feat: 添加 Mermaid Shortcode 支持（推荐方式）
- 新增 [mermaid]...[/mermaid] shortcode - 支持 theme、width、height、align 参数 - 不依赖 WP-Markdown 的处理方式 - 不会被 WordPress 自动格式化破坏 - 在原生编辑器中清晰可见 - 添加完整的使用指南和示例
This commit is contained in:
365
tests/test-shortcode-example.md
Normal file
365
tests/test-shortcode-example.md
Normal file
@@ -0,0 +1,365 @@
|
||||
# AI 垃圾评论检测系统架构
|
||||
|
||||
## 系统概述
|
||||
|
||||
本文档展示 Argon 主题的 AI 垃圾评论检测系统的完整架构和流程。
|
||||
|
||||
## 主流程图
|
||||
|
||||
[mermaid theme="default" width="100%" align="center"]
|
||||
flowchart TD
|
||||
Start([用户提交评论]) --> PreProcess[预处理评论]
|
||||
PreProcess --> CheckEnabled{启用 AI 检测?}
|
||||
|
||||
CheckEnabled -->|否| DirectSave[直接保存评论]
|
||||
CheckEnabled -->|是| CheckMode{检测模式?}
|
||||
|
||||
CheckMode -->|manual| DirectSave
|
||||
CheckMode -->|keyword| CheckKeyword[检查关键字]
|
||||
CheckMode -->|sample| CheckSample[智能抽样]
|
||||
CheckMode -->|all| NeedCheck[需要检测]
|
||||
|
||||
CheckKeyword --> KeywordMatch{匹配关键字?}
|
||||
KeywordMatch -->|是| NeedCheck
|
||||
KeywordMatch -->|否| DirectSave
|
||||
|
||||
CheckSample --> SampleDecision{抽中?}
|
||||
SampleDecision -->|是| NeedCheck
|
||||
SampleDecision -->|否| DirectSave
|
||||
|
||||
NeedCheck --> SetPending[设置待审核状态]
|
||||
SetPending --> SaveWithFlag[保存评论并标记]
|
||||
SaveWithFlag --> TriggerDetect[触发 AI 检测]
|
||||
|
||||
TriggerDetect --> CallAPI[调用 AI API]
|
||||
CallAPI --> ParseResult[解析结果]
|
||||
|
||||
ParseResult --> CheckConfidence{置信度?}
|
||||
CheckConfidence -->|高| AutoHandle[自动处理]
|
||||
CheckConfidence -->|中| ManualReview[人工审核]
|
||||
CheckConfidence -->|低| MarkNormal[标记正常]
|
||||
|
||||
AutoHandle --> CheckAction{处理方式?}
|
||||
CheckAction -->|trash| MoveTrash[移入回收站]
|
||||
CheckAction -->|hold| KeepPending[保持待审核]
|
||||
CheckAction -->|mark| JustMark[仅标记]
|
||||
|
||||
MoveTrash --> NotifyAdmin[通知管理员]
|
||||
KeepPending --> NotifyAdmin
|
||||
JustMark --> SaveResult[保存结果]
|
||||
ManualReview --> SaveResult
|
||||
MarkNormal --> SaveResult
|
||||
NotifyAdmin --> SaveResult
|
||||
|
||||
SaveResult --> AILearn{启用学习?}
|
||||
AILearn -->|是| ExtractKeywords[提取关键词]
|
||||
AILearn -->|否| End([结束])
|
||||
ExtractKeywords --> UpdateDB[更新词库]
|
||||
UpdateDB --> End
|
||||
|
||||
DirectSave --> End
|
||||
|
||||
style Start fill:#e1f5e1,stroke:#2e7d32,stroke-width:2px
|
||||
style End fill:#e1f5e1,stroke:#2e7d32,stroke-width:2px
|
||||
style MoveTrash fill:#ff6b6b,stroke:#c62828,stroke-width:2px
|
||||
style MarkNormal fill:#95e1d3,stroke:#2e7d32,stroke-width:2px
|
||||
style ManualReview fill:#ffa500,stroke:#ff8f00,stroke-width:2px
|
||||
[/mermaid]
|
||||
|
||||
## 检测模式详解
|
||||
|
||||
[mermaid theme="default" width="80%" align="center"]
|
||||
flowchart LR
|
||||
subgraph Manual[Manual 模式]
|
||||
M1[关闭实时检测]
|
||||
M2[仅手动扫描]
|
||||
end
|
||||
|
||||
subgraph Keyword[Keyword 模式]
|
||||
K1[关键字触发]
|
||||
K2[精准检测]
|
||||
K3[低成本]
|
||||
end
|
||||
|
||||
subgraph Sample[Sample 模式]
|
||||
S1[智能抽样]
|
||||
S2[平衡准确性]
|
||||
S3[控制成本]
|
||||
end
|
||||
|
||||
subgraph All[All 模式]
|
||||
A1[全量检测]
|
||||
A2[最高准确性]
|
||||
A3[高成本]
|
||||
end
|
||||
|
||||
style Manual fill:#e3f2fd
|
