在线问诊 Python、FastAPI、Neo4j — 创建 节点关系

• 关系：症状-检查
• 关系：疾病-症状
• 代码重构

relationship_data.csv

症状,检查,疾病,药品,宜吃,忌吃
"上下楼梯疼,不能久站,感觉有点肿","膝关节核磁","右膝髌上囊及关节腔少量积液","扶他林","西红柿,香蕉","辣椒,大蒜"
"眼睛胀痛,干涩,畏光,眼胀,眼痛,看东西有时候清楚有时候不清楚","视力,眼底","干眼","施图伦","胡萝卜,核桃仁,菠菜","海鲜,芥末"

关系：症状-检查

def generate_examine() -> list:
    """
    关系：疾病-检查
    """
    rels_check = []
    df = pd.read_csv('relationship_data.csv')
    for idx, row in df.iterrows():
        for symptom in row['检查'].split(','):
            for exam in row['症状'].split(','):
                rels_check.append([exam, symptom])
    rels_check = deduplicate(rels_check)
    return rels_check

def relationship_rels_check():
    """
    # 创建关系
    match(p:Symptom),(q:Examine) where p.name='上下楼梯疼' and q.name='膝关节核磁' create (p)-[rel:need_check{name:'症状检查'}]->(q)

    # 删除关系
    MATCH(p: Symptom)-[r: need_check]-(q:Examine)
    WHERE p.name = '上下楼梯疼' and q.name = '膝关节核磁'
    DELETE r
    """
    cql = "MATCH(p:Symptom)-[r:need_check]-(q:Examine) DELETE r"
    neo4j.execute_write(cql)
    print("删除成功 => need_check")
    # 症状 需要  做哪些检查
    rels_check = generate_examine()
    print(rels_check)
    cql_list = generate_cql('Symptom', 'Examine', rels_check, 'need_check', '症状检查')
    for cql in cql_list:
        neo4j.execute_write(cql)
        print(cql)

关系：疾病-症状

def generate_symptom() -> list:
    """
    关系：疾病-症状 (疾病有哪些症状）
    """
    rels_check = []
    df = pd.read_csv('relationship_data.csv')
    for idx, row in df.iterrows():
        for symptom in row['症状'].split(','):
            for exam in row['疾病'].split(','):
                rels_check.append([exam, symptom])
    rels_check = deduplicate(rels_check)
    return rels_check

代码重构

包括疾病用药，食物能吃，食物不能吃的关系。
详细代码如下

import logging
from utils.neo4j_provider import neo4j
import pandas as pd

logging.root.setLevel(logging.INFO)


# 关系去重函数
def deduplicate(relation_old) -> list:
    relation_new = []
    for each in relation_old:
        if each not in relation_new:
            relation_new.append(each)
    return relation_new


def generate_cql(start_node, end_node, edges, rel_type, rel_name) -> str:
    """
    生成 CQL
    """
    cql = []
    for edge in edges:
        p = edge[0]
        q = edge[1]
        # 创建关系的 Cypher 语句
        cql.append(
            "MATCH(p:%s),(q:%s) WHERE p.name='%s' and q.name='%s' CREATE (p)-[rel:%s{name:'%s'}]->(q)" % (start_node, end_node, p, q, rel_type, rel_name))
        print('创建关系 {}-{}->{}'.format(p, rel_type, q))
    return cql


def generate_relation(l_name, r_name) -> list:
    relation_list = []
    df = pd.read_csv('relationship_data.csv')
    for idx, row in df.iterrows():
        for l_node in row[l_name].split(','):
            for r_node in row[r_name].split(','):
                relation_list.append([l_node, r_node])
    return deduplicate(relation_list)


def create_relationship(l_node, r_node, relationship, l_data_name, r_data_name, relation_name):
    """
    创建关系
    :param l_node: 左节点 name
    :param r_node: 右节点 name
    :param relationship: 关系
    :param l_data_name: 左数据列名
    :param r_data_name: 右数据列名
    :param relation_name: 关系 name
    :return:
    """
    neo4j.delete_relationship(l_node, r_node, relationship)

    relation_list = generate_relation(l_data_name, r_data_name)
    print(relation_list)

    cql_list = generate_cql(l_node, r_node, relation_list, relationship, relation_name)
    for cql in cql_list:
        neo4j.execute_write(cql)
        print(cql)


def relationship_relation_check():
    l_node = "Symptom"
    r_node = "Examine"
    relationship = "need_check"
    l_data_name = '症状'
    r_data_name = '检查'
    rel_name = '症状检查'
    create_relationship(l_node, r_node, relationship, l_data_name, r_data_name, rel_name)


def relationship_has_symptom():
    l_node = "Disease"
    r_node = "Symptom"
    relationship = "has_symptom"
    l_data_name = '疾病'
    r_data_name = '症状'
    rel_name = '症状'
    create_relationship(l_node, r_node, relationship, l_data_name, r_data_name, rel_name)


def relationship_used_drugs():
    l_node = "Disease"
    r_node = "Drug"
    relationship = "used_drugs"
    l_data_name = '疾病'
    r_data_name = '药品'
    rel_name = '常用药品'
    create_relationship(l_node, r_node, relationship, l_data_name, r_data_name, rel_name)


def relationship_doeat_foods():
    l_node = "Disease"
    r_node = "Foods"
    relationship = "doeat_foods"
    l_data_name = '疾病'
    r_data_name = '宜吃'
    rel_name = '推荐食物'
    create_relationship(l_node, r_node, relationship, l_data_name, r_data_name, rel_name)


def relationship_noteat_foods():
    l_node = "Disease"
    r_node = "Foods"
    relationship = "noteat_foods"
    l_data_name = '疾病'
    r_data_name = '忌吃'
    rel_name = '忌吃食物'
    create_relationship(l_node, r_node, relationship, l_data_name, r_data_name, rel_name)


if __name__ == "__main__":
    # 有症状需要做哪些检查
    relationship_relation_check()

    # 疾病有哪些症状
    relationship_has_symptom()

    # 疾病常用药物
    relationship_used_drugs()

    # 推荐饮食
    relationship_doeat_foods()

    # 不宜饮食
    relationship_noteat_foods()

源代码地址：https://gitee.com/VipSoft/VipQA