auto-coder 0.1.309__py3-none-any.whl → 0.1.310__py3-none-any.whl

{auto_coder-0.1.309.dist-info → auto_coder-0.1.310.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: auto-coder
-Version: 0.1.309
+Version: 0.1.310
 Summary: AutoCoder: AutoCoder
 Author: allwefantasy
 Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence

{auto_coder-0.1.309.dist-info → auto_coder-0.1.310.dist-info}/RECORD

@@ -14,7 +14,7 @@ autocoder/command_parser.py,sha256=fx1g9E6GaM273lGTcJqaFQ-hoksS_Ik2glBMnVltPCE,1
 autocoder/lang.py,sha256=U6AjVV8Rs1uLyjFCZ8sT6WWuNUxMBqkXXIOs4S120uk,14511
 autocoder/models.py,sha256=AyoZ-Pzy0oyYUmWCxOIRiOImsqboSfRET7LO9-UOuxI,11172
 autocoder/run_context.py,sha256=IUfSO6_gp2Wt1blFWAmOpN0b0nDrTTk4LmtCYUBIoro,1643
-autocoder/version.py,sha256=pYQUD_mrZAZNxc9Lw3mOsln_cgf30rD3GGdAewKlpVU,23
+autocoder/version.py,sha256=Yva2ub3_rI3hSMIe4yqnO-D1-Taf21vJw07BOhnUd5E,23
 autocoder/agent/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 autocoder/agent/auto_demand_organizer.py,sha256=URAq0gSEiHeV_W4zwhOI_83kHz0Ryfj1gcfh5jwCv_w,6501
 autocoder/agent/auto_filegroup.py,sha256=pBsAkBcpFTff-9L5OwI8xhf2xPKpl-aZwz-skF2B6dc,6296
@@ -74,6 +74,7 @@ autocoder/common/mcp_server.py,sha256=1SCtpBRmN299xWX-0aV0imWS2CX6zBUOZBocbV_J6B
 autocoder/common/mcp_tools.py,sha256=YdEhDzRnwAr2J3D-23ExIQFWbrNO-EUpIxg179qs9Sw,12666
 autocoder/common/memory_manager.py,sha256=Xx6Yv0ULxVfcFfmD36hdHFFhxCgRAs-5fTd0fLHJrpQ,3773
 autocoder/common/model_speed_test.py,sha256=U48xUUpOnbwUal1cdij4YAn_H2PD2pNaqrMHaYtQRfI,15200
+autocoder/common/openai_content.py,sha256=M_V_UyHrqNVWjgrYvxfAupZw2I0Nr3iilYv6SxSvfLA,8091
 autocoder/common/printer.py,sha256=P1WU0QjlfnjqTP5uA55GkHZCpFzRPFkc34DMMandreg,2023
 autocoder/common/recall_validation.py,sha256=Avt9Q9dX3kG6Pf2zsdlOHmsjd-OeSj7U1PFBDp_Cve0,1700
 autocoder/common/result_manager.py,sha256=nBcFRj5reBC7vp13M91f4B8iPW8B8OehayHlUdeAt1g,3776
@@ -137,7 +138,7 @@ autocoder/rag/doc_filter.py,sha256=UduVO2mlrngwJICrefjDJTYfdmQ4GcRXrfWDQ7xXksk,1
 autocoder/rag/document_retriever.py,sha256=5BDqKVJqLPScEnua5S5suXhWuCaALIfPf5obXeJoWfs,8461
 autocoder/rag/lang.py,sha256=_jmUtxZDG1fmF4b2mhMJbYS1YQDb2ZE8nyAn5_vrvjA,3350
 autocoder/rag/llm_wrapper.py,sha256=Ht5GF5yJtrztoliujsZzx_ooWZmHkd5xLZKcGEiicZw,4303
-autocoder/rag/long_context_rag.py,sha256=6rqq0pvYe9N4TvyLwd2OB21ZUrPC4FfxZuks0weAz4A,41935
+autocoder/rag/long_context_rag.py,sha256=RE4xse3XxSC_HQA5erqrx6MhanP_29mBRdYOTJQZYGc,42106
 autocoder/rag/qa_conversation_strategy.py,sha256=1AcHV0MU00yTls20LlCPO-Un_OhSrr_p-H5lxLleAq4,6060
 autocoder/rag/rag_config.py,sha256=8LwFcTd8OJWWwi1_WY4IzjqgtT6RyE2j4PjxS5cCTDE,802
 autocoder/rag/rag_entry.py,sha256=6TKtErZ0Us9XSV6HgRKXA6yR3SiZGPHpynOKSaR1wgE,2463
@@ -198,9 +199,9 @@ autocoder/utils/types.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 autocoder/utils/auto_coder_utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 autocoder/utils/auto_coder_utils/chat_stream_out.py,sha256=xuBeWD0YOckqRo8JB1WkVIMOYH6c24m7JfV4svBfPDo,15113
 autocoder/utils/chat_auto_coder_utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-auto_coder-0.1.309.dist-info/LICENSE,sha256=HrhfyXIkWY2tGFK11kg7vPCqhgh5DcxleloqdhrpyMY,11558
-auto_coder-0.1.309.dist-info/METADATA,sha256=PK_1Bf18b0XqyZNBitO_y_KFkBlR2sdlwDwYuJ2HjJM,2747
-auto_coder-0.1.309.dist-info/WHEEL,sha256=GV9aMThwP_4oNCtvEC2ec3qUYutgWeAzklro_0m4WJQ,91
-auto_coder-0.1.309.dist-info/entry_points.txt,sha256=0nzHtHH4pNcM7xq4EBA2toS28Qelrvcbrr59GqD_0Ak,350
-auto_coder-0.1.309.dist-info/top_level.txt,sha256=Jqc0_uJSw2GwoFQAa9iJxYns-2mWla-9ok_Y3Gcznjk,10
-auto_coder-0.1.309.dist-info/RECORD,,
+auto_coder-0.1.310.dist-info/LICENSE,sha256=HrhfyXIkWY2tGFK11kg7vPCqhgh5DcxleloqdhrpyMY,11558
+auto_coder-0.1.310.dist-info/METADATA,sha256=zh8Gtsl5ahulVrleWS6UchHXOiJfeG-8SFeikpSffSg,2747
+auto_coder-0.1.310.dist-info/WHEEL,sha256=GV9aMThwP_4oNCtvEC2ec3qUYutgWeAzklro_0m4WJQ,91
+auto_coder-0.1.310.dist-info/entry_points.txt,sha256=0nzHtHH4pNcM7xq4EBA2toS28Qelrvcbrr59GqD_0Ak,350
+auto_coder-0.1.310.dist-info/top_level.txt,sha256=Jqc0_uJSw2GwoFQAa9iJxYns-2mWla-9ok_Y3Gcznjk,10
+auto_coder-0.1.310.dist-info/RECORD,,

