title-embedding-base-on-milvus/CollectionTrans.py

from pymilvus import Collection, CollectionSchema, FieldSchema, DataType, connections

# 连接到 Milvus
connections.connect("default", host="127.0.0.1", port="19530")

# 定义新的字段和 schema
dim = 768  # 假设 dim 已经定义
new_fields = [
    # 旧字段
    FieldSchema(name="id", dtype=DataType.INT64, is_primary=True),
    FieldSchema(name="title_vector",
                dtype=DataType.FLOAT_VECTOR, dim=dim),
    FieldSchema(name="title", dtype=DataType.VARCHAR, max_length=256,
                description="视频标题"),
    FieldSchema(name="preview_times",
                dtype=DataType.INT64, description="预曝光次数"),
    FieldSchema(name="preview_users", dtype=DataType.INT64,
                description="预曝光用户数"),
    FieldSchema(name="view_times", dtype=DataType.INT64,
                description="曝光次数"),
    FieldSchema(name="view_users", dtype=DataType.INT64,
                description="曝光用户数"),
    FieldSchema(name="play_times", dtype=DataType.INT64,
                description="播放次数"),
    FieldSchema(name="play_users", dtype=DataType.INT64,
                description="播放用户数"),
    FieldSchema(name="share_times", dtype=DataType.INT64,
                description="分享次数"),
    FieldSchema(name="share_users", dtype=DataType.INT64,
                description="分享用户数"),
    FieldSchema(name="return_times",
                dtype=DataType.INT64, description="回看次数"),
    FieldSchema(name="return_users",
                dtype=DataType.INT64, description="回看用户数"),
    # 新字段
    FieldSchema(name="create_time", dtype=DataType.INT64, description="创建时间戳")
]
new_schema = CollectionSchema(new_fields, description="两年内的分发过的视频标题")

# 创建新 collection
new_collection_name = "two_year_all_viewed_videos"
new_collection = Collection(name=new_collection_name, schema=new_schema)

# 从旧 collection 中读取数据并迁移
old_collection_name = "two_year_all_video_titles"
old_collection = Collection(name=old_collection_name)
old_data = ...  # 读取旧 collection 数据的逻辑

# 为旧数据添加 '创建时间' 字段
new_data = ...
for data in old_data:
    data['创建时间'] = ...  # 设置 '创建时间' 字段的值
    new_data.append(data)

# 插入数据到新 collection
new_collection.insert(new_data)

# 验证新 collection 数据正确后，可以选择删除旧 collection
# old_collection.drop()