gproxy-protocol 4.0.0

GPROXY v4 connection modelling, operation declarations, vendor wire types and pairwise conversion
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

use super::{EmbeddingBatchError, EmbeddingBatchResult, EmbeddingFailure, at, packing};
use crate::{
    WireRequest,
    adapt::{JsonInvocation, invoke_json},
    capability::Upstream,
    codec::{self, CodecLimits},
    transform::{
        Converted, TransformError,
        embeddings::{self, OpenAiUsageFacts},
    },
    wire::{gemini::embeddings as g, openai::embeddings as o},
};

/// One OpenAI input through Gemini embedContent, retaining requested encoding
/// and validating the returned dimension and factual usage.
pub async fn openai_to_gemini_single<U: Upstream>(
    upstream: &U,
    target: &U::Target,
    template: WireRequest<()>,
    input: o::CreateEmbeddingRequestBody,
    target_model: &str,
    limits: CodecLimits,
    usage: Option<OpenAiUsageFacts>,
) -> Result<EmbeddingBatchResult<o::CreateEmbeddingResponseBody>, EmbeddingBatchError> {
    if usage.is_some_and(|v| v.prompt_tokens < 0 || v.total_tokens < v.prompt_tokens) {
        return Err(TransformError::shape(
            "embedding.usage",
            "usage must be nonnegative and total must cover prompt tokens",
        )
        .into());
    }
    let prepared = embeddings::openai_to_gemini_single(input, target_model)?;
    let WireRequest {
        method,
        path,
        query,
        headers,
        body: (),
    } = template;
    let response: JsonInvocation<g::EmbedContentResponseBody> = invoke_json(
        upstream,
        target,
        WireRequest {
            method,
            path,
            query,
            headers,
            body: prepared.value.body,
        },
        limits,
    )
    .await?;
    let response = match response {
        JsonInvocation::Success(response) => response.body,
        JsonInvocation::Rejected(response) => {
            return Err(EmbeddingBatchError {
                completed_calls: 0,
                completed_items: 0,
                failure: EmbeddingFailure::Rejected(Box::new(response)),
            });
        }
    };
    let mut output = prepared
        .value
        .response
        .finish_single(response, usage)
        .map_err(|error| at(error, 1, 0))?;
    let mut report = prepared.report;
    report.diagnostics.append(&mut output.report.diagnostics);
    output.report = report;
    codec::encode_json(&output.value, limits)
        .map_err(|error| at(packing::encoding(error), 1, 0))?;
    Ok(EmbeddingBatchResult {
        output,
        completed_calls: 1,
    })
}

/// One Gemini input through OpenAI embeddings. The selected target request
/// carries its explicit model and declared dimensionality.
pub async fn gemini_single_to_openai<U: Upstream>(
    upstream: &U,
    target: &U::Target,
    template: WireRequest<()>,
    input: g::EmbedContentRequestBody,
    target_model: &str,
    limits: CodecLimits,
) -> Result<EmbeddingBatchResult<g::EmbedContentResponseBody>, EmbeddingBatchError> {
    let prepared = embeddings::gemini_single_to_openai(input, target_model)?;
    let dimensions = prepared.value.dimensions;
    let WireRequest {
        method,
        path,
        query,
        headers,
        body: (),
    } = template;
    let response: JsonInvocation<o::CreateEmbeddingResponseBody> = invoke_json(
        upstream,
        target,
        WireRequest {
            method,
            path,
            query,
            headers,
            body: prepared.value,
        },
        limits,
    )
    .await?;
    let response = match response {
        JsonInvocation::Success(response) => response.body,
        JsonInvocation::Rejected(response) => {
            return Err(EmbeddingBatchError {
                completed_calls: 0,
                completed_items: 0,
                failure: EmbeddingFailure::Rejected(Box::new(response)),
            });
        }
    };
    let converted =
        embeddings::openai_single_response_to_gemini(response).map_err(|error| at(error, 1, 0))?;
    if let Some(expected) = dimensions {
        let actual = converted
            .value
            .embedding
            .as_ref()
            .and_then(|v| v.values.as_ref())
            .and_then(|v| i64::try_from(v.len()).ok());
        if actual != Some(expected) {
            return Err(at(
                TransformError::invalid_result(
                    "embedding.dimensions",
                    "response dimensions differ from request",
                ),
                1,
                0,
            ));
        }
    }
    let mut report = prepared.report;
    report.diagnostics.extend(converted.report.diagnostics);
    codec::encode_json(&converted.value, limits)
        .map_err(|error| at(packing::encoding(error), 1, 0))?;
    Ok(EmbeddingBatchResult {
        output: Converted {
            value: converted.value,
            report,
        },
        completed_calls: 1,
    })
}