eu_parl/streamlit_app_talking_ep.py

import difflib
import re
import traceback
from datetime import datetime

import altair as alt
import matplotlib.pyplot as plt
import pandas as pd
import streamlit as st


from arango_things import arango_db
from things import normalize_party_names
from streamlit_info import (
    party_colors,
    explainer,
    limit_warning,
    party_colors_lighten,
    css,
    results_limit,
    parties,
    summary_note,
)

from langchain.schema import Document


class Params:
    """Class containing parameters for URL.

    Attributes:
        params (dict): A dictionary containing the parameters.
        q (str): The search query.
        parties (list): A list of political parties.
        persons (list): A list of persons.
        from_year (int): The start year for the search.
        to_year (int): The end year for the search.
        debates (list): A list of debates.
        excerpt_page (int): The page number for the excerpt.
    """

    def __init__(self, params):
        """Initializes the Params class.

        Args:
            params (dict): A dictionary containing the parameters.
        """
        self.params = params
        # Set parameters.
        self.q = self.set_param("q")
        self.parties = self.set_param("parties")
        self.persons = self.set_param("persons")
        self.from_year = self.set_param("from_year")
        self.to_year = self.set_param("to_year")
        self.debates = self.set_param("debates")
        self.excerpt_page = self.set_param("excerpt_page")

    def set_param(self, key):
        """Returns the value of a parameter if it exists.

        Args:
            key (str): The key of the parameter.

        Returns:
            The value of the parameter if it exists, otherwise a default value.
        """
        if key in self.params:
            if key in ["parties", "persons", "debates"]:
                value = self.params[key][0].split(",")
            else:
                value = self.params[key][0]

        else:
            value = []
            if key == "q":
                value = ""
            elif key == "from_year":
                value = 2019  # Catch all.
            elif key == "to_year":
                value = 2023  # Catch all.
            elif key == "excerpt_page":
                value = 0
        return value

    def update(self):
        """Updates the URL parameters."""
        st.experimental_set_query_params(
            q=self.q,
            from_year=self.from_year,
            to_year=self.to_year,
            parties=",".join(self.parties),
            debates=",".join(self.debates),
            persons=",".join(self.persons),
            excerpt_page=self.excerpt_page,
        )

    def update_param(self, key, value):
        """Updates a single parameter.

        Args:
            key (str): The key of the parameter to update.
            value (str or int or list): The new value of the parameter.
        """
        self.params[key] = value
        self.update()

    def reset(self, q=False):
        """Resets all parameters to their default values.

        Args:
            q (str): The new search query (optional).
        """
        for key in self.params:
            self.params[key] = []
        if q:
            self.q = q


@st.cache_data(show_spinner=False)
def summarize(df_party, user_input):
    """
    Summarizes the standpoints of a political party in the EU parliament based on a user search.

    Args:
        df_party (pandas.DataFrame): A DataFrame containing speeches by parliamentarians from the political party in the EU parliament.
        user_input (str): The user's search query.

    Returns:
        str: A short summary of the standpoints of the political party in the EU parliament, in relation to the user search.
    """

    texts = []

    documents = []
    for _, row in df_party.iterrows():
        documents.append(
            Document(
                page_content=f"**{row['Speaker']} {row['Date']}**\n{row['Summary']}",
                metadata=row.to_dict(),
            )
        )

    # https://python.langchain.com/docs/use_cases/question_answering/qa_citations

    # Get the 12 most relevant speeches (relevance defined by BM25 in the arango query).
    if df_party.shape[0] > 10:
        df_party = df_party.sort_values(by="relevance_score", ascending=False)[:10]

    if df_party.shape[0] > 5:
        # Sample 6 rows with weights proportional to the year counts
        sample_size = 5
        weights = df_party["Year"].value_counts(normalize=True)
        df_party = df_party.groupby("Year", group_keys=False).apply(
            lambda x: x.sample(n=int(sample_size * weights[x.name]))
        )

    # Sort the dataframe by date.
    df_party.sort_values(by="Date", ascending=False, inplace=True)
    # Loop through the speeches and add its text to the list.
    for _, row in df_party.sort_values(by="Date", ascending=False)[:10].iterrows():
        texts.append(f"**{row['Speaker']} {row['Date']}**\n{row['Summary']}")
    # Join the texts into a single string.
    # Create a prompt for the user to summarize the party's standpoints.
    prompt = re.sub(
        r"\W\W\W+",
        "\n",
        f"""
    A user have made this search in a database: <{user_input}>
    
    That resulted in the lists below, consisting of speeches by parliamentarians from the {party} party in the EU parliament:

    {texts}

    Please make a very short summary of the standpoints of the {party} party in the EU parliament, in relation to the user search. Make note of the speaker and the dates for the speeches.
    
