naufalnashif
commited on
Commit
•
8abf6b7
1
Parent(s):
9cb5b93
Delete app.py
Browse files
app.py
DELETED
@@ -1,47 +0,0 @@
|
|
1 |
-
|
2 |
-
import streamlit as st
|
3 |
-
import tweepy
|
4 |
-
import pandas as pd
|
5 |
-
import snscrape.modules.twitter as sntwitter
|
6 |
-
from tqdm.notebook import tqdm
|
7 |
-
|
8 |
-
|
9 |
-
st.title("Twitter Data Scraping App")
|
10 |
-
|
11 |
-
# Input kata kunci pencarian
|
12 |
-
search_query = st.text_input("Masukkan kata kunci pencarian:", "")
|
13 |
-
|
14 |
-
# Rentang waktu
|
15 |
-
start_date = st.date_input("Tanggal Mulai")
|
16 |
-
end_date = st.date_input("Tanggal Selesai")
|
17 |
-
|
18 |
-
def scrape_twitter_data(search_query, start_date, end_date):
|
19 |
-
# Mencari tweet dengan kata kunci dan rentang waktu
|
20 |
-
query = f'{search_query} until:{end_date} since:{start_date}'
|
21 |
-
tweets = sntwitter.TwitterSearchScraper(query)
|
22 |
-
|
23 |
-
tweets = []
|
24 |
-
n_tweets = 4_000
|
25 |
-
for i,tweet in tqdm(enumerate(tweets.get_items()), total=n_tweets):
|
26 |
-
data = [
|
27 |
-
tweet.date,
|
28 |
-
tweet.id,
|
29 |
-
tweet.rawContent,
|
30 |
-
tweet.user.username,
|
31 |
-
tweet.likeCount,
|
32 |
-
tweet.retweetCount,
|
33 |
-
]
|
34 |
-
tweets.append(data)
|
35 |
-
if i > n_tweets :
|
36 |
-
break
|
37 |
-
|
38 |
-
df = pd.DataFrame(
|
39 |
-
tweets, columns=["date", "id", "content", "username", "like_count", "retweet_count" ]
|
40 |
-
)
|
41 |
-
return df
|
42 |
-
|
43 |
-
if st.button("Scrape Data"):
|
44 |
-
df = scrape_twitter_data(search_query, start_date, end_date)
|
45 |
-
st.write("Hasil Pencarian:")
|
46 |
-
st.write(df)
|
47 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|