||||
style Keyword fill:#fff3e0
|
||||
style Sample fill:#f3e5f5
|
||||
style All fill:#fce4ec
|
||||
[/mermaid]
|
||||
|
||||
## AI 检测流程
|
||||
|
||||
[mermaid theme="default" width="90%"]
|
||||
sequenceDiagram
|
||||
participant User as 用户
|
||||
participant WP as WordPress
|
||||
participant Argon as Argon 主题
|
||||
participant AI as AI API
|
||||
participant DB as 数据库
|
||||
|
||||
User->>WP: 提交评论
|
||||
WP->>Argon: 触发 preprocess_comment
|
||||
Argon->>Argon: 检查检测规则
|
||||
|
||||
alt 需要检测
|
||||
Argon->>DB: 保存评论（待审核）
|
||||
Argon->>AI: 发送检测请求
|
||||
AI-->>Argon: 返回检测结果
|
||||
Argon->>Argon: 解析结果和置信度
|
||||
|
||||
alt 高置信度垃圾评论
|
||||
Argon->>DB: 移入回收站
|
||||
Argon->>WP: 发送通知邮件
|
||||
else 中等置信度
|
||||
Argon->>DB: 标记待人工审核
|
||||
else 正常评论
|
||||
Argon->>DB: 标记为正常
|
||||
end
|
||||
|
||||
Argon->>DB: 保存检测记录
|
||||
|
||||
opt 启用 AI 学习
|
||||
Argon->>AI: 提取关键词
|
||||
AI-->>Argon: 返回关键词
|
||||
Argon->>DB: 更新学习词库
|
||||
end
|
||||
else 跳过检测
|
||||
Argon->>DB: 直接保存评论
|
||||
end
|
||||
|
||||
WP-->>User: 显示提交结果
|
||||
[/mermaid]
|
||||
|
||||
## 数据库结构
|
||||
|
||||
[mermaid theme="default" width="85%"]
|
||||
erDiagram
|
||||
COMMENT ||--o{ COMMENT_META : has
|
||||
COMMENT {
|
||||
bigint comment_ID PK
|
||||
bigint comment_post_ID FK
|
||||
text comment_content
|
||||
varchar comment_author
|
||||
varchar comment_author_email
|
||||
varchar comment_author_IP
|
||||
datetime comment_date
|
||||
varchar comment_approved
|
||||
}
|
||||
|
||||
COMMENT_META {
|
||||
bigint meta_id PK
|
||||
bigint comment_id FK
|
||||
varchar meta_key
|
||||
longtext meta_value
|
||||
}
|
||||
|
||||
SPAM_DETECTION ||--|| COMMENT : detects
|
||||
SPAM_DETECTION {
|
||||
bigint id PK
|
||||
bigint comment_id FK
|
||||
varchar result
|
||||
float confidence
|
||||
text reason
|
||||
text keywords
|
||||
datetime detected_at
|
||||
varchar detection_code
|
||||
}
|
||||
|
||||
LEARNED_KEYWORDS ||--o{ SPAM_DETECTION : learns_from
|
||||
LEARNED_KEYWORDS {
|
||||
bigint id PK
|
||||
varchar keyword
|
||||
int weight
|
||||
varchar category
|
||||
datetime created_at
|
||||
datetime updated_at
|
||||
}
|
||||
[/mermaid]
|
||||
|
||||
## 系统状态机
|
||||
|
||||
[mermaid theme="default" width="70%"]
|
||||
stateDiagram-v2
|
||||
[*] --> Submitted: 用户提交
|
||||
|
||||
Submitted --> Pending: 需要检测
|
||||
Submitted --> Approved: 跳过检测
|
||||
|
||||
Pending --> Detecting: 开始检测
|
||||
Detecting --> Analyzed: 检测完成
|
||||
|
||||
Analyzed --> Spam: 高置信度垃圾
|
||||
Analyzed --> Suspicious: 中等置信度
|
||||
Analyzed --> Clean: 低置信度/正常
|
||||
|
||||
Spam --> Trash: 自动处理
|
||||
Spam --> Hold: 保持待审核
|
||||
Spam --> Marked: 仅标记
|
||||
|
||||
Suspicious --> ManualReview: 等待人工审核
|
||||
Clean --> Approved: 自动通过
|
||||
|
||||
ManualReview --> Approved: 管理员批准
|
||||
ManualReview --> Trash: 管理员拒绝
|
||||
|
||||
Hold --> Approved: 管理员批准
|
||||
Hold --> Trash: 管理员拒绝
|
||||
|
||||
Marked --> Approved: 管理员批准
|
||||
Marked --> Trash: 管理员拒绝
|
||||
|
||||
Trash --> [*]
|
||||
Approved --> [*]
|
||||
|
||||
note right of Detecting
|
||||
调用 AI API
|
||||
解析返回结果
|
||||
计算置信度
|
||||
end note
|
||||
|
||||
