components.py

import streamlit as st
import pandas as pd
from connectors import *
import numpy as np
from processing import *
from util import *
import plotly.express as px


def hero_section():
    st.title('📗 lexiscore')
    st.markdown('A nutritional label for food for thought.')


def sidebar_section():
    st.sidebar.header('settings')
    st.session_state['conceptarium_url'] = st.sidebar.text_input(
        'conceptarium URL')
    st.session_state['access_token'] = st.sidebar.text_input(
        'access token', type='password')


def add_section(parent):
    parent.markdown('#### ➕ add food for thought')

    days = None
    folder = None

    type = parent.selectbox('What type of food for thought would you like to add?', (
        '📣 RSS (OPML)', '🔖 Bookmarks (HTML Export)', '📄 PDF', '📕 EPUB', '📝 Plain Text'))

    if type == '📣 RSS (OPML)':
        days = parent.number_input(
            'How many days does this stuff keep?', step=1)
        files = parent.file_uploader(
            'Place your food for thought here:', type=['xml', 'opml'])
    elif type == '🔖 Bookmarks (HTML Export)':
        folder = parent.text_input('How\'s the bookmark folder called?')
        days = parent.number_input(
            'How many days does this stuff keep?', step=1)
        files = parent.file_uploader(
            'Place your food for thought here:', type=['html'])
    elif type == '📄 PDF':
        files = parent.file_uploader(
            'Place your food for thought here:', accept_multiple_files=True, type='pdf')
    elif type == '📕 EPUB':
        files = parent.file_uploader(
            'Place your food for thought here:', accept_multiple_files=True, type='epub')
    elif type == '📝 Plain Text':
        files = parent.file_uploader(
            'Place your food for thought here:', accept_multiple_files=True, type=['text', 'txt', 'md'])

    parent.caption('')

    if not os.path.exists('tmp'):
        os.makedirs('tmp')

    if parent.button('add item'):
        if files == None:
            parent.warning('Please add file!')
        else:
            if not isinstance(files, list):
                files = [files]

            for file in files:
                f = open(os.path.join('tmp', file.name), 'wb+')
                f.write(file.getbuffer())
                f.close()

                path = os.path.join('tmp', file.name)

                with st.spinner('Fetching content...'):
                    if type == '📣 RSS (OPML)':
                        item_type = '📣 RSS'
                        data = fetch_from_opml(path, days)
                    elif type == '🔖 Bookmarks (HTML Export)':
                        item_type = '🔖 Bookmark'
                        data = fetch_from_bookmarks(path, folder, days)
                    elif type == '📄 PDF':
                        item_type = '📄 PDF'
                        data = fetch_from_pdf(path)
                    elif type == '📕 EPUB':
                        item_type = '📕 EPUB'
                        data = fetch_from_epub(path)
                    elif type == '📝 Plain Text':
                        item_type = '📝 Plain Text'
                        data = fetch_from_plaintext(path)

                for k, v in data.items():
                    if len(v[0].split()) / 250 > 1:
                        new_entry = pd.DataFrame([[item_type, k, round(len(v[0].split()) / 250), None, None, None, v[0], v[1], v[2]]], columns=[
                                                 'type', 'title', 'reading time', 'skill', 'challenge', 'lexiscore', 'text', 'raw', 'filename'])
                        st.session_state['data'] = st.session_state['data'].append(
                            new_entry, ignore_index=True)

            st.experimental_rerun()


def cart_section(parent):
    parent.markdown('#### 🛒 cart')
    parent.markdown('')
    parent.table(st.session_state['data'][[
                 'type', 'title', 'reading time', 'skill', 'challenge', 'lexiscore']])

    if st.session_state['data'].shape[0] > 0:
        with parent.expander('distribution'):
            fig = px.scatter(st.session_state['data'], x='skill', y='challenge', hover_data=[
                             'title', 'lexiscore'], color_discrete_sequence=['#228b22'])
            st.plotly_chart(fig)
            st.download_button(
                'download raw', st.session_state['data'].to_csv(), 'lexiscore.csv')
        if parent.button('start labeling'):
            for idx, row in st.session_state['data'].iterrows():
                if row['lexiscore'] is None:
                    with st.spinner('Determining the nutritional value of "' + row['title'] + '"...'):
                        content_paragraphs = get_paragraphs(row['text'])
                        content_embeddings = get_embeddings(content_paragraphs)

