mirror of
https://github.com/router-for-me/CLIProxyAPI.git
synced 2026-02-03 04:50:52 +08:00
feat(usage): implement usage tracking infrastructure across executors
- Added `LoggerPlugin` to log usage metrics for observability. - Introduced a new `Manager` to handle usage record queuing and plugin registration. - Integrated new usage reporter and detailed metrics parsing into executors, covering providers like OpenAI, Codex, Claude, and Gemini. - Improved token usage breakdown across streaming and non-streaming responses.
This commit is contained in:
Luis Pater
@@ -6,7 +6,6 @@
|
||||
package chat_completions
|
||||
|
||||
import (
|
||||
"bufio"
|
||||
"bytes"
|
||||
"context"
|
||||
"encoding/json"
|
||||
@@ -278,19 +277,14 @@ func mapAnthropicStopReasonToOpenAI(anthropicReason string) string {
|
||||
// Returns:
|
||||
// - string: An OpenAI-compatible JSON response containing all message content and metadata
|
||||
func ConvertClaudeResponseToOpenAINonStream(_ context.Context, _ string, originalRequestRawJSON, requestRawJSON, rawJSON []byte, _ *any) string {
|
||||
|
||||
chunks := make([][]byte, 0)
|
||||
|
||||
scanner := bufio.NewScanner(bytes.NewReader(rawJSON))
|
||||
buffer := make([]byte, 10240*1024)
|
||||
scanner.Buffer(buffer, 10240*1024)
|
||||
for scanner.Scan() {
|
||||
line := scanner.Bytes()
|
||||
// log.Debug(string(line))
|
||||
lines := bytes.Split(rawJSON, []byte("\n"))
|
||||
for _, line := range lines {
|
||||
if !bytes.HasPrefix(line, dataTag) {
|
||||
continue
|
||||
}
|
||||
chunks = append(chunks, bytes.TrimSpace(rawJSON[5:]))
|
||||
chunks = append(chunks, bytes.TrimSpace(line[5:]))
|
||||
}
|
||||
|
||||
// Base OpenAI non-streaming response template
|
||||
|
||||
@@ -6,7 +6,6 @@
|
||||
package chat_completions
|
||||
|
||||
import (
|
||||
"bufio"
|
||||
"bytes"
|
||||
"context"
|
||||
"time"
|
||||
@@ -166,153 +165,141 @@ func ConvertCodexResponseToOpenAI(_ context.Context, modelName string, originalR
|
||||
// Returns:
|
||||
// - string: An OpenAI-compatible JSON response containing all message content and metadata
|
||||
func ConvertCodexResponseToOpenAINonStream(_ context.Context, _ string, originalRequestRawJSON, requestRawJSON, rawJSON []byte, _ *any) string {
|
||||
scanner := bufio.NewScanner(bytes.NewReader(rawJSON))
|
||||
buffer := make([]byte, 10240*1024)
|
||||
scanner.Buffer(buffer, 10240*1024)
|
||||
for scanner.Scan() {
|
||||
line := scanner.Bytes()
|
||||
// log.Debug(string(line))
|
||||
if !bytes.HasPrefix(line, dataTag) {
|
||||
continue
|
||||
}
|
||||
rawJSON = bytes.TrimSpace(rawJSON[5:])
|
||||
|
||||
rootResult := gjson.ParseBytes(rawJSON)
|
||||
// Verify this is a response.completed event
|
||||
if rootResult.Get("type").String() != "response.completed" {
|
||||
continue
|
||||
}
|
||||
unixTimestamp := time.Now().Unix()
|
||||
|
||||
responseResult := rootResult.Get("response")
|
||||
|
||||
template := `{"id":"","object":"chat.completion","created":123456,"model":"model","choices":[{"index":0,"message":{"role":"assistant","content":null,"reasoning_content":null,"tool_calls":null},"finish_reason":null,"native_finish_reason":null}]}`
|
||||
|
||||
// Extract and set the model version.
|
||||
if modelResult := responseResult.Get("model"); modelResult.Exists() {
|
||||
template, _ = sjson.Set(template, "model", modelResult.String())
|
||||
}
|
||||
|
||||
// Extract and set the creation timestamp.
