npcrs 0.1.8

Rust core for the NPC system — agent kernel, jinx executor, LLM client
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
143
144
145
146
147
148
149
150
151
152
153

use crate::error::Result;
use crate::r#gen::Message;
use std::collections::HashMap;
use std::time::Instant;

#[derive(Debug, Clone)]
pub struct InferResult {
    pub model: String,
    pub provider: String,
    pub response: String,
    pub tokens: u64,
    pub cost: f64,
    pub latency_ms: u64,
}

pub async fn infer_matrix(
    prompt: &str,
    models: &[(String, String)],
    system_prompt: Option<&str>,
) -> Result<Vec<InferResult>> {
    let mut results = Vec::with_capacity(models.len());

    for (model, provider) in models {
        let start = Instant::now();

        let mut messages = Vec::new();
        if let Some(s) = system_prompt {
            messages.push(Message::system(s));
        }
        messages.push(Message::user(prompt));

        let result = crate::r#gen::get_genai_response(
            provider, model, &messages, None, None, None, None, None, false, None,
        )
        .await;

        let elapsed = start.elapsed().as_millis() as u64;

        match result {
            Ok(resp) => {
                let text = resp.message.content.unwrap_or_default();
                let (tokens, cost) = if let Some(ref usage) = resp.usage {
                    let total = usage.total_tokens;
                    let c = crate::r#gen::cost::calculate_cost(
                        model,
                        usage.prompt_tokens,
                        usage.completion_tokens,
                    );
                    (total, c)
                } else {
                    (0, 0.0)
                };

                results.push(InferResult {
                    model: model.clone(),
                    provider: provider.clone(),
                    response: text,
                    tokens,
                    cost,
                    latency_ms: elapsed,
                });
            }
            Err(e) => {
                results.push(InferResult {
                    model: model.clone(),
                    provider: provider.clone(),
                    response: format!("[ERROR] {e}"),
                    tokens: 0,
                    cost: 0.0,
                    latency_ms: elapsed,
                });
            }
        }
    }

    Ok(results)
}

pub fn ensemble_vote(results: &[InferResult]) -> String {
    if results.is_empty() {
        return String::new();
    }

    let mut counts: HashMap<String, usize> = HashMap::new();
    let mut order: Vec<String> = Vec::new();

    for r in results {
        if r.response.starts_with("[ERROR]") {
            continue;
        }

        let key = r.response.trim().to_lowercase();
        let count = counts.entry(key.clone()).or_insert(0);
        if *count == 0 {
            order.push(key);
        }
        *count += 1;
    }

    if order.is_empty() {
        return results[0].response.clone();
    }

    let best = order
        .into_iter()
        .max_by_key(|k| *counts.get(k).unwrap_or(&0))
        .unwrap_or_default();

    results
        .iter()
        .find(|r| r.response.trim().to_lowercase() == best)
        .map(|r| r.response.clone())
        .unwrap_or_default()
}

#[derive(Debug, Clone)]
pub struct MatrixStats {
    pub total_models: usize,
    pub successful: usize,
    pub failed: usize,
    pub total_tokens: u64,
    pub total_cost: f64,
    pub avg_latency_ms: f64,
    pub min_latency_ms: u64,
    pub max_latency_ms: u64,
}

pub fn matrix_stats(results: &[InferResult]) -> MatrixStats {
    let successful = results
        .iter()
        .filter(|r| !r.response.starts_with("[ERROR]"))
        .count();
    let failed = results.len() - successful;
    let total_tokens: u64 = results.iter().map(|r| r.tokens).sum();
    let total_cost: f64 = results.iter().map(|r| r.cost).sum();
    let latencies: Vec<u64> = results.iter().map(|r| r.latency_ms).collect();
    let avg_latency = if latencies.is_empty() {
        0.0
    } else {
        latencies.iter().sum::<u64>() as f64 / latencies.len() as f64
    };

    MatrixStats {
        total_models: results.len(),
        successful,
        failed,
        total_tokens,
        total_cost,
        avg_latency_ms: avg_latency,
        min_latency_ms: latencies.iter().copied().min().unwrap_or(0),
        max_latency_ms: latencies.iter().copied().max().unwrap_or(0),
    }
}