Files
jupyter/文件管理1.ipynb
T
2022-04-27 16:34:40 +08:00

61 KiB
Executable File

文件管理

数字文件名转换为文本文件名

In [1]:
import os,sys,shutil
import openpyxl
import math

fi_xls = os.getcwd() + '/data/岳阳兴长.xlsx'
fi_path = os.getcwd() + '/file/220427'
old = []
dict1 = {}

wb = openpyxl.load_workbook(fi_xls)
sheet = wb.active
depart = []
for n in range(2,sheet.max_row+1):    
    m_name = str(sheet.cell(n,1).value)
    m_depart = sheet.cell(n,5).value
    if m_depart not in depart:
        depart.append(sheet.cell(n,5).value)
    dict1[int(m_name.split('.')[0])] = [sheet.cell(n,2).value.strip(),sheet.cell(n,5).value.strip()]
    
# 创建部门办公室    
m_path = os.getcwd()+'/file/220427/new'
for pn in depart:
    if not os.path.exists(m_path + '/' + pn):
        os.mkdir(m_path + '/' + pn)
fl=os.listdir(fi_path)
for fn in fl:
    if os.path.isfile(fi_path + '/' + fn):
        ofn = int(fn.split('.')[0])
        old.append(ofn) 
old.sort()
for n in old:    
    o_name = f'{fi_path}/{n}.pdf'
    n_name = f'{fi_path}/new/{dict1[n][1]}/{str(n).rjust(5,"0")}-{dict1[n][0]}.pdf'
    if not os.path.exists(n_name):
        shutil.copyfile(o_name,n_name)
        print(n_name)
#print(dict1)
/home/songyi/python/jupyterlab/jupyter/file/220427/new/公司高管/00003-李湘波.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/公司高管/00005-霍国良.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/安全环保部/00006-王巍.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/安全环保部/00007-邹高尚.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/安全环保部/00008-黄伟全.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/安全环保部/00009-戴忠.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/安全环保部/00010-黄铮.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/安全环保部/00011-黄勇.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/安全环保部/00012-朱志南.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/安全环保部/00013-徐应湘.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/安全环保部/00014-杨细东.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/安全环保部/00015-梁维民.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/安全环保部/00016-王忠平.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/安全环保部/00017-周罗继.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/安全环保部/00018-李碧宏.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工生产部/00019-刘湘.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工生产部/00021-王维肇.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工生产部/00022-瞿翔龙.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工生产部/00023-曹晖.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工生产部/00024-周广平.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工生产部/00025-刘立成.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工生产部/00026-张运生.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工生产部/00027-陶刚.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工生产部/00028-蒋孝华.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工生产部/00029-陈光辉.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工生产部/00030-彭海波.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工生产部/00031-程憬.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工生产部/00032-任香元.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工生产部/00033-甘亚玲.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工生产部/00034-胡凯华.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工生产部/00035-黄晓华.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工生产部/00036-宋莉莉.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工生产部/00037-喻勇霞.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工生产部/00038-湛意平.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工生产部/00039-钟礼旭.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00040-谭明凤.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00041-曾啸峰.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00042-杜文学.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00043-方先炬.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00044-湛会英.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00045-谢定.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00046-胡伟.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00047-刘雄晖.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00049-甘强.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00050-吴岭.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00051-赵武.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00052-龚朝辉.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00054-林芳.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00055-严腾飞.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00058-曾武军.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00059-郭艳.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00060-郭俐.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00061-余丽利.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00062-刘朝晖.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00063-周芳.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00064-李想.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00065-谢丽.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00066-何红.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00067-陈克萍.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00068-胡勇.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00069-周文.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00072-刘永波.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00073-段积荣.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00074-毛玉湘.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00075-杨勇.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00076-朱辉.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00077-郭爱君.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00078-肖文.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00079-戴群.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00080-刘云.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00081-张建.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00083-乔新伟.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00086-袁建平.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00089-郭华.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00090-余意.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00091-方成宇.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00092-李俊.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00093-曾成艺.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00094-刘翔.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00095-肖靖.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00096-郭佳君.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00097-周景.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00098-李红.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00099-贺琴.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00100-李爱军.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00101-胡燕梅.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00102-刘玲.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00103-雷波.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00104-桑波.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00105-李桃霞.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00106-瞿强国.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00107-姚军霞.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00108-刘先锋.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00109-谢江南.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00110-张开良.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00112-杨秀清.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00113-易丽英.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00114-程云.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00115-许真.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00116-易宏彬.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00117-文科.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00118-龚永峰.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00119-李德金.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00120-张元姿.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00121-宋军.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00122-侯芳.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00123-朱昊.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00124-杨海林.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00125-周丽.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00127-张嫦燕.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00128-郑娟.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00129-李斌.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00130-侯谊.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00131-魏桂芳.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00132-廖丽华.