|
||||
if createdAtResult := responseResult.Get("created_at"); createdAtResult.Exists() {
|
||||
template, _ = sjson.Set(template, "created", createdAtResult.Int())
|
||||
} else {
|
||||
template, _ = sjson.Set(template, "created", unixTimestamp)
|
||||
}
|
||||
|
||||
// Extract and set the response ID.
|
||||
if idResult := responseResult.Get("id"); idResult.Exists() {
|
||||
template, _ = sjson.Set(template, "id", idResult.String())
|
||||
}
|
||||
|
||||
// Extract and set usage metadata (token counts).
|
||||
if usageResult := responseResult.Get("usage"); usageResult.Exists() {
|
||||
if outputTokensResult := usageResult.Get("output_tokens"); outputTokensResult.Exists() {
|
||||
template, _ = sjson.Set(template, "usage.completion_tokens", outputTokensResult.Int())
|
||||
}
|
||||
if totalTokensResult := usageResult.Get("total_tokens"); totalTokensResult.Exists() {
|
||||
template, _ = sjson.Set(template, "usage.total_tokens", totalTokensResult.Int())
|
||||
}
|
||||
if inputTokensResult := usageResult.Get("input_tokens"); inputTokensResult.Exists() {
|
||||
template, _ = sjson.Set(template, "usage.prompt_tokens", inputTokensResult.Int())
|
||||
}
|
||||
if reasoningTokensResult := usageResult.Get("output_tokens_details.reasoning_tokens"); reasoningTokensResult.Exists() {
|
||||
template, _ = sjson.Set(template, "usage.completion_tokens_details.reasoning_tokens", reasoningTokensResult.Int())
|
||||
}
|
||||
}
|
||||
|
||||
// Process the output array for content and function calls
|
||||
outputResult := responseResult.Get("output")
|
||||
if outputResult.IsArray() {
|
||||
outputArray := outputResult.Array()
|
||||
var contentText string
|
||||
var reasoningText string
|
||||
var toolCalls []string
|
||||
|
||||
for _, outputItem := range outputArray {
|
||||
outputType := outputItem.Get("type").String()
|
||||
|
||||
switch outputType {
|
||||
case "reasoning":
|
||||
// Extract reasoning content from summary
|
||||
if summaryResult := outputItem.Get("summary"); summaryResult.IsArray() {
|
||||
summaryArray := summaryResult.Array()
|
||||
for _, summaryItem := range summaryArray {
|
||||
if summaryItem.Get("type").String() == "summary_text" {
|
||||
reasoningText = summaryItem.Get("text").String()
|
||||
break
|
||||
}
|
||||
}
|
||||
}
|
||||
case "message":
|
||||
// Extract message content
|
||||
if contentResult := outputItem.Get("content"); contentResult.IsArray() {
|
||||
contentArray := contentResult.Array()
|
||||
for _, contentItem := range contentArray {
|
||||
if contentItem.Get("type").String() == "output_text" {
|
||||
contentText = contentItem.Get("text").String()
|
||||
break
|
||||
}
|
||||
}
|
||||
}
|
||||
case "function_call":
|
||||
// Handle function call content
|
||||
functionCallTemplate := `{"id": "","type": "function","function": {"name": "","arguments": ""}}`
|
||||
|
||||
if callIdResult := outputItem.Get("call_id"); callIdResult.Exists() {
|
||||
functionCallTemplate, _ = sjson.Set(functionCallTemplate, "id", callIdResult.String())
|
||||
}
|
||||
|
||||
if nameResult := outputItem.Get("name"); nameResult.Exists() {
|
||||
n := nameResult.String()
|
||||
