google_scholar_web_search.cpython-310.pyc•4.41 kB
o
�#�g � @ s� d dl Z d dlmZ d dlZd dlmZ d dd�Zd dd�Zed kr�d
Zeedd �Z e
d � e D ]*Z e
d
e d � �� e
de d � �� e
de d � �� e
de d � �� e
d� q0d
Z ee dddd�Z
e
d� e
D ]*Z e
d
e d � �� e
de d � �� e
de d � �� e
de d � �� e
d� qke�d�Zee�Ze�e� e�e�Ze�e� ed d Ze�e�Ze�e� d d � ed D �Ze
e� dS dS ) � N)�
BeautifulSoup)� scholarly� c C s d| � dd�� �}ddi}tj||d�}|jdkr#td|j� �� g S t|jd �}g }d
}|jd d d
�D ]V}||kr= |S |jddd
�} | rJ| � � nd}
| rZ| �d�rZ| �d�d nd} |jd dd
�} | ri| � � nd}
|jd dd
�}|rx|� � nd}|
|
|| d�}|�
|� |d7 }q4|S )a'
Function to search Google Scholar using a simple keyword query.
Parameters:
query (str): The search query (e.g., paper title or author).
num_results (int): The number of results to retrieve.
Returns:
list: A list of dictionaries containing search results.
z%https://scholar.google.com/scholar?q=� �+�
User-Agent�sMozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36��headers�� �(Failed to fetch data. HTTP Status code: � html.parserr �div�gs_ri��class_�h3�gs_rt�No title available�a�href�No link available�gs_a�No authors available�gs_rs�No abstract available��Title�Authors�Abstract�URL� N) �replace�requests�get� status_code�printr �text�find_all�find�get_text�append)�query� num_results�
search_urlr
�response�soup�results�count�item� title_tag�title�link� authors_tag�authors� abstract_tag�abstract� result_data� r< �O/home/zjlab/code/mcp_hub/Google-Scholar-MCP-Server/google_scholar_web_search.py�google_scholar_search s: �
� �
r> c C s` d}d| � dd�i}|r||d< |r |\}}||d< ||d< |d�d d
� |�� D ��7 }d d i}tj||d
�} | jdkrFtd| j� �� g S t| jd�}
g } d} |
j ddd�D ]V}
| |kr` | S |
j
ddd�}|rm|� � nd}|r}|�
d�r}|�
d�d nd}|
j
ddd�}|r�|� � nd }|
j
dd d�}|r�|� � nd }||||d �}| � |� | d 7 } qW| S )"a
Function to search Google Scholar using advanced search filters (e.g., author, year range).
Parameters:
query (str): The search query (e.g., paper title or topic).
author (str): The author's name to filter the results (default is None).
year_range (tuple): A tuple (start_year, end_year) to filter the results by publication year (default is None).
num_results (int): The number of results to retrieve.
Returns:
list: A list of dictionaries containing search results.
z#https://scholar.google.com/scholar?�qr r Zas_authZas_yloZas_yhi�&c S s g | ] \}}|� d |� ��qS )�=r< )�.0�key�valuer< r< r= �
<listcomp>_ s z2advanced_google_scholar_search.<locals>.<listcomp>r r r r r r
r r r r r r r r r r r r r r r r! N)
r" �join�itemsr# r$ r% r& r r' r( r) r* r+ )r, �author�
year_ranger- r. Z
search_paramsZ
start_yearZend_yearr
r/ r0 r1 r2 r3 r4 r5 r6 r7 r8 r9 r: r; r< r< r= � advanced_google_scholar_searchE sJ �
� �
rJ �__main__zmachine learning)r- zResults for keyword search:z
Title: r z Authors: r z
Abstract: r zURL: r zP--------------------------------------------------------------------------------zIan Goodfellow)i� i� )rH rI r- z
Results for advanced search:zSteven A CholewiakZ publicationsc C s g | ]}|d d �qS )�bibr5 r< )rB Zpubr< r< r= rE � s rE )r )NNr )r# Zbs4r �timer r> rJ �__name__r, r1 r& �resultZadvanced_queryZadvanced_resultsZ
search_authorZ search_query�nextZfirst_author_result�pprint�fillrH Zfirst_publicationZfirst_publication_filledZpublication_titlesr< r< r< r= �<module> sH
>L
