botrun-flow-lang 5.12.263__py3-none-any.whl → 6.2.21__py3-none-any.whl

botrun_flow_lang/langgraph_agents/agents/util/plotly_util.py

@@ -1,59 +1,59 @@
-import plotly.graph_objects as go
-import os
-from tempfile import NamedTemporaryFile
-from typing import Dict, Any, Optional
-from .local_files import upload_and_get_tmp_public_url, upload_html_and_get_public_url
-
-
-async def generate_plotly_files(
-    figure_data: Dict[str, Any],
-    botrun_flow_lang_url: str,
-    user_id: str,
-    title: Optional[str] = None,
-) -> str:
-    """
-    Generate plotly HTML file from figure data and upload it to GCS.
-
-    Args:
-        figure_data: Dictionary containing plotly figure data and layout
-        botrun_flow_lang_url: URL for the botrun flow lang API
-        user_id: User ID for file upload
-        title: Optional title for the plot
-
-    Returns:
-        str: URL for the HTML file or error message starting with "Error: "
-    """
-    try:
-        # Create plotly figure from data and layout
-        data = figure_data.get("data", [])
-        layout = figure_data.get("layout", {})
-
-        # If title is provided, add it to the layout before creating the figure
-        if title:
-            layout["title"] = {"text": title}
-
-        # Create figure with data and layout
-        fig = go.Figure(data=data, layout=layout)
-
-    except Exception as e:
-        return f"Error: {str(e)}"
-
-    # Create temporary file
-    with NamedTemporaryFile(suffix=".html", delete=False) as html_temp:
-        try:
-            # Save HTML with plotly.js included
-            fig.write_html(html_temp.name, include_plotlyjs=True, full_html=True)
-
-            # Upload file to GCS
-            html_url = await upload_html_and_get_public_url(
-                html_temp.name, botrun_flow_lang_url, user_id
-            )
-
-            # Clean up temporary file
-            os.unlink(html_temp.name)
-
-            return html_url
-        except Exception as e:
-            # Clean up temporary file in case of error
-            os.unlink(html_temp.name)
-            return f"Error: {str(e)}"
+import plotly.graph_objects as go
+import os
+from tempfile import NamedTemporaryFile
+from typing import Dict, Any, Optional
+from .local_files import upload_and_get_tmp_public_url, upload_html_and_get_public_url
+
+
+async def generate_plotly_files(
+    figure_data: Dict[str, Any],
+    botrun_flow_lang_url: str,
+    user_id: str,
+    title: Optional[str] = None,
+) -> str:
+    """
+    Generate plotly HTML file from figure data and upload it to GCS.
+
+    Args:
+        figure_data: Dictionary containing plotly figure data and layout
+        botrun_flow_lang_url: URL for the botrun flow lang API
+        user_id: User ID for file upload
+        title: Optional title for the plot
+
+    Returns:
+        str: URL for the HTML file or error message starting with "Error: "
+    """
+    try:
+        # Create plotly figure from data and layout
+        data = figure_data.get("data", [])
+        layout = figure_data.get("layout", {})
+
+        # If title is provided, add it to the layout before creating the figure
+        if title:
+            layout["title"] = {"text": title}
+
+        # Create figure with data and layout
+        fig = go.Figure(data=data, layout=layout)
+
+    except Exception as e:
+        return f"Error: {str(e)}"
+
+    # Create temporary file
+    with NamedTemporaryFile(suffix=".html", delete=False) as html_temp:
+        try:
+            # Save HTML with plotly.js included
+            fig.write_html(html_temp.name, include_plotlyjs=True, full_html=True)
+
+            # Upload file to GCS
+            html_url = await upload_html_and_get_public_url(
+                html_temp.name, botrun_flow_lang_url, user_id
+            )
+
+            # Clean up temporary file