    Example 1:
    In 2020, Parliamentarian NN, ({party}) wanted to decrease the budget for the EU parliament. Later, 2022, she wanted to incrase the budget.
    
    Example 2:
    Parliamentarians from {party} have been very active in the debate about the EU budget. They have been both for and against the budget.
    
    Short summary: 
    """,
    )

    # Generate a summary of the party's standpoints.
    # Return the summaries dictionary.
    system_prompt = "You are a journalist and have been asked to write a short summary of the standpoints of the party in the EU parliament."

    return ollama(
        prompt=prompt, system_prompt=system_prompt, temperature=0.5, print_tokens=False
    )


def make_snippet(text, input_tokens, token_text_list, token_input_list):
    """
    Find the word searched for and give it some context.

    Args:
        text (str): The text to search for the input tokens.
        input_tokens (str): The input tokens to search for in the text.
        token_text_list (list): A list of tokens in the text.
        token_input_list (list): A list of input tokens.

    Returns:
        str: A snippet of text containing the input tokens and some context.
    """
    # If the input token is "speaker", return the first 300 characters of the text.
    if input_tokens == "speaker":
        snippet = str(text[:300])
        if len(text) > 300:
            snippet += "..."
    else:
        snippet = []
        text_lower = text.lower()
        # Calculate snippet length in words.
        snippet_length = 40 * int(8 / len(input_tokens))  # * Change to another value?

        # Loop through each input token.
        for token in input_tokens:
            # Remove any non-alphanumeric characters and convert to lowercase.
            token = re.sub(r"[^a-zA-Z0-9\s]", "", token).lower()
            if token in text_lower:
                # Find the position of the token in the text.
                position = text_lower.find(token)
                position_start = position - snippet_length
                # Find the start of the snippet by looking for the previous space character.
                while True and position_start > 0:
                    position_start -= 1
                    if text_lower[position_start] == " ":
                        break

                position_end = position + int(snippet_length / 2)
                # Find the end of the snippet by looking for the next space character.
                while True and position_end > 0 and position_end < len(text_lower) - 1:
                    position_end += 1
                    if text_lower[position_end] == " ":
                        break

                text_before = "".join(text[position_start:position])

                if position_end > len(text):
                    position_end = len(text)
                text_after = "".join(text[position + len(token) + 1 : position_end])
                word_context = f"{text_before} <b>{token}</b> {text_after}"
                snippet.append(word_context)

            if len(snippet) == 2:
                break

        if snippet == [""]:
            # If the input token is not found in the text, find the closest match in the token list.
            text_list = text.split(" ")
            for token in token_input_list:
                token = "".join(token)
                if token in token_text_list:
                    position = token_text_list.index(token)
                    w = 20
                    # Get a list of words around the token, and find the closest match to the token in that list.
                    list_from_text = text_list[position - w : position + w]
                    closest_match = difflib.get_close_matches(
                        token, list_from_text, n=1
                    )
                    # If no match is found, expand the list of words and try again.
                    if not closest_match:
                        w = 50
                        list_from_text = text_list[position - w : position + w]
                        closest_match = difflib.get_close_matches(
                            token, list_from_text, n=1
                        )
                    if closest_match:
                        word = closest_match[0]
                        # Calculate the difference between the position of the token and the position of the closest match.
                        diff_token_to_word = list_from_text.index(word) - w
                        position = position + diff_token_to_word
                    else:
                        word = text_list[position]
                    # Get the context of the word, including the words before and after it.
                    position_start = position - snippet_length
                    if position_start < 0:
                        position_start = 0
                    position_end = position + int(snippet_length / 2)
                    text_before = " ".join(text_list[position_start:position])
                    if position_end > len(text_list):
                        position_end = len(text_list)
                    text_after = " ".join(text_list[position + 1 : position_end])
                    word_context = f"{text_before} <b>{word}</b> {text_after}"
                    snippet.append(word_context)

        snippet = "|".join(snippet)
        snippet = f"...{snippet}..."
    return snippet


