From 3eaabc8c359596ffc9c2999ba0125def2b316f14 Mon Sep 17 00:00:00 2001
From: Cyberes <cyberes@evulid.cc>
Date: Mon, 25 Sep 2023 12:38:02 -0600
Subject: [PATCH] fix copied code

---
 llm_server/routes/openai/chat_completions.py | 34 +++++++++++---------
 llm_server/routes/v1/generate_stream.py      |  6 ++--
 2 files changed, 21 insertions(+), 19 deletions(-)

diff --git a/llm_server/routes/openai/chat_completions.py b/llm_server/routes/openai/chat_completions.py
index 5b2a83b..4069475 100644
--- a/llm_server/routes/openai/chat_completions.py
+++ b/llm_server/routes/openai/chat_completions.py
@@ -65,25 +65,27 @@ def openai_chat_completions():
                                             json_obj = json.loads(json_str.decode())
                                             new = json_obj['text'][0].split(handler.prompt + generated_text)[1]
                                             generated_text = generated_text + new
-                                            data = {
-                                                "id": f"chatcmpl-{generate_oai_string(30)}",
-                                                "object": "chat.completion.chunk",
-                                                "created": int(time.time()),
-                                                "model": model,
-                                                "choices": [
-                                                    {
-                                                        "index": 0,
-                                                        "delta": {
-                                                            "content": new
-                                                        },
-                                                        "finish_reason": None
-                                                    }
-                                                ]
-                                            }
-                                            yield f'data: {json.dumps(data)}\n\n'
                                         except IndexError:
+                                            # ????
                                             continue
 
+                                        data = {
+                                            "id": f"chatcmpl-{generate_oai_string(30)}",
+                                            "object": "chat.completion.chunk",
+                                            "created": int(time.time()),
+                                            "model": model,
+                                            "choices": [
+                                                {
+                                                    "index": 0,
+                                                    "delta": {
+                                                        "content": new
+                                                    },
+                                                    "finish_reason": None
+                                                }
+                                            ]
+                                        }
+                                        yield f'data: {json.dumps(data)}\n\n'
+
                         yield 'data: [DONE]\n\n'
                         end_time = time.time()
                         elapsed_time = end_time - start_time
diff --git a/llm_server/routes/v1/generate_stream.py b/llm_server/routes/v1/generate_stream.py
index ed112f6..2e2c188 100644
--- a/llm_server/routes/v1/generate_stream.py
+++ b/llm_server/routes/v1/generate_stream.py
@@ -9,7 +9,7 @@ from ..helpers.client import format_sillytavern_err
 from ..helpers.http import require_api_key, validate_json
 from ..ooba_request_handler import OobaRequestHandler
 from ... import opts
-from ...database.database import increment_token_uses, log_prompt
+from ...database.database import log_prompt
 from ...llm.generator import generator
 from ...llm.vllm import tokenize
 from ...stream import sock
@@ -81,7 +81,9 @@ def stream(ws):
                             for json_str in json_strs:
                                 if json_str:
                                     try:
+                                        json_obj = json.loads(json_str.decode())
                                         new = json_obj['text'][0].split(input_prompt + generated_text)[1]
+                                        generated_text = generated_text + new
                                     except IndexError:
                                         # ????
                                         continue
@@ -92,8 +94,6 @@ def stream(ws):
                                         'text': new
                                     }))
                                     message_num += 1
-
-                                    generated_text = generated_text + new
                                     partial_response = b''  # Reset the partial response
 
                         # If there is no more data, break the loop