import os
import re
import xml.etree.ElementTree as ET


class Mail:
    def __init__(self, v, tresc = ''):
        self.addres = v
        self.tresc = tresc
        self.spam = 0
        self.ham = 0


SPAM = {}
HAM = {}
maile = []

doc = ET.parse('spam/dict.xml')
root = doc.getroot()
for child in root:
    t = child.text
    p = child.attrib['probabilty']
    if child.attrib['type'] == 'spam':
        SPAM[t] = p
    elif child.attrib['type'] == 'ham':
        HAM[t] = p

print(SPAM)
print(HAM)

for dirname, dirnames, filenames in os.walk('spam/'):
    for filename in filenames:
        if '.txt' in filename:
            #print(os.path.join(dirname, filename))
            with open(os.path.join(dirname, filename)) as f:
                data = f.read()
                r = re.search(r"[a-zA-Z0-9_.+-]+@[a-zA-Z0-9-]+\.[a-zA-Z0-9-.]+", data).group(0)
                #print(r)
                maile.append(Mail(r, data))

                for k, v in SPAM.items():
                    count = len(re.findall('^('+k+')*', data.lower()))
                    print(count)

                for k, v in HAM.items():
                    count = len(re.findall('^('+k+')*', data.lower()))
                    print(count)