stasis-rs 0.1.0

Durable AI orchestration framework with runtime jobs, lineage, and memory integration
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122

use std::sync::Arc;

use genai::chat::{ChatMessage, ChatRequest, ChatResponse};
use tokio::sync::mpsc;

use crate::domain::errors::{Result, StasisError};
use crate::ports::outbound::ai_chat_client::AiChatClient;
use crate::ports::outbound::ai_chat_client::StreamDelta;

#[derive(Clone, Debug, Default)]
pub struct PromptExecutionContext {
    pub trace_id: Option<String>,
    pub correlation_id: Option<String>,
    pub policy_profile: Option<String>,
    pub model_hint: Option<String>,
}

#[derive(Clone, Debug)]
pub struct PromptExecutionRequest {
    pub system_prompt: Option<String>,
    pub user_prompt: String,
    pub context: PromptExecutionContext,
}

impl PromptExecutionRequest {
    pub fn from_user_prompt(prompt: impl Into<String>) -> Self {
        Self {
            system_prompt: None,
            user_prompt: prompt.into(),
            context: PromptExecutionContext::default(),
        }
    }

    pub fn with_system_prompt(mut self, prompt: impl Into<String>) -> Self {
        self.system_prompt = Some(prompt.into());
        self
    }

    pub fn with_context(mut self, context: PromptExecutionContext) -> Self {
        self.context = context;
        self
    }
}

#[derive(Clone, Debug)]
pub struct PromptExecutionResponse {
    pub text: String,
    pub metadata: PromptExecutionContext,
}

#[derive(Clone, Debug)]
pub struct PromptChatCompletion {
    pub response: ChatResponse,
    pub metadata: PromptExecutionContext,
}

#[derive(Clone)]
pub struct PromptExecutionPipeline {
    chat_client: Arc<dyn AiChatClient>,
}

impl PromptExecutionPipeline {
    pub fn new(chat_client: Arc<dyn AiChatClient>) -> Self {
        Self { chat_client }
    }

    pub async fn complete_chat(
        &self,
        request: ChatRequest,
        context: PromptExecutionContext,
    ) -> Result<PromptChatCompletion> {
        let response = self.chat_client.complete(request, None).await?;
        Ok(PromptChatCompletion {
            response,
            metadata: context,
        })
    }

    pub async fn complete_chat_stream(
        &self,
        request: ChatRequest,
        context: PromptExecutionContext,
        chunk_tx: Option<&mpsc::UnboundedSender<StreamDelta>>,
    ) -> Result<PromptChatCompletion> {
        let response = self
            .chat_client
            .complete_stream(request, None, chunk_tx)
            .await?;
        Ok(PromptChatCompletion {
            response,
            metadata: context,
        })
    }

    pub async fn execute(
        &self,
        request: PromptExecutionRequest,
    ) -> Result<PromptExecutionResponse> {
        let context = request.context.clone();
        let mut messages = Vec::with_capacity(2);
        if let Some(system_prompt) = request.system_prompt {
            messages.push(ChatMessage::system(system_prompt));
        }
        messages.push(ChatMessage::user(request.user_prompt));

        let chat_response = self
            .complete_chat(ChatRequest::new(messages), context.clone())
            .await?
            .response;

        let text = chat_response
            .into_first_text()
            .map(|value| value.trim().to_string())
            .filter(|value| !value.is_empty())
            .ok_or_else(|| StasisError::PortFailure("chat response was empty".to_string()))?;

        Ok(PromptExecutionResponse {
            text,
            metadata: context,
        })
    }
}