rev := buildReverseMapFromOriginalOpenAI(originalRequestRawJSON)
|
||||
if orig, ok := rev[n]; ok {
|
||||
n = orig
|
||||
}
|
||||
functionCallTemplate, _ = sjson.Set(functionCallTemplate, "function.name", n)
|
||||
}
|
||||
|
||||
if argsResult := outputItem.Get("arguments"); argsResult.Exists() {
|
||||
functionCallTemplate, _ = sjson.Set(functionCallTemplate, "function.arguments", argsResult.String())
|
||||
}
|
||||
|
||||
toolCalls = append(toolCalls, functionCallTemplate)
|
||||
}
|
||||
}
|
||||
|
||||
// Set content and reasoning content if found
|
||||
if contentText != "" {
|
||||
template, _ = sjson.Set(template, "choices.0.message.content", contentText)
|
||||
template, _ = sjson.Set(template, "choices.0.message.role", "assistant")
|
||||
}
|
||||
|
||||
if reasoningText != "" {
|
||||
template, _ = sjson.Set(template, "choices.0.message.reasoning_content", reasoningText)
|
||||
template, _ = sjson.Set(template, "choices.0.message.role", "assistant")
|
||||
}
|
||||
|
||||
// Add tool calls if any
|
||||
if len(toolCalls) > 0 {
|
||||
template, _ = sjson.SetRaw(template, "choices.0.message.tool_calls", `[]`)
|
||||
for _, toolCall := range toolCalls {
|
||||
template, _ = sjson.SetRaw(template, "choices.0.message.tool_calls.-1", toolCall)
|
||||
}
|
||||
template, _ = sjson.Set(template, "choices.0.message.role", "assistant")
|
||||
}
|
||||
}
|
||||
|
||||
// Extract and set the finish reason based on status
|
||||
if statusResult := responseResult.Get("status"); statusResult.Exists() {
|
||||
status := statusResult.String()
|
||||
if status == "completed" {
|
||||
template, _ = sjson.Set(template, "choices.0.finish_reason", "stop")
|
||||
template, _ = sjson.Set(template, "choices.0.native_finish_reason", "stop")
|
||||
}
|
||||
}
|
||||
|
||||
return template
|
||||
rootResult := gjson.ParseBytes(rawJSON)
|
||||
// Verify this is a response.completed event
|
||||
if rootResult.Get("type").String() != "response.completed" {
|
||||
return ""
|
||||
}
|
||||
return ""
|
||||
|
||||
unixTimestamp := time.Now().Unix()
|
||||
|
||||
responseResult := rootResult.Get("response")
|
||||
|
||||
template := `{"id":"","object":"chat.completion","created":123456,"model":"model","choices":[{"index":0,"message":{"role":"assistant","content":null,"reasoning_content":null,"tool_calls":null},"finish_reason":null,"native_finish_reason":null}]}`
|
||||
|
||||
// Extract and set the model version.
|
||||
if modelResult := responseResult.Get("model"); modelResult.Exists() {
|
||||
template, _ = sjson.Set(template, "model", modelResult.String())
|
||||
}
|
||||
|
||||
// Extract and set the creation timestamp.
|
||||
if createdAtResult := responseResult.Get("created_at"); createdAtResult.Exists() {
|
||||
template, _ = sjson.Set(template, "created", createdAtResult.Int())
|
||||
} else {
|
||||
template, _ = sjson.Set(template, "created", unixTimestamp)
|
||||
}
|
||||
|
||||
// Extract and set the response ID.
|
||||
if idResult := responseResult.Get("id"); idResult.Exists() {
|
||||
template, _ = sjson.Set(template, "id", idResult.String())
|
||||
}
|
||||
|
||||
// Extract and set usage metadata (token counts).