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00133-任君慧.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00134-张友香.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00135-彭继东.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00136-谢艳.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00137-李荣.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00138-贾杰夫.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00140-高建人.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工一区/00142-刘梦婷.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00143-杨浩然.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00144-姚登雄.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00145-陈雄.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00146-骆炎峰.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00147-石磊.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00148-苏丰.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00149-欧阳斌.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00150-符胜军.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00151-毛朝晖.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00152-郭品质.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00153-黄辉.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00154-李伟锋.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00155-李宁.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00156-程晓.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00157-廖冬华.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00158-王代明.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00159-冯小红.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00160-欧鹏.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00161-程锦.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00162-余阳林.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00163-胡志斌.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00164-钟图富.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00165-季艳平.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00166-钟华旺.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00167-尹德斌.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00168-谭必军.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00169-李玲.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00170-胡彪.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00171-龙登.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00172-龚毅.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00173-李明明.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00174-陈刚.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00175-马灯山.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00177-喻中建.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00178-胡永红.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00179-张永红.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00181-何可.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00184-方猛辉.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00186-冯东明.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00188-彭中豪.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00189-罗林.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00190-何芳.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00191-李丹.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00192-舒勇.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/化工二区/00193-季成.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00195-吕国葆.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00196-汪华.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00197-熊云.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00198-叶立.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00199-刘琪.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00200-陈斌.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00201-饶俊.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00202-毛伟.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00203-李少华.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00204-马庆.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00205-何岘.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00206-杨扬.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00207-金灿.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00208-胡小兵.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00209-何海祥.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00210-余新.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00211-田嘉.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00212-麻玉娟.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00213-杨春.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00214-吴剑.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00215-胡剑.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00216-金波.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00217-祁友云.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00218-李航.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00219-喻陕辉.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00220-赵晓红.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00221-黄建民.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00222-徐德云.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00223-罗友望.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00224-钟维钢.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00225-李阁泽.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00226-何炙.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00227-邓卫东.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00228-韩飞飞.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00229-张文远.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00230-刘金花.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00231-莫凝.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00232-段红英.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00234-胡康仁.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00237-黄阿梅.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00238-梁勇.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00241-胡武岗.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00242-何向阳.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00244-江海涛.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00246-陈列喜.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00248-彭军荣.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00249-刘亚武.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/企业运营部/00251-周勇.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/企业运营部/00252-白羽.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/企业运营部/00255-王卫平.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/企业运营部/00256-刘晨.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/企业运营部/00257-吴蓉.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/企业运营部/00258-李亮凯.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/营销中心/00259-谢炜.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/营销中心/00260-陈建永.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/营销中心/00261-蒋小霞.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/营销中心/00262-赖晓波.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/营销中心/00263-余炎.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/营销中心/00264-赵轩.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/营销中心/00265-段为.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/湖南立为/00266-蒋佳.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/湖南立为/00267-李勇.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/湖南立为/00268-申凯.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/湖南立为/00269-贺君.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/湖南立为/00270-周末夫.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/湖南立为/00271-胡魁.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/湖南立为/00272-杨壹博.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/湖南立为/00273-刘曲波.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/湖南立为/00274-彭勇专.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/湖南立为/00275-刘昊.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/长进公司/00276-卢小虎.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/长进公司/00277-彭卫.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/长进公司/00278-谢学森.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/长进公司/00279-王湘军.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/长进公司/00280-赵文辉.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/长进公司/00281-何花.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/长进公司/00282-周烽.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/长进公司/00283-彭勇.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/长进公司/00284-周华.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/长进公司/00285-龙正金.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/长进公司/00286-谭炜.