多个文件.txt文件转为一个.xml文件:
#! /usr/bin/python
import os, sys
import glob
import xml.dom.minidom
from PIL import Image
src_txt_dir = "/home/lrj/picture/ICDAR2013/Challenge2_Test_Task1_GT/"
img_basenames = [] # e.g. 100.jpg
#for item in img_basenames:
fileList=os.listdir(src_txt_dir)
xml_file=open((src_txt_dir+'gt.xml'), 'w')
xml_file.write('<?xml version="1.0" encoding="UTF-8"?>\n')
xml_file.write('<tagset>\n')
total_num=len(fileList)
idx=range(total_num-1)
i=1
# control the number of txt
for i in idx:
stri='%d'%i
gt = open(src_txt_dir + 'gt_img_' + stri+ '.txt').read().splitlines()
xml_file.write(' <image>\n')
xml_file.write(' <imageName>'+'img_'+stri+'.jpg'+'</imageName>\n')
xml_file.write(' <taggedRectangles>\n')
for img_each_label in gt:
spt = img_each_label.split(',')
width=int(str(spt[2]))-int(str(spt[0]))
w='%d'%width
height=int(str(spt[3]))-int(str(spt[1]))
h='%d'%height
xml_file.write(' <taggedRectangle x='+'"'+str(spt[0])+'"'+ ' y='+'"'+str(spt[1])+'"'+' width='+'"'+w+'"'+' height='+'"'+h+'"' +' offset='+'"'+'%d'%0 + '"'+' />\n')
xml_file.write(' </taggedRectangles>\n')
xml_file.write(' </image>\n')
xml_file.write('</tagset>\n')