def build_style_parties(parties):
    """Build a CSS styl for party names buttons."""
    style = "<style> "
    for party in parties:
        style += f' span[data-baseweb="tag"][aria-label="{party}, close by backspace"]{{ background-color: {party_colors[party]}}} .st-eg {{min-width: 14px;}} '  # max-width: 328px;
    style += "</style>"
    return style


def build_style_mps(mps):
    """Build a CSS style for party names buttons."""
    style = "<style> "
    for party in mps:
        party = fix_party(party)
        try:
            style += f' span[data-baseweb="tag"][aria-label="{party}, close by backspace"]{{ background-color: {party_colors[party]};}} .st-eg {{min-width: 14px;}} '  # max-width: 328px;
        except KeyError:
            style += f' span[data-baseweb="tag"][aria-label="{party}, close by backspace"]{{ background-color: {party_colors["NA"]};}} .st-eg {{min-width: 14px;}} '
    style += "</style>"
    return style


def fix_party(party):
    """Replace old party codes with new ones."""
    party = party.upper().replace("KDS", "KD").replace("FP", "L")
    return party


def build_style_debate_types(debates):
    """Build a CSS style for debate type buttons."""
    style = "<style> "

    for debate in debates:
        style += f' span[data-baseweb="tag"][aria-label="{debate}, close by backspace"]{{ background-color: #767676;}} .st-eg {{min-width: 14px;}}'  # max-width: 328px;
    style += "</style>"
    return style


def highlight_cells(party):
    if party in party_colors.keys():
        color = party_colors[party]
        return f"background-color: {color}; font-weight: 'bold'"


@st.cache_data
def options_persons(df):
    d = {}
    for i in df.groupby("Speaker"):
        d[i[0]] = i[1].shape[0]
    return [f"{key} - {value}" for key, value in d.items()]


@st.cache_data
def get_data(aql, input_tokens):
    """Get data from SQL database.

    Args:
        sql (str): A SQL query string.

    Returns:
        DataFrame: Dataframe with some adjustments to the data fetched from the DB.
    """

    cursor = arango_db.aql.execute(aql, count=True)

    if cursor.__len__() == 0:
        df = None
        status = "no hits"
    else:
        if cursor.__len__() == results_limit:
            status = "limit"
        else:
            status = "ok"
        df = pd.DataFrame([doc for doc in cursor])
        df.mep_id = df.mep_id.astype(int)
        df["Year"] = df["Date"].apply(lambda x: int(x[:4]))
        # df.drop_duplicates(ignore_index=True, inplace=True)
        df["Party"] = df["Party"].apply(lambda x: normalize_party_names(x))

        df.sort_values(["Date", "number"], axis=0, ascending=True, inplace=True)

        df["len_translation"] = df["Text"].apply(lambda x: len(x.split(" ")))
        df["len_tokens"] = df["tokens"].apply(lambda x: len(x))
        df.sort_values(["Date", "speech_id", "number"], axis=0, inplace=True)

    return df, status


def user_input_to_db(user_input):
    """Writes user input to db for debugging."""
    arango_db.collection("streamlit_user_input").insert(
        {"timestamp": datetime.timestamp(datetime.now()), "input": user_input}
    )


def create_aql_query(input_tokens):
    """Returns a valid sql query."""
    # Split input into tokens

    # Build AQL query
    aql_query = "FOR doc IN speeches_view\nSEARCH "

    # Add SEARCH PHRASE clause
    search_phrase = next((token for token in input_tokens if " " in token), None)
    if search_phrase:
        aql_query += f'PHRASE(doc.translation, {search_phrase}, "text_en")\n'
        input_tokens.remove(search_phrase)

    # Add ANALYZER clause for each token
    for token in input_tokens:
        if token.startswith("-"):
            if "*" in token:
                aql_query += (
                    f'AND NOT LIKE(doc.translation, "{token[1:].replace("*", "%")}")\n'
                )
            else:
                aql_query += f'AND NOT ANALYZER(TOKENS("{token[1:]}", "text_en") ANY == doc.translation, "text_en")\n'
        elif "*" in token:
            aql_query += f'AND LIKE(doc.translation, "{token.replace("*", "%")}")\n'
        else:
            aql_query += f'AND ANALYZER(TOKENS("{token}", "text_en") ANY == doc.translation, "text_en")\n'

