U
�,�i; � @ s� d Z ddlZddlmZmZ zddlmZ W n ek
rD dZY nX ddlm Z m
Z
m Z de e
dddfee e e ee e
eed �d
d �ZdS ) z+Content extraction tool using DeepSeek API.� N)�Any�Optional)�OpenAI� )�DEEPSEEK_API_BASE�DEEPSEEK_DEFAULT_MODEL�DEEPSEEK_DEFAULT_PROMPT�< T)�content�prompt�model�api_key�timeout� use_proxy�returnc
C s6 | s t d��d}| }t|�|krLt�dt|�� d|� d�� |d|� d }t|t|d�}d } d
|� d |� d �}
t�d
|� �� zp|jjj |d| d�d|
d�gddid�} | j
d j j } | j
j| j
j| j
jd�}
t�d|
d � �� | ||
d�W S tk
�r0 } z t�d� tdt|�� ���W 5 d}~X Y nX dS )a� Extract structured information from content using DeepSeek API.
This function extracts structured information from provided content using the
DeepSeek API. It supports custom extraction prompts and model selection.
Note: This tool does NOT fetch URLs. Use fetch_page tool first to get content,
then pass that content to this tool for extraction.
Args:
content: The content to extract information from (e.g., from fetch_page).
prompt: Custom extraction instruction.
model: DeepSeek model name.
api_key: DeepSeek API key.
timeout: Request timeout in seconds.
use_proxy: Whether to use proxy (for API call).
Returns:
Dictionary containing:
- extracted: The extracted structured information (JSON object)
- model: The model used for extraction
- usage: Token usage information
Raises:
ValueError: If content is not provided.
Exception: On API errors.
z'content' must be providedi�� zContent truncated from z to z charactersNz)
[Content truncated due to length limits])r
Zbase_urlr a You are a content extraction assistant. Extract structured information from the provided content according to the user's instruction. Return the result as a valid JSON object. If the content is not in the expected language, translate and extract accordingly.zExtraction instruction: z
Content to extract from:
---
zg
---
Please extract the information according to the instruction above and return it as a JSON object.z!Calling DeepSeek API with model: �system)Zroler
�user�typeZ json_object)r �messagesZresponse_formatr )�
prompt_tokens�completion_tokens� total_tokensz#Extraction completed. Tokens used: r )� extractedr �usagezDeepSeek API call failedzExtraction failed: )�
ValueError�len�logging�warningr r �info�chat� completions�create�choices�messager
r r r r � Exception� exception�str)r
r r r
r r Zmax_content_lengthZ page_content�clientZ
system_promptZ user_prompt�responseZextracted_contentr �e� r* �$src\mcp_server_ddgs\tools\extract.py�extract_content sN # ��� ��
��
r, )�__doc__r �typingr r �openair � ImportError� constantsr r r r& �int�boolr, r* r* r* r+ �<module> s,
��
