move examples gateway

AudioQnA/audioqna.py

@@ -4,7 +4,10 @@
 import asyncio
 import os
 
-from comps import AudioQnAGateway, MicroService, ServiceOrchestrator, ServiceType
+from comps import Gateway, MegaServiceEndpoint, MicroService, ServiceOrchestrator, ServiceType
+from comps.cores.proto.api_protocol import AudioChatCompletionRequest, ChatCompletionResponse
+from comps.cores.proto.docarray import LLMParams
+from fastapi import Request
 
 MEGA_SERVICE_HOST_IP = os.getenv("MEGA_SERVICE_HOST_IP", "0.0.0.0")
 MEGA_SERVICE_PORT = int(os.getenv("MEGA_SERVICE_PORT", 8888))
@@ -16,11 +19,18 @@
 TTS_SERVICE_PORT = int(os.getenv("TTS_SERVICE_PORT", 9088))
 
 
-class AudioQnAService:
+class AudioQnAService(Gateway):
     def __init__(self, host="0.0.0.0", port=8000):
         self.host = host
         self.port = port
-        self.megaservice = ServiceOrchestrator()
+        super().__init__(
+            megaservice=ServiceOrchestrator(),
+            host=self.host,
+            port=self.port,
+            endpoint=str(MegaServiceEndpoint.AUDIO_QNA),
+            input_datatype=AudioChatCompletionRequest,
+            output_datatype=ChatCompletionResponse,
+        )
 
     def add_remote_service(self):
         asr = MicroService(
@@ -50,7 +60,30 @@ def add_remote_service(self):
         self.megaservice.add(asr).add(llm).add(tts)
         self.megaservice.flow_to(asr, llm)
         self.megaservice.flow_to(llm, tts)
-        self.gateway = AudioQnAGateway(megaservice=self.megaservice, host="0.0.0.0", port=self.port)
+
+    async def handle_request(self, request: Request):
+        data = await request.json()
+
+        chat_request = AudioChatCompletionRequest.parse_obj(data)
+        parameters = LLMParams(
+            # relatively lower max_tokens for audio conversation
+            max_tokens=chat_request.max_tokens if chat_request.max_tokens else 128,
+            top_k=chat_request.top_k if chat_request.top_k else 10,
+            top_p=chat_request.top_p if chat_request.top_p else 0.95,
+            temperature=chat_request.temperature if chat_request.temperature else 0.01,
+            frequency_penalty=chat_request.frequency_penalty if chat_request.frequency_penalty else 0.0,
+            presence_penalty=chat_request.presence_penalty if chat_request.presence_penalty else 0.0,
+            repetition_penalty=chat_request.repetition_penalty if chat_request.repetition_penalty else 1.03,
+            streaming=False,  # TODO add streaming LLM output as input to TTS
+        )
+        result_dict, runtime_graph = await self.megaservice.schedule(
+            initial_inputs={"byte_str": chat_request.audio}, llm_parameters=parameters
+        )
+
+        last_node = runtime_graph.all_leaves()[-1]
+        response = result_dict[last_node]["byte_str"]
+
+        return response
 
 
 if __name__ == "__main__":

ChatQnA/chatqna.py

@@ -6,7 +6,17 @@
 import os
 import re
 
-from comps import ChatQnAGateway, MicroService, ServiceOrchestrator, ServiceType
+from comps import Gateway, MegaServiceEndpoint, MicroService, ServiceOrchestrator, ServiceType
+from comps.cores.proto.api_protocol import (
+    ChatCompletionRequest,
+    ChatCompletionResponse,
+    ChatCompletionResponseChoice,
+    ChatMessage,
+    UsageInfo,
+)
+from comps.cores.proto.docarray import LLMParams, RerankerParms, RetrieverParms
+from fastapi import Request
+from fastapi.responses import StreamingResponse
 from langchain_core.prompts import PromptTemplate
 
 
@@ -173,14 +183,22 @@ def align_generator(self, gen, **kwargs):
     yield "data: [DONE]\n\n"
 