+            os.unlink(html_temp.name)
+
+            return html_url
+        except Exception as e:
+            # Clean up temporary file in case of error
+            os.unlink(html_temp.name)
+            return f"Error: {str(e)}"

botrun_flow_lang/langgraph_agents/agents/util/tavily_search.py

@@ -1,199 +1,199 @@
-"""
-Tavily search utility for government research
-參考 open_deep_research 的最佳實踐
-"""
-
-import os
-import logging
-from typing import List, Dict, Any, AsyncGenerator
-from dataclasses import dataclass
-
-try:
-    from tavily import TavilyClient, AsyncTavilyClient
-except ImportError:
-    TavilyClient = None
-    AsyncTavilyClient = None
-
-logger = logging.getLogger(__name__)
-
-
-@dataclass
-class SearchEvent:
-    """搜尋事件結構，與 perplexity_search 保持一致"""
-
-    chunk: str
-    raw_json: Dict[str, Any] = None
-
-
-async def respond_with_tavily_search(
-    query: str,
-    user_prompt_prefix: str = "",
-    messages: List[Dict[str, str]] = None,
-    domain_filter: List[str] = None,
-    stream: bool = False,
-    model_name: str = "tavily",
-) -> AsyncGenerator[SearchEvent, None]:
-    """
-    使用 Tavily 進行搜尋（參考 open_deep_research 實作）
-
-    Args:
-        query: 搜尋查詢
-        user_prompt_prefix: 用戶提示前綴（保持與 perplexity 一致的介面）
-        messages: 訊息列表（保持介面一致，但 Tavily 只使用 query）
-        domain_filter: 領域過濾
-        stream: 是否串流（Tavily 不支援，但保持介面一致）
-        model_name: 模型名稱（保持介面一致）
-
-    Yields:
-        SearchEvent: 搜尋事件
-    """
-
-    if AsyncTavilyClient is None:
-        logger.error("Tavily client not available. Please install tavily-python")
-        yield SearchEvent(
-            chunk="錯誤：Tavily 客戶端未安裝",
-            raw_json={"error": "tavily-python not installed"},
-        )
-        return
-
-    api_key = os.getenv("TAVILY_API_KEY")
-    if not api_key:
-        logger.error("TAVILY_API_KEY not found in environment variables")
-        yield SearchEvent(
-            chunk="錯誤：未設定 TAVILY_API_KEY",
-            raw_json={"error": "TAVILY_API_KEY not set"},
-        )
-        return
-
-    try:
-        # 使用 AsyncTavilyClient 進行搜尋（現代寫法）
-        logger.info(f"使用 Tavily 搜尋: {query}")
-
-        # 初始化 async client
-        async_client = AsyncTavilyClient(api_key=api_key)
-
-        # 準備搜尋參數
-        search_params = {
-            "query": query,
-            "search_depth": "advanced",
-            # "include_raw_content": True,
-            # "include_answer": "advanced",
-            "max_results": 5,
-        }
-
-        # 如果有 domain filter，加入參數
-        if domain_filter:
-            search_params["include_domains"] = domain_filter
-
-        # 執行 async 搜尋
-        response = await async_client.search(**search_params)
-
-        # 處理搜尋結果（參考 open_deep_research 格式化）
-        if response and "results" in response and len(response["results"]) > 0:
-            content_parts = []
-            sources = []
-
-            # 去重複 URL（參考 open_deep_research）
-            seen_urls = set()
-            unique_results = []
-            for result in response["results"]:
-                url = result.get("url", "")
-                if url and url not in seen_urls:
-                    seen_urls.add(url)
-                    unique_results.append(result)
-
-            # 格式化結果
-            for i, result in enumerate(unique_results[:5]):  # 限制結果數量
-                title = result.get("title", "")
-                content = result.get("content", "")
-                raw_content = result.get("raw_content", "")
-                url = result.get("url", "")
-
-                # 優先使用 raw_content，回退到 content
-                # display_content = raw_content if content else raw_content
-
-                if title and content:
-                    # 限制每個結果的長度（參考 open_deep_research）
-                    # if len(display_content) > 1000:
-                    #     display_content = display_content[:1000] + "..."