autocoder/common/openai_content.py

@@ -0,0 +1,256 @@
+from typing import Any, Dict, List, Optional, Union
+import base64
+import os
+from enum import Enum
+from pydantic import BaseModel, Field, validator
+
+
+class ContentType(str, Enum):
+    """Type of content in the OpenAI chat message."""
+    TEXT = "text"
+    IMAGE_URL = "image_url"
+
+
+class ImageUrl(BaseModel):
+    """Image URL structure in OpenAI chat messages."""
+    url: str = Field(..., description="URL of the image, can be http(s) or data URI")
+
+    @validator('url')
+    def validate_url(cls, v):
+        """Validate that URL is either an http(s) URL or a valid data URI."""
+        if v.startswith(('http://', 'https://')):
+            return v
+        elif v.startswith('data:image/'):
+            return v
+        else:
+            raise ValueError("Image URL must be http(s) or data URI format")
+
+
+class TextContent(BaseModel):
+    """Text content in OpenAI chat messages."""
+    type: str = ContentType.TEXT
+    text: str
+
+
+class ImageUrlContent(BaseModel):
+    """Image URL content in OpenAI chat messages."""
+    type: str = ContentType.IMAGE_URL
+    image_url: Union[str, ImageUrl]
+
+    @validator('image_url')
+    def validate_image_url(cls, v):
+        """Convert string to ImageUrl if necessary."""
+        if isinstance(v, str):
+            return ImageUrl(url=v)
+        return v
+
+
+ContentItem = Union[TextContent, ImageUrlContent]
+
+
+class OpenAIMessage(BaseModel):
+    """Model for an OpenAI chat message."""
+    role: str
+    content: Union[str, List[ContentItem]]
+    name: Optional[str] = None
+
+
+class OpenAIConversation(BaseModel):
+    """Model for a conversation with OpenAI."""
+    messages: List[OpenAIMessage]
+
+
+def is_structured_content(content: Any) -> bool:
+    """
+    Check if the content is structured (list of items with type field).
+    
+    Args:
+        content: The content to check
+        
+    Returns:
+        bool: True if the content is structured, False otherwise
+    """
+    if not isinstance(content, list):
+        return False
+    
+    if not content:
+        return False
+    
+    # Check if all items have a 'type' field
+    return all(isinstance(item, dict) and 'type' in item for item in content)
+
+
+def encode_image_to_base64(image_path: str) -> str:
+    """
+    Encode an image file to base64.
+    
+    Args:
+        image_path: Path to the image file
+        
+    Returns:
+        str: Base64-encoded image data
+    """
+    if not os.path.exists(image_path):
+        raise FileNotFoundError(f"Image file not found: {image_path}")
+    
+    with open(image_path, "rb") as image_file:
+        encoded_string = base64.b64encode(image_file.read()).decode('utf-8')
+    
+    # Determine content type based on file extension
+    file_ext = os.path.splitext(image_path)[1].lower()
+    content_type = {
+        '.jpg': 'image/jpeg',
+        '.jpeg': 'image/jpeg',
+        '.png': 'image/png',
+        '.gif': 'image/gif',
+        '.webp': 'image/webp',
+    }.get(file_ext, 'image/jpeg')
+    
+    return f"data:{content_type};base64,{encoded_string}"
+
+
+def create_text_content(text: str) -> TextContent:
+    """
+    Create a text content item.
+    
+    Args:
+        text: The text content
+        
+    Returns:
+        TextContent: A text content item
+    """
+    return TextContent(text=text)
+
+
+def create_image_content(image_path_or_url: str) -> ImageUrlContent:
+    """
+    Create an image content item from a file path or URL.
+    
+    Args:
+        image_path_or_url: Path to the image file or an image URL
+        
+    Returns:
+        ImageUrlContent: An image content item
+    """
+    # If it's a URL already, use it directly
+    if image_path_or_url.startswith(('http://', 'https://', 'data:')):
+        return ImageUrlContent(image_url=image_path_or_url)
+    
+    # Otherwise, treat it as a file path and encode it