 
-class ChatQnAService:
+class ChatQnAService(Gateway):
     def __init__(self, host="0.0.0.0", port=8000):
         self.host = host
         self.port = port
         ServiceOrchestrator.align_inputs = align_inputs
         ServiceOrchestrator.align_outputs = align_outputs
         ServiceOrchestrator.align_generator = align_generator
-        self.megaservice = ServiceOrchestrator()
+
+        super().__init__(
+            megaservice=ServiceOrchestrator(),
+            host=self.host,
+            port=self.port,
+            endpoint=str(MegaServiceEndpoint.CHAT_QNA),
+            input_datatype=ChatCompletionRequest,
+            output_datatype=ChatCompletionResponse,
+        )
 
     def add_remote_service(self):
 
@@ -223,7 +241,6 @@ def add_remote_service(self):
         self.megaservice.flow_to(embedding, retriever)
         self.megaservice.flow_to(retriever, rerank)
         self.megaservice.flow_to(rerank, llm)
-        self.gateway = ChatQnAGateway(megaservice=self.megaservice, host="0.0.0.0", port=self.port)
 
     def add_remote_service_without_rerank(self):
 
@@ -256,7 +273,6 @@ def add_remote_service_without_rerank(self):
         self.megaservice.add(embedding).add(retriever).add(llm)
         self.megaservice.flow_to(embedding, retriever)
         self.megaservice.flow_to(retriever, llm)
-        self.gateway = ChatQnAGateway(megaservice=self.megaservice, host="0.0.0.0", port=self.port)
 
     def add_remote_service_with_guardrails(self):
         guardrail_in = MicroService(
@@ -314,7 +330,55 @@ def add_remote_service_with_guardrails(self):
         self.megaservice.flow_to(retriever, rerank)
         self.megaservice.flow_to(rerank, llm)
         # self.megaservice.flow_to(llm, guardrail_out)
-        self.gateway = ChatQnAGateway(megaservice=self.megaservice, host="0.0.0.0", port=self.port)
+
+    async def handle_request(self, request: Request):
+        data = await request.json()
+        stream_opt = data.get("stream", True)
+        chat_request = ChatCompletionRequest.parse_obj(data)
+        prompt = self._handle_message(chat_request.messages)
+        parameters = LLMParams(
+            max_tokens=chat_request.max_tokens if chat_request.max_tokens else 1024,
+            top_k=chat_request.top_k if chat_request.top_k else 10,
+            top_p=chat_request.top_p if chat_request.top_p else 0.95,
+            temperature=chat_request.temperature if chat_request.temperature else 0.01,
+            frequency_penalty=chat_request.frequency_penalty if chat_request.frequency_penalty else 0.0,
+            presence_penalty=chat_request.presence_penalty if chat_request.presence_penalty else 0.0,
+            repetition_penalty=chat_request.repetition_penalty if chat_request.repetition_penalty else 1.03,
+            streaming=stream_opt,
+            chat_template=chat_request.chat_template if chat_request.chat_template else None,
+        )
+        retriever_parameters = RetrieverParms(
+            search_type=chat_request.search_type if chat_request.search_type else "similarity",
+            k=chat_request.k if chat_request.k else 4,
+            distance_threshold=chat_request.distance_threshold if chat_request.distance_threshold else None,
+            fetch_k=chat_request.fetch_k if chat_request.fetch_k else 20,
+            lambda_mult=chat_request.lambda_mult if chat_request.lambda_mult else 0.5,
+            score_threshold=chat_request.score_threshold if chat_request.score_threshold else 0.2,
+        )
+        reranker_parameters = RerankerParms(
+            top_n=chat_request.top_n if chat_request.top_n else 1,
+        )
+        result_dict, runtime_graph = await self.megaservice.schedule(
+            initial_inputs={"text": prompt},
+            llm_parameters=parameters,
+            retriever_parameters=retriever_parameters,
+            reranker_parameters=reranker_parameters,
+        )
+        for node, response in result_dict.items():
+            if isinstance(response, StreamingResponse):
+                return response
+        last_node = runtime_graph.all_leaves()[-1]
+        response = result_dict[last_node]["text"]
+        choices = []
+        usage = UsageInfo()
+        choices.append(
+            ChatCompletionResponseChoice(
+                index=0,
+                message=ChatMessage(role="assistant", content=response),
+                finish_reason="stop",
+            )
+        )
+        return ChatCompletionResponse(model="chatqna", choices=choices, usage=usage)
 
