Update ai statistics #1303
Update ai statistics #1303
Conversation
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## main #1303 +/- ##
==========================================
+ Coverage 35.91% 44.23% +8.32%
==========================================
Files 69 76 +7
Lines 11576 9895 -1681
==========================================
+ Hits 4157 4377 +220
+ Misses 7104 5183 -1921
- Partials 315 335 +20 |
| logAttributes[logAttribute.Key] = value | ||
| } | ||
| case "request_body": | ||
| raw := gjson.GetBytes(body, logAttribute.Value).Raw |
There was a problem hiding this comment.
Raw取的是序列化后的数据，用.String()的话，打印日志的时候会有格式问题
There was a problem hiding this comment.
{"ai_log":"{"consumer":"21321r9fncsb2dq","question":"用三行python实现求和函数","answer":"在Python中，你可以使用内置的sum()函数来计算一个数字列表的总和。这是一个简单的实现：\n\npython\\ndef sum_function(numbers):\\n return sum(numbers)\\n\\n# 测试函数\\nnumbers = [1, 2, 3, 4, 5]\\nprint(sum_function(numbers)) # 输出：15\\n\n\n如果你想要自己手动实现一个求和函数，可以使用以下代码：\n\npython\\ndef sum_function(numbers):\\n total = 0\\n for number in numbers:\\n total += number\\n return total\\n\\n# 测试函数\\nnumbers = [1, 2, 3, 4, 5]\\nprint(sum_function(numbers)) # 输出：15\\n\n\n在这两个例子中，我们定义了一个名为sum_function的函数，它接受一个数字列表作为参数，并返回该列表中所有数字的总和。","output_token":"188","llm_first_token_duration":"237","llm_service_duration":"4341","model":"qwen-turbo","input_token":"16"}"}
There was a problem hiding this comment.
上面这个是一个例子，如果LLM返回的内容中有换行符之类的，容易有问题，我直接复制粘贴被github打印了。。原始的是这样的：
There was a problem hiding this comment.
{"ai_log":"{\"consumer\":\"21321r9fncsb2dq\",\"question\":\"用三行python实现求和函数\",\"answer\":\"在Python中，你可以使用内置的`sum()`函数来计算一个数字列表的总和。这是一个简单的实现：\\n\\n```python\\ndef sum_function(numbers):\\n return sum(numbers)\\n\\n# 测试函数\\nnumbers = [1, 2, 3, 4, 5]\\nprint(sum_function(numbers)) # 输出：15\\n```\\n\\n如果你想要自己手动实现一个求和函数，可以使用以下代码：\\n\\n```python\\ndef sum_function(numbers):\\n total = 0\\n for number in numbers:\\n total += number\\n return total\\n\\n# 测试函数\\nnumbers = [1, 2, 3, 4, 5]\\nprint(sum_function(numbers)) # 输出：15\\n```\\n\\n在这两个例子中，我们定义了一个名为`sum_function`的函数，它接受一个数字列表作为参数，并返回该列表中所有数字的总和。\",\"output_token\":\"188\",\"llm_first_token_duration\":\"237\",\"llm_service_duration\":\"4341\",\"model\":\"qwen-turbo\",\"input_token\":\"16\"}"}| return data | ||
| } | ||
| setTraceAttributeValueBySource(config, "response_streaming_body", streamingBodyBuffer, log) | ||
| setLogAttributeValueBySource(ctx, config, "response_streaming_body", streamingBodyBuffer, log) |
There was a problem hiding this comment.
为什么 endOfStream 的时候记录的 attribute 还是 “response_streaming_body” source 的？
增加用户扩展 log & span attribute 机制