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/长进公司/00287-吴三平.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/长进公司/00288-彭毅超.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/长进公司/00289-曾海清.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/长进公司/00291-王永良.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/长进公司/00292-谭建霞.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/长进公司/00293-王心卫.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/长进公司/00294-殷鸿.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/长进公司/00296-徐扬兵.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/长进公司/00297-吴世钢.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/长进公司/00298-袁汉波.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/长进公司/00299-汤海平.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/长进公司/00300-罗浩.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/投资发展部/00301-邓桥.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/投资发展部/00302-龚琛荥.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/投资发展部/00304-李柯.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/投资发展部/00305-谢磊.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/投资发展部/00306-罗萱平.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/投资发展部/00307-谢溢.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/投资发展部/00308-何思宇.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/研发中心/00311-刘允德.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/研发中心/00312-张泉.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/研发中心/00313-李凡.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/研发中心/00314-苏彦斌.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/研发中心/00315-胥琦.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/研发中心/00316-谢以杰.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/研发中心/00317-刘建华.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/研发中心/00318-张海科.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/研发中心/00319-潘闰.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/研发中心/00320-芦易.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/研发中心/00323-谢连.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/研发中心/00324-罗菁菁.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/财务资产部/00325-谈敏.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/财务资产部/00326-王金玲.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/财务资产部/00327-李永利.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/财务资产部/00328-黄一鹤.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/财务资产部/00329-徐旦.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/财务资产部/00330-何圭.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/财务资产部/00331-雷桂萍.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/财务资产部/00332-汪庆梅.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/财务资产部/00333-蒋逸芬.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/湖南立泰/00336-叶小六.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/湖南立泰/00338-周蒙.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/湖南立泰/00340-王宗勇.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/湖南立泰/00342-罗禹莉.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/湖南立泰/00345-向丽琴.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/湖南立泰/00346-李扬.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/湖南立泰/00347-李劲.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/质检中心/00348-彭朝晖.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/质检中心/00349-刘劲.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/质检中心/00350-邹文利.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/质检中心/00351-卢洁.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/质检中心/00353-文静.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/质检中心/00354-李莎莉.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/质检中心/00357-王伟.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/质检中心/00358-涂琼.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/质检中心/00359-王自湘.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/质检中心/00360-朱冬萍.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/质检中心/00361-刘临华.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/质检中心/00363-彭聪.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/质检中心/00365-游旭红.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/质检中心/00366-周欲勇.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/质检中心/00367-郭芳.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/质检中心/00368-孙亮.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/质检中心/00369-许婉莹.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/质检中心/00370-郭金.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/质检中心/00372-陶泓宇.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/质检中心/00376-曾婧一.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/质检中心/00378-梁艳青.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/质检中心/00379-杨彩琳.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/质检中心/00380-杨光.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/质检中心/00381-袁红霞.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/质检中心/00383-宋恬.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/行政管理部/00384-李斌.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/行政管理部/00385-李毅.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/行政管理部/00387-蒋福林.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/行政管理部/00388-杨红文.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/行政管理部/00389-陈蕊晗.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/行政管理部/00390-王海峰.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/行政管理部/00391-王纯初.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/行政管理部/00392-刘玲.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/行政管理部/00394-王珏.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/行政管理部/00397-韩双双.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/行政管理部/00400-关杰.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/行政管理部/00401-刘丽丹.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/行政管理部/00402-曹盛兰.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/行政管理部/00403-佘新明.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/行政管理部/00404-宋卫凤.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/行政管理部/00405-颜婷婷.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00406-陈相.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00407-倪文强.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00408-刘林.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00409-黄优东.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00410-王钟庆.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00411-陶沛.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00412-蒋帅.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00413-杨向阳.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00414-龚华.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00415-吴斌.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00416-李莉.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00417-吴雄.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00418-吴曦.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00419-李亮.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00420-马颂.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00421-曹竞.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00423-喻杰军.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00424-粟来富.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00425-郭维.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00426-周艳.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00427-文松.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00428-张应龙.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00429-熊千里.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00430-陈小军.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00431-何卓.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00432-朱金华.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00433-郭意.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00434-马琳.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00435-张亮.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00436-杨亚.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00437-王琼.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00438-周涵海.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00439-许丹伟.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00440-李昱瑾.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00441-周弘毅.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/销售华中长岭油库/00443-肖慧.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/能源公司/00501-袁正兴.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/行政管理部/00502-刘华.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/安全环保部/00503-王其海.pdf
/home/songyi/python/jupyterlab/jupyter/file/220427/new/安全环保部/00504-何强.pdf

