wine_survey.py

import streamlit as st
import pandas as pd
from st_aggrid import AgGrid, GridUpdateMode
from st_aggrid.grid_options_builder import GridOptionsBuilder
import pickle
from st_on_hover_tabs import on_hover_tabs
import collections


st.set_page_config(layout="wide")

merchant = 'winemag'

# Load Dataframe
def load_wine_dataframe(merchant):
    path_to_file = merchant+'_wine_dataframe_varieties_regions.txt'
    with open(path_to_file,'rb') as file_object:
        raw_data = file_object.read()
    wine_dataframe = pickle.loads(raw_data)
    return wine_dataframe

def wine_selector():
    wine_dataframe = load_wine_dataframe(merchant)
    wd = wine_dataframe[['country','region_1','province','variety','title','normalized_descriptors']]


    by_country = wd.groupby(['country'])['title'].count().sort_values(ascending=False).reset_index()
    df_c = pd.DataFrame(by_country)
    gd = GridOptionsBuilder.from_dataframe(df_c)
    gd.configure_selection(selection_mode='multiple', use_checkbox=True)
    gridoptions = gd.build()
    grid_table_c = AgGrid(df_c, height=400, width=400, gridOptions=gridoptions,fit_columns_on_grid_load=True,
                       update_mode=GridUpdateMode.SELECTION_CHANGED)

    selected_row = grid_table_c["selected_rows"]
    if selected_row:
        df_result = pd.DataFrame(selected_row)
        print(df_result.head())
        st.dataframe(df_result['country'])

        filtered_by_country = wd.loc[(wd['country'].isin(list(df_result['country'])))][['country','province','title','variety','normalized_descriptors']]
        country_list = filtered_by_country.groupby(['country','province'])['title'].count().sort_values(ascending=False).reset_index()

        gd = GridOptionsBuilder.from_dataframe(country_list)
        gd.configure_selection(selection_mode='multiple', use_checkbox=True)
        gridoptions = gd.build()
        st.write("### Now choose some provinces from selected countries")
        grid_table_r = AgGrid(country_list, height=400, width=400, gridOptions=gridoptions,fit_columns_on_grid_load=True,
                           update_mode=GridUpdateMode.SELECTION_CHANGED)
        
        selected_row_province = grid_table_r["selected_rows"]
        if selected_row_province:
            df_result_p = pd.DataFrame(selected_row_province)
            st.dataframe(df_result_p['province'])

            filtered_by_province = filtered_by_country.loc[(filtered_by_country['province'].isin(list(df_result_p['province'])))][['country','province','title','variety','normalized_descriptors']]
            province_list = filtered_by_province.groupby(['province','variety'])['title'].count().sort_values(ascending=False).reset_index()

            gd = GridOptionsBuilder.from_dataframe(province_list)
            gd.configure_selection(selection_mode='multiple', use_checkbox=True)
            gridoptions = gd.build()
            st.write("### Now select the varieties you preffer most:")
            grid_table_v = AgGrid(province_list, height=400, width=400, gridOptions=gridoptions,fit_columns_on_grid_load=True,
                            update_mode=GridUpdateMode.SELECTION_CHANGED)            

            selected_row_variety = grid_table_v["selected_rows"]
            df_result_v = pd.DataFrame(selected_row_variety)

            filtered_by_variety  = filtered_by_province.loc[(filtered_by_province['variety'].isin(list(df_result_v['variety'])))][['country','province','title','variety','normalized_descriptors']]

            descriptor_mapping = pd.DataFrame()
            descriptor_mapping = pd.read_csv('descriptor_mapping.csv')
            level3 = descriptor_mapping['level_3'].to_list()


            desc_common = pd.DataFrame()
            for index, row in df_result_v.iterrows():
                # TODO: add region filter
                query = filtered_by_variety.loc[(filtered_by_variety['variety'].isin([row['variety']]))][['normalized_descriptors']]
                result_list = []
                for d_list in query['normalized_descriptors']:
                    result_list = result_list + d_list

                res_list = [tag for tag in result_list if tag in level3 ]
                counter = collections.Counter(res_list)                    
                common_desc = str([(i, round((counter[i] / len(res_list) * 100.0),2)) for i, count in counter.most_common(20)])

                dict2 = {'province': [row['province']], 'variety': [row['variety']], 'common_desc': [common_desc]}
                df2 = pd.DataFrame(dict2)
                desc_common = pd.concat([desc_common,df2])



            # st.dataframe(grid_table_v["selected_rows"])
            st.write("## Tasting notes")
            st.dataframe(desc_common,use_container_width=True)

### UI
st.markdown('<style>' + open('./style.css').read() + '</style>', unsafe_allow_html=True)

with st.sidebar:
    tabs = on_hover_tabs(tabName=['My Profile', 'Suggested For Me', 'Explore'], 
                         iconName=['dashboard', 'money', 'economy'], default_choice=0)

if tabs =='My Profile':
    st.title("My Profile")
    st.header("The wine from what countries do you like?")
    wine_selector()

elif tabs == 'Suggested For Me':
    st.title("Paper")
    st.write('Name of option is {}'.format(tabs))

elif tabs == 'Explore':
    st.title("Tom")
    st.write('Name of option is {}'.format(tabs))