-
-                    content_parts.append(
-                        f"[{i+1}] Title: {title}\nContent: {content}\nURL: {url}"
-                    )
-                    sources.append(url)
-
-            # 組合最終內容
-            full_content = "\n\n".join(content_parts)
-
-            # 限制總長度（參考 open_deep_research 的 30000 字元限制）
-            # if len(full_content) > 10000:  # 適合政府研究的長度
-            #     full_content = full_content[:10000] + "\n\n[內容已截斷以符合長度限制]"
-
-            # 構建回應 JSON（與 Perplexity 格式保持一致）
-            response_json = {
-                "message": {"content": full_content},
-                "usage": {
-                    "prompt_tokens": len(query.split()),
-                    "completion_tokens": len(full_content.split()),
-                    "total_tokens": len(query.split()) + len(full_content.split()),
-                },
-                "model": "tavily-search",
-                "sources": sources,
-                "results_count": len(unique_results),
-            }
-
-            yield SearchEvent(chunk=full_content, raw_json=response_json)
-
-        else:
-            logger.warning("Tavily 搜尋回應中沒有結果")
-            yield SearchEvent(
-                chunk="未找到搜尋結果", raw_json={"error": "no results found"}
-            )
-
-    except Exception as e:
-        logger.error(f"Tavily 搜尋失敗: {e}")
-        yield SearchEvent(chunk=f"搜尋失敗: {str(e)}", raw_json={"error": str(e)})
-
-
-# 為了保持一致性，提供同步版本
-def search_with_tavily(
-    query: str, domain_filter: List[str] = None, max_results: int = 5
-) -> Dict[str, Any]:
-    """
-    同步版本的 Tavily 搜尋（參考 open_deep_research）
-
-    Args:
-        query: 搜尋查詢
-        domain_filter: 領域過濾
-        max_results: 最大結果數量
-
-    Returns:
-        搜尋結果字典
-    """
-
-    if TavilyClient is None:
-        return {"error": "tavily-python not installed"}
-
-    api_key = os.getenv("TAVILY_API_KEY")
-    if not api_key:
-        return {"error": "TAVILY_API_KEY not set"}
-
-    try:
-        client = TavilyClient(api_key=api_key)
-
-        search_params = {
-            "query": query,
-            "search_depth": "advanced",
-            "include_raw_content": True,  # 參考 open_deep_research
-            "include_domains": domain_filter if domain_filter else None,
-            "max_results": max_results,
-        }
-
-        search_params = {k: v for k, v in search_params.items() if v is not None}
-
-        response = client.search(**search_params)
-        return response
-
-    except Exception as e:
-        logger.error(f"Tavily 搜尋失敗: {e}")
-        return {"error": str(e)}
+"""
+Tavily search utility for government research
+參考 open_deep_research 的最佳實踐
+"""
+
+import os
+import logging
+from typing import List, Dict, Any, AsyncGenerator
+from dataclasses import dataclass
+
+try:
+    from tavily import TavilyClient, AsyncTavilyClient
+except ImportError:
+    TavilyClient = None
+    AsyncTavilyClient = None
+
+logger = logging.getLogger(__name__)
+
+
+@dataclass
+class SearchEvent:
+    """搜尋事件結構，與 perplexity_search 保持一致"""
+
+    chunk: str
+    raw_json: Dict[str, Any] = None
+
+
+async def respond_with_tavily_search(
+    query: str,
+    user_prompt_prefix: str = "",
+    messages: List[Dict[str, str]] = None,
+    domain_filter: List[str] = None,
+    stream: bool = False,
+    model_name: str = "tavily",
+) -> AsyncGenerator[SearchEvent, None]:
+    """
+    使用 Tavily 進行搜尋（參考 open_deep_research 實作）
+
+    Args:
+        query: 搜尋查詢
+        user_prompt_prefix: 用戶提示前綴（保持與 perplexity 一致的介面）
+        messages: 訊息列表（保持介面一致，但 Tavily 只使用 query）
+        domain_filter: 領域過濾
+        stream: 是否串流（Tavily 不支援，但保持介面一致）
+        model_name: 模型名稱（保持介面一致）
+
+    Yields:
+        SearchEvent: 搜尋事件
+    """
+
+    if AsyncTavilyClient is None:
+        logger.error("Tavily client not available. Please install tavily-python")
+        yield SearchEvent(
+            chunk="錯誤：Tavily 客戶端未安裝",
+            raw_json={"error": "tavily-python not installed"},
+        )
+        return
+
+    api_key = os.getenv("TAVILY_API_KEY")
+    if not api_key:
+        logger.error("TAVILY_API_KEY not found in environment variables")
+        yield SearchEvent(
+            chunk="錯誤：未設定 TAVILY_API_KEY",
+            raw_json={"error": "TAVILY_API_KEY not set"},
+        )
+        return
+
+    try:
+        # 使用 AsyncTavilyClient 進行搜尋（現代寫法）
+        logger.info(f"使用 Tavily 搜尋: {query}")
+
+        # 初始化 async client
+        async_client = AsyncTavilyClient(api_key=api_key)
+
+        # 準備搜尋參數
+        search_params = {
+            "query": query,
+            "search_depth": "advanced",
+            # "include_raw_content": True,
+            # "include_answer": "advanced",
+            "max_results": 5,
+        }
+
+        # 如果有 domain filter，加入參數
+        if domain_filter:
+            search_params["include_domains"] = domain_filter
+
+        # 執行 async 搜尋
+        response = await async_client.search(**search_params)
+
+        # 處理搜尋結果（參考 open_deep_research 格式化）
+        if response and "results" in response and len(response["results"]) > 0:
+            content_parts = []
+            sources = []
+
+            # 去重複 URL（參考 open_deep_research）
+            seen_urls = set()
+            unique_results = []
+            for result in response["results"]:
+                url = result.get("url", "")
+                if url and url not in seen_urls:
+                    seen_urls.add(url)
+                    unique_results.append(result)
+
+            # 格式化結果
+            for i, result in enumerate(unique_results[:5]):  # 限制結果數量
+                title = result.get("title", "")
+                content = result.get("content", "")
+                raw_content = result.get("raw_content", "")
+                url = result.get("url", "")
+
+                # 優先使用 raw_content，回退到 content
+                # display_content = raw_content if content else raw_content
+
+                if title and content:
+                    # 限制每個結果的長度（參考 open_deep_research）
+                    # if len(display_content) > 1000:
+                    #     display_content = display_content[:1000] + "..."