PDF文件压缩

In [ ]:
import fitz
import os


def covert2pic(zoom):
    if os.path.exists('.pdf'):       # 临时文件,需为空
         os.removedirs('.pdf')
    os.mkdir('.pdf')
    for pg in range(totaling):
        page = doc[pg]
        zoom = int(zoom)            #值越大,分辨率越高,文件越清晰
        rotate = int(0)
        print(page)
        trans = fitz.Matrix(zoom / 100.0, zoom / 100.0).preRotate(rotate)
        pm = page.getPixmap(matrix=trans, alpha=False)
      
        lurl='.pdf/%s.jpg' % str(pg+1)
        pm.writePNG(lurl)
    doc.close()

def pic2pdf(obj):
    doc = fitz.open()
    for pg in range(totaling):
        img = '.pdf/%s.jpg' % str(pg+1)
        imgdoc = fitz.open(img)                 # 打开图片
        pdfbytes = imgdoc.convertToPDF()        # 使用图片创建单页的 PDF
        os.remove(img)  
        imgpdf = fitz.open("pdf", pdfbytes)
        doc.insertPDF(imgpdf)                   # 将当前页插入文档
    if os.path.exists(obj):         # 若文件存在先删除
        os.remove(obj)
    doc.save(obj)                   # 保存pdf文件
    doc.close()


def pdfz(sor, obj, zoom):    
    covert2pic(zoom)
    pic2pdf(obj)
    


sor = "5.pdf"              # 需要压缩的PDF文件
obj = "new-" + sor
doc = fitz.open(sor) 
totaling = doc.pageCount

zoom = 150                     # 清晰度调节,缩放比率
pdfz(sor, obj, zoom)
os.removedirs('.pdf')
In [ ]:
from pdf2image import convert_from_path, convert_from_bytes
import os,sys
import tempfile
from pdf2image.exceptions import (
    PDFInfoNotInstalledError,
    PDFPageCountError,
    PDFSyntaxError
)
#images = convert_from_path('1.pdf',dpi=200,fmt='jpg', output_folder='./sample_data')
with tempfile.TemporaryDirectory() as path:
    images_from_path = convert_from_path('5.pdf', dpi=100,fmt='jpg', output_folder='./pic')
print(path)
In [ ]:
import img2pdf  
import os,sys
import glob

fl=glob.glob('./pic/*.jpg')
fl.sort()
a4inpt = (img2pdf.mm_to_pt(210),img2pdf.mm_to_pt(297))
layout_fun = img2pdf.get_layout_fun(a4inpt)
with open("name.pdf","wb") as f:
    f.write(img2pdf.convert(fl,layout_fun=layout_fun))
In [ ]:
import glob
import fitz  # 导入本模块需安装pymupdf库
import os,sys

def pic2pdf_1(img_path, pdf_path, pdf_name):
    doc = fitz.open()
    fl=os.listdir(img_path)
    fl.sort()
    width, height = fitz.PaperSize("a4")
    for img in fl:
        fn = img_path+'/'+img
        if os.path.isfile(fn):
            imgdoc = fitz.open(img_path+'/'+img)
            pdfbytes = imgdoc.convertToPDF()
            imgpdf = fitz.open("pdf", pdfbytes,width = width, height = height)
            doc.insertPDF(imgpdf)
    doc.save(pdf_path +'/'+ pdf_name)
    doc.close()

img_path = os.getcwd() +'/pic'
pdf_path = os.getcwd()
pic2pdf_1(img_path=img_path, pdf_path=pdf_path, pdf_name='1.pdf')

