【BugFix】completion接口echo回显支持 (#3245)

* wenxin-tools-511,修复v1/completion无法回显的问题。

* 支持多prompt的回显

* 支持多prompt情况下的流式回显

* 补充了 completion 接口支持 echo 的单元测试

* pre-commit

* 移除了多余的test文件

* 修复了completion接口echo支持的单测方法

* 补充了单元测试文件

* 补充单测

* unittest

* 补充单测

* 修复单测

* 删除不必要的assert.

* 重新提交

* 更新测试方法

* ut

* 验证是否是正确思路单测

* 验证是否是正确思路单测

* 验证是否是正确思路单测3

* 优化单测代码，有针对性地缩小单测范围。

* 优化单测代码2，有针对性地缩小单测范围。

* 优化单测代码3，有针对性地缩小单测范围。

* support 'echo' in chat/completion.

* update

* update

* update

* update

* update

* update

* 补充了关于tokenid的单元测试

* update

* 修正index错误

* 修正index错误

fastdeploy/entrypoints/openai/serving_completion.py

@@ -241,6 +241,14 @@ class OpenAIServingCompletion:
             if dealer is not None:
                 dealer.close()
 
+    async def _echo_back_prompt(self, request, res, idx):
+        if res["outputs"].get("send_idx", -1) == 0 and request.echo:
+            if isinstance(request.prompt, list):
+                prompt_text = request.prompt[idx]
+            else:
+                prompt_text = request.prompt
+            res["outputs"]["text"] = prompt_text + (res["outputs"]["text"] or "")
+
     def calc_finish_reason(self, max_tokens, token_num, output, tool_called):
         if max_tokens is None or token_num != max_tokens:
             if tool_called or output.get("tool_call"):
@@ -338,6 +346,7 @@ class OpenAIServingCompletion:
                     else:
                         arrival_time = res["metrics"]["arrival_time"] - inference_start_time[idx]
 
+                    await self._echo_back_prompt(request, res, idx)
                     output = res["outputs"]
                     output_top_logprobs = output["top_logprobs"]
                     logprobs_res: Optional[CompletionLogprobs] = None
@@ -450,7 +459,7 @@ class OpenAIServingCompletion:
             final_res = final_res_batch[idx]
             prompt_token_ids = prompt_batched_token_ids[idx]
             assert prompt_token_ids is not None
-            prompt_text = final_res["prompt"]
+            prompt_text = request.prompt
             completion_token_ids = completion_batched_token_ids[idx]
 
             output = final_res["outputs"]
@@ -468,16 +477,14 @@ class OpenAIServingCompletion:
 
             if request.echo:
                 assert prompt_text is not None
-                if request.max_tokens == 0:
-                    token_ids = prompt_token_ids
-                    output_text = prompt_text
+                token_ids = [*prompt_token_ids, *output["token_ids"]]
+                if isinstance(prompt_text, list):
+                    output_text = prompt_text[idx] + output["text"]
                 else:
-                    token_ids = [*prompt_token_ids, *output["token_ids"]]
-                    output_text = prompt_text + output["text"]
+                    output_text = str(prompt_text) + output["text"]
             else:
                 token_ids = output["token_ids"]
                 output_text = output["text"]
-
             finish_reason = self.calc_finish_reason(request.max_tokens, final_res["output_token_ids"], output, False)
 
             choice_data = CompletionResponseChoice(