flyllm 0.4.0

A Rust library for unifying LLM backends as an abstraction layer with load balancing.
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
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142

use serde::{Serialize, Deserialize};
use std::pin::Pin;
use futures::Stream;
use crate::errors::LlmError;

/// Enum representing the different LLM providers supported
#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize, Hash)]
pub enum ProviderType {
    Anthropic,
    OpenAI,
    Mistral,
    Google,
    Ollama,
    LMStudio,
    Groq,
    Cohere,
    TogetherAI,
    Perplexity,
}

/// Unified request structure used across all providers
#[derive(Debug, Serialize, Deserialize, Clone)]
pub struct LlmRequest {
    pub messages: Vec<Message>,
    pub model: Option<String>,
    pub max_tokens: Option<u32>,
    pub temperature: Option<f32>,
}

/// Standard message format used across providers
#[derive(Debug, Serialize, Deserialize, Clone)]
pub struct Message {
    pub role: String,
    pub content: String,
}

/// Unified response structure returned by all providers
#[derive(Debug, Serialize, Deserialize, Clone)]
pub struct LlmResponse {
    pub content: String,
    pub model: String,
    pub usage: Option<TokenUsage>,
}

/// Token usage information returned by providers
#[derive(Debug, Serialize, Deserialize, Clone)]
pub struct TokenUsage {
    pub prompt_tokens: u32,
    pub completion_tokens: u32,
    pub total_tokens: u32,
}

impl Default for TokenUsage {
    fn default() -> Self {
        Self {
            prompt_tokens: 0,
            completion_tokens: 0,
            total_tokens: 0
        }
    }
}

/// Information about an LLM model
#[derive(Debug, Serialize, Deserialize, Clone)]
pub struct ModelInfo {
    pub name: String,
    pub provider: ProviderType,
}

/// Display implementation for ProviderType
impl std::fmt::Display for ProviderType {
    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
        match self {
            ProviderType::Anthropic => write!(f, "Anthropic"),
            ProviderType::OpenAI => write!(f, "OpenAI"),
            ProviderType::Mistral => write!(f, "Mistral"),
            ProviderType::Google => write!(f, "Google"),
            ProviderType::Ollama => write!(f, "Ollama"),
            ProviderType::LMStudio => write!(f, "LMStudio"),
            ProviderType::Groq => write!(f, "Groq"),
            ProviderType::Cohere => write!(f, "Cohere"),
            ProviderType::TogetherAI => write!(f, "TogetherAI"),
            ProviderType::Perplexity => write!(f, "Perplexity"),
        }
    }
}

impl From<&str> for ProviderType {
    fn from(value: &str) -> Self {
        match value.to_lowercase().as_str() {
            "anthropic" => ProviderType::Anthropic,
            "openai" => ProviderType::OpenAI,
            "mistral" => ProviderType::Mistral,
            "google" => ProviderType::Google,
            "ollama" => ProviderType::Ollama,
            "lmstudio" => ProviderType::LMStudio,
            "groq" => ProviderType::Groq,
            "cohere" => ProviderType::Cohere,
            "togetherai" => ProviderType::TogetherAI,
            "perplexity" => ProviderType::Perplexity,
            _ => panic!("Unknown provider: {}", value),
        }
    }
}

/// A chunk of streamed content from an LLM provider
#[derive(Debug, Clone)]
pub struct StreamChunk {
    /// The text content of this chunk
    pub content: String,
    /// The model that generated this chunk (may be empty until final chunk)
    pub model: Option<String>,
    /// Whether this is the final chunk in the stream
    pub is_final: bool,
    /// Token usage information (typically only available in final chunk)
    pub usage: Option<TokenUsage>,
}

impl StreamChunk {
    /// Create a new content chunk
    pub fn content(text: impl Into<String>) -> Self {
        Self {
            content: text.into(),
            model: None,
            is_final: false,
            usage: None,
        }
    }

    /// Create a final chunk with usage information
    pub fn final_chunk(model: impl Into<String>, usage: Option<TokenUsage>) -> Self {
        Self {
            content: String::new(),
            model: Some(model.into()),
            is_final: true,
            usage,
        }
    }
}

/// Type alias for a boxed stream of chunks
pub type LlmStream = Pin<Box<dyn Stream<Item = Result<StreamChunk, LlmError>> + Send>>;