feat: 抖音指定创作者done

This commit is contained in:
Relakkes
2024-05-28 01:07:19 +08:00
parent df1e4a7b02
commit 478db4cc4b
8 changed files with 241 additions and 19 deletions

View File

@@ -92,3 +92,31 @@ async def update_dy_aweme_comment(aweme_id: str, comment_item: Dict):
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)
async def save_creator(user_id: str, creator: Dict):
user_info = creator.get('user', {})
gender_map = {
0: '未知',
1: '',
2: ''
}
avatar_uri = user_info.get('avatar_300x300', {}).get('uri')
local_db_item = {
'user_id': user_id,
'nickname': user_info.get('nickname'),
'gender': gender_map.get(user_info.get('gender'), '未知'),
'avatar': f"https://p3-pc.douyinpic.com/img/{avatar_uri}" + r"~c5_300x300.jpeg?from=2956013662",
'desc': user_info.get('signature'),
'ip_location': user_info.get('ip_location'),
'follows': user_info.get("following_count", 0),
'fans': user_info.get("max_follower_count", 0),
'interaction': user_info.get("total_favorited", 0),
'videos_count': user_info.get("aweme_count", 0),
"last_modify_ts": utils.get_current_timestamp(),
}
utils.logger.info(f"[store.douyin.save_creator] creator:{local_db_item}")
await DouyinStoreFactory.create_store().store_creator(local_db_item)

View File

@@ -26,13 +26,14 @@ def calculate_number_of_files(file_store_path: str) -> int:
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
return max([int(file_name.split("_")[0]) for file_name in os.listdir(file_store_path)]) + 1
except ValueError:
return 1
class DouyinCsvStoreImplement(AbstractStore):
csv_store_path: str = "data/douyin"
file_count:int=calculate_number_of_files(csv_store_path)
file_count: int = calculate_number_of_files(csv_store_path)
def make_save_file_name(self, store_type: str) -> str:
"""
@@ -65,7 +66,7 @@ class DouyinCsvStoreImplement(AbstractStore):
async def store_content(self, content_item: Dict):
"""
Xiaohongshu content CSV storage implementation
Douyin content CSV storage implementation
Args:
content_item: note item dict
@@ -76,7 +77,7 @@ class DouyinCsvStoreImplement(AbstractStore):
async def store_comment(self, comment_item: Dict):
"""
Xiaohongshu comment CSV storage implementation
Douyin comment CSV storage implementation
Args:
comment_item: comment item dict
@@ -85,6 +86,17 @@ class DouyinCsvStoreImplement(AbstractStore):
"""
await self.save_data_to_csv(save_item=comment_item, store_type="comments")
async def store_creator(self, creator: Dict):
"""
Douyin creator CSV storage implementation
Args:
creator: creator item dict
Returns:
"""
await self.save_data_to_csv(save_item=creator, store_type="creator")
class DouyinDbStoreImplement(AbstractStore):
async def store_content(self, content_item: Dict):
@@ -109,7 +121,6 @@ class DouyinDbStoreImplement(AbstractStore):
else:
await update_content_by_content_id(aweme_id, content_item=content_item)
async def store_comment(self, comment_item: Dict):
"""
Douyin content DB storage implementation
@@ -130,11 +141,29 @@ class DouyinDbStoreImplement(AbstractStore):
else:
await update_comment_by_comment_id(comment_id, comment_item=comment_item)
async def store_creator(self, creator: Dict):
"""
Douyin content DB storage implementation
Args:
creator: creator dict
Returns:
"""
from .douyin_store_sql import (add_new_creator, query_creator_by_user_id,
update_creator_by_user_id)
user_id = creator.get("user_id")
user_detail: Dict = await query_creator_by_user_id(user_id)
if not user_detail:
creator["add_ts"] = utils.get_current_timestamp()
await add_new_creator(creator)
else:
await update_creator_by_user_id(user_id, creator)
class DouyinJsonStoreImplement(AbstractStore):
json_store_path: str = "data/douyin"
lock = asyncio.Lock()
file_count:int=calculate_number_of_files(json_store_path)
file_count: int = calculate_number_of_files(json_store_path)
def make_save_file_name(self, store_type: str) -> str:
"""
@@ -146,7 +175,6 @@ class DouyinJsonStoreImplement(AbstractStore):
"""
return f"{self.json_store_path}/{self.file_count}_{crawler_type_var.get()}_{store_type}_{utils.get_current_date()}.json"
async def save_data_to_json(self, save_item: Dict, store_type: str):
@@ -193,3 +221,15 @@ class DouyinJsonStoreImplement(AbstractStore):
"""
await self.save_data_to_json(comment_item, "comments")
async def store_creator(self, creator: Dict):
"""
Douyin creator CSV storage implementation
Args:
creator: creator item dict
Returns:
"""
await self.save_data_to_json(save_item=creator, store_type="creator")

View File

@@ -100,3 +100,49 @@ async def update_comment_by_comment_id(comment_id: str, comment_item: Dict) -> i
async_db_conn: AsyncMysqlDB = media_crawler_db_var.get()
effect_row: int = await async_db_conn.update_table("douyin_aweme_comment", comment_item, "comment_id", comment_id)
return effect_row
async def query_creator_by_user_id(user_id: str) -> Dict:
"""
查询一条创作者记录
Args:
user_id:
Returns:
"""
async_db_conn: AsyncMysqlDB = media_crawler_db_var.get()
sql: str = f"select * from dy_creator where user_id = '{user_id}'"
rows: List[Dict] = await async_db_conn.query(sql)
if len(rows) > 0:
return rows[0]
return dict()
async def add_new_creator(creator_item: Dict) -> int:
"""
新增一条创作者信息
Args:
creator_item:
Returns:
"""
async_db_conn: AsyncMysqlDB = media_crawler_db_var.get()
last_row_id: int = await async_db_conn.item_to_table("dy_creator", creator_item)
return last_row_id
async def update_creator_by_user_id(user_id: str, creator_item: Dict) -> int:
"""
更新一条创作者信息
Args:
user_id:
creator_item:
Returns:
"""
async_db_conn: AsyncMysqlDB = media_crawler_db_var.get()
effect_row: int = await async_db_conn.update_table("dy_creator", creator_item, "user_id", user_id)
return effect_row