PubTator_search.cpython-310.pyc•12.3 kB
o
ȿ�g�@ � @ s� d dl Z d dlZd dlZd dlZd dlmZmZmZm Z m
Z
G dd� d�Z e dk�rid dl
Z
d dlZd dlmZ ejejdd� e�e �Ze
jd d
�Zejd g d �d
dd� e�� Z�zz�e ddd�Zdd� Zdd� Zdd� Z dd� Z eee e d�Z ej d
kr�e � � D ]l\Z!Z"ze�#d e!� �� e"� Z$e�#d e!� d �� W q� e%y� Z& ze�'d e!� d e(e&�� d �� W Y dZ&[&q�dZ&[&ww ze ej � Z$e�#d ej � �� W n e%y� Z& ze�'d ej � d e(e&�� �� W Y dZ&[&ndZ&[&ww W nQ e j)j*�y Z& ze�'d!e&� �� W Y dZ&[&nAdZ&[&w e+�y( Z& ze�'d"e&� �� W Y dZ&[&n/dZ&[&w e%�yB Z& ze�'d#e&� �� W Y dZ&[&n dZ&[&ww W ej,�-� dS W ej,�-� dS W ej,�-� dS W ej,�-� dS ej,�-� w dS )$� N)�List�Dict�Optional�Union� Generatorc @ s� e Zd ZdZdZd2dedefdd�Zd d
� Z
d3dee de de de
de e e f f
dd�Z
d4de dee dee de fdd�Z d5de d ee d ee d ee de f
d d �Z ! "d6de d#ed$ee d%edee ddf f
d&d'�Z ( d7d)e d e d*ee d$ee dee ddf f
d+d,�Zd-e dee fd.d/�Z
"d8de de d$ee de
d%edee e e f ddf f d0d1�ZdS )9� PubTator3APIz3https://www.ncbi.nlm.nih.gov/research/pubtator3-apiz8https://www.ncbi.nlm.nih.gov/CBBresearch/Lu/Demo/RESTful� � � max_retries�timeoutc C s: t �� | _| jj�ddd�� d| _|| _|| _d| _d S )Nz PubTator3API Python Client/2.1zapplication/json)z
User-Agent�Acceptg��(\���?r ) �requestsZSession�session�headers�update�
request_delayr
r �_last_request_time)�selfr
r � r �9D:\code\github\mcp\PubTator-MCP-Server\pubtator_search.py�__init__ s
�
zPubTator3API.__init__c O s� t � � }|| j }|| jk rt �| j| � | j|d< t| j�D ]?}z| jj|g|�R i |��}t � � | _|W S t j
j y_ } z|| jd krK� t �t d| d�� W Y d}~q d}~ww dS )z5Request method with rate limiting and retry mechanismr � � �
N)
�timer r �sleepr �ranger
r �requestr
�
exceptions�RequestException�min) r �method�args�kwargs� current_time�elapsed�attempt�response�er r r �_rate_limited_request s"
���z"PubTator3API._rate_limited_request�pmid�biocjsonF�ids�id_type�format� full_text�returnc
C s� |st d��|dvrt d��|dvrt d��zN|dkr$| j� d|� �}n| j� d|� �}|� d �d
�|�i}|r@|d kr@d |d
< | jd||d�}|�� |dkra|�� }t|t�r^d|iW S |W S |jW S t j
j y{ } z t dt
| �� ���d} ~ ww )u�
导出一组文献的标注结果
参数:
ids: 文献ID列表(pmids或pmcids)
id_type: ID类型，可以是"pmid"或"pmcid"
format: 返回格式("pubtator", "biocxml"或"biocjson")
full_text: 是否获取全文(仅适用于biocxml/biocjson格式)
返回:
标注结果(JSON字典或XML字符串)
zIDs list cannot be empty)r* �pmcidz!id_type must be 'pmid' or 'pmcid')�pubtatorZbiocxmlr+ z2format must be one of: pubtator, biocxml, biocjsonr1 z/publications/pmc_export/z/publications/export/�s�,r2 �true�full�GET��paramsr+ Z documentsz Failed to export publications: N)�
ValueError�BASE_URL�joinr) �raise_for_status�json�
isinstance�list�textr
r r � Exception�str)