|
||||
if usageResult := responseResult.Get("usage"); usageResult.Exists() {
|
||||
if outputTokensResult := usageResult.Get("output_tokens"); outputTokensResult.Exists() {
|
||||
template, _ = sjson.Set(template, "usage.completion_tokens", outputTokensResult.Int())
|
||||
}
|
||||
if totalTokensResult := usageResult.Get("total_tokens"); totalTokensResult.Exists() {
|
||||
template, _ = sjson.Set(template, "usage.total_tokens", totalTokensResult.Int())
|
||||
}
|
||||
if inputTokensResult := usageResult.Get("input_tokens"); inputTokensResult.Exists() {
|
||||
template, _ = sjson.Set(template, "usage.prompt_tokens", inputTokensResult.Int())
|
||||
}
|
||||
if reasoningTokensResult := usageResult.Get("output_tokens_details.reasoning_tokens"); reasoningTokensResult.Exists() {
|
||||
template, _ = sjson.Set(template, "usage.completion_tokens_details.reasoning_tokens", reasoningTokensResult.Int())
|
||||
}
|
||||
}
|
||||
|
||||
// Process the output array for content and function calls
|
||||
outputResult := responseResult.Get("output")
|
||||
if outputResult.IsArray() {
|
||||
outputArray := outputResult.Array()
|
||||
var contentText string
|
||||
var reasoningText string
|
||||
var toolCalls []string
|
||||
|
||||
for _, outputItem := range outputArray {
|
||||
outputType := outputItem.Get("type").String()
|
||||
|
||||
switch outputType {
|
||||
case "reasoning":
|
||||
// Extract reasoning content from summary
|
||||
if summaryResult := outputItem.Get("summary"); summaryResult.IsArray() {
|
||||
summaryArray := summaryResult.Array()
|
||||
for _, summaryItem := range summaryArray {
|
||||
if summaryItem.Get("type").String() == "summary_text" {
|
||||
reasoningText = summaryItem.Get("text").String()
|
||||
break
|
||||
}
|
||||
}
|
||||
}
|
||||
case "message":
|
||||
// Extract message content
|
||||
if contentResult := outputItem.Get("content"); contentResult.IsArray() {
|
||||
contentArray := contentResult.Array()
|
||||
for _, contentItem := range contentArray {
|
||||
if contentItem.Get("type").String() == "output_text" {
|
||||
contentText = contentItem.Get("text").String()
|
||||
break
|
||||
}
|
||||
}
|
||||
}
|
||||
case "function_call":
|
||||
// Handle function call content
|
||||
functionCallTemplate := `{"id": "","type": "function","function": {"name": "","arguments": ""}}`
|
||||
|
||||
if callIdResult := outputItem.Get("call_id"); callIdResult.Exists() {
|
||||
functionCallTemplate, _ = sjson.Set(functionCallTemplate, "id", callIdResult.String())
|
||||
}
|
||||
|
||||
if nameResult := outputItem.Get("name"); nameResult.Exists() {
|
||||
n := nameResult.String()
|
||||
rev := buildReverseMapFromOriginalOpenAI(originalRequestRawJSON)
|
||||
if orig, ok := rev[n]; ok {
|
||||
n = orig
|
||||
}
|
||||
functionCallTemplate, _ = sjson.Set(functionCallTemplate, "function.name", n)
|
||||
}
|
||||
|
||||
if argsResult := outputItem.Get("arguments"); argsResult.Exists() {
|
||||
functionCallTemplate, _ = sjson.Set(functionCallTemplate, "function.arguments", argsResult.String())
|
||||
}
|
||||
|
||||
toolCalls = append(toolCalls, functionCallTemplate)
|
||||
}
|
||||
}
|
||||
|
||||
// Set content and reasoning content if found
|
||||
if contentText != "" {
|
||||
template, _ = sjson.Set(template, "choices.0.message.content", contentText)
|
||||
template, _ = sjson.Set(template, "choices.0.message.role", "assistant")
|
||||
}
|
||||
|
||||
if reasoningText != "" {
|
||||
template, _ = sjson.Set(template, "choices.0.message.reasoning_content", reasoningText)
|
||||
template, _ = sjson.Set(template, "choices.0.message.role", "assistant")
|
||||
}
|
||||
|
||||
// Add tool calls if any
|
||||
if len(toolCalls) > 0 {
|
||||
template, _ = sjson.SetRaw(template, "choices.0.message.tool_calls", `[]`)
|
||||
for _, toolCall := range toolCalls {
|
||||
template, _ = sjson.SetRaw(template, "choices.0.message.tool_calls.-1", toolCall)
|
||||
}
|
||||
template, _ = sjson.Set(template, "choices.0.message.role", "assistant")
|
||||
}
|
||||
}
|
||||
|
||||
// Extract and set the finish reason based on status
|
||||
if statusResult := responseResult.Get("status"); statusResult.Exists() {
|
||||
status := statusResult.String()
|
||||
if status == "completed" {
|
||||
template, _ = sjson.Set(template, "choices.0.finish_reason", "stop")
|
||||
template, _ = sjson.Set(template, "choices.0.native_finish_reason", "stop")
|
||||
}
|
||||
}
|
||||
|
||||
return template
|
||||
}
|
||||
|
||||
// buildReverseMapFromOriginalOpenAI builds a map of shortened tool name -> original tool name
|
||||
|
||||
Reference in New Issue
Block a user