    # Add fields for relevance and tokens.

    str_tokens = str([re.sub(r"[^a-zA-Z0-9\s]", "", token) for token in input_tokens])
    select_fields = f"""
                {{
                "speech_id": doc._key,
                "Text": doc.translation,
                "number": doc.speech_number,
                "debatetype": doc.debate_type,
                "Speaker": doc.name,
                "Date": doc.date,
                "url": doc.url, 
                "Party": doc.party,
                "mep_id": doc.mep_id,
                "Summary": doc.summary,
                "debate_id": doc.debate_id,
                'relevance_score': BM25(doc),
                'tokens': TOKENS(doc.translation, 'text_en'),
                'input_tokens': TOKENS({str_tokens}, 'text_en')}}
                """

    # Add LIMIT and RETURN clause
    aql_query += f"""
    FILTER char_length(doc.summary) > 10
    LIMIT {results_limit}
    SORT BM25(doc) DESC
    RETURN {select_fields}
    """

    return aql_query.replace("SEARCH AND", "SEARCH ")


def error2db(error, user_input, engine):
    """Write error to DB for debugging."""
    doc = (
        {
            "error": error,
            "time": datetime.date(datetime.now()),
            "user_input": str(user_input),
        },
    )
    arango_db.collection("errors").inser(doc)


@st.cache_data
def get_speakers():
    """Get all meps."""
    df = pd.DataFrame(
        [
            i
            for i in arango_db.aql.execute(
                "for doc in meps return {'name': doc.name, 'mep_id': doc._key, mep_url: doc.mep_URI}"
            )
        ]
    )
    df.mep_id = df.mep_id.astype(int)
    return df


def search_person(user_input, list_persons):
    """Returns SQL query made for searching everything a defined speaker has said.

    Args:
        user_input (str): The string resulting from user input (input()).

    Returns:
        list: List of search terms.
    """
    # List all alternatives.
    options = [i for i in list_persons if i.lower() == user_input.lower()]
    options = [f"Yes, search for {i}" for i in options]
    no_option = f"No, I want to search for what have been said about {user_input}."
    options += [no_option, "Choose an alternative"]
    preselected_option = len(options) - 1
    # Let the user select a person or no_alternative.
    speaker = st.selectbox(
        ":red[Do you want to search for what a specific parliamentarian has said?]",
        options,
        index=preselected_option,
    )

    if speaker == "Choose an alternative":
        st.stop()
    if speaker == no_option:
        # Return "normal" query if no_alternative.
        aql = create_aql_query(user_input)
    else:
        speaker = speaker.replace("Yes, search for ", "")
        aql = f"FOR doc IN meps FILTER doc.name == '{speaker}' RETURN doc"
    return aql


def show_excerpts():
    """
    Displays excerpts from a dataframe in a Streamlit app.

    The function uses the global variables `df`, `input_tokens`, `party_colors_lighten`, and `excerpts_container`.
    It also uses the session state variables `excerpt_page`, `df_excerpts`, `disable_next_page_button`, and `text_next_page_button`.

    The function displays a specified number of excerpts per page, and allows the user to navigate between pages.
    Each excerpt includes the speaker, date, party, and a snippet of text.
    The user can click a button to view the full text of an excerpt.

    Returns:
        None
    """

    results_per_page = 10  # Number of excerpts to display per page

    with excerpts_container:
        n = results_per_page * int(st.session_state["excerpt_page"])

        # Make snippets from the text field (short and long).
        df_excepts = df.iloc[
            int(st.session_state["excerpt_page"])
            * results_per_page : int(st.session_state["excerpt_page"])
            * results_per_page
            + results_per_page
        ].copy()

        df_excepts["Excerpt"] = df_excepts.apply(
            lambda x: make_snippet(
                x["Text"], input_tokens, x["tokens"], x["input_tokens"]
            ),
            axis=1,
        )

        if "df_excerpts" in st.session_state:
            df_excepts = pd.concat([st.session_state["df_excerpts"], df_excepts])
            st.session_state["df_excerpts"] = df_excepts
        else:
            st.session_state["df_excerpts"] = df_excepts

        if len(df_excepts) == len(df):
            st.session_state["disable_next_page_button"] = True
            st.session_state["text_next_page_button"] = "No more results"

        new_debate = True
        debate_id = None

        for _, row in df_excepts.iterrows():
            n += 1
            # Find out if it's a new debate.
            if row["debate_id"] == debate_id:
                new_debate = False
            else:
                new_debate = True
            debate_id = row["debate_id"]

