1use crate::{Result, types::Content};
2use async_trait::async_trait;
3use futures::stream::Stream;
4use serde::{Deserialize, Serialize};
5use std::collections::HashMap;
6use std::pin::Pin;
7
8pub type LlmResponseStream = Pin<Box<dyn Stream<Item = Result<LlmResponse>> + Send>>;
9
10#[async_trait]
11pub trait Llm: Send + Sync {
12 fn name(&self) -> &str;
13 async fn generate_content(&self, req: LlmRequest, stream: bool) -> Result<LlmResponseStream>;
14}
15
16#[derive(Debug, Clone, Serialize, Deserialize)]
17pub struct LlmRequest {
18 pub model: String,
19 pub contents: Vec<Content>,
20 pub config: Option<GenerateContentConfig>,
21 #[serde(skip)]
22 pub tools: HashMap<String, serde_json::Value>,
23}
24
25#[derive(Debug, Clone, Serialize, Deserialize)]
26pub struct GenerateContentConfig {
27 pub temperature: Option<f32>,
28 pub top_p: Option<f32>,
29 pub top_k: Option<i32>,
30 pub max_output_tokens: Option<i32>,
31 #[serde(skip_serializing_if = "Option::is_none")]
32 pub response_schema: Option<serde_json::Value>,
33}
34
35#[derive(Debug, Clone, Default, Serialize, Deserialize)]
36pub struct LlmResponse {
37 pub content: Option<Content>,
38 pub usage_metadata: Option<UsageMetadata>,
39 pub finish_reason: Option<FinishReason>,
40 pub partial: bool,
41 pub turn_complete: bool,
42 pub interrupted: bool,
43 pub error_code: Option<String>,
44 pub error_message: Option<String>,
45}
46
47#[derive(Debug, Clone, Serialize, Deserialize)]
48pub struct UsageMetadata {
49 pub prompt_token_count: i32,
50 pub candidates_token_count: i32,
51 pub total_token_count: i32,
52}
53
54#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize)]
55pub enum FinishReason {
56 Stop,
57 MaxTokens,
58 Safety,
59 Recitation,
60 Other,
61}
62
63impl LlmRequest {
64 pub fn new(model: impl Into<String>, contents: Vec<Content>) -> Self {
65 Self { model: model.into(), contents, config: None, tools: HashMap::new() }
66 }
67
68 pub fn with_response_schema(mut self, schema: serde_json::Value) -> Self {
70 let config = self.config.get_or_insert(GenerateContentConfig {
71 temperature: None,
72 top_p: None,
73 top_k: None,
74 max_output_tokens: None,
75 response_schema: None,
76 });
77 config.response_schema = Some(schema);
78 self
79 }
80
81 pub fn with_config(mut self, config: GenerateContentConfig) -> Self {
83 self.config = Some(config);
84 self
85 }
86}
87
88impl LlmResponse {
89 pub fn new(content: Content) -> Self {
90 Self {
91 content: Some(content),
92 usage_metadata: None,
93 finish_reason: Some(FinishReason::Stop),
94 partial: false,
95 turn_complete: true,
96 interrupted: false,
97 error_code: None,
98 error_message: None,
99 }
100 }
101}
102
103#[cfg(test)]
104mod tests {
105 use super::*;
106
107 #[test]
108 fn test_llm_request_creation() {
109 let req = LlmRequest::new("test-model", vec![]);
110 assert_eq!(req.model, "test-model");
111 assert!(req.contents.is_empty());
112 }
113
114 #[test]
115 fn test_llm_request_with_response_schema() {
116 let schema = serde_json::json!({
117 "type": "object",
118 "properties": {
119 "name": { "type": "string" }
120 }
121 });
122 let req = LlmRequest::new("test-model", vec![]).with_response_schema(schema.clone());
123
124 assert!(req.config.is_some());
125 let config = req.config.unwrap();
126 assert!(config.response_schema.is_some());
127 assert_eq!(config.response_schema.unwrap(), schema);
128 }
129
130 #[test]
131 fn test_llm_request_with_config() {
132 let config = GenerateContentConfig {
133 temperature: Some(0.7),
134 top_p: Some(0.9),
135 top_k: Some(40),
136 max_output_tokens: Some(1024),
137 response_schema: None,
138 };
139 let req = LlmRequest::new("test-model", vec![]).with_config(config);
140
141 assert!(req.config.is_some());
142 let config = req.config.unwrap();
143 assert_eq!(config.temperature, Some(0.7));
144 assert_eq!(config.max_output_tokens, Some(1024));
145 }
146
147 #[test]
148 fn test_llm_response_creation() {
149 let content = Content::new("assistant");
150 let resp = LlmResponse::new(content);
151 assert!(resp.content.is_some());
152 assert!(resp.turn_complete);
153 assert!(!resp.partial);
154 assert_eq!(resp.finish_reason, Some(FinishReason::Stop));
155 }
156
157 #[test]
158 fn test_finish_reason() {
159 assert_eq!(FinishReason::Stop, FinishReason::Stop);
160 assert_ne!(FinishReason::Stop, FinishReason::MaxTokens);
161 }
162}