208 lines
4.6 KiB
SQL
208 lines
4.6 KiB
SQL
-- 创建新的events表
|
||
CREATE TABLE IF NOT EXISTS events (
|
||
-- 基本信息
|
||
event_id UUID DEFAULT generateUUIDv4(),
|
||
timestamp DateTime DEFAULT now(),
|
||
date Date DEFAULT toDate(now()),
|
||
hour UInt8 DEFAULT toHour(now()),
|
||
-- 实体关联
|
||
user_id String,
|
||
influencer_id String,
|
||
content_id String,
|
||
project_id String,
|
||
-- 事件信息
|
||
event_type Enum8(
|
||
'view' = 1,
|
||
-- 浏览
|
||
'like' = 2,
|
||
-- 点赞
|
||
'unlike' = 3,
|
||
-- 取消点赞
|
||
'follow' = 4,
|
||
-- 关注
|
||
'unfollow' = 5,
|
||
-- 取消关注
|
||
'comment' = 6,
|
||
-- 评论
|
||
'share' = 7,
|
||
-- 分享
|
||
'click' = 8,
|
||
-- 点击链接
|
||
'impression' = 9,
|
||
-- 曝光
|
||
'purchase' = 10,
|
||
-- 购买
|
||
'signup' = 11 -- 注册
|
||
),
|
||
-- 转化漏斗
|
||
funnel_stage Enum8(
|
||
'exposure' = 1,
|
||
-- 曝光
|
||
'interest' = 2,
|
||
-- 兴趣
|
||
'consideration' = 3,
|
||
-- 考虑
|
||
'intent' = 4,
|
||
-- 意向
|
||
'evaluation' = 5,
|
||
-- 评估
|
||
'purchase' = 6 -- 购买
|
||
),
|
||
-- 内容信息
|
||
platform String,
|
||
-- 社交平台
|
||
content_type Enum8(
|
||
'video' = 1,
|
||
'image' = 2,
|
||
'text' = 3,
|
||
'story' = 4,
|
||
'reel' = 5,
|
||
'live' = 6
|
||
),
|
||
content_status Enum8(
|
||
-- 审核状态
|
||
'approved' = 1,
|
||
'pending' = 2,
|
||
'rejected' = 3
|
||
),
|
||
-- 互动分析
|
||
sentiment Enum8(
|
||
-- 情感分析
|
||
'positive' = 1,
|
||
'neutral' = 2,
|
||
'negative' = 3
|
||
),
|
||
comment_text String,
|
||
-- 评论文本
|
||
keywords Array(String),
|
||
-- 关键词
|
||
-- 数值指标
|
||
interaction_value Float64,
|
||
-- 互动价值
|
||
followers_count UInt32,
|
||
-- 粉丝数
|
||
followers_change Int32,
|
||
-- 粉丝变化量
|
||
likes_count UInt32,
|
||
-- 点赞数
|
||
likes_change Int32,
|
||
-- 点赞变化量
|
||
views_count UInt32,
|
||
-- 观看数
|
||
-- 设备信息
|
||
ip String,
|
||
user_agent String,
|
||
device_type String,
|
||
referrer String,
|
||
-- 地理信息
|
||
geo_country String,
|
||
geo_city String,
|
||
-- 会话信息
|
||
session_id String
|
||
) ENGINE = MergeTree() PARTITION BY toYYYYMM(timestamp)
|
||
ORDER BY
|
||
(event_type, influencer_id, date, hour) SETTINGS index_granularity = 8192;
|
||
|
||
-- 创建influencers表
|
||
CREATE TABLE IF NOT EXISTS influencers (
|
||
influencer_id String,
|
||
name String,
|
||
platform String,
|
||
profile_url String,
|
||
project_id String,
|
||
date Date DEFAULT toDate(now()),
|
||
followers UInt32 DEFAULT 0,
|
||
PRIMARY KEY (influencer_id)
|
||
) ENGINE = MergeTree();
|
||
|
||
-- 创建posts表
|
||
CREATE TABLE IF NOT EXISTS posts (
|
||
post_id String,
|
||
title String,
|
||
influencer_id String,
|
||
project_id String,
|
||
platform String,
|
||
type String DEFAULT 'post',
|
||
format String DEFAULT 'text',
|
||
date Date DEFAULT toDate(now()),
|
||
timestamp DateTime DEFAULT now(),
|
||
created_at DateTime DEFAULT now(),
|
||
views UInt32 DEFAULT 0,
|
||
likes UInt32 DEFAULT 0,
|
||
comments UInt32 DEFAULT 0,
|
||
shares UInt32 DEFAULT 0,
|
||
PRIMARY KEY (post_id)
|
||
) ENGINE = MergeTree();
|
||
|
||
CREATE TABLE IF NOT EXISTS sync_logs (
|
||
timestamp DateTime DEFAULT now(),
|
||
duration_ms UInt64,
|
||
posts_synced UInt32,
|
||
comments_synced UInt32,
|
||
influencer_changes_synced UInt32,
|
||
projects_synced UInt32,
|
||
success UInt8,
|
||
error_messages String
|
||
) ENGINE = MergeTree()
|
||
ORDER BY
|
||
(timestamp);
|
||
|
||
-- 创建专门的comments表存储评论数据
|
||
CREATE TABLE IF NOT EXISTS comments (
|
||
-- 基本标识信息
|
||
comment_id String,
|
||
-- 评论唯一ID
|
||
post_id String,
|
||
-- 关联的帖子ID
|
||
user_id String,
|
||
-- 发表评论的用户ID
|
||
-- 时间信息
|
||
created_at DateTime DEFAULT now(),
|
||
-- 评论创建时间
|
||
date Date DEFAULT toDate(created_at),
|
||
-- 日期(用于分区)
|
||
-- 评论内容信息
|
||
content String,
|
||
-- 评论内容
|
||
sentiment_score Float64,
|
||
-- 情感分数(-1到1)
|
||
sentiment Enum8(
|
||
-- 情感分类
|
||
'positive' = 1,
|
||
'neutral' = 2,
|
||
'negative' = 3
|
||
),
|
||
-- 关联信息
|
||
project_id String,
|
||
-- 项目ID
|
||
influencer_id String,
|
||
-- 网红ID
|
||
platform String,
|
||
-- 评论所在平台
|
||
-- 互动信息
|
||
likes_count UInt32 DEFAULT 0,
|
||
-- 点赞数量
|
||
replies_count UInt32 DEFAULT 0,
|
||
-- 回复数量
|
||
-- 元数据
|
||
parent_comment_id String DEFAULT '',
|
||
-- 父评论ID(用于回复)
|
||
is_reply UInt8 DEFAULT 0,
|
||
-- 是否为回复(0=否，1=是)
|
||
-- 同步信息
|
||
is_synced UInt8 DEFAULT 1,
|
||
-- 是否已同步(0=否，1=是)
|
||
last_updated DateTime DEFAULT now(),
|
||
-- 最后更新时间
|
||
-- 分析信息
|
||
keywords Array(String),
|
||
-- 提取的关键词
|
||
topics Array(String),
|
||
-- 关联的话题
|
||
-- 内部处理信息
|
||
is_active UInt8 DEFAULT 1,
|
||
-- 是否活跃(0=已删除/隐藏，1=活跃)
|
||
is_spam UInt8 DEFAULT 0 -- 是否为垃圾评论(0=否，1=是)
|
||
) ENGINE = MergeTree() PARTITION BY toYYYYMM(date)
|
||
ORDER BY
|
||
(post_id, created_at, comment_id) SETTINGS index_granularity = 8192; |