            # TODO Clean names from titles etc.
            # Write to table.

            if new_debate:
                # st.write("---", unsafe_allow_html=True)
                st.markdown(
                    f""" <span style="font-weight: bold;">{row['Date']}</span> """,
                    unsafe_allow_html=True,
                )
            col1, col2, col3 = st.columns([2, 7, 2])
            with col1:
                if row["mep_id"] in meps_ids:
                    st.markdown(
                        f"""<a href="https://www.europarl.europa.eu/meps/en/{row["mep_id"]}" target="_blank">{row['Speaker']}</a>""",
                        unsafe_allow_html=True,
                    )
                else:
                    st.markdown(row["Speaker"])

            with col2:
                snippet = (
                    row["Excerpt"]
                    .replace(":", "\:")
                    .replace("<p>", "")
                    .replace("</p>", "")
                )

                st.markdown(
                    f""" <span style="background-color:{party_colors_lighten[row['Party']]}; color:black;">{snippet}</span> """,
                    unsafe_allow_html=True,
                )
            with col3:
                full_text = st.button("Full text", key=row["speech_id"])
                if full_text:
                    with st.sidebar:
                        url_person = (
                            f'https://www.europarl.europa.eu/meps/en/{row["mep_id"]}'
                        )

                        st.markdown(
                            f""" <span class="{row['Party']}" style="font-weight: bold;">[ {row['Speaker']} ]({url_person})</span> """,
                            unsafe_allow_html=True,
                        )
                        st.markdown(
                            f""" <span style="font-style: italic;">{row["Date"]}</span> """,
                            unsafe_allow_html=True,
                        )
                        st.write(row["Text"].replace(":", "\:"), unsafe_allow_html=True)

                        st.markdown(f"📝 [Read the protocol]({row['url']})")


###* PAGE LOADS *###
# Title and explainer for streamlit
st.set_page_config(
    page_title="EP Debates",
    # page_icon="favicon.png",
    initial_sidebar_state="auto",
)
title = "What are they talking about in the EU?"
st.title(title)
st.markdown(css, unsafe_allow_html=True)
# Get params from url.
params = Params(st.experimental_get_query_params())
params = Params({})

# The official colors of the parties
partycodes = list(party_colors.keys())  # List of partycodes

# Max hits returned by db.
return_limit = 10000

# Ask for word to search for.
user_input = st.text_input(
    " ",
    value=params.q,
    placeholder="Search something",
    # label_visibility="hidden",
    help='You can use asterix (*), minus (-), quotationmarks ("") and OR.',
)


if len(user_input) > 3:
    params.q = user_input

    # print(user_input.upper())
    # print(ollama(prompt=f'''A user wants to search in a database containing debates in the European Parliament and have made the input below. Take that input and write three questions would generate a good result if used for quering a vector database. Answer with a python style list containing the three questions.

    # User input: {user_input}

    # Questions: '''))

    try:  #! When in procution, uncomment this.
        user_input = user_input.replace("'", '"')
        input_tokens = re.findall(r'(?:"[^"]*"|\S)+', user_input)

        # Put user input in session state (first run).
        if "user_input" not in st.session_state:
            st.session_state["user_input"] = user_input
            user_input_to_db(user_input)

        else:
            if st.session_state["user_input"] != user_input:
                # Write user input to DB.
                st.session_state["user_input"] = user_input
                user_input_to_db(user_input)
                # Reset url parameters.
                params.reset(q=user_input)

        params.update()

        meps_df = get_speakers()
        meps_ids = meps_df.mep_id.to_list()

        # Check if user has searched for a specific politician.
        if len(user_input.split(" ")) in [
            2,
            3,
            4,
        ]:  # TODO Better way of telling if name?
            list_persons = meps_df["name"].to_list()
            if user_input.lower() in list_persons:
                sql = search_person(user_input, list_persons)
                search_terms = "speaker"

        aql = create_aql_query(input_tokens)

        ## Fetch data from DB.
        df, status = get_data(aql, input_tokens)

        if status == "no hits":  # If no hits.
            st.write("No hits. Try again!")
            st.stop()
        elif status == "limit":
            st.write(limit_warning)
            st.stop()

        party_talks = pd.DataFrame(df["Party"].value_counts())

        party_labels = party_talks.index.to_list()  # List with active parties.
        if type(party_labels) == "list":
            party_labels.sort()

        if user_input != "speaker":
            # Let the user select parties to be included.
            container_parties = st.container()
            with container_parties:
                style_Partyes = build_style_parties(
                    party_labels
                )  # Make the options the right colors.
                st.markdown(style_Partyes, unsafe_allow_html=True)
                params.parties = st.multiselect(
                    label="Filter on parties.",
                    options=party_labels,
                    default=party_labels,
                )
            if params.parties != []:
                df = df.loc[df["Party"].isin(params.parties)]
                if len(df) == 0:
                    st.stop()