note right of ManualReview
|
||||
显示在后台
|
||||
等待管理员操作
|
||||
end note
|
||||
[/mermaid]
|
||||
|
||||
## 性能优化策略
|
||||
|
||||
[mermaid theme="default" width="95%"]
|
||||
flowchart TD
|
||||
subgraph Input[输入优化]
|
||||
I1[关键字预过滤]
|
||||
I2[白名单用户]
|
||||
I3[智能抽样]
|
||||
end
|
||||
|
||||
subgraph Process[处理优化]
|
||||
P1[异步检测]
|
||||
P2[批量处理]
|
||||
P3[缓存结果]
|
||||
end
|
||||
|
||||
subgraph API[API 优化]
|
||||
A1[Prompt 优化]
|
||||
A2[Token 控制]
|
||||
A3[超时处理]
|
||||
end
|
||||
|
||||
subgraph Storage[存储优化]
|
||||
S1[索引优化]
|
||||
S2[定期清理]
|
||||
S3[归档历史]
|
||||
end
|
||||
|
||||
Input --> Process
|
||||
Process --> API
|
||||
API --> Storage
|
||||
|
||||
style Input fill:#e8f5e9
|
||||
style Process fill:#e3f2fd
|
||||
style API fill:#fff3e0
|
||||
style Storage fill:#f3e5f5
|
||||
[/mermaid]
|
||||
|
||||
## 配置建议矩阵
|
||||
|
||||
[mermaid theme="default" width="100%"]
|
||||
graph TB
|
||||
subgraph Small[小型博客 < 100评论/天]
|
||||
S1[检测模式: keyword]
|
||||
S2[Prompt: 标准模式]
|
||||
S3[置信度阈值: > 0.9]
|
||||
S4[自动处理: trash]
|
||||
end
|
||||
|
||||
subgraph Medium[中型博客 100-500评论/天]
|
||||
M1[检测模式: sample 30%]
|
||||
M2[Prompt: 标准模式]
|
||||
M3[置信度阈值: > 0.85]
|
||||
M4[自动处理: hold]
|
||||
end
|
||||
|
||||
subgraph Large[大型博客 > 500评论/天]
|
||||
L1[检测模式: sample 40%]
|
||||
L2[Prompt: 极简模式]
|
||||
L3[置信度阈值: > 0.8]
|
||||
L4[自动处理: mark]
|
||||
L5[定期批量扫描]
|
||||
end
|
||||
|
||||
style Small fill:#c8e6c9
|
||||
style Medium fill:#fff9c4
|
||||
style Large fill:#ffccbc
|
||||
[/mermaid]
|
||||
|
||||
## 错误处理流程
|
||||
|
||||
[mermaid theme="default" width="85%"]
|
||||
flowchart TD
|
||||
Start([检测开始]) --> CallAPI[调用 AI API]
|
||||
|
||||
CallAPI --> CheckResponse{响应状态?}
|
||||
|
||||
CheckResponse -->|成功| ParseJSON[解析 JSON]
|
||||
CheckResponse -->|超时| Timeout[超时处理]
|
||||
CheckResponse -->|错误| APIError[API 错误]
|
||||
|
||||
ParseJSON --> ValidateData{数据有效?}
|
||||
ValidateData -->|是| ProcessResult[处理结果]
|
||||
ValidateData -->|否| DataError[数据错误]
|
||||
|
||||
Timeout --> RetryCheck{重试次数?}
|
||||
RetryCheck -->|< 3| Retry[重试请求]
|
||||
RetryCheck -->|>= 3| FallbackPending[降级：待审核]
|
||||
|
||||
Retry --> CallAPI
|
||||
|
||||
APIError --> LogError[记录错误日志]
|
||||
DataError --> LogError
|
||||
|
||||
LogError --> NotifyAdmin[通知管理员]
|
||||
NotifyAdmin --> FallbackPending
|
||||
|
||||
ProcessResult --> SaveResult[保存结果]
|
||||
FallbackPending --> SaveResult
|
||||
|
||||
SaveResult --> End([结束])
|
||||
|
||||
style Timeout fill:#ffccbc
|
||||
style APIError fill:#ffccbc
|
||||
style DataError fill:#ffccbc
|
||||
style FallbackPending fill:#fff9c4
|
||||
style ProcessResult fill:#c8e6c9
|
||||
[/mermaid]
|
||||
|
||||
## 总结
|
||||
|
||||
通过以上流程图，我们可以清晰地看到：
|
||||
|
||||
1. **主流程**：从用户提交到最终处理的完整流程
|
||||
2. **检测模式**：四种模式的特点和适用场景
|
||||
3. **时序交互**：各组件之间的交互顺序
|
||||
4. **数据结构**：数据库表关系和字段定义
|
||||
5. **状态机**：评论的各种状态转换
|
||||
6. **性能优化**：多层次的优化策略
|
||||
7. **配置建议**：不同规模博客的推荐配置
|
||||
8. **错误处理**：完善的异常处理机制
|
||||
|
||||
这个系统设计充分考虑了：
|
||||
- ✅ 准确性：多级置信度判断
|
||||
- ✅ 性能：异步处理、智能抽样
|
||||
- ✅ 成本：灵活的检测模式
|
||||
- ✅ 可靠性：完善的错误处理
|
||||
- ✅ 可维护性：清晰的架构设计
|
||||
Reference in New Issue
Block a user