from bs4 import BeautifulSoup
from wordcloud import WordCloud, STOPWORDS, ImageColorGenerator
from more_words import words
from PIL import Image
import numpy as np

msg_file_name = "175.html"
u1 = u"A***** B*****"
u2 = u"H***** B*******"

msg_file = open(msg_file_name)

soup = BeautifulSoup(msg_file, "html.parser")

u1_msgs = []
u2_msgs = []

msg = soup.find("p")

while True:
    if msg == None:
        break
    name = msg.find_previous("div", class_="message").span.text

    if u1 == name:
        u1_msgs.append(msg.text)
    elif u2 == name:
        u2_msgs.append(msg.text)

    msg = msg.find_next("p")

stopwords = set(STOPWORDS)
mask = np.array(Image.open("example.jpg"))
image_colors = ImageColorGenerator(mask)

for x in words:
    stopwords.add(x)
#stopwords.add("said")

wc = WordCloud(background_color="white", mask=mask, stopwords=stopwords)
wc.generate(" ".join(u1_msgs))
wc.recolor(color_func=image_colors)
wc.to_file("u1.png")

wc.generate(" ".join(u2_msgs))
wc.recolor(color_func=image_colors)
wc.to_file("u2.png")