        # Let the user select type of debate.
        container_debate = st.container()
        with container_debate:
            debates = df["debatetype"].unique().tolist()
            debates.sort()

            style = build_style_debate_types(debates)
            st.markdown(style, unsafe_allow_html=True)
            params.debates = st.multiselect(
                label="Select type of debate",
                options=debates,
                default=debates,
            )
        if params.debates != []:
            df = df.loc[df["debatetype"].isin(params.debates)]
            if len(df) == 0:
                st.stop()
        params.update()

        # Let the user select a range of years.
        from_year = int(params.from_year)
        to_year = int(params.to_year)
        df_ = df.loc[
            df["Year"].isin([i for i in range(from_year, to_year)])
        ]  # TODO Ugly.
        years = list(range(int(df["Year"].min()), int(df["Year"].max()) + 1))
        if len(years) > 1:
            params.from_year, params.to_year = st.select_slider(
                "Select years",
                list(range(int(df["Year"].min()), int(df["Year"].max()) + 1)),
                value=(years[0], years[-1]),
            )
            df = df.loc[
                df["Year"].isin(list(range(params.from_year, params.to_year + 1)))
            ]
        elif len(years) == 1:
            df = df.loc[df["Year"] == years[0]]

        params.update()

        if user_input != "speaker":
            # Let the user select talkers.
            options = options_persons(df)
            style_mps = build_style_mps(options)  # Make the options the right colors.
            st.markdown(style_mps, unsafe_allow_html=True)
            col1_persons, col2_persons = st.columns([5, 2])
            # Sort alternatives in column to the right.
            with col2_persons:
                sort = st.selectbox(
                    "Sort result by", options=["Alphabetical order", "Most speeches"]
                )
                if sort == "Most speeches":
                    options = sorted(
                        options,
                        key=lambda x: [int(i) for i in x.split() if i.isdigit()][-1],
                        reverse=True,
                    )
                else:
                    options.sort()
            # Present options in column to the left.
            with col1_persons:
                expand_persons = st.container()
                with expand_persons:
                    params.persons = st.multiselect(
                        label="Filter on parlamentarians",
                        options=options,
                        default=[],
                    )

            # Filter df.
            if params.persons != []:
                params.persons = [i[: i.find(" - ")] for i in params.persons]
                df = df.loc[df["Speaker"].isin(params.persons)]
        params.update()

        # Give df an index.
        df.index = range(1, df.shape[0] + 1)

        ##* Start render. *##

        st.markdown("---")  # Draw line after filtering.
        st.write(f"**Hits: {df.shape[0]}**")

        if "hits" not in st.session_state:
            st.session_state["hits"] = df.shape[0]
        else:
            if st.session_state["hits"] != df.shape[0]:
                del st.session_state["df_excerpts"]
                del st.session_state["excerpt_page"]
                del st.session_state["text_next_page_button"]
                del st.session_state["disable_next_page_button"]
                st.session_state["hits"] = df.shape[0]

        ##! Show snippets.
        expand = st.expander(
            "Show excerpts from the speeches.",
            expanded=False,
        )
        with expand:
            if "excerpt_page" not in st.session_state:
                st.session_state["excerpt_page"] = 0
            excerpts_container = st.container()
            show_excerpts()
            st.session_state["excerpt_page"] += 1

            if "disable_next_page_button" not in st.session_state:
                st.session_state["disable_next_page_button"] = False
            if "text_next_page_button" not in st.session_state:
                st.session_state["text_next_page_button"] = "10 more please"
            if st.button(
                st.session_state["text_next_page_button"],
                key=f"next_page_{st.session_state['excerpt_page']}",
                disabled=st.session_state["disable_next_page_button"],
            ):
                st.session_state["excerpt_page"] += 1
                show_excerpts()

        # * Download all data in df.
        st.download_button(
            "Download the data as CSV",
            data=df.to_csv(
                index=False,
                sep=";",
                columns=[
                    "speech_id",
                    "Text",
                    "Party",
                    "Speaker",
                    "Date",
                    "url",
                ],
            ).encode("utf-8"),
            file_name=f"{user_input}.csv",
            mime="text/csv",
        )

