pages/0_📙_Dictionary_(Search).py

import streamlit as st
import pandas as pd
import gdown
import os
import pickle
from collections import defaultdict, Counter

from PyDictionary import PyDictionary

dictionary = PyDictionary()

st.set_page_config(layout="wide", page_title="ACl23 Secret Language")

hide_expander_border = """
<style>
.st-bd {border-style: none;}
</style>
"""

# st.title("ACl23 Secret Language")

# sidebar
st.sidebar.header("📙 Dictionary")
_data = st.experimental_get_query_params()
default_title = 'Asian'
if _data:
    default_title = _data['word'][0]

title = st.sidebar.text_input(":red[Search secret languages given the following word (case-sensitive)]", default_title)

if ord(title[0]) in list(range(48, 57)):
    file_name = 'num_dict.pkl'
elif ord(title[0]) in list(range(97, 122)) + list(range(65, 90)):
    file_name = f'{ord(title[0])}_dict.pkl'
else:
    file_name = 'other_dict.pkl'

datas = pickle.load(open(f'all_secret_langauge_by_fist/{file_name}', 'rb'))
if title in datas:
    st.title(title)
    st.markdown(f"## {title}'s meaning in English[¹](#jump)")

    # write the meaning of input word
    try:
        title_mean = dictionary.meaning(title)
        _string = '>'
        for key in title_mean:
            _string += f':violet[{key}]: {";".join(title_mean[key])}<br>'
        st.markdown(_string, unsafe_allow_html=True)
    except:
        st.error(f'We cannot find the meaning of {title} in English (PyDictionary), which might be due to the bug.', icon="🚨")

    st.markdown(f"---")
    st.markdown(f"## {title}'s secret languages")
    data_title = datas[title]
    title_secret_languages = list(sorted(list(set(data_title["secret languages"]))))
    # dataframe = pd.DataFrame(datas[title])
    # st.markdown(f'### We found {len(set(dataframe.loc[:, "secret languages"]))} secret languages of {title}.', unsafe_allow_html=True)
    st.markdown(f'Overall, we found :red[{len(title_secret_languages)}] secret languages of :blue[{title}].', unsafe_allow_html=True)
    special = '"'
    # _title_secret_languages = [f'[{i}](#{i.strip().replace("(", ",,").replace(")", "..").replace("[", ",,,").replace("]", "...").replace(special, "././")})'
    #                            for i in title_secret_languages]
    # st.markdown('>' + ', '.join(_title_secret_languages).replace('<s>', '\<s\>').replace('$', '\$').replace('~', '\~'),
    #             unsafe_allow_html=True)

    secret_language_by_task = {
        'QA':[],
        'NLI':[],
        'paraphrase':[],
    }
    for i in range(len(data_title['secret languages'])):
        secret_language_by_task[data_title['tasks'][i]].append(data_title['secret languages'][i])
    for k in secret_language_by_task:
        secret_language_by_task[k] = list(set(secret_language_by_task[k]))

    def present_sl_task(secret_language_by_task, task):
        all_sl = sorted(secret_language_by_task[task])
        st.markdown(f':red[{len(all_sl)}] secret languages of :blue[{title}] on {task.replace("paraphrase", "Paraphrase")}', unsafe_allow_html=True)
        special = '"'
        _title_secret_languages = [f'[{i}](#{i.strip().replace("(", ",,").replace(")", "..").replace("[", ",,,").replace("]", "...").replace(special, "././")}_{task})'
                               for i in all_sl]
        st.markdown('>' + ', '.join(_title_secret_languages).replace('<s>', '\<s\>').replace('$', '\$').replace('~', '\~'),
                unsafe_allow_html=True)
    present_sl_task(secret_language_by_task, 'NLI')
    present_sl_task(secret_language_by_task, 'QA')
    present_sl_task(secret_language_by_task, 'paraphrase')

    st.markdown(f"*Hyperlinks only function when the corresponding tab is open. "
                f"For example, the hyperlinks in the paraphrase section will only work when the paraphrase tab is open.*")
    st.markdown(f"---")
    st.markdown(f"## Examples of replaced sentences")

    # st.text(','.join(title_secret_languages).replace('<s>', '\<s\>'))
    # st.dataframe(dataframe)
    _num = Counter(data_title['tasks'])
    tab1, tab2, tab3 = st.tabs([f'NLI ({_num["NLI"]})', f'QA ({_num["QA"]})', f'Paraphrase ({_num["paraphrase"]})'])


