auto-coder 0.1.191__tar.gz → 0.1.192__tar.gz

{auto_coder-0.1.191 → auto_coder-0.1.192}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: auto-coder
-Version: 0.1.191
+Version: 0.1.192
 Summary: AutoCoder: AutoCoder
 Author: allwefantasy
 Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
@@ -26,7 +26,7 @@ Requires-Dist: tabulate
 Requires-Dist: jupyter_client
 Requires-Dist: prompt-toolkit
 Requires-Dist: tokenizers
-Requires-Dist: byzerllm[saas]>=0.1.138
+Requires-Dist: byzerllm[saas]>=0.1.139
 Requires-Dist: patch
 Requires-Dist: diff_match_patch
 Requires-Dist: GitPython

{auto_coder-0.1.191 → auto_coder-0.1.192}/src/auto_coder.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: auto-coder
-Version: 0.1.191
+Version: 0.1.192
 Summary: AutoCoder: AutoCoder
 Author: allwefantasy
 Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
@@ -26,7 +26,7 @@ Requires-Dist: tabulate
 Requires-Dist: jupyter_client
 Requires-Dist: prompt-toolkit
 Requires-Dist: tokenizers
-Requires-Dist: byzerllm[saas]>=0.1.138
+Requires-Dist: byzerllm[saas]>=0.1.139
 Requires-Dist: patch
 Requires-Dist: diff_match_patch
 Requires-Dist: GitPython

{auto_coder-0.1.191 → auto_coder-0.1.192}/src/auto_coder.egg-info/SOURCES.txt

@@ -94,6 +94,9 @@ src/autocoder/rag/loaders/docx_loader.py
 src/autocoder/rag/loaders/excel_loader.py
 src/autocoder/rag/loaders/pdf_loader.py
 src/autocoder/rag/loaders/ppt_loader.py
+src/autocoder/rag/stream_event/__init__.py
+src/autocoder/rag/stream_event/event_writer.py
+src/autocoder/rag/stream_event/types.py
 src/autocoder/regexproject/__init__.py
 src/autocoder/suffixproject/__init__.py
 src/autocoder/tsproject/__init__.py

{auto_coder-0.1.191 → auto_coder-0.1.192}/src/auto_coder.egg-info/requires.txt

@@ -16,7 +16,7 @@ tabulate
 jupyter_client
 prompt-toolkit
 tokenizers
-byzerllm[saas]>=0.1.138
+byzerllm[saas]>=0.1.139
 patch
 diff_match_patch
 GitPython

{auto_coder-0.1.191 → auto_coder-0.1.192}/src/autocoder/auto_coder_rag.py

@@ -264,7 +264,7 @@ def main(input_args: Optional[List[str]] = None):
     serve_parser.add_argument(
         "--inference_compute_precision",
         type=int,
-        default=6,
+        default=64,
         help="The precision of the inference compute",
     )
 

{auto_coder-0.1.191 → auto_coder-0.1.192}/src/autocoder/common/__init__.py

@@ -317,7 +317,7 @@ class AutoCoderArgs(pydantic.BaseModel):
     disable_inference_enhance: Optional[bool] = False
     inference_deep_thought: Optional[bool] = False
     inference_slow_without_deep_thought: Optional[bool] = False
-    inference_compute_precision: int = 6
+    inference_compute_precision: int = 64
     without_contexts: Optional[bool] = False
 
     class Config:

{auto_coder-0.1.191 → auto_coder-0.1.192}/src/autocoder/rag/api_server.py

@@ -19,10 +19,10 @@ from byzerllm.utils.client.entrypoints.openai.protocol import (
     ChatCompletionRequest,
     ErrorResponse,
     CompletionRequest,
-    Embeddings,
-    EmbeddingsOutput,
-    EmbeddingsData,
-    EmbeddingsUsage,
+    EmbeddingCompletionRequest,
+    EmbeddingResponse,
+    EmbeddingResponseData,
+    UsageInfo,
 )
 from pydantic import BaseModel
 from typing import List,Optional
@@ -122,34 +122,49 @@ async def create_chat_completion(
 
 
 @router_app.post("/v1/embeddings")
-async def embed(body: Embeddings):
-    """Given a prompt, the model will return one embedding.
-
+async def embed(body: EmbeddingCompletionRequest):
+    """Generate embeddings for given input text.
+    
+    Args:
+        body: The embedding request containing input text and parameters.
+        
     Returns:
-        A response object with an embedding.
+        EmbeddingResponse with embeddings and usage statistics.
     """
     embedding_id = f"embed-{random_uuid()}"
-
-    results_list = llm_client.emb(body.model, request=LLMRequest(instruction=body.input))
-    tokens = 0
-
-    return EmbeddingsOutput(
-        data=[
-            EmbeddingsData(
-                embedding=results.output,
-                index=i,
-                object="embedding",
-            )
-            for i, results in enumerate(results_list)
-        ],
-        id=embedding_id,
-        object="list",
-        created=int(time.time()),
+    
+    # Handle both string and list inputs
+    inputs = body.input if isinstance(body.input, list) else [body.input]
+    
+    # Generate embeddings for each input
+    results_list = []
+    for text in inputs:
+        result = llm_client.emb(body.model, request=LLMRequest(instruction=text))
+        results_list.extend(result)
+
+    # Build response data
+    data = [
+        EmbeddingResponseData(
+            embedding=result.output,
+            index=i,
+            object="embedding"
+        )
+        for i, result in enumerate(results_list)
+    ]
+    
+    # Calculate token usage (simplified)
+    token_count = sum(len(str(input).split()) for input in inputs)
+    
+    return EmbeddingResponse(
+        data=data,
         model=body.model,
-        usage=EmbeddingsUsage(
-            prompt_tokens=tokens,
-            total_tokens=tokens,
+        object="list",
+        usage=UsageInfo(
+            prompt_tokens=token_count,
+            total_tokens=token_count
         ),
+        created=int(time.time()),
+        id=embedding_id
     )
 
 class ServerArgs(BaseModel):

