empreinte_carbone.py

import streamlit as st
from comparateur import get_table_empreintes_detailed
from comparateur import *
import base64
import pandas as pd
import altair as alt

# Function to read and encode an SVG file to Base64
def load_svg_as_base64(file_path):
    with open(file_path, "rb") as f:
        svg_data = f.read()
    return base64.b64encode(svg_data).decode()

def color_scale(val):
    if val == '-':
        return 'background-color: {color}'
    elif val <= 1:
        color = '#008571' #'rgba(0,238,0,0.5)'  # green with opacity
    elif val <= 10:
        color = '#83c2b8'  # light green with opacity
    elif val <= 50:
        color = '#efcd82'  # light yellow with opacity
    elif val <= 100:
        color = '#f2aa56'  # light orange with opacity
    else:
        color = '#e87a58'  # light red with opacity
    return f'background-color: {color};color:white'

def display_cf_comparison(stm: st):
    svg_file_path = "feuille.svg"
    svg_base64 = load_svg_as_base64(svg_file_path)
    stm.markdown(
    f"""
    **Votre consommation carbone**
        <img src='data:image/svg+xml;base64,{svg_base64}' alt='svg' width='15' height='15' style='margin-left: 10px;'>
    """,
    unsafe_allow_html=True
    )

    serveur_emission = st.session_state['emission'].stop()
    emission_api = sum([value["el"] for value in st.session_state["partial_emissions"].values()])

    if serveur_emission is None :
        serveur_emission = 0
    if emission_api is None :
        emission_api = 0
    total_emission = serveur_emission + emission_api

    if total_emission == 0:
        pourcentage_api = 0
        pourcentage_serveur = 0
    else:
        pourcentage_api = emission_api / total_emission
        pourcentage_serveur = serveur_emission / total_emission
    
    

    stm.markdown(f"<div style='text-align: center; margin-bottom: 10px;'><b>{total_emission*1000:.2f}</b> g eq. CO2</div>", unsafe_allow_html=True)
    stm.markdown("Dont :")
    stm.markdown(f"- Empreinte serveur (via CodeCarbon) : **{serveur_emission*1000:.2f}** g eq. CO2 ({pourcentage_serveur:.2%})")
    stm.write(f"- Empreinte IA (via EcoLogits) : **{emission_api*1000:.2f}** g eq. CO2 ({pourcentage_api:.2%})")
    # stm.markdown("(avec l'outil CodeCarbon)")
    c1,c2,c3 = stm.columns([1,1,1])

    c2.write("---")

    stm.markdown("**Votre équivalence**")
    col1,col2,col3 = stm.columns([1,1,1])
    display_comparaison(col1,total_emission,dict_comparaison_1kgCO2["eau en litre"][0]*1000,dict_comparaison_1kgCO2["eau en litre"][1],"ml")
    display_comparaison(col2,total_emission,dict_comparaison_1kgCO2["tgv en km"][0],dict_comparaison_1kgCO2["tgv en km"][1],"km")
    display_comparaison(col3,total_emission,dict_comparaison_1kgCO2["voiture en km"][0]*1000,dict_comparaison_1kgCO2["voiture en km"][1],"m")
    stm.markdown("\n")
    stm.markdown(
    f"""
    Powered by **ADEME**
        <a href='https://www.ademe.fr' target='_blank'><img src='https://www.ademe.fr/wp-content/uploads/2022/11/ademe-logo-2022-1.svg' alt='svg' width='30' height='30' style='margin-left: 10px;'>
    """,
    unsafe_allow_html=True
    )

def display_carbon_footprint():
    st.title("EMPREINTE ÉNERGÉTIQUE DE L'APPLICATION IA CARTO RSE")
    display_cf_comparison(st)
    table = get_table_empreintes_detailed()
    # table[['Consommation Totale']] = table[['Consommation Totale']].map('${:,.2f}'.format)
   
    table.replace({0.00: '-'}, inplace=True)
    #just 2 digits after the comma
    styled_df = table[['Consommation Totale']].rename(columns={'Consommation Totale': 'Consommation totale (g eqCo2)'})
    styled_df = styled_df.round(2)
    
    styled_df = styled_df.style.applymap(color_scale, subset=['Consommation totale (g eqCo2)'])
    st.markdown("---")
    st.markdown("### DÉTAIL PAR TÂCHE")
    st.table(styled_df)
    with st.expander("Plus de détails"):
        st.table(table)
    
    st.markdown("### SYNTHESE (Dialogue IA et non IA)")

    serveur_emission = st.session_state['emission'].stop()
    emission_api = sum([value["el"] for value in st.session_state["partial_emissions"].values()])
    print(serveur_emission, emission_api)
    total_emission = serveur_emission + emission_api

    pourcentage_api = emission_api / total_emission
    pourcentage_serveur = serveur_emission / total_emission

    df = pd.DataFrame({"Catégorie": ["Identification + dessin","IA (extraction pp + dialogue)"], "valeur": [pourcentage_serveur, pourcentage_api]})
    color_scale_alt = alt.Scale(domain=['Identification + dessin', 'IA (extraction pp + dialogue)'], range=['#011166', '#63abdf'])

    base=alt.Chart(df).encode(
    theta=alt.Theta(field="valeur", type="quantitative", stack=True),
    color=alt.Color(field="Catégorie", type="nominal", scale=color_scale_alt),
)


    pie = base.mark_arc(outerRadius=100)
    text = base.mark_text(radius=150,fill= "black",align='center', baseline='middle',fontSize=20).encode(alt.Text(field="valeur", type="quantitative", format=".2%"))

    chart = alt.layer(pie, text, data=df).resolve_scale(theta="independent")
    st.altair_chart(chart, use_container_width=True)