        # Remove talks from same party within the same session to make the
        # statistics more representative.

        df_ = df[["speech_id", "Party", "Year"]].drop_duplicates()

        if user_input != "speaker":
            ## Make pie chart.

            party_talks = pd.DataFrame(df_["Party"].value_counts())
            party_labels = party_talks.index.to_list()
            fig, ax1 = plt.subplots()

            total = party_talks["Party"].sum()
            mentions = party_talks["Party"]  #!
            ax1.pie(
                mentions,
                labels=party_labels,
                autopct=lambda p: "{:.0f}".format(p * total / 100),
                colors=[party_colors[key] for key in party_labels],
                startangle=90,
            )

        # Make bars per year.
        years = set(df["Year"].tolist())

        df_years = pd.DataFrame(columns=["Party", "Year"])

        for year, df_i in df.groupby("Year"):
            dff = pd.DataFrame(data=df_i["Party"].value_counts())
            dff["Year"] = str(year)
            df_years = pd.concat([df_years, dff])
        df_years["party_code"] = df_years.index
        # df_years = df_years.groupby(["party_code", "Year"]).size()
        df_years["color"] = df_years["party_code"].apply(lambda x: party_colors[x])
        # df_years.drop(["Party"], axis=1, inplace=True)

        df_years.rename(
            columns={"Party": "Mentions", "party_code": "Party"}, inplace=True
        )
        chart = (
            alt.Chart(df_years)
            .mark_bar()
            .encode(
                x="Year",
                y="Mentions",
                color=alt.Color("color", scale=None),
                tooltip=["Party", "Mentions"],
            )
        )

        if user_input == "speaker":
            st.altair_chart(chart, use_container_width=True)

        else:
            # Put the charts in a table.
            fig1, fig2 = st.columns(2)
            with fig1:
                st.pyplot(fig)
            with fig2:
                st.altair_chart(chart, use_container_width=True)

        # Create an empty container
        container = st.empty()

        # * Make a summary of the standpoints of the parties.

        # A general note about the summaries. #TODO Make better.
        st.markdown(summary_note)

        # Count the number of speeches by party.
        party_counts = df["Party"].value_counts()
        # Sort the dataframe by party.
        df_sorted = df.loc[df["Party"].isin(party_counts.index)].sort_values(
            by=["Party"]
        )
        # Create an empty dictionary to store the summaries.
        summaries = {}
        # Loop through each party and their respective speeches.
        for party, df_party in df_sorted.groupby("Party"):
            # Create an empty list to store the texts of each speech.

            with st.spinner(f"Summarizing speeches by **{party}** parliamentarians..."):
                summary = summarize(df_party, user_input)
                st.markdown(
                    f"##### <span style='color:{party_colors[party]}'>{parties[party]}  ({party})</span>",
                    unsafe_allow_html=True,
                )
            st.write(summary)

        # A general note about the summaries. #TODO Make better.

        # * Get feedback.
        st.empty()
        feedback_container = st.empty()

        with feedback_container.container():
            feedback = st.text_area(
                "*Feel free to write suggestions for functions and improvements here!*"
            )
            send = st.button("Send")
            if len(feedback) > 2 and send:
                doc = {
                    "feedback": feedback,
                    "params": st.experimental_get_query_params(),
                    "where": ("streamlit", title),
                    "timestamp": str(datetime.now()),
                    "date": str(datetime.date(datetime.now())),
                }
                arango_db.collection("feedback").insert(doc)

                feedback_container.write("*Thanks*")
        params.update()

        # st.markdown("##")

    except Exception as e:
        if (
            e == "streamlit.runtime.scriptrunner.script_runner.StopException"
        ):  # If st.stop() is used.
            pass
        else:
            print(traceback.format_exc())
            st.markdown(
                ":red[Something has gone wrong, I'm trying to fix it as soon as possible. Feel free to try searching for something else.]"
            )
            arango_db.collection("errors").insert(
                {
                    "timestamp": str(datetime.now()),
                    "error": traceback.format_exc(),
                    "params": st.experimental_get_query_params(),
                    "where": ("streamlit", title),
                }
            )

expand_explainer = st.expander(
    "*What is this? Where does the data come from? How do I do?*"
)
with expand_explainer:
    st.markdown(explainer)