adm-ntuh-net/forteo/scrape.py
2024-12-12 10:19:16 +08:00

50 lines
1.1 KiB
Python
Executable file

import re
# from openpyxl import load_workbook
from pandas import read_excel
from pymongo import MongoClient
SHEETS = (
("神外-總院_1070920.xlsx", "總院"),
# ("(骨穩)總院_分科1070928.xlsx", "工作表1")
("(骨穩)總院_分科1071002.xlsx", "工作表1")
)
client = MongoClient("mongodb.xiao.tw", 27017)
db = client.forteo
posts = db.posts
# print(posts.find_one())
# exit()
from ntuhgov.portal_selenium import *
from ntuhgov.myutil import *
for file_name, sheet_name in SHEETS:
data = read_excel(file_name, sheet_name)
for chartno in data.病歷號:
if not re.search("\d+", str(chartno)):
continue
print(chartno)
post = posts.find_one({"_id": chartno})
if post is None:
post = {"_id": chartno}
if 'drug' not in post:
post['drug'] = ShowMedicalRecordDrug(chartno)
if 'report' not in post:
post['report'] = ElectronicMedicalReportViewer(chartno)
if 'op' not in post:
post['op'] = OPNoteList(chartno)
posts.update({"_id": chartno}, {"$set": post}, upsert=True)