import os
import sys
import time

pdfs = (pdfs for pdfs in os.listdir('.') if pdfs.endswith('.pdf'))

for pdf1 in pdfs:
 pdf = pdf1.replace(' ', '_').replace('-', '_').replace('&', '_')
 os.rename(pdf1, pdf)
 print('='*30)
 print(pdf)
 
 txt = pdf[:-4] + '.txt'
 exe = '"' + sys.executable + '" "'
 pdf2txt = os.path.dirname(sys.executable)
 pdf2txt = pdf2txt + '\\scripts\\pdf2txt.py" -o '
 try:
 #调用命令行工具pdf2txt.py进行转换
 #如果pdf加密过可以改写下面的代码
 #在-o前面使用-P来指定密码
 cmd = exe + pdf2txt + txt + ' ' + pdf
 os.popen(cmd)
 #转换需要一定时间，一般小文件2秒钟足够了
 time.sleep(2)
 #输出转换后的文本，前200个字符
 with open(txt, encoding='utf8') as fp:
 print(fp.read(200))
 except:
 pass