+    return ImageUrlContent(image_url=encode_image_to_base64(image_path_or_url))
+
+
+def normalize_content(content: Any) -> Union[str, List[ContentItem]]:
+    """
+    Normalize content to either a string or a list of structured content items.
+    
+    Args:
+        content: The content to normalize
+        
+    Returns:
+        Union[str, List[ContentItem]]: Normalized content
+    """
+    if isinstance(content, str):
+        return content
+    
+    if is_structured_content(content):
+        normalized_items = []
+        for item in content:
+            if item['type'] == ContentType.TEXT:
+                normalized_items.append(create_text_content(item['text']))
+            elif item['type'] == ContentType.IMAGE_URL:
+                normalized_items.append(ImageUrlContent(image_url=item['image_url']))
+        return normalized_items
+    
+    # If it's neither a string nor structured content, convert to string
+    return str(content)
+
+
+def create_message(role: str, content: Union[str, List[ContentItem]], name: Optional[str] = None) -> OpenAIMessage:
+    """
+    Create an OpenAI chat message.
+    
+    Args:
+        role: Role of the message sender (system, user, assistant)
+        content: Content of the message (string or structured content)
+        name: Optional name of the sender
+        
+    Returns:
+        OpenAIMessage: An OpenAI chat message
+    """
+    return OpenAIMessage(
+        role=role,
+        content=normalize_content(content),
+        name=name
+    )
+
+
+def process_conversations(conversations: List[Dict[str, Any]]) -> List[Dict[str, Any]]:
+    """
+    处理会话列表，确保每个消息都符合标准格式，并只保留文本内容。
+    
+    Args:
+        conversations: 会话列表，可能包含各种格式的消息
+        
+    Returns:
+        List[Dict[str, Any]]: 标准化后的会话列表，每个消息都有 role 和 content 字段
+    
+    例子:
+        >>> conversations = [
+        ...     {"role": "user", "content": "Hello"},
+        ...     {"role": "assistant", "content": "Hi, how can I help?"},
+        ...     {"role": "user", "content": [
+        ...         {"type": "text", "text": "What's in this image?"},
+        ...         {"type": "image_url", "image_url": "data:image/jpeg;base64,/9j/4AAQ..."}
+        ...     ]}
+        ... ]
+        >>> processed = process_conversations(conversations)
+        >>> # 结果保持相同的结构，但确保格式一致性
+
+        输出格式要是这样的：
+        [
+            {"role": "user", "content": "Hello"},
+            {"role": "assistant", "content": "Hi, how can I help?"},
+            {"role": "user", "content": "What's in this image?"}         
+        ]
+
+        只保留 text 内容。如果有多个 text 内容，用换行符连接弄成一个。
+        
+    """
+    processed_conversations = []
+    
+    for message in conversations:
+        # 确保消息有 role 字段
+        if "role" not in message:
+            raise ValueError(f"Message missing 'role' field: {message}")
+        
+        role = message["role"]
+        
+        # 处理 content 字段，确保存在
+        if "content" not in message:
+            processed_content = ""  # 如果不存在，设置为空字符串
+        else:
+            content = message["content"]
+            
+            # 处理结构化内容
+            if isinstance(content, list) and is_structured_content(content):
+                # 提取所有文本内容并用换行符连接
+                text_contents = []
+                for item in content:
+                    if item.get('type') == ContentType.TEXT and 'text' in item:
+                        text_contents.append(item['text'])
+                processed_content = '\n'.join(text_contents)
+            else:
+                # 如果是字符串或其他类型，确保转换为字符串
+                processed_content = str(content) if content is not None else ""
+        
+        # 构建标准化的消息
+        processed_message = {"role": role, "content": processed_content}
+        
+        # 如果原消息有 name 字段，也加入
+        if "name" in message and message["name"]:
+            processed_message["name"] = message["name"]
+        
+        processed_conversations.append(processed_message)
+    
+    return processed_conversations 

