doctest1 / app.py
Vadim212's picture
Upload 8 files
0416ac9 verified
raw
history blame contribute delete
No virus
972 Bytes
import csv
import xml.etree.ElementTree as ET
import glob, os
rootFolder = "c:/317"
file = open(rootFolder + "/result.csv", "w", encoding="utf-8")
file.write("prompt,text,rejected_text\n")
def parseXML(xmlFile):
prompt = xmlFile.replace("Using_", "").replace(".xml", "").replace(".", " ").replace("_", " ")
text = ""
try:
tree = ET.parse(rootFolder + "/" + xmlFile)
root = tree.getroot()
for item in root.findall(".//text"):
text += item.text
if text.find("а") == -1:
#file.write("### prompt\n")
file.write(prompt + "," + text.replace(",", " ") + "\n")
#file.write("### text\n")
#file.write(text.replace(",", " "))
except:
print("=======")
os.chdir(rootFolder)
for xmlFile in glob.glob("*.xml"):
print(xmlFile)
parseXML(xmlFile)
## parseXML('Using_WinRT_Viewer.Search_Panel.xml')
file.close()