Spaces:
Running
Running
File size: 4,199 Bytes
3f76ff4 837e3ac 3f76ff4 837e3ac 3f76ff4 837e3ac 3f76ff4 837e3ac 3f76ff4 837e3ac 3f76ff4 | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 | import {
isAnthropicCompatibleBaseUrl,
normalizeLlmConfig,
type LlmConfig,
} from "./llm-config";
import { compactTracePreview } from "./chat-trace";
export type LlmCallResult = {
text: string;
endpoint: string;
requestPreview: string;
responsePreview: string;
};
function extractJsonText(value: unknown): string | undefined {
if (typeof value === "string") return value;
if (Array.isArray(value)) {
const joined = value
.map((item) => {
if (typeof item === "string") return item;
if (
item &&
typeof item === "object" &&
"type" in item &&
"text" in item &&
item.type === "text" &&
typeof item.text === "string"
) {
return item.text;
}
return "";
})
.filter(Boolean)
.join("\n");
return joined || undefined;
}
return undefined;
}
function extractOpenAiText(payload: unknown): string | undefined {
const data = payload as
| {
choices?: Array<{
message?: { content?: unknown };
text?: unknown;
}>;
}
| undefined;
const choice = data?.choices?.[0];
return extractJsonText(choice?.message?.content) ?? extractJsonText(choice?.text);
}
function extractAnthropicText(payload: unknown): string | undefined {
const data = payload as
| {
content?: Array<{
type?: string;
text?: string;
}>;
}
| undefined;
return data?.content
?.filter((item) => item.type === "text" && typeof item.text === "string")
.map((item) => item.text)
.join("\n");
}
export async function generateLlmText(args: {
config: LlmConfig;
systemPrompt: string;
userPrompt: string;
temperature?: number;
maxTokens?: number;
responseFormat?: "json_object";
}): Promise<LlmCallResult> {
const { systemPrompt, userPrompt } = args;
const config = normalizeLlmConfig(args.config);
const temperature = args.temperature ?? 0.2;
const maxTokens = args.maxTokens ?? 2200;
const anthropicStyle = isAnthropicCompatibleBaseUrl(config.baseUrl);
const responseFormat = args.responseFormat;
const endpoint = anthropicStyle
? `${config.baseUrl}/v1/messages`
: `${config.baseUrl}/chat/completions`;
const response = await fetch(endpoint, {
method: "POST",
headers: anthropicStyle
? {
"content-type": "application/json",
"x-api-key": config.apiKey,
"anthropic-version": "2023-06-01",
}
: {
"content-type": "application/json",
authorization: `Bearer ${config.apiKey}`,
},
body: JSON.stringify(
anthropicStyle
? {
model: config.model,
max_tokens: maxTokens,
temperature,
system: systemPrompt,
messages: [
{
role: "user",
content: [
{
type: "text",
text: userPrompt,
},
],
},
],
}
: {
model: config.model,
temperature,
messages: [
{ role: "system", content: systemPrompt },
{ role: "user", content: userPrompt },
],
...(responseFormat
? { response_format: { type: responseFormat } }
: {}),
},
),
});
if (!response.ok) {
const text = await response.text().catch(() => "");
throw new Error(`LLM request failed (${response.status}): ${text}`.slice(0, 1200));
}
const rawText = await response.text();
let payload: unknown = null;
if (rawText.trim()) {
try {
payload = JSON.parse(rawText);
} catch {
payload = null;
}
}
const text = anthropicStyle
? extractAnthropicText(payload)
: extractOpenAiText(payload) ?? rawText.trim();
if (!text) {
throw new Error("LLM returned an empty response.");
}
return {
text,
endpoint,
requestPreview: compactTracePreview(
`SYSTEM:\n${systemPrompt}\n\nUSER:\n${userPrompt}`,
),
responsePreview: compactTracePreview(text),
};
}
|