
com.simiacryptus.jopenai.ClientUtil.kt Maven / Gradle / Ivy
package com.simiacryptus.jopenai
import com.google.gson.Gson
import com.google.gson.JsonObject
import com.simiacryptus.jopenai.exceptions.*
import java.io.File
import java.io.IOException
import java.nio.charset.Charset
import java.util.regex.Pattern
object ClientUtil {
var auxiliaryLog: File? = null
fun checkError(result: String) {
try {
val jsonObject = Gson().fromJson(
result,
JsonObject::class.java
)
if (null == jsonObject) return
if (jsonObject.has("error")) {
val errorObject = jsonObject.getAsJsonObject("error")
val errorMessage = errorObject["message"].asString
if (errorMessage.startsWith("That model is currently overloaded with other requests.")) {
throw RequestOverloadException(errorMessage)
}
maxTokenErrorMessage.find { it.matcher(errorMessage).matches() }?.let {
val matcher = it.matcher(errorMessage)
if (matcher.find()) {
val modelMax = matcher.group(1).toInt()
val request = matcher.group(2).toInt()
val messages = matcher.group(3).toInt()
val completion = matcher.group(4).toInt()
throw ModelMaxException(modelMax, request, messages, completion)
}
}
rateLimitErrorMessage.matcher(errorMessage).let {
if (it.matches()) {
val org = it.group(3)
val limit = it.group(4).toInt()
val delay = it.group(5).toLong()
throw RateLimitException(org, limit, delay)
}
}
quotaErrorMessage.matcher(errorMessage).let {
if (it.matches()) {
throw QuotaException()
}
}
invalidModelException.matcher(errorMessage).let {
if (it.matches()) {
val model = it.group(1)
throw InvalidModelException(model)
}
}
invalidValueException.matcher(errorMessage).let {
if (it.matches()) {
val field = it.group(1)
val value = it.group(2)
throw InvalidValueException(field, value)
}
}
throw IOException(errorMessage)
}
} catch (e: com.google.gson.JsonSyntaxException) {
throw IOException("Invalid JSON response: $result")
}
}
private var _keyTxt: String? = null
var keyTxt: String
get() {
if (null != _keyTxt) return _keyTxt!!
val resourceAsStream = OpenAIClient::class.java.getResourceAsStream("/openai.key")
if (null != resourceAsStream) return resourceAsStream.readAllBytes().toString(Charsets.UTF_8).trim()
val keyFile = File(File(System.getProperty("user.home")), "openai.key")
if (keyFile.exists()) return keyFile.readText().trim()
if (System.getenv().containsKey("OPENAI_KEY")) return System.getenv("OPENAI_KEY").trim()
return ""
}
set(value) {
_keyTxt = value
}
fun String.toContentList() = listOf(this).map { ApiModel.ContentPart(text = it, type = "text") }
fun String.toChatMessage(role: ApiModel.Role = ApiModel.Role.user) =
ApiModel.ChatMessage(role = role, content = toContentList())
val allowedCharset: Charset = Charset.forName("ASCII")
val maxTokenErrorMessage = listOf(
Pattern.compile(
"""This model's maximum context length is (\d+) tokens. However, you requested (\d+) tokens \((\d+) in the messages, (\d+) in the completion\).*"""
),
// This model's maximum context length is 4097 tokens, however you requested 80052 tokens (52 in your prompt; 80000 for the completion). Please reduce your prompt; or completion length.
Pattern.compile(
"""This model's maximum context length is (\d+) tokens, however you requested (\d+) tokens \((\d+) in your prompt; (\d+) for the completion\).*"""
)
)
val rateLimitErrorMessage = Pattern.compile(
"""Rate limit reached for (\d+)KTPM-(\d+)RPM in organization (\S+) on tokens per min. Limit: (\d+) / min. Please try again in (\d+)ms. Contact us through our help center at help.openai.com if you continue to have issues."""
)
val quotaErrorMessage = Pattern.compile(
"""You exceeded your current quota, please check your plan and billing details."""
)
val invalidModelException = Pattern.compile(
"""The model `(\S+)` does not exist or you do not have access to it."""
)
val invalidValueException = Pattern.compile(
"""Invalid value for '(\S+)': (\S+)"""
)
}
© 2015 - 2025 Weber Informatics LLC | Privacy Policy