feat: re-add prompt messages to result and chunks in llm (#17883)

Signed-off-by: -LAN- <laipz8200@outlook.com>

api/core/model_manager.py

@@ -177,7 +177,7 @@ class ModelInstance:
         )
 
     def get_llm_num_tokens(
-        self, prompt_messages: list[PromptMessage], tools: Optional[list[PromptMessageTool]] = None
+        self, prompt_messages: Sequence[PromptMessage], tools: Optional[Sequence[PromptMessageTool]] = None
     ) -> int:
         """
         Get number of tokens for llm

api/core/model_runtime/callbacks/base_callback.py

@@ -58,7 +58,7 @@ class Callback(ABC):
         chunk: LLMResultChunk,
         model: str,
         credentials: dict,
-        prompt_messages: list[PromptMessage],
+        prompt_messages: Sequence[PromptMessage],
         model_parameters: dict,
         tools: Optional[list[PromptMessageTool]] = None,
         stop: Optional[Sequence[str]] = None,
@@ -88,7 +88,7 @@ class Callback(ABC):
         result: LLMResult,
         model: str,
         credentials: dict,
-        prompt_messages: list[PromptMessage],
+        prompt_messages: Sequence[PromptMessage],
         model_parameters: dict,
         tools: Optional[list[PromptMessageTool]] = None,
         stop: Optional[Sequence[str]] = None,

api/core/model_runtime/callbacks/logging_callback.py

@@ -74,7 +74,7 @@ class LoggingCallback(Callback):
         chunk: LLMResultChunk,
         model: str,
         credentials: dict,
-        prompt_messages: list[PromptMessage],
+        prompt_messages: Sequence[PromptMessage],
         model_parameters: dict,
         tools: Optional[list[PromptMessageTool]] = None,
         stop: Optional[Sequence[str]] = None,
@@ -104,7 +104,7 @@ class LoggingCallback(Callback):
         result: LLMResult,
         model: str,
         credentials: dict,
-        prompt_messages: list[PromptMessage],
+        prompt_messages: Sequence[PromptMessage],
         model_parameters: dict,
         tools: Optional[list[PromptMessageTool]] = None,
         stop: Optional[Sequence[str]] = None,

api/core/model_runtime/entities/llm_entities.py

@@ -1,8 +1,9 @@
+from collections.abc import Sequence
 from decimal import Decimal
 from enum import StrEnum
 from typing import Optional
 
-from pydantic import BaseModel
+from pydantic import BaseModel, Field
 
 from core.model_runtime.entities.message_entities import AssistantPromptMessage, PromptMessage
 from core.model_runtime.entities.model_entities import ModelUsage, PriceInfo
@@ -107,7 +108,7 @@ class LLMResult(BaseModel):
 
     id: Optional[str] = None
     model: str
-    prompt_messages: list[PromptMessage]
+    prompt_messages: Sequence[PromptMessage] = Field(default_factory=list)
     message: AssistantPromptMessage
     usage: LLMUsage
     system_fingerprint: Optional[str] = None
@@ -130,7 +131,7 @@ class LLMResultChunk(BaseModel):
     """
 
     model: str
-    prompt_messages: list[PromptMessage]
+    prompt_messages: Sequence[PromptMessage] = Field(default_factory=list)
     system_fingerprint: Optional[str] = None
     delta: LLMResultChunkDelta
 

api/core/model_runtime/model_providers/__base/large_language_model.py

@@ -45,7 +45,7 @@ class LargeLanguageModel(AIModel):
         stream: bool = True,
         user: Optional[str] = None,
         callbacks: Optional[list[Callback]] = None,
-    ) -> Union[LLMResult, Generator]:
+    ) -> Union[LLMResult, Generator[LLMResultChunk, None, None]]:
         """
         Invoke large language model
 
@@ -205,22 +205,26 @@ class LargeLanguageModel(AIModel):
                 user=user,
                 callbacks=callbacks,
             )
-
+            # Following https://github.com/langgenius/dify/issues/17799,
-        return result
+            # we removed the prompt_messages from the chunk on the plugin daemon side.
+            # To ensure compatibility, we add the prompt_messages back here.
+            result.prompt_messages = prompt_messages
+            return result
+        raise NotImplementedError("unsupported invoke result type", type(result))
 
     def _invoke_result_generator(
         self,
         model: str,
         result: Generator,
         credentials: dict,
-        prompt_messages: list[PromptMessage],
+        prompt_messages: Sequence[PromptMessage],
         model_parameters: dict,
         tools: Optional[list[PromptMessageTool]] = None,
         stop: Optional[Sequence[str]] = None,
         stream: bool = True,
         user: Optional[str] = None,
         callbacks: Optional[list[Callback]] = None,
-    ) -> Generator:
+    ) -> Generator[LLMResultChunk, None, None]:
         """
         Invoke result generator
 
@@ -235,6 +239,10 @@ class LargeLanguageModel(AIModel):
 
         try:
             for chunk in result:
+                # Following https://github.com/langgenius/dify/issues/17799,
+                # we removed the prompt_messages from the chunk on the plugin daemon side.
+                # To ensure compatibility, we add the prompt_messages back here.
+                chunk.prompt_messages = prompt_messages
                 yield chunk
 
                 self._trigger_new_chunk_callbacks(
@@ -403,7 +411,7 @@ class LargeLanguageModel(AIModel):
         chunk: LLMResultChunk,
         model: str,
         credentials: dict,
-        prompt_messages: list[PromptMessage],
+        prompt_messages: Sequence[PromptMessage],
         model_parameters: dict,
         tools: Optional[list[PromptMessageTool]] = None,
         stop: Optional[Sequence[str]] = None,
@@ -450,7 +458,7 @@ class LargeLanguageModel(AIModel):
         model: str,
         result: LLMResult,
         credentials: dict,
-        prompt_messages: list[PromptMessage],
+        prompt_messages: Sequence[PromptMessage],
         model_parameters: dict,
         tools: Optional[list[PromptMessageTool]] = None,
         stop: Optional[Sequence[str]] = None,