lowannann/assignment_1

1#R11142005 紀柔安2import streamlit as st3from twNLP-app.src.views.components.spinner import dowload_ckip_package, download_cwn_drivers4import pandas as pd5import requests 6import bs47from snownlp import SnowNLP8 9 10def run_app(ckip_nlp_models, cwn_upgrade) -> None:11    # need to download first because CWN packages will first check whether12    # there is .cwn_graph folder in the root directory.13    download_cwn_drivers(cwn_upgrade)14    dowload_ckip_package(ckip_nlp_models)15 16    from views.components.sidebar import visualize_side_bar17    from views.containers import display_cwn, display_ckip, display_data_form18 19    st.title("NLP app for PTT")20    st.write("這是一個針對PTT語料的 情緒分析｜中文NLP管線處理🔎")21    st.image("/Users/joannechi/nlpWeb/myApp/nlpweb/nlp_assignment_1/img/Mo-PTT-Logo.png", width=200)22 23    #menu = ["Text","Sentences"]24    #choice = st.sidebar.selectbox("Menu",menu) 25 26 27    #spectra = st.file_uploader("upload your file", type={"csv", "txt"})28    #if spectra is not None:29    #    spectra_df = pd.read_csv(spectra) #讀取csv30    #    st.write(spectra_df)31 32    #~~web crawler~~33    st.subheader("PTT Crawler 🐛")34    st.text('目前看板有：HatePolitics｜Gossiping｜Military｜Stock')35    selected = st.selectbox('請選擇看板：', 36      ['HatePolitics', 'Gossiping','Military','Stock']) 37    if selected=='HatePolitics':38        URL = "https://www.ptt.cc/bbs/HatePolitics/index.html"39    elif selected=='Gossiping':40        URL = "https://www.ptt.cc/bbs/Gossiping/index.html"41    elif selected=='Military':42        URL = "https://www.ptt.cc/bbs/Military/index.html"43    else:44        URL = "https://www.ptt.cc/bbs/Stock/index.html"45 46    my_headers = {'cookie': 'over18=1;'}47    response = requests.get(URL, headers = my_headers)48    soup = bs4.BeautifulSoup(response.text,"html.parser")49    list_results=[]50    for t in soup.find_all('div','title'):51        find_a=t.find('a')52        find_href="https://www.ptt.cc"+find_a.get("href")53        title=t.text54        results={55            "title":title,56            "url":find_href57        }58        list_results.append(results)        59    my_df=pd.DataFrame(list_results)60    print(my_df)61    st.write(my_df)62    #~~web crawler~~63 64    #~~sentiment analysis~~65    st.subheader("情緒分析")66    with st.form(key="nlpForm"):67        raw_text=st.text_area("請輸入句子✏️")68        submit_button=st.form_submit_button(label="確定")69        70    if submit_button:71        72        st.info("sentiment")73        sentiment=SnowNLP(SnowNLP(raw_text).han) #轉簡體74        sentiment_han=sentiment.sentiments75        st.write(sentiment_han)76 77            #emoji78        if sentiment_han>0:79            st.markdown("Sentiment:: Positive :smiley: ")80        elif sentiment_han<0:81            st.markdown("Sentiment:: Negative :angry: ")82        else:83            st.markdown("Sentiment:: Neutral :neutral: ")84 85 86 87        #with col2:88            #st.info("category")89            #category=SnowNLP(SnowNLP(raw_text).han) #轉簡體90            #category_han=list(category.tags)91            #st.write(category_han)92 93    #~~sentiment analysis~~94 95    st.subheader("中文 NLP 管線處理")96 97    input_data = display_data_form()98    model, pipeline, active_visualizers = visualize_side_bar(ckip_nlp_models) 99                                        #return model_options, pipeline_options, active_visualizers100 101    display_factories = {"CKIP": display_ckip, "CWN": display_cwn}102 103    if "input_data" in st.session_state:104        display_factories[pipeline](105            model, active_visualizers, st.session_state["input_data"]106        )107 108 109if __name__ == "__main__":110    ckip_nlp_models = ["bert-base", "albert-tiny", "bert-tiny", "albert-base"]111    run_app(ckip_nlp_models, cwn_upgrade=False)112