Skip to main content

sim_codec_chat/providers/openai/
encode.rs

1use serde_json::{Value, json};
2use sim_kernel::{CodecId, Error, Expr, Result};
3
4use crate::{is_model_request_expr, validate_chat_transcript};
5
6use super::OpenAiRequestOptions;
7use super::common::{
8    codec_error, codec_eval_to_codec, flatten_expr, list_field, map_field, marker_is_true,
9    optional_u64_field, string_field, symbol_field,
10};
11
12/// Encodes a model-request transcript into OpenAI chat-completion JSON.
13pub fn encode_openai_request(expr: &Expr, options: &OpenAiRequestOptions) -> Result<Vec<u8>> {
14    if !is_model_request_expr(expr) {
15        return Err(Error::Eval(
16            "openai codec expects a model-request transcript".to_owned(),
17        ));
18    }
19    validate_chat_transcript(expr)?;
20    let mut payload = json!({
21        "model": options.model,
22        "stream": options.stream,
23        "messages": transcript_messages(expr)?,
24        "tools": if options.tools { Value::Array(Vec::new()) } else { Value::Null },
25    });
26    if options.stream
27        && let Some(object) = payload.as_object_mut()
28    {
29        object.insert("stream_options".to_owned(), json!({"include_usage": true}));
30    }
31    serde_json::to_vec(&payload)
32        .map_err(|err| Error::Eval(format!("openai codec failed to encode request: {err}")))
33}
34
35/// Encodes a model-response transcript into OpenAI chat-completion JSON.
36pub fn encode_openai_response(expr: &Expr) -> Result<Vec<u8>> {
37    let value = response_json(expr)?;
38    serde_json::to_vec(&value)
39        .map_err(|err| Error::Eval(format!("openai codec failed to encode response: {err}")))
40}
41
42pub(super) fn encode_openai_response_for_codec(codec: CodecId, expr: &Expr) -> Result<String> {
43    if !marker_is_true(expr, "model-response") {
44        return Err(codec_error(
45            codec,
46            "openai codec expects a model-response transcript",
47        ));
48    }
49    validate_chat_transcript(expr).map_err(|err| codec_eval_to_codec(codec, err))?;
50    let value = response_json(expr).map_err(|err| codec_eval_to_codec(codec, err))?;
51    serde_json::to_string(&value).map_err(|err| codec_error(codec, err))
52}
53
54fn transcript_messages(expr: &Expr) -> Result<Vec<Value>> {
55    let Expr::Map(entries) = expr else {
56        return Err(Error::Eval(
57            "openai codec expects request transcript as a map".to_owned(),
58        ));
59    };
60    let mut messages = list_field(map_field(entries, "messages")?)?
61        .iter()
62        .map(message_to_json)
63        .collect::<Result<Vec<_>>>()?;
64    messages.push(json!({
65        "role": "user",
66        "content": [{
67            "type": "text",
68            "text": flatten_expr(map_field(entries, "task")?),
69        }],
70    }));
71    Ok(messages)
72}
73
74fn message_to_json(expr: &Expr) -> Result<Value> {
75    let Expr::Map(entries) = expr else {
76        return Err(Error::Eval("openai codec message must be a map".to_owned()));
77    };
78    Ok(json!({
79        "role": symbol_field(entries, "role")?,
80        "content": list_field(map_field(entries, "content")?)?
81            .iter()
82            .map(content_part_to_json)
83            .collect::<Result<Vec<_>>>()?,
84    }))
85}
86
87fn content_part_to_json(expr: &Expr) -> Result<Value> {
88    let Expr::Map(entries) = expr else {
89        return Err(Error::Eval(
90            "openai codec content part must be a map".to_owned(),
91        ));
92    };
93    match symbol_field(entries, "type")?.as_str() {
94        "text" => Ok(json!({
95            "type": "text",
96            "text": string_field(entries, "text")?,
97        })),
98        other => Err(Error::Eval(format!(
99            "openai codec does not support content part type {other}"
100        ))),
101    }
102}
103
104fn response_json(expr: &Expr) -> Result<Value> {
105    let Expr::Map(entries) = expr else {
106        return Err(Error::Eval(
107            "openai codec expects response transcript as a map".to_owned(),
108        ));
109    };
110    let model = string_field(entries, "model")?;
111    let finish_reason = symbol_field(entries, "stop-reason")?;
112    Ok(json!({
113        "id": "chatcmpl-sim",
114        "object": "chat.completion",
115        "created": 0,
116        "model": model,
117        "choices": [{
118            "index": 0,
119            "message": {
120                "role": "assistant",
121                "content": response_text(entries)?,
122            },
123            "finish_reason": finish_reason,
124        }],
125        "usage": response_usage(entries)?,
126    }))
127}
128
129fn response_text(entries: &[(Expr, Expr)]) -> Result<String> {
130    list_field(map_field(entries, "content")?)?
131        .iter()
132        .map(text_content)
133        .collect::<Result<Vec<_>>>()
134        .map(|parts| parts.join(""))
135}
136
137fn text_content(expr: &Expr) -> Result<String> {
138    let Expr::Map(entries) = expr else {
139        return Err(Error::Eval(
140            "openai codec content part must be a map".to_owned(),
141        ));
142    };
143    match symbol_field(entries, "type")?.as_str() {
144        "text" => string_field(entries, "text"),
145        other => Err(Error::Eval(format!(
146            "openai codec does not support content part type {other}"
147        ))),
148    }
149}
150
151fn response_usage(entries: &[(Expr, Expr)]) -> Result<Value> {
152    let Some(usage) = entries.iter().find_map(|(field, value)| match field {
153        Expr::Symbol(symbol) if symbol.name.as_ref() == "usage" => Some(value),
154        _ => None,
155    }) else {
156        return Ok(Value::Null);
157    };
158    let Expr::Map(fields) = usage else {
159        return Err(Error::Eval(
160            "openai codec usage field must be a map".to_owned(),
161        ));
162    };
163    let prompt = optional_u64_field(fields, "input-tokens")?;
164    let completion = optional_u64_field(fields, "output-tokens")?;
165    let total = optional_u64_field(fields, "total-tokens")?.or_else(|| {
166        prompt
167            .zip(completion)
168            .map(|(left, right)| left.saturating_add(right))
169    });
170    Ok(json!({
171        "prompt_tokens": prompt.unwrap_or(0),
172        "completion_tokens": completion.unwrap_or(0),
173        "total_tokens": total.unwrap_or(0),
174    }))
175}