    def present_dataframe(dataframe, key, title):
        new_dataframe = dataframe.loc[dataframe['tasks'] == key].reset_index()
        new_dataframe['replaced sentences'] = new_dataframe['replaced sentences'].str.replace('<s>', '[POS]')
        if len(new_dataframe):
            new_dataframe = new_dataframe.drop(columns=['tasks', 'index'])
            # st.markdown(new_dataframe.columns)
            for i in range(len(new_dataframe)):
                _title = f'{i + 1}\. **[{new_dataframe.loc[i, "secret languages"]}]**'
                with st.expander(_title):
                    # _string = f'{i + 1}. :red[{new_dataframe.loc[i, "secret languages"]}]'
                    _string = 'Original '
                    if key == 'NLI':
                        _string += 'hypothesis: :'
                    elif key == 'QA':
                        _string += 'question: :'
                    elif key == 'Paraphrase':
                        _string += 'sentence 1: :'
                    _string += f'blue[{new_dataframe.loc[i, "original sentences"]}]'.replace(":", "[colon]")
                    _string += '<br>Replaced '
                    if key == 'NLI':
                        _string += 'hypothesis: :'
                    elif key == 'QA':
                        _string += 'question: :'
                    elif key == 'Paraphrase':
                        _string += 'sentence 1: :'
                    _string += f'red[{new_dataframe.loc[i, "replaced sentences"]}]'.replace(":", "[colon]")
                    if key == 'NLI':
                        _string += '<br>premise: :'
                    elif key == 'QA':
                        _string += '<br>text: :'
                    elif key == 'Paraphrase':
                        _string += '<br>sentence 2: :'
                    _string += f'blue[{new_dataframe.loc[i, "premise / sentence 2 / text"]}]'.replace(":", "[colon]")
                    st.markdown(_string, unsafe_allow_html=True)
                    # st.text(f'Examples: :blue[{new_dataframe.loc[i, "replaced sentences".replace(":", "[colon]")]}]')
            # st.dataframe(new_dataframe)
            st.markdown(hide_expander_border, unsafe_allow_html=True)
        else:
            st.error(f'We did not find any Secret Language of {title} on {key}.')


    def present_dict(_dict, task):
        # st.text(set(_dict['tasks']))
        _all = defaultdict(int)
        for i in range(len(_dict['secret languages'])):
            if _dict['tasks'][i] == task:
                _sl = _dict['secret languages'][i]
                if type(_all[_sl]) == int:
                    _all[_sl] = {
                        'Original hypothesis': [],
                        'Replaced hypothesis': [],
                        'Premise': []
                    }
                _all[_sl]['Original hypothesis'].append(_dict['original sentences'][i])
                if task == 'QA':
                    _all[_sl]['Replaced hypothesis'].append(_dict['replaced sentences'][i].replace('<s>', ''))
                else:
                    _all[_sl]['Replaced hypothesis'].append(_dict['replaced sentences'][i])
                _all[_sl]['Premise'].append(_dict['premise / sentence 2 / text'][i])
        if len(_all.keys()):
            all_keys = sorted(list(_all.keys()))
            for i in range(len(all_keys)):
                _sl = all_keys[i]
                _sl_in_span = _sl.strip().replace("(", ",,").replace(")", "..").replace("[", ",,,").replace("]", "...").replace(special, "././")
                _title = f'{i + 1}. <span id="{_sl_in_span}_{task}"> **:red[{_sl}]**</span>'
                # with st.expander(_title, expanded=True):
                _string = _title + '<br>Examples:<br>'
                # st.markdown(_title, unsafe_allow_html=True)
                # st.markdown(f'Examples:', unsafe_allow_html=True)
                _string += '<blockquote><ol>'
                for j in range(len(_all[_sl]['Original hypothesis'])):
                    # _string += f'{j+1}. Original '
                    _string += f'<li> **Original '
                    if task == 'NLI':
                        _string += 'hypothesis**: :'
                    elif task == 'QA':
                        _string += 'question**: :'
                    elif task == 'paraphrase':
                        _string += 'sentence 1**: :'
                    _string += f'blue[{_all[_sl]["Original hypothesis"][j]}]'.replace(":", "[colon]")
                    _string += '<br> **Replaced '
                    if task == 'NLI':
                        _string += 'hypothesis**: :'
                    elif task == 'QA':
                        _string += 'question**: :'
                    elif task == 'paraphrase':
                        _string += 'sentence 1**: :'
                    _string += f'red[{_all[_sl]["Replaced hypothesis"][j]}]'.replace(":", "[colon]")
                    if task == 'NLI':
                        _string += '<br> **premise**: :'
                    elif task == 'QA':
                        _string += '<br> **text**: :'
                    elif task == 'paraphrase':
                        _string += '<br> **sentence 2**: :'
                    _string += f'blue[{_all[_sl]["Premise"][j]}]'.replace(":", "[colon]")
                    _string += '<br></li>'
                _string += '</ol></blockquote>'
                st.markdown(_string.replace('<s>', '\<s\>').replace('$', '\$').replace('~', '\~'), unsafe_allow_html=True)
                # st.text(f'Examples: :blue[{new_dataframe.loc[i, "replaced sentences".replace(":", "[colon]")]}]')
            # st.dataframe(new_dataframe)
            st.markdown(hide_expander_border, unsafe_allow_html=True)
        else:
            st.error(f'We did not find any Secret Language of {title} on {task}.', icon="⚠️")


    with tab1:
        # st.header("NLI")
        # present(dataframe, 'NLI', title)
        present_dict(data_title, 'NLI')
    with tab2:
        # st.header("QA")
        # present(dataframe, 'QA', title)
        present_dict(data_title, 'QA')
    with tab3:
        # present(dataframe, 'Paraphrase', title)
        present_dict(data_title, 'paraphrase')
    st.markdown(
        f'<span id="jump">¹</span>*Enlish meaning is supported by [PyDictionary](https://pypi.org/project/PyDictionary/).*',
        unsafe_allow_html=True)
else:
    st.error(f'{title} is not in the dictionary of Secret Language.', icon="⚠️")