Spaces:
Sleeping
Sleeping
aliicemill
commited on
Commit
•
40fbfee
1
Parent(s):
40eeb61
Upload app.py
Browse files
app.py
ADDED
@@ -0,0 +1,106 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# Importing Libraries
|
2 |
+
# author:@aliicemill
|
3 |
+
import nltk
|
4 |
+
from PIL import Image
|
5 |
+
import matplotlib.pyplot as plt
|
6 |
+
import streamlit as st
|
7 |
+
from nltk.corpus import stopwords
|
8 |
+
from nltk.tokenize import word_tokenize
|
9 |
+
import string
|
10 |
+
import stylecloud
|
11 |
+
|
12 |
+
# Downloading word clusters
|
13 |
+
nltk.download('stopwords')
|
14 |
+
nltk.download('punkt')
|
15 |
+
|
16 |
+
def preprocess_and_create_stylecloud(file_path, output_name='stylecloud.png',
|
17 |
+
icon_name='fas fa-laptop', lang='english'):
|
18 |
+
# read file
|
19 |
+
with open(file_path, 'r', encoding='utf-8') as f:
|
20 |
+
text = f.read()
|
21 |
+
|
22 |
+
# represent stopwords
|
23 |
+
stop_words = set(stopwords.words(lang))
|
24 |
+
|
25 |
+
# Remove punctuation marks
|
26 |
+
translator = str.maketrans('', '', string.punctuation)
|
27 |
+
text = text.translate(translator)
|
28 |
+
|
29 |
+
# Tokenize the text and convert to lowercase
|
30 |
+
tokens = word_tokenize(text.lower(), language=lang)
|
31 |
+
|
32 |
+
# Filter out stopwords
|
33 |
+
filtered_tokens = [word for word in tokens if word not in stop_words]
|
34 |
+
|
35 |
+
# Join filtered tokens
|
36 |
+
processed_text = ' '.join(filtered_tokens)
|
37 |
+
|
38 |
+
# Create StyleCloud
|
39 |
+
stylecloud.gen_stylecloud(text=processed_text,
|
40 |
+
icon_name=icon_name,
|
41 |
+
output_name=output_name)
|
42 |
+
# Show the generated StyleCloud
|
43 |
+
im = Image.open(output_name)
|
44 |
+
plt.figure(figsize=(10, 10))
|
45 |
+
plt.imshow(im)
|
46 |
+
plt.axis('off') # Hide axes
|
47 |
+
plt.show()
|
48 |
+
|
49 |
+
def create_stylecloud(text, language, icon):
|
50 |
+
output_file = "stylecloud.png"
|
51 |
+
|
52 |
+
stylecloud.gen_stylecloud(text=text,
|
53 |
+
icon_name=icon,
|
54 |
+
output_name=output_file)
|
55 |
+
|
56 |
+
return output_file
|
57 |
+
|
58 |
+
st.title("WordCloud Creator")
|
59 |
+
|
60 |
+
file = st.file_uploader("Import txt file", type=["txt"])
|
61 |
+
|
62 |
+
if file is not None:
|
63 |
+
text = file.getvalue().decode("utf-8")
|
64 |
+
|
65 |
+
language = st.radio("Language", ["tr", "en","fr","de"])
|
66 |
+
|
67 |
+
icon_options = ['fa-address-book', 'fa-address-card', 'fa-angry', 'fa-arrow-alt-circle-down', 'fa-arrow-alt-circle-left',
|
68 |
+
'fa-arrow-alt-circle-right', 'fa-arrow-alt-circle-up', 'fa-bell', 'fa-bell-slash', 'fa-bookmark', 'fa-building',
|
69 |
+
'fa-calendar', 'fa-calendar-alt', 'fa-calendar-check', 'fa-calendar-minus', 'fa-calendar-plus', 'fa-calendar-times',
|
70 |
+
'fa-caret-square-down', 'fa-caret-square-left', 'fa-caret-square-right', 'fa-caret-square-up', 'fa-chart-bar',
|
71 |
+
