官方wiki
https://docs.python.org/2/library/csv.html
写的把 csv表转化成特定格式的小工具
#! /usr/bin/env python
#coding=utf8
import os
import sys
import json
import codecs // 打开不同编码的文件
import csv // 解析 csv
def dd_data(confpath=None,inputfilepath=None,outfilepath=None):
if not confpath or not inputfilepath or not outfilepath:
print "请保证路径正确！"
syt.exit()
conf_dic=json.load(open(confpath,"r"))
out_f = open(outfilepath,"w+")
map_dic = conf_dic.get("map_dic",{"zyid":0,"patientid":0})
encoding = conf_dic.get("encoding","utf-8")
split_str = conf_dic.get("split",",")
if isinstance(split_str,unicode):
split_str = split_str.encode("utf8")
def utf_8_encoder(unicode_csv_data):
for line in unicode_csv_data:
yield line.encode('utf-8')
with codecs.open(inputfilepath,"rb",encoding) as f:
reader = csv.reader(utf_8_encoder(f))
for row in reader :
#line= line.encode("utf-8")
#line = line.strip().replace("\n","@@@@")
#list = line.split(split_str)
list = row if row else []
if list :
zy_id =list[map_dic.get("zyid",0)]
patient_id = list[map_dic.get("patientid",0)]
title= list[map_dic.get("title",0)] if "title" in map_dic else ""
timestamp = list[map_dic.get("timestamp",len(list)-1)] if "timestamp" in map_dic else ""
content = "<html>"
for key,index in map_dic.items():
if key in ["zyid","patientid","timestatmp","title"]:
continue
try:
single_content = list[index].strip().replace("\n","@@@@")
except:
print "conf 配置索引值超出解析的列数，请检查配置！"
break
#sys.exit()
single_content = single_content.strip().replace("\n","@@@@")
#print type(single_content)
single_content = u"<p><p>%s:%s</p></p>"%(key,single_content.decode("utf8"))
content = content +single_content
content = content +"</html>"
out_f.write((u"%s\t%s\t%s\t%s\t%s\n"%(zy_id,patient_id,title,content,timestamp)).encode("utf8"))
f.close()
out_f.close()
print "success out_path_file=%s"%(outfilepath)
if __name__ == "__main__":
import argparse
parser =argparse.ArgumentParser(description="--conf= --infilee= --outfile=")
parser.add_argument("--conf",type=str,default=None)
parser.add_argument("--infile",type=str,default=None)
parser.add_argument("--outfile",type=str,default=None)
#parser.add_argument("--help",type=str,defalut=None)
args = parser.parse_args()
if not args.conf or not args.outfile or not args.infile:
print "请输入正确启动参数，参考 python mapper.py --conf= --infile= --outfile="
example_conf = {"encoding":"utf8","map_dic":{"zyid(必须字段)":0,"patientid（必须字段）":1,"title":2,"主诉（内容字段中文名）":3,"timestamp（时间戳）":4}}
print "\nconf实例\n%s"%(json.dumps(example_conf,ensure_ascii=False,indent=4))
sys.exit()
dd_data(args.conf,args.infile,args.outfile)
