Create a wordcloud from Codex telegram group chat

codex_logo.png

codex_wordcloud.png

gen.py

from bs4 import BeautifulSoup as bs4
import os
from wordcloud import WordCloud
from nltk import word_tokenize
from nltk.corpus import stopwords
from string import punctuation
import numpy as np
from PIL import Image
import random


# returns a random cyan looking color
def grey_color_func(word, font_size, position, orientation, random_state=None, **kwargs):
	return "rgb(0, {}, {})".format(random.randint(200, 255),random.randint(200, 255))

# parse data from Telegram export chat data. Data is stored in html format
def parse_data():
	files = os.listdir(".")
	html_files = [x for x in files if x.endswith("html")]

	text = ""

	for f in html_files:
		print(f, end="\r")

		with open(f, "r", encoding="utf-8") as fdata:
			soup = bs4(fdata, "html.parser")
			txts = soup.findAll("div", {"class": "text"})
			for txt in txts:
				text += txt.get_text()

	text = text.replace("aposs", "'")
	text = text.replace("apost", "'")
	text = text.replace("aposll", "'")
	text = text.replace("https", "'")

	stop = set(stopwords.words('english'))
	stop.update(set(punctuation))

	words = [w for w in word_tokenize(text.lower()) if w not in stop]

	return words # the final wordset

words = parse_data()
print(f"{len(words)} words\t\t")
codex_logo = np.array(Image.open("codex_logo.png")) # mask to use

wc = WordCloud(width=1920, height=1080, mask=codex_logo, max_words=2000, background_color="rgb(44,44,44)")
	 #,contour_width=3, contour_color="white")

wc.generate(" ".join(words))
wc.recolor(color_func=grey_color_func, random_state=3)

wc.to_file("codex_wordcloud.png")