r r, r- r. r/ �urlr9 r' �resultr( r r r �export_publications* s2
��z PubTator3API.export_publicationsN�query�concept�limitc C s\ | j � d�}d|i}|r|dvrtd��||d< |r ||d< | jd||d�}|�� |�� S )
u}
通过自由文本查询查找生物概念的标识符
参数:
query: 查询文本
concept: 可选，指定生物概念类型
(如"gene", "disease", "chemical", "species", "mutation")
limit: 可选，限制返回结果数量
返回:
包含实体ID的JSON字典
z/entity/autocomplete/rG )�gene�disease�chemicalZspeciesZmutationzInvalid concept typerH rI r7 r8 N)r; r: r) r= r> )r rG rH rI rD r9 r' r r r �find_entity_id] s zPubTator3API.find_entity_id� entity_id�
relation_type�target_entity_type� max_resultsc C s� | j � d�}|�d�std��d|i}g d�}|r)||vr%|dkr%td��||d< |r7|d vr3td
��||d < |r=||d < | jd
||d�}|�� |�� S )u
查找相关实体
参数:
entity_id: 实体ID(通过find_entity_id获取)
relation_type: 可选，指定关系类型
(如"treat", "cause", "interact", "associate")
target_entity_type: 可选，指定目标实体类型
(如"gene", "disease", "chemical")
max_results: 可选，限制返回结果的最大数量
返回:
相关实体结果的JSON字典
z
/relations�@zeInvalid entity ID format, should start with '@', e.g., '@CHEMICAL_remdesivir' or '@DISEASE_Neoplasms'�e1)
�treat�causeZcotreat�convert�compareZinteractZ associateZpositive_correlateZnegative_correlateZpreventZinhibitZ stimulateZ
drug_interact�ANYzInvalid relation type�type)rJ rK rL �variantzInvalid target entity type�e2rI r7 r8 N)r; �
startswithr: r) r= r> ) r rN rO rP rQ rD r9 Zvalid_relationsr' r r r �find_related_entities} s$
z"PubTator3API.find_related_entitiesr �d �page� max_pages�
batch_sizec
c s6 � |}d}d}|du s||kr�z.| j � d�}||d�} | jd|| d�}
|
�� |
�� } | �d�s3W dS d}| V |d 7 }W nQ tjjyk } z |d 7 }||krWtd
|� d ���t �
t d | d
�� W Y d} ~ qd} ~ w tj y� |d 7 }||kr�td
|� d���t �
t d | d
�� Y qw |du s||ksdS dS )uy 增强的搜索功能，支持自动分页和错误重试
参数:
query: 查询内容(自由文本/实体ID/关系查询)
page: 起始页码
max_pages: 最大获取页数(None表示无限制)
batch_size: 每批处理的PMID数量
返回:
生成器，逐页产生搜索结果
r r Nz/search/)rA r_ r7 r8 �resultsr zSearch terminated after z consecutive request failuresr r z& consecutive response parsing failures)
r; r) r= r> �getr
r r rB r r r �JSONDecodeError)
r rG r_ r` ra Z current_pageZconsecutive_errorsZmax_consecutive_errorsrD r9 r' �datar( r r r �search� s> �
���zPubTator3API.searchrX �entity1�entity2c c sJ � |du rd|� d|� �}n d|� d|� d|� �}| j ||d�E dH dS )ua
专门的关系查询方法
参数:
entity1: 第一个实体ID
relation_type: 关系类型(ANY/treat/cause等)
entity2: 第二个实体ID或类型(如"DISEASE")
max_pages: 最大获取页数
返回:
生成器，逐页产生关系搜索结果
Nz
relations:�|�r` )rf )r rg rO rh r` rG r r r �search_relations� s
�z PubTator3API.search_relationsrb c C s dd� |� dg �D �S )u�
从搜索结果中提取PMID列表
参数:
results: 单页搜索结果
返回:
PMID列表
c S s g | ] }d |v rt |d ��qS )r* �rC )�.0rE r r r �
