Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -358,6 +358,7 @@ def handsome_chat_completions():
|
|
358 |
first_token_time = first_chunk_time - start_time if first_chunk_time else 0
|
359 |
total_time = end_time - start_time
|
360 |
|
|
|
361 |
prompt_tokens = 0
|
362 |
completion_tokens = 0
|
363 |
response_content = ""
|
@@ -369,6 +370,7 @@ def handsome_chat_completions():
|
|
369 |
try:
|
370 |
response_json = json.loads(line)
|
371 |
|
|
|
372 |
if "usage" in response_json and "completion_tokens" in response_json["usage"]:
|
373 |
completion_tokens = response_json["usage"]["completion_tokens"]
|
374 |
|
@@ -381,6 +383,7 @@ def handsome_chat_completions():
|
|
381 |
except (KeyError, ValueError, IndexError) as e:
|
382 |
logging.error(f"解析流式响应单行 JSON 失败: {e}, 行内容: {line}")
|
383 |
|
|
|
384 |
user_content = ""
|
385 |
messages = data.get("messages", [])
|
386 |
for message in messages:
|
@@ -388,16 +391,19 @@ def handsome_chat_completions():
|
|
388 |
user_content += message["content"] + " "
|
389 |
user_content = user_content.strip()
|
390 |
|
|
|
391 |
logging.info(
|
392 |
-
f"使用的key: {api_key}, 提示token: {prompt_tokens}, 输出token: {completion_tokens}, 首字用时:
|
393 |
)
|
394 |
return Response(stream_with_context(generate()), content_type=response.headers['Content-Type'])
|
395 |
else:
|
|
|
396 |
response.raise_for_status()
|
397 |
end_time = time.time()
|
398 |
response_json = response.json()
|
399 |
total_time = end_time - start_time
|
400 |
|
|
|
401 |
try:
|
402 |
prompt_tokens = response_json["usage"]["prompt_tokens"]
|
403 |
completion_tokens = response_json["usage"]["completion_tokens"]
|
@@ -408,6 +414,7 @@ def handsome_chat_completions():
|
|
408 |
completion_tokens = 0
|
409 |
response_content = ""
|
410 |
|
|
|
411 |
user_content = ""
|
412 |
messages = data.get("messages", [])
|
413 |
for message in messages:
|
@@ -415,8 +422,9 @@ def handsome_chat_completions():
|
|
415 |
user_content += message["content"] + " "
|
416 |
user_content = user_content.strip()
|
417 |
|
|
|
418 |
logging.info(
|
419 |
-
f"使用的key: {api_key}, 提示token: {prompt_tokens}, 输出token: {completion_tokens}, 首字用时: 0, 总共用时: {total_time:.4f}秒, 使用的模型: {model_name}, 用户的内容: {user_content.replace(chr(10),
|
420 |
)
|
421 |
return jsonify(response_json)
|
422 |
|
|
|
358 |
first_token_time = first_chunk_time - start_time if first_chunk_time else 0
|
359 |
total_time = end_time - start_time
|
360 |
|
361 |
+
# 处理流式响应,逐行解析 JSON
|
362 |
prompt_tokens = 0
|
363 |
completion_tokens = 0
|
364 |
response_content = ""
|
|
|
370 |
try:
|
371 |
response_json = json.loads(line)
|
372 |
|
373 |
+
# 提取信息,这里只累加 completion_tokens 和 content
|
374 |
if "usage" in response_json and "completion_tokens" in response_json["usage"]:
|
375 |
completion_tokens = response_json["usage"]["completion_tokens"]
|
376 |
|
|
|
383 |
except (KeyError, ValueError, IndexError) as e:
|
384 |
logging.error(f"解析流式响应单行 JSON 失败: {e}, 行内容: {line}")
|
385 |
|
386 |
+
# 提取用户输入的内容,忽略非文本内容
|
387 |
user_content = ""
|
388 |
messages = data.get("messages", [])
|
389 |
for message in messages:
|
|
|
391 |
user_content += message["content"] + " "
|
392 |
user_content = user_content.strip()
|
393 |
|
394 |
+
# 记录日志,将换行符替换为 \n
|
395 |
logging.info(
|
396 |
+
f"使用的key: {api_key}, 提示token: {prompt_tokens}, 输出token: {completion_tokens}, 首字用时: {first_token_time:.4f}秒, 总共用时: {total_time:.4f}秒, 使用的模型: {model_name}, 用户的内容: {user_content.replace(chr(10), '\\n').replace(chr(13), '\\n')}, 输出的内容: {response_content.replace(chr(10), '\\n').replace(chr(13), '\\n')}"
|
397 |
)
|
398 |
return Response(stream_with_context(generate()), content_type=response.headers['Content-Type'])
|
399 |
else:
|
400 |
+
# 非流式响应处理... (保持原样)
|
401 |
response.raise_for_status()
|
402 |
end_time = time.time()
|
403 |
response_json = response.json()
|
404 |
total_time = end_time - start_time
|
405 |
|
406 |
+
# 从响应中提取信息
|
407 |
try:
|
408 |
prompt_tokens = response_json["usage"]["prompt_tokens"]
|
409 |
completion_tokens = response_json["usage"]["completion_tokens"]
|
|
|
414 |
completion_tokens = 0
|
415 |
response_content = ""
|
416 |
|
417 |
+
# 提取用户输入的内容,忽略非文本内容
|
418 |
user_content = ""
|
419 |
messages = data.get("messages", [])
|
420 |
for message in messages:
|
|
|
422 |
user_content += message["content"] + " "
|
423 |
user_content = user_content.strip()
|
424 |
|
425 |
+
# 记录日志,将换行符替换为 \n
|
426 |
logging.info(
|
427 |
+
f"使用的key: {api_key}, 提示token: {prompt_tokens}, 输出token: {completion_tokens}, 首字用时: 0, 总共用时: {total_time:.4f}秒, 使用的模型: {model_name}, 用户的内容: {user_content.replace(chr(10), '\\n').replace(chr(13), '\\n')}, 输出的内容: {response_content.replace(chr(10), '\\n').replace(chr(13), '\\n')}"
|
428 |
)
|
429 |
return jsonify(response_json)
|
430 |
|