 
 if __name__ == "__main__":
@@ -331,3 +395,5 @@ def add_remote_service_with_guardrails(self):
         chatqna.add_remote_service_with_guardrails()
     else:
         chatqna.add_remote_service()
+
+    # chatqna.start()

CodeGen/codegen.py

@@ -4,19 +4,36 @@
 import asyncio
 import os
 
-from comps import CodeGenGateway, MicroService, ServiceOrchestrator, ServiceType
+from comps import Gateway, MegaServiceEndpoint, MicroService, ServiceOrchestrator, ServiceType
+from comps.cores.proto.api_protocol import (
+    ChatCompletionRequest,
+    ChatCompletionResponse,
+    ChatCompletionResponseChoice,
+    ChatMessage,
+    UsageInfo,
+)
+from comps.cores.proto.docarray import LLMParams
+from fastapi import Request
+from fastapi.responses import StreamingResponse
 
 MEGA_SERVICE_HOST_IP = os.getenv("MEGA_SERVICE_HOST_IP", "0.0.0.0")
 MEGA_SERVICE_PORT = int(os.getenv("MEGA_SERVICE_PORT", 7778))
 LLM_SERVICE_HOST_IP = os.getenv("LLM_SERVICE_HOST_IP", "0.0.0.0")
 LLM_SERVICE_PORT = int(os.getenv("LLM_SERVICE_PORT", 9000))
 
 
-class CodeGenService:
+class CodeGenService(Gateway):
     def __init__(self, host="0.0.0.0", port=8000):
         self.host = host
         self.port = port
-        self.megaservice = ServiceOrchestrator()
+        super().__init__(
+            megaservice=ServiceOrchestrator(),
+            host=self.host,
+            port=self.port,
+            endpoint=str(MegaServiceEndpoint.CODE_GEN),
+            input_datatype=ChatCompletionRequest,
+            output_datatype=ChatCompletionResponse,
+        )
 
     def add_remote_service(self):
         llm = MicroService(
@@ -28,7 +45,45 @@ def add_remote_service(self):
             service_type=ServiceType.LLM,
         )
         self.megaservice.add(llm)
-        self.gateway = CodeGenGateway(megaservice=self.megaservice, host="0.0.0.0", port=self.port)
+
+    async def handle_request(self, request: Request):
+        data = await request.json()
+        stream_opt = data.get("stream", True)
+        chat_request = ChatCompletionRequest.parse_obj(data)
+        prompt = self._handle_message(chat_request.messages)
+        parameters = LLMParams(
+            max_tokens=chat_request.max_tokens if chat_request.max_tokens else 1024,
+            top_k=chat_request.top_k if chat_request.top_k else 10,
+            top_p=chat_request.top_p if chat_request.top_p else 0.95,
+            temperature=chat_request.temperature if chat_request.temperature else 0.01,
+            frequency_penalty=chat_request.frequency_penalty if chat_request.frequency_penalty else 0.0,
+            presence_penalty=chat_request.presence_penalty if chat_request.presence_penalty else 0.0,
+            repetition_penalty=chat_request.repetition_penalty if chat_request.repetition_penalty else 1.03,
+            streaming=stream_opt,
+        )
+        result_dict, runtime_graph = await self.megaservice.schedule(
+            initial_inputs={"query": prompt}, llm_parameters=parameters
+        )
+        for node, response in result_dict.items():
+            # Here it suppose the last microservice in the megaservice is LLM.
+            if (
+                isinstance(response, StreamingResponse)
+                and node == list(self.megaservice.services.keys())[-1]
+                and self.megaservice.services[node].service_type == ServiceType.LLM
+            ):
+                return response
+        last_node = runtime_graph.all_leaves()[-1]
+        response = result_dict[last_node]["text"]
+        choices = []
+        usage = UsageInfo()
+        choices.append(
+            ChatCompletionResponseChoice(
+                index=0,
+                message=ChatMessage(role="assistant", content=response),
+                finish_reason="stop",
+            )
+        )
+        return ChatCompletionResponse(model="codegen", choices=choices, usage=usage)
 