<listcomp> s z;PubTator3API.extract_pmids_from_results.<locals>.<listcomp>rb N)rc )r rb r r r �extract_pmids_from_results� s
z'PubTator3API.extract_pmids_from_resultsc c s| � z�g }| j ||d�D ]d}| �|�}|�|� t|�|kro|d|� } zdd� | D �} | �| d||�}
|
V ||d� }W n+ tjjyh } z tdt| �� dt | �� �� |dkrc|d }W Y d} ~ q� d} ~ ww t|�|ks q |r�zd
d� |D �}| �|d||�}
|
V W W dS tjjy� } ztd t|�� dt | �� �� � d} ~ ww W dS t
y� } z t
d t | �� ���d} ~ ww )
uy 搜索并批量导出文献，支持分批处理和错误重试
参数:
query: 搜索查询
format: 导出格式
max_pages: 最大搜索页数
full_text: 是否导出全文
batch_size: 每批处理的PMID数量
返回:
生成器，产生导出的文献内容
rj Nc S � g | ]}t |��qS r rl �rm r* r r r rn ( � z9PubTator3API.batch_export_from_search.<locals>.<listcomp>r* u 批量导出失败(批次大小:z): r r c S rp r rl rq r r r rn > rr u 处理剩余PMID失败(数量:u# 批量导出过程中发生错误: ) rf ro �extend�lenrF r
r r �printrC rB ) r rG r. r` r/ ra Z all_pmids� page_resultZ
page_pmids�batchrE r( r r r �batch_export_from_search s` �
� �� ��� �����z%PubTator3API.batch_export_from_search)r r )r* r+ F)NN)NNN)r Nr^ )rX NN)r+ r Fr^ )�__name__�
__module__� __qualname__r; Z ANNOTATE_URL�intr r) r rC �boolr r rF r rM r] r rf rk ro rx r r r r r s� �����
�6����
�#�����
�5�����
�7�����
��������r �__main__)�datetimez)%(asctime)s - %(levelname)s - %(message)s)�levelr. zPubTator3API Test Program)� descriptionz--test)�all�export�entity�relationZannotaterf r� zSelect test case to run)�choices�default�helpr r )r
r c C s: t �d� ddg} tj| dd�}t �dt| �� d�� |S )Nz,Testing export publications functionality...Z25359968Z25359969r+ )r. zSuccessfully exported z
publications)�logger�info�apirF rt )ZpmidsrE r r r �test_export_publicationsb s
r� c C �2 t �d� d} tj| ddd�}t �d| � �� |S )Nz'Testing find entity ID functionality...ZdiabetesrK � )rH rI zSuccessfully found entity: )r� r� r� rM )rG rE r r r �test_find_entity_idi s
r� c C r� )Nz.Testing find related entities functionality...z@DISEASE_Diabetes_MellitusrT rL )rO rP z'Successfully found entities related to )r� r� r� r] )rN rE r r r �test_find_related_entitiesp s
�r� c C sD t �d� d} g }tj| dd�D ]}|�|� qt �d| � �� |S )Nz Testing search functionality...zdiabetes treatmentr rj zSuccessfully searched: )r� r� r� rf �append)rG rb rv r r r � test_search} s
r� )r� r� r� rf z
Executing test: zTest successful: �
z Test failed z: zNetwork request error: zParameter error: zUnknown error: ).r
r> r � urllib.parse�urllib�typingr r r r r r ry �argparse�loggingr � basicConfig�INFO� getLoggerr� �ArgumentParser�parser� add_argument�
parse_argsr" r� r� r� r� r� Z
test_funcs�test�items�name�funcr� rE rB r( �errorrC r r r: r �closer r r r �<module> s�
H �
�
�
(�� (��� � � �������