图片文件扫描识别

In [ ]:
#将指定目录下图片文件进行文字识别
import os,sys
from aip import AipOcr
import glob

""" 你的 APPID AK SK """
APP_ID = '17553946'
API_KEY = 'IMC1ss3Tyo3vEAdVH6jgcdv2'
SECRET_KEY = 'Iua6KdhuDtz159rjzGeFGpqgjhnoU0zZ'

client = AipOcr(APP_ID, API_KEY, SECRET_KEY)
def get_file_content(filePath):
    with open(filePath, 'rb') as fp:
        return fp.read()

options = {}
options["language_type"] = "CHN_ENG"
options["detect_direction"] = "true"
options["detect_language"] = "true"
options["probability"] = "true"

fi_path = os.getcwd()+'/pic'
fl = glob.glob(f'{fi_path}/*.jpg')
fl.sort()
for fl1 in fl:
    file_name =  fl1
    image = get_file_content(file_name)
    result= client.basicGeneral(image, options)
    if 'words_result' in result:
        print('\n'.join([w['words'] for w in result['words_result']]))
        print('\n')
        
        

word文档读取

In [ ]:
import docx
import json
Doc = docx.Document(r"症状对症处方选穴.docx")
#print("檔案內含段落數:",len(Doc.paragraphs),"\n")
testList = []
text1 = []
dict1 = {}
for text in Doc.paragraphs:
    testList.append(text)
i = 0
for pg in testList:
    
    if len(pg.text) >0:
        s = ''.join(pg.text.split())        
        if s[0:1] == '第':
            i += 1
            n = 0
            
            dict1.setdefault(i,{})
            dict1[i]['title'] = pg.text.split()[1]
            dict1[i]['sub'] = {}
            
            #dict1[i]['title'].setdefault(i,{})
        elif s[0:1] == '(':
            sub = s.split(')')[1]
            dict2 = {}
            dict1[i]['sub'].setdefault(sub,[])
        else:
            dict1[i]['sub'][sub].append(s)
filename = '症状对症处方选穴.json'
with open(filename,'w') as fl:
    json.dump(dict1, fl,ensure_ascii=False) 
print('ok')  
In [ ]:
import docx
import json
Doc = docx.Document(r"常见疾病辨病处方选穴.docx")
#print("檔案內含段落數:",len(Doc.paragraphs),"\n")
testList = []
text1 = []
dict1 = {}
for text in Doc.paragraphs:
    testList.append(text)
i = 0
for pg in testList:
    
    if len(pg.text) >0:
        s = ''.join(pg.text.split())        
        if s[0:1] == '第':
            i += 1
            n = 0
            
            dict1.setdefault(i,{})
            dict1[i]['title'] = pg.text.split('节')[1]
            dict1[i]['sub'] = {}
            
            #dict1[i]['title'].setdefault(i,{})
        elif s[0:1] == '(':
            sub = s.split(')')[1]
            dict2 = {}
            dict1[i]['sub'].setdefault(sub,[])
        else:
            dict1[i]['sub'][sub].append(s)
filename = '常见疾病辨病处方选穴.json'
with open(filename,'w') as fl:
    json.dump(dict1, fl,ensure_ascii=False)
print('ok')  
In [ ]:
import json
import re
filename = '症状对症处方选穴.json'
pattern = r'[\d\.]'
mo =r'[\u4e00-\u9fa5]+'
dict2 = {}
with open(filename,'r') as fl:
    dict1 = json.load(fl) 
for k,v in dict1.items():
    m_title = v['title']
    for k1,v1 in v['sub'].items():
        sub_title = k1            
        s = ''
        for mx in v1:            
            s = s + mx
        list1 = re.split(pattern, s)
        list2 = []
        for ss in list1:
            if ss != '':
                list3 = re.findall(mo,ss)
                list2.append(list3)
                
