|
|
|
import glob, os, sys;
|
|
sys.path.append('../utils')
|
|
|
|
|
|
import seaborn as sns
|
|
import matplotlib.pyplot as plt
|
|
import numpy as np
|
|
import pandas as pd
|
|
import streamlit as st
|
|
from utils.netzero_classifier import load_netzeroClassifier, netzero_classification
|
|
import logging
|
|
logger = logging.getLogger(__name__)
|
|
from utils.config import get_classifier_params
|
|
from io import BytesIO
|
|
import xlsxwriter
|
|
import plotly.express as px
|
|
|
|
|
|
|
|
classifier_identifier = 'netzero'
|
|
params = get_classifier_params(classifier_identifier)
|
|
|
|
|
|
_lab_dict = {
|
|
'NEGATIVE':'NO NETZERO TARGET',
|
|
'NA':'NOT APPLICABLE',
|
|
'NETZERO':'NETZERO TARGET',
|
|
}
|
|
|
|
|
|
@st.cache_data
|
|
def to_excel(df):
|
|
len_df = len(df)
|
|
output = BytesIO()
|
|
writer = pd.ExcelWriter(output, engine='xlsxwriter')
|
|
df.to_excel(writer, index=False, sheet_name='Sheet1')
|
|
workbook = writer.book
|
|
worksheet = writer.sheets['Sheet1']
|
|
worksheet.data_validation('E2:E{}'.format(len_df),
|
|
{'validate': 'list',
|
|
'source': ['No', 'Yes', 'Discard']})
|
|
writer.save()
|
|
processed_data = output.getvalue()
|
|
return processed_data
|
|
|
|
def app():
|
|
|
|
with st.container():
|
|
if 'key1' in st.session_state:
|
|
df = st.session_state.key1
|
|
|
|
|
|
classifier = load_netzeroClassifier(classifier_name=params['model_name'])
|
|
st.session_state['{}_classifier'.format(classifier_identifier)] = classifier
|
|
|
|
if sum(df['Target Label'] == 'TARGET') > 100:
|
|
warning_msg = ": This might take sometime, please sit back and relax."
|
|
else:
|
|
warning_msg = ""
|
|
|
|
df = netzero_classification(haystack_doc=df,
|
|
threshold= params['threshold'])
|
|
st.session_state.key1 = df
|
|
|
|
|
|
def netzero_display():
|
|
if 'key1' in st.session_state:
|
|
df = st.session_state.key2
|
|
hits = df[df['Netzero Label'] == 'NETZERO']
|
|
range_val = min(5,len(hits))
|
|
if range_val !=0:
|
|
count_df = df['Netzero Label'].value_counts()
|
|
count_df = count_df.rename('count')
|
|
count_df = count_df.rename_axis('Netzero Label').reset_index()
|
|
count_df['Label_def'] = count_df['Netzero Label'].apply(lambda x: _lab_dict[x])
|
|
|
|
fig = px.bar(count_df, y="Label_def", x="count", orientation='h', height =200)
|
|
c1, c2 = st.columns([1,1])
|
|
with c1:
|
|
st.plotly_chart(fig,use_container_width= True)
|
|
|
|
hits = hits.sort_values(by=['Netzero Score'], ascending=False)
|
|
st.write("")
|
|
st.markdown("###### Top few NetZero Target Classified paragraph/text results ######")
|
|
range_val = min(5,len(hits))
|
|
for i in range(range_val):
|
|
|
|
|
|
st.write('**Result {}** `page {}` (Relevancy Score: {:.2f})'.format(i+1,hits.iloc[i]['page'],hits.iloc[i]['Netzero Score']))
|
|
st.write("\t Text: \t{}".format(hits.iloc[i]['text']))
|
|
else:
|
|
st.info("π€ No Netzero target found")
|
|
|
|
|
|
|