                        # print('---')
                        # print('\n\n'.join(content_paragraphs))

                        if len(content_paragraphs) > 1:
                            results = get_closest_thoughts(content_embeddings)
                            skill = get_skill(results)
                            challenge = get_challenge(
                                results, content_paragraphs)
                            raw_challenge = get_raw_challenge(
                                content_paragraphs)
                            challenge = -(raw_challenge -
                                          challenge) / raw_challenge

                            alpha = np.arctan2(
                                (challenge + 0.15), (skill - 0.375))
                            lexiscore = np.abs(alpha - 0.8) // 0.25

                            if lexiscore >= 4:
                                lexiscore = 'E'
                            elif lexiscore == 3:
                                lexiscore = 'D'
                            elif lexiscore == 2:
                                lexiscore = 'C'
                            elif lexiscore == 1:
                                lexiscore = 'B'
                            else:
                                lexiscore = 'A'

                            st.session_state['data'].loc[idx]['skill'] = skill
                            st.session_state['data'].loc[idx]['challenge'] = challenge
                            st.session_state['data'].loc[idx]['lexiscore'] = lexiscore
                        else:
                            print('no paragraphs:',
                                  row['title'], '---', row['text'], '---')

            st.experimental_rerun()


def meal_prep_section(parent):
    if st.session_state['data'].shape[0] > 0:
        parent.markdown('')
        parent.markdown('#### 🍱 meal prep')

        parent.image('assets/lexiscores.png')

        lexiscores = ['A', 'B', 'C', 'D', 'E']
        min_lexiscore = parent.select_slider(
            'Specify the minimum lexiscore to use for meal prep:', lexiscores)
        allowed_lexiscores = lexiscores[:lexiscores.index(min_lexiscore) + 1]
        include_predictions = parent.checkbox(
            'Include labels in output document?', True)

        if parent.button('start'):
            selection = [
                e in allowed_lexiscores for e in st.session_state['data']['lexiscore']]
            selection = st.session_state['data'][selection]

            if selection.shape[0] == 0:
                parent.warning(
                    'No item qualifies for this selection, please lower the bar!')
            else:
                html = '<h1>🍱 meal prep</h1><hr><div><br/><br/></div>'

                for idx, row in selection.iterrows():
                    if include_predictions:
                        html += '<img width="20%" src="file://' + \
                            os.path.abspath(
                                'assets/' + row['lexiscore'] + '.png') + '"><br/>'
                    html += '<h1>🥗 ' + row['title'] + '</h1>'
                    html += '<li><b>⏱️ ' + \
                        str(row['reading time']) + '</b> minutes</li>'
                    #html += '<li>📗 lexiscore <b>' + row['lexiscore'] + '</b></li>'
                    html += '<hr><div><br/><br/></div>'

                    if row['type'] != '📄 PDF':
                        html += row['raw']
                    else:
                        pix_paths = pdf_to_images(row['filename'])

                        for pix_path in pix_paths:
                            html += '<img src="file://' + \
                                os.path.abspath(pix_path) + '"><br/>'

                    html += '<div><br/><br/></div><hr>'

                # f = open(os.path.abspath('mealprep.html'), 'w+')
                # f.write(html)

                parent.info(
                    'Meal prep complete! Please use the button below to download the results.')
                parent.download_button(
                    label='download', data=html, file_name='mealprep.html', mime='text/html')


def footer_section():
    hide_streamlit_style = '''
                <style>
                #MainMenu {visibility: hidden;}
                footer {visibility: hidden;}
                </style>
                '''
    st.markdown(hide_streamlit_style, unsafe_allow_html=True)

    footer = '''
    ---
    <style>
    button {
        border: 4px solid;
        border-color: #228b22;
        border-radius: 4px;
        background-color: #228b22;
        color: #fffffd;
        font-weight: bold;
        padding-left: 5px;
        padding-right: 5px;
    }
    </style>
    <center>
        <div>
            <a href="https://paulbricman.com/contact"><button>send feedback</button></a>
            <a href="https://github.com/paulbricman/lexiscore"><button>learn more</button></a>
            <a href="https://github.com/sponsors/paulbricman"><button>support me 🤍</button></a>
        </div>
    </center>
    '''

    st.markdown(footer, unsafe_allow_html=True)


'''
>>> data['lexiscore'] = np.abs(data['alpha']-0.6) // (np.std(data['alpha'] / 2))
>>> px.scatter(data, x='skill', y='challenge', hover_data=['title', 'lexiscore']).show()
'''