        #print(m_title,sub_title,list2)
        dict2.setdefault(m_title,{})
        dict2[m_title][sub_title] = list2
filename = 'new_症状对症处方选穴.json'
with open(filename,'w') as fl:
    json.dump(dict2, fl,ensure_ascii=False) 
print('ok')  
       

word使用模板生成文档

In [ ]:
import openpyxl
from docxtpl import DocxTemplate

wb = openpyxl.load_workbook('file/火车餐厅人员信息表1.xlsx')
sheet = wb.active
# sheets = wb.sheetnames
list1 = []
context = {}
for n in range(2, sheet.max_row+1):
    dict1 = {}
    dict1['xh'] = sheet.cell(n,1).value
    dict1['xm'] = sheet.cell(n,2).value
    dict1['sfzhm'] = sheet.cell(n,3).value
    dict1['dh'] = sheet.cell(n,4).value
    dict1['cx'] = sheet.cell(n,5).value
    dict1['xzdz'] = sheet.cell(n,6).value
    list1.append(dict1)
context['info'] = list1
tpl = DocxTemplate("file/火车餐厅人员信息表.docx")
tpl.render(context)
tpl.save('file/test1.docx')

PDF文件读取

pdfminer读取PDF文件

In [ ]:
from newpdfminer.pdfparser import PDFParser, PDFDocument
from newpdfminer.pdfinterp import PDFResourceManager, PDFPageInterpreter
from newpdfminer.converter import PDFPageAggregator
from newpdfminer.layout import LAParams, LTTextBox
from newpdfminer.pdfinterp import PDFTextExtractionNotAllowed

path = "file/211027/114.pdf"

# 用文件对象来创建一个pdf文档分析器
praser = PDFParser(open(path, 'rb'))
# 创建一个PDF文档
doc = PDFDocument()
# 连接分析器 与文档对象
praser.set_document(doc)
doc.set_parser(praser)

# 提供初始化密码
# 如果没有密码 就创建一个空的字符串
doc.initialize()

# 检测文档是否提供txt转换,不提供就忽略
if not doc.is_extractable:
    raise PDFTextExtractionNotAllowed
else:
    # 创建PDf 资源管理器 来管理共享资源
    rsrcmgr = PDFResourceManager()
    # 创建一个PDF设备对象
    laparams = LAParams()
    device = PDFPageAggregator(rsrcmgr, laparams=laparams)
    # 创建一个PDF解释器对象
    interpreter = PDFPageInterpreter(rsrcmgr, device)

    # 循环遍历列表,每次处理一个page的内容
    for page in doc.get_pages():
        interpreter.process_page(page)                        
        # 接受该页面的LTPage对象
        layout = device.get_result()
        # 这里layout是一个LTPage对象,里面存放着这个 page 解析出的各种对象
        # 包括 LTTextBox, LTFigure, LTImage, LTTextBoxHorizontal 等                            
        for x in layout:
            if isinstance(x, LTTextBox):
                print(x.get_text().strip())

pdfplumber读取PDF文件

In [ ]:
import pdfplumber
from PIL import Image

# 读取 PDF 文档
pdf = pdfplumber.open("5.pdf")

# 获取页数
print("总页数:",len(pdf.pages))
print("-----------------------------------------")

# 读取第 4 页;索引从 1 开始
page = pdf.pages[0] 
print("本页:",page.page_number + 1)
print("-----------------------------------------")
# 单页文件存储为图片
#im = page.to_image()
#im.save('./5_1.png', format='PNG')
# 单页文件中原始图片读取,不能读取图表文件
#imgs = page.images
#for i, img in enumerate(imgs):
#    size = img['width'], img['height']
#    data = img['stream'].get_data()
#    out_path = f'003pdf_images_{i}.png'
#    with open(out_path, 'wb') as fimg_out:
#        fimg_out.write(data)
# 导出第 4 页文本
text = page.extract_text()
print(text)

fitz1.8读取PDF文件中的图片

In [ ]:
import fitz
import re
import os

file_path = '5.pdf'  # PDF 文件路径
dir_path = './data' # 存放图片的文件夹

def pdf2image1(path, pic_path):
    checkIM = r"/Subtype(?= */Image)"
    pdf = fitz.open(path)
    lenXREF = pdf.xref_length()
    count = 1
    for i in range(1, lenXREF):
        text = pdf.xref_object(i)
        isImage = re.search(checkIM, text)
        if not isImage:
            continue
        pix = fitz.Pixmap(pdf, i)
        new_name = f"img_{count}.png"
        pix.writePNG(os.path.join(pic_path, new_name))
        count += 1
        pix = None

pdf2image1(file_path, dir_path)

