.get("avatar_thumb", {}).get("url_list", [""])[0],
|
||||
"liked_count": str(interact_info.get("digg_count")),
|
||||
"collected_count": str(interact_info.get("collect_count")),
|
||||
"comment_count": str(interact_info.get("comment_count")),
|
||||
"share_count": str(interact_info.get("share_count")),
|
||||
"ip_location": aweme_item.get("ip_label", ""),
|
||||
"last_modify_ts": utils.get_current_timestamp(),
|
||||
"aweme_url": f"https://www.douyin.com/video/{aweme_id}"
|
||||
}
|
||||
utils.logger.info(
|
||||
f"[store.douyin.update_douyin_aweme] douyin aweme id:{aweme_id}, title:{save_content_item.get('title')}")
|
||||
await DouyinStoreFactory.create_store().store_content(content_item=save_content_item)
|
||||
|
||||
|
||||
async def batch_update_dy_aweme_comments(aweme_id: str, comments: List[Dict]):
|
||||
if not comments:
|
||||
return
|
||||
for comment_item in comments:
|
||||
await update_dy_aweme_comment(aweme_id, comment_item)
|
||||
|
||||
|
||||
async def update_dy_aweme_comment(aweme_id: str, comment_item: Dict):
|
||||
comment_aweme_id = comment_item.get("aweme_id")
|
||||
if aweme_id != comment_aweme_id:
|
||||
utils.logger.error(
|
||||
f"[store.douyin.update_dy_aweme_comment] comment_aweme_id: {comment_aweme_id} != aweme_id: {aweme_id}")
|
||||
return
|
||||
user_info = comment_item.get("user", {})
|
||||
comment_id = comment_item.get("cid")
|
||||
avatar_info = user_info.get("avatar_medium", {}) or user_info.get("avatar_300x300", {}) or user_info.get(
|
||||
"avatar_168x168", {}) or user_info.get("avatar_thumb", {}) or {}
|
||||
save_comment_item = {
|
||||
"comment_id": comment_id,
|
||||
"create_time": comment_item.get("create_time"),
|
||||
"ip_location": comment_item.get("ip_label", ""),
|
||||
"aweme_id": aweme_id,
|
||||
"content": comment_item.get("text"),
|
||||
"user_id": user_info.get("uid"),
|
||||
"sec_uid": user_info.get("sec_uid"),
|
||||
"short_user_id": user_info.get("short_id"),
|
||||
"user_unique_id": user_info.get("unique_id"),
|
||||
"user_signature": user_info.get("signature"),
|
||||
"nickname": user_info.get("nickname"),
|
||||
"avatar": avatar_info.get("url_list", [""])[0],
|
||||
"sub_comment_count": str(comment_item.get("reply_comment_total", 0)),
|
||||
"last_modify_ts": utils.get_current_timestamp(),
|
||||
}
|
||||
utils.logger.info(
|
||||
f"[store.douyin.update_dy_aweme_comment] douyin aweme comment: {comment_id}, content: {save_comment_item.get('content')}")
|
||||
|
||||
await DouyinStoreFactory.create_store().store_comment(comment_item=save_comment_item)
|
||||
60
store/douyin/douyin_store_db_types.py
Normal file
60
store/douyin/douyin_store_db_types.py
Normal file
@@ -0,0 +1,60 @@
|
||||
# -*- coding: utf-8 -*-
|
||||
# @Author : relakkes@gmail.com
|
||||
# @Time : 2024/1/14 18:50
|
||||
# @Desc : 抖音存储到DB的模型类集合
|
||||
|
||||
|
||||
from tortoise import fields
|
||||
from tortoise.models import Model
|
||||
|
||||
|
||||
class DouyinBaseModel(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")
|
||||
sec_uid = fields.CharField(null=True, max_length=128, description="用户sec_uid")
|
||||
short_user_id = fields.CharField(null=True, max_length=64, description="用户短ID")
|
||||
user_unique_id = fields.CharField(null=True, max_length=64, description="用户唯一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")
|
||||
user_signature = fields.CharField(null=True, max_length=500, description="用户签名")
|
||||
ip_location = fields.CharField(null=True, max_length=255, description="评论时的IP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 DouyinAweme(DouyinBaseModel):
|
||||
aweme_id = fields.CharField(max_length=64, index=True, description="视频ID")
|
||||
aweme_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")
|
||||
comment_count = fields.CharField(null=True, max_length=16, description="视频评论数")
|
||||
share_count = fields.CharField(null=True, max_length=16, description="视频分享数")
|
||||