'fa-check-circle', 'fa-check-square', 'fa-circle', 'fa-clipboard', 'fa-clock', 'fa-clone', 'fa-closed-captioning',
|
72 |
+
'fa-comment', 'fa-comment-alt', 'fa-comment-dots', 'fa-comments', 'fa-compass', 'fa-copy', 'fa-copyright', 'fa-credit-card',
|
73 |
+
'fa-dizzy', 'fa-dot-circle', 'fa-edit', 'fa-envelope', 'fa-envelope-open', 'fa-eye', 'fa-eye-slash', 'fa-file',
|
74 |
+
'fa-file-alt', 'fa-file-archive', 'fa-file-audio', 'fa-file-code', 'fa-file-excel', 'fa-file-image', 'fa-file-pdf',
|
75 |
+
'fa-file-powerpoint', 'fa-file-video', 'fa-file-word', 'fa-flag', 'fa-flushed', 'fa-folder', 'fa-folder-open', 'fa-frown',
|
76 |
+
'fa-frown-open', 'fa-futbol', 'fa-gem', 'fa-grimace', 'fa-grin', 'fa-grin-alt', 'fa-grin-beam', 'fa-grin-beam-sweat',
|
77 |
+
'fa-grin-hearts', 'fa-grin-squint', 'fa-grin-squint-tears', 'fa-grin-stars', 'fa-grin-tears', 'fa-grin-tongue',
|
78 |
+
'fa-grin-tongue-squint', 'fa-grin-tongue-wink', 'fa-grin-wink', 'fa-hand-lizard', 'fa-hand-paper', 'fa-hand-peace',
|
79 |
+
'fa-hand-point-down', 'fa-hand-point-left', 'fa-hand-point-right', 'fa-hand-point-up', 'fa-hand-pointer', 'fa-hand-rock',
|
80 |
+
'fa-hand-scissors', 'fa-hand-spock', 'fa-handshake', 'fa-hdd', 'fa-heart', 'fa-hospital', 'fa-hourglass', 'fa-id-badge',
|
81 |
+
'fa-id-card', 'fa-image', 'fa-images', 'fa-keyboard', 'fa-kiss', 'fa-kiss-beam', 'fa-kiss-wink-heart', 'fa-laugh',
|
82 |
+
'fa-laugh-beam', 'fa-laugh-squint', 'fa-laugh-wink', 'fa-lemon', 'fa-life-ring', 'fa-lightbulb', 'fa-list-alt',
|
83 |
+
'fa-map', 'fa-meh', 'fa-meh-blank', 'fa-meh-rolling-eyes', 'fa-minus-square', 'fa-money-bill-alt', 'fa-moon',
|
84 |
+
'fa-newspaper', 'fa-object-group', 'fa-object-ungroup', 'fa-paper-plane', 'fa-pause-circle', 'fa-play-circle',
|
85 |
+
'fa-plus-square', 'fa-question-circle', 'fa-registered', 'fa-sad-cry', 'fa-sad-tear', 'fa-save', 'fa-share-square',
|
86 |
+
'fa-smile', 'fa-smile-beam', 'fa-smile-wink', 'fa-snowflake', 'fa-square', 'fa-star', 'fa-star-half', 'fa-sticky-note',
|
87 |
+
'fa-stop-circle', 'fa-sun', 'fa-surprise', 'fa-thumbs-down', 'fa-thumbs-up', 'fa-times-circle', 'fa-tired', 'fa-trash-alt',
|
88 |
+
'fa-user', 'fa-user-circle', 'fa-window-close', 'fa-window-maximize', 'fa-window-minimize', 'fa-window-restore']
|
89 |
+
icon_options = list(map(lambda x : "fas "+x,icon_options))
|
90 |
+
|
91 |
+
icon = st.selectbox("Icon Selection", icon_options, index=1)
|
92 |
+
|
93 |
+
if st.button("Create"):
|
94 |
+
output_file = create_stylecloud(text, language, icon)
|
95 |
+
st.success(f"Here is your file: {output_file}")
|
96 |
+
|
97 |
+
image = Image.open(output_file)
|
98 |
+
st.image(image, caption='WordCloud', use_column_width=True)
|
99 |
+
|
100 |
+
with open(output_file, "rb") as file:
|
101 |
+
btn = st.download_button(
|
102 |
+
label="Download WordCloud",
|
103 |
+
data=file,
|
104 |
+
file_name=output_file,
|
105 |
+
mime="image/png"
|
106 |
+
)
|