autocoder/rag/long_context_rag.py

@@ -40,6 +40,7 @@ from autocoder.rag.lang import get_message_with_format_and_newline
 from autocoder.rag.qa_conversation_strategy import get_qa_strategy
 from autocoder.rag.searchable import SearchableResults
 from autocoder.rag.conversation_to_queries import extract_search_queries
+from autocoder.common import openai_content as OpenAIContentProcessor
 try:
     from autocoder_pro.rag.llm_compute import LLMComputeEngine
     pro_version = version("auto-coder-pro")
@@ -348,7 +349,7 @@ class LongContextRAG:
         role_mapping=None,
         llm_config: Dict[str, Any] = {},
         extra_request_params: Dict[str, Any] = {}
-    ):
+    ):        
         try:
             return self._stream_chat_oai(
                 conversations,
@@ -399,6 +400,7 @@ class LongContextRAG:
         llm_config: Dict[str, Any] = {},
         extra_request_params: Dict[str, Any] = {}
     ):
+        conversations = OpenAIContentProcessor.process_conversations(conversations)
         if self.client:
             model = model or self.args.model
             response = self.client.chat.completions.create(
@@ -415,6 +417,7 @@ class LongContextRAG:
             target_llm = self.llm.get_sub_client("qa_model")
 
         query = conversations[-1]["content"]
+        
         context = []
 
         if (

autocoder/version.py

@@ -1 +1 @@
-__version__ = "0.1.309"
+__version__ = "0.1.310"