collected_count = fields.CharField(null=True, max_length=16, description="视频收藏数")
|
||||
aweme_url = fields.CharField(null=True, max_length=255, description="视频详情页URL")
|
||||
|
||||
class Meta:
|
||||
table = "douyin_aweme"
|
||||
table_description = "抖音视频"
|
||||
|
||||
def __str__(self):
|
||||
return f"{self.aweme_id} - {self.title}"
|
||||
|
||||
|
||||
class DouyinAwemeComment(DouyinBaseModel):
|
||||
comment_id = fields.CharField(max_length=64, index=True, description="评论ID")
|
||||
aweme_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 = "douyin_aweme_comment"
|
||||
table_description = "抖音视频评论"
|
||||
|
||||
def __str__(self):
|
||||
return f"{self.comment_id} - {self.content}"
|
||||
130
store/douyin/douyin_store_impl.py
Normal file
130
store/douyin/douyin_store_impl.py
Normal file
@@ -0,0 +1,130 @@
|
||||
# -*- coding: utf-8 -*-
|
||||
# @Author : relakkes@gmail.com
|
||||
# @Time : 2024/1/14 18:46
|
||||
# @Desc : 抖音存储实现类
|
||||
|
||||
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 DouyinCsvStoreImplement(AbstractStore):
|
||||
csv_store_path: str = "data/douyin"
|
||||
|
||||
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/douyin/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 comments（contents | 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):
|
||||
"""
|
||||
Xiaohongshu 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):
|
||||
"""
|
||||
Xiaohongshu 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 DouyinDbStoreImplement(AbstractStore):
|
||||
async def store_content(self, content_item: Dict):
|
||||
"""
|
||||
Douyin content DB storage implementation
|
||||
Args:
|
||||
content_item: content item dict
|
||||
|
||||
Returns:
|
||||
|
||||
"""
|
||||
from .douyin_store_db_types import DouyinAweme
|
||||
aweme_id = content_item.get("aweme_id")
|
||||
if not await DouyinAweme.filter(aweme_id=aweme_id).exists():
|
||||
content_item["add_ts"] = utils.get_current_timestamp()
|
||||
douyin_aweme_pydantic = pydantic_model_creator(DouyinAweme, name='DouyinAwemeCreate', exclude=('id',))
|
||||
douyin_data = douyin_aweme_pydantic(**content_item)
|
||||
douyin_aweme_pydantic.model_validate(douyin_data)
|
||||
await DouyinAweme.create(**douyin_data.dict())
|
||||
else:
|
||||
douyin_aweme_pydantic = pydantic_model_creator(DouyinAweme, name='DouyinAwemeUpdate',
|
||||
exclude=('id', 'add_ts'))
|
||||
douyin_data = douyin_aweme_pydantic(**content_item)
|
||||
douyin_aweme_pydantic.model_validate(douyin_data)
|
||||
await DouyinAweme.filter(aweme_id=aweme_id).update(**douyin_data.model_dump())
|
||||
|
||||
async def store_comment(self, comment_item: Dict):
|
||||
"""
|
||||
Douyin content DB storage implementation
|
||||
Args:
|
||||
comment_item: comment item dict
|
||||
|
||||
Returns:
|
||||
|
||||
"""
|
||||
from .douyin_store_db_types import DouyinAwemeComment
|
||||
comment_id = comment_item.get("comment_id")
|
||||
if not await DouyinAwemeComment.filter(comment_id=comment_id).exists():
|
||||
comment_item["add_ts"] = utils.get_current_timestamp()
|
||||
comment_pydantic = pydantic_model_creator(DouyinAwemeComment, name='DouyinAwemeCommentCreate',
|
||||
exclude=('id',))
|
||||
comment_data = comment_pydantic(**comment_item)
|
||||
comment_pydantic.model_validate(comment_data)
|
||||
await DouyinAwemeComment.create(**comment_data.model_dump())
|
||||
else:
|
||||
comment_pydantic = pydantic_model_creator(DouyinAwemeComment, name='DouyinAwemeCommentUpdate',
|
||||
exclude=('id', 'add_ts'))
|
||||
comment_data = comment_pydantic(**comment_item)
|
||||
comment_pydantic.model_validate(comment_data)
|
||||
await DouyinAwemeComment.filter(comment_id=comment_item).update(**comment_data.model_dump())
|
||||
|
||||
|
||||
class DouyinJsonStoreImplement(AbstractStore):
|
||||
async def store_content(self, content_item: Dict):
|
||||
pass
|
||||
|
||||
async def store_comment(self, comment_item: Dict):
|
||||
pass
|
||||
|
||||
Reference in New Issue
Block a user