 
 if __name__ == "__main__":

CodeTrans/code_translation.py

@@ -4,19 +4,35 @@
 import asyncio
 import os
 
-from comps import CodeTransGateway, MicroService, ServiceOrchestrator
+from comps import Gateway, MegaServiceEndpoint, MicroService, ServiceOrchestrator, ServiceType
+from comps.cores.proto.api_protocol import (
+    ChatCompletionRequest,
+    ChatCompletionResponse,
+    ChatCompletionResponseChoice,
+    ChatMessage,
+    UsageInfo,
+)
+from fastapi import Request
+from fastapi.responses import StreamingResponse
 
 MEGA_SERVICE_HOST_IP = os.getenv("MEGA_SERVICE_HOST_IP", "0.0.0.0")
 MEGA_SERVICE_PORT = int(os.getenv("MEGA_SERVICE_PORT", 7777))
 LLM_SERVICE_HOST_IP = os.getenv("LLM_SERVICE_HOST_IP", "0.0.0.0")
 LLM_SERVICE_PORT = int(os.getenv("LLM_SERVICE_PORT", 9000))
 
 
-class CodeTransService:
+class CodeTransService(Gateway):
     def __init__(self, host="0.0.0.0", port=8000):
         self.host = host
         self.port = port
-        self.megaservice = ServiceOrchestrator()
+        super().__init__(
+            megaservice=ServiceOrchestrator(),
+            host=self.host,
+            port=self.port,
+            endpoint=str(MegaServiceEndpoint.CODE_TRANS),
+            input_datatype=ChatCompletionRequest,
+            output_datatype=ChatCompletionResponse,
+        )
 
     def add_remote_service(self):
         llm = MicroService(
@@ -27,7 +43,46 @@ def add_remote_service(self):
             use_remote_service=True,
         )
         self.megaservice.add(llm)
-        self.gateway = CodeTransGateway(megaservice=self.megaservice, host="0.0.0.0", port=self.port)
+
+    async def handle_request(self, request: Request):
+        data = await request.json()
+        language_from = data["language_from"]
+        language_to = data["language_to"]
+        source_code = data["source_code"]
+        prompt_template = """
+            ### System: Please translate the following {language_from} codes into {language_to} codes.
+
+            ### Original codes:
+            '''{language_from}
+
+            {source_code}
+
+            '''
+
+            ### Translated codes:
+        """
+        prompt = prompt_template.format(language_from=language_from, language_to=language_to, source_code=source_code)
+        result_dict, runtime_graph = await self.megaservice.schedule(initial_inputs={"query": prompt})
+        for node, response in result_dict.items():
+            # Here it suppose the last microservice in the megaservice is LLM.
+            if (
+                isinstance(response, StreamingResponse)
+                and node == list(self.megaservice.services.keys())[-1]
+                and self.megaservice.services[node].service_type == ServiceType.LLM
+            ):
+                return response
+        last_node = runtime_graph.all_leaves()[-1]
+        response = result_dict[last_node]["text"]
+        choices = []
+        usage = UsageInfo()
+        choices.append(
+            ChatCompletionResponseChoice(
+                index=0,
+                message=ChatMessage(role="assistant", content=response),
+                finish_reason="stop",
+            )
+        )
+        return ChatCompletionResponse(model="codetrans", choices=choices, usage=usage)
 
 
 if __name__ == "__main__":