Pdf文件拆分

In [ ]:
import os, sys
from PyPDF2 import PdfFileWriter, PdfFileReader, PdfFileMerger

file = "file/211027/114.pdf"
pdf_reader = PdfFileReader(file)
n = pdf_reader.numPages
print(n)
pdfWriter = PdfFileWriter()
pageObj = pdf_reader.getPage(n-2)
pageObj.scaleBy(1)
pdfWriter.addPage(pageObj)
with open('split_114.pdf', 'wb') as pdfOutputFile:
    pdfWriter.write(pdfOutputFile)
In [ ]:
import fitz
import os

file = "file/211027/114.pdf"
doc = fitz.open(file)
n = doc.page_count
print(n)
doc2 = fitz.open()
page = doc[n - 2]
#doc2.insert_pdf(doc, to_page = n-2)  # first 10 pages
doc2.insert_pdf(doc, from_page = n-2,to_page = n-2)
#print(type(page))
#doc2.insert_pdf(page) # last 10 pages
doc2.save("first-and-last-10.pdf")

PDF文件读取表格

In [ ]:
import pdfplumber
import json
import re

file = "file/1_中石化北海炼化2020年团体体检报告.pdf"
pdf = pdfplumber.open(file)
list1 = []
dict1 = {}
n = 1
for page in pdf.pages:
    # print(page.extract_text())
    
    
    for pdf_table in page.extract_tables():
        list2 = []
        table = []
        cells = []
        for row in pdf_table:
            if not any(row):
                # 如果一行全为空,则视为一条记录结束
                if any(cells):
                    table.append(cells)
                    cells = []
            elif all(row):
                # 如果一行全不为空,则本条为新行,上一条结束
                if any(cells):
                    table.append(cells)
                    cells = []
                table.append(row)
            else:
                if len(cells) == 0:
                    cells = row
                else:
                    for i in range(len(row)):
                        if row[i] is not None and not cells[i]:
                            cells[i] = row[i]
                        elif row[i] is not None and cells[i]:                            
                            table.append(cells)
                            cells = row
                            #cells[i] = row[i]
                            break
                        elif row[i] is None and cells[i]:
                            continue
                        elif row[i] is None and not cells[i]:
                            #cells[i] = row[i]
                            continue                        
                        #    cells[i] = row[i] if cells[i] is None else cells[i] + row[i]
       
        list2 = []                    
        for row in table:
            data =[re.sub('\s+', '', cell) if cell is not None else None for cell in row]
            print(data)
            
            #print(json.dumps(data_list, indent=2, ensure_ascii=False))
            #with open('Test1.json','a',encoding="utf-8") as file:   # json文件的存放位置
            #    json.dump(data_list, file, ensure_ascii=False)
            list2.append(data)
        if len(list2) > 0:
            dict1[n] = list2
        #print(dict1)
        #list1.append(list2)
        n +=1
with open('Test1.json','w',encoding="utf-8") as file:
    json.dump(dict1, file, ensure_ascii=False)
            
pdf.close()
In [ ]:
import pdfplumber
import json
import re

file = "file/1_中石化北海炼化2020年团体体检报告.pdf"
pdf = pdfplumber.open(file)
list1 = []
dict1 = {}
n = 1
for page in pdf.pages:
    for pdf_table in page.extract_tables():
        list2 = []
        table = []
        cells = []
        for row in pdf_table:
            print(row)
            print('******')
        print('--------')
        
In [ ]:
import camelot
import json
import re

file = "file/1_中石化北海炼化2020年团体体检报告.pdf"
tables = camelot.read_pdf(file, pages='3',flavor='stream')
# 2.导出pdf所有的表格为csv文件
tables.export('foo.json', f='json')
print('ok!')
In [ ]: