Package com.axelor.ai.chat

Class AiUsageLogger

java.lang.Object

com.axelor.ai.chat.AiUsageLogger

public class AiUsageLogger
extends Object

Records AI usage metrics. All logging is fire-and-forget in a background thread.

Constructor Summary

Constructors

Constructor	Description
AiUsageLogger()

Method Summary

Modifier and Type	Method	Description
static void	log(AiUsageSource source, String modelCode, String providerCode, String agentCode, String chatId, Integer inputTokens, Integer outputTokens, Integer cacheReadTokens, Integer cacheWriteTokens, Integer latencyMs, AiUsageStatus status, String errorMessage)
static void	logError(AiUsageSource source, String modelCode, String providerCode, String agentCode, String chatId, Integer latencyMs, String errorMessage)
static <T> T	logged(AiUsageSource source, String modelCode, String providerCode, String agentCode, String chatId, Supplier<T> call, Function<T,dev.langchain4j.model.output.TokenUsage> tokens)	Makes a model call outside the chat stream, writing its usage row: the tokens the response reports, or the error it failed with.
static void	logSuccess(AiUsageSource source, String modelCode, String providerCode, String agentCode, String chatId, Integer inputTokens, Integer outputTokens, Integer cacheReadTokens, Integer cacheWriteTokens, Integer latencyMs)
static void	refused(AiUsageSource source, ChatRequest request)	A turn refused, before its first model call or between two: its model may not receive what the chat holds.
static void	refused(AiUsageSource source, ChatRequest request, String reason)	A call refused for what the chat holds, said as the given reason: a consultation's differs from a turn's, the turn going on without it.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Details

AiUsageLogger

public AiUsageLogger()

Method Details

log

public static void log(AiUsageSource source,
 String modelCode,
 String providerCode,
 String agentCode,
 String chatId,
 Integer inputTokens,
 Integer outputTokens,
 Integer cacheReadTokens,
 Integer cacheWriteTokens,
 Integer latencyMs,
 AiUsageStatus status,
 String errorMessage)

refused

public static void refused(AiUsageSource source,
 ChatRequest request)

A turn refused, before its first model call or between two: its model may not receive what the chat holds. Logged under the model it would have gone to.

refused

public static void refused(AiUsageSource source,
 ChatRequest request,
 String reason)

A call refused for what the chat holds, said as the given reason: a consultation's differs from a turn's, the turn going on without it.

logSuccess

public static void logSuccess(AiUsageSource source,
 String modelCode,
 String providerCode,
 String agentCode,
 String chatId,
 Integer inputTokens,
 Integer outputTokens,
 Integer cacheReadTokens,
 Integer cacheWriteTokens,
 Integer latencyMs)

logError

public static void logError(AiUsageSource source,
 String modelCode,
 String providerCode,
 String agentCode,
 String chatId,
 Integer latencyMs,
 String errorMessage)

logged

public static <T> T logged(AiUsageSource source,
 String modelCode,
 String providerCode,
 String agentCode,
 String chatId,
 Supplier<T> call,
 Function<T,dev.langchain4j.model.output.TokenUsage> tokens)

Makes a model call outside the chat stream, writing its usage row: the tokens the response reports, or the error it failed with.