+
+                    content_parts.append(
+                        f"[{i+1}] Title: {title}\nContent: {content}\nURL: {url}"
+                    )
+                    sources.append(url)
+
+            # 組合最終內容
+            full_content = "\n\n".join(content_parts)
+
+            # 限制總長度（參考 open_deep_research 的 30000 字元限制）
+            # if len(full_content) > 10000:  # 適合政府研究的長度
+            #     full_content = full_content[:10000] + "\n\n[內容已截斷以符合長度限制]"
+
+            # 構建回應 JSON（與 Perplexity 格式保持一致）
+            response_json = {
+                "message": {"content": full_content},
+                "usage": {
+                    "prompt_tokens": len(query.split()),
+                    "completion_tokens": len(full_content.split()),
+                    "total_tokens": len(query.split()) + len(full_content.split()),
+                },
+                "model": "tavily-search",
+                "sources": sources,
+                "results_count": len(unique_results),
+            }
+
+            yield SearchEvent(chunk=full_content, raw_json=response_json)
+
+        else:
+            logger.warning("Tavily 搜尋回應中沒有結果")
+            yield SearchEvent(
+                chunk="未找到搜尋結果", raw_json={"error": "no results found"}
+            )
+
+    except Exception as e:
+        logger.error(f"Tavily 搜尋失敗: {e}")
+        yield SearchEvent(chunk=f"搜尋失敗: {str(e)}", raw_json={"error": str(e)})
+
+
+# 為了保持一致性，提供同步版本
+def search_with_tavily(
+    query: str, domain_filter: List[str] = None, max_results: int = 5
+) -> Dict[str, Any]:
+    """
+    同步版本的 Tavily 搜尋（參考 open_deep_research）
+
+    Args:
+        query: 搜尋查詢
+        domain_filter: 領域過濾
+        max_results: 最大結果數量
+
+    Returns:
+        搜尋結果字典
+    """
+
+    if TavilyClient is None:
+        return {"error": "tavily-python not installed"}
+
+    api_key = os.getenv("TAVILY_API_KEY")
+    if not api_key:
+        return {"error": "TAVILY_API_KEY not set"}
+
+    try:
+        client = TavilyClient(api_key=api_key)
+
+        search_params = {
+            "query": query,
+            "search_depth": "advanced",
+            "include_raw_content": True,  # 參考 open_deep_research
+            "include_domains": domain_filter if domain_filter else None,
+            "max_results": max_results,
+        }
+
+        search_params = {k: v for k, v in search_params.items() if v is not None}
+
+        response = client.search(**search_params)
+        return response
+
+    except Exception as e:
+        logger.error(f"Tavily 搜尋失敗: {e}")
+        return {"error": str(e)}

botrun_flow_lang/langgraph_agents/agents/util/usage_metadata.py

@@ -0,0 +1,34 @@
+"""
+Usage Metadata 模組
+
+提供 LLM 呼叫的 token 使用量追蹤功能。
+"""
+
+from dataclasses import dataclass, asdict
+from typing import Dict, Any
+
+
+@dataclass
+class UsageMetadata:
+    """Token usage metadata that matches the expected parsing format."""
+    prompt_tokens: int = 0
+    completion_tokens: int = 0
+    total_tokens: int = 0
+    cache_creation_input_tokens: int = 0
+    cache_read_input_tokens: int = 0
+    model: str = ""
+
+    def __add__(self, other: "UsageMetadata") -> "UsageMetadata":
+        """Combine two UsageMetadata objects."""
+        return UsageMetadata(
+            prompt_tokens=self.prompt_tokens + other.prompt_tokens,
+            completion_tokens=self.completion_tokens + other.completion_tokens,
+            total_tokens=self.total_tokens + other.total_tokens,
+            cache_creation_input_tokens=self.cache_creation_input_tokens + other.cache_creation_input_tokens,
+            cache_read_input_tokens=self.cache_read_input_tokens + other.cache_read_input_tokens,
+            model=self.model or other.model,
+        )
+
+    def to_dict(self) -> Dict[str, Any]:
+        """Convert to dictionary."""
+        return asdict(self)