{auto_coder-0.1.191 → auto_coder-0.1.192}/src/autocoder/rag/llm_wrapper.py

@@ -9,6 +9,7 @@ from byzerllm.utils.client import LLMResponse
 from byzerllm.utils.types import SingleOutputMeta
 from autocoder.rag.simple_rag import SimpleRAG
 from loguru import logger
+from byzerllm.utils.langutil import asyncfy_with_semaphore
 
 class LLWrapper:
 
@@ -31,7 +32,7 @@ class LLWrapper:
                  model:Optional[str] = None,
                  role_mapping=None,llm_config:Dict[str,Any]={}
                  )->Union[List[LLMResponse],List[LLMFunctionCallResponse],List[LLMClassResponse]]: 
-        res,contexts = self.rag.stream_chat_oai(conversations)
+        res,contexts = self.rag.stream_chat_oai(conversations,llm_config=llm_config)
         s = "".join(res)        
         return [LLMResponse(output=s,metadata={},input="")]
 
@@ -40,7 +41,7 @@ class LLWrapper:
                         role_mapping=None,
                         delta_mode=False,
                         llm_config:Dict[str,Any]={}): 
-        res,contexts = self.rag.stream_chat_oai(conversations)        
+        res,contexts = self.rag.stream_chat_oai(conversations,llm_config=llm_config)        
         for t in res:                        
             yield (t,SingleOutputMeta(0,0))
 
@@ -49,7 +50,7 @@ class LLWrapper:
                         role_mapping=None,
                         delta_mode=False,
                         llm_config:Dict[str,Any]={}): 
-        res,contexts = self.rag.stream_chat_oai(conversations)                
+        res,contexts = await asyncfy_with_semaphore(lambda: self.rag.stream_chat_oai(conversations,llm_config=llm_config))()                
         for t in res:                                    
             yield (t,SingleOutputMeta(0,0))
                          

{auto_coder-0.1.191 → auto_coder-0.1.192}/src/autocoder/rag/long_context_rag.py

@@ -30,6 +30,7 @@ from autocoder.rag.token_limiter import TokenLimiter
 from tokenizers import Tokenizer
 from autocoder.rag.variable_holder import VariableHolder
 from importlib.metadata import version
+from autocoder.rag.stream_event import event_writer
 
 try:        
     from autocoder_pro.rag.llm_compute import LLMComputeEngine
@@ -346,6 +347,29 @@ class LongContextRAG:
                     delta_mode=True,
                 )
                 return (chunk[0] for chunk in chunks), context
+            
+            try:
+                request_params = json.loads(query)
+                if "request_id" in request_params:                    
+                    request_id = request_params["request_id"]
+                    index = request_params["index"]
+                    
+                    file_path = event_writer.get_event_file_path(request_id)                    
+                    logger.info(f"Get events for request_id: {request_id} index: {index} file_path: {file_path}")
+                    events = []
+                    if not os.path.exists(file_path):
+                        return [],context
+
+                    with open(file_path, "r") as f:
+                        for line in f:
+                            event = json.loads(line)
+                            if event["index"] >= index:
+                                events.append(event)
+                    return [json.dumps({
+                        "events": [event for event in events],                        
+                    },ensure_ascii=False)], context                
+            except json.JSONDecodeError:
+                pass
 
             if self.args.without_contexts and LLMComputeEngine is not None:
                 llm_compute_engine = LLMComputeEngine(

auto_coder-0.1.192/src/autocoder/rag/stream_event/event_writer.py

@@ -0,0 +1,12 @@
+import os
+from autocoder.rag.stream_event.types import Event
+
+def write_event(event: Event,base_path: str="events"):
+    os.makedirs(base_path, exist_ok=True)
+    with open(f"{base_path}/{event.request_id}.jsonl", "a") as f:
+        f.write(event.model_dump_json() + "\n")
+
+def get_event_file_path(request_id: str,base_path: str="events") -> str:    
+    return f"{base_path}/{request_id}.jsonl"
+        
+    

auto_coder-0.1.192/src/autocoder/rag/stream_event/types.py

@@ -0,0 +1,16 @@
+from typing import List, Dict, Any
+from enum import Enum
+from pydantic import BaseModel
+
+class EventType(Enum):
+    START = "start"
+    THOUGHT = "thought"
+    CHUNK = "chunk"
+    DONE = "done"
+    ERROR = "error"
+
+class Event(BaseModel):
+    request_id: str
+    event_type: EventType
+    content: str  
+    index: int

auto_coder-0.1.192/src/autocoder/version.py

@@ -0,0 +1 @@
+__version__ = "0.1.192"

auto_coder-0.1.191/src/autocoder/version.py

@@ -1 +0,0 @@
-__version__ = "0.1.191"