tcutil.py

'''
TREC coding utilities - Common functions to several pages
'''

import numpy as np
import streamlit as st


class Labels():
    '''
    Used to carry labels information. Contain information for the constant labels (blank and sheet)
    If modifiable labels (annot, correct, predicted) are not set yet, initialize them.
    All labels are stored in a labels attribute, which is a dict. Only used for result sheet data (containing 200
    symbols).
    '''
    def __init__(self):
        self.labels = {}
        self.labels['blank_labels'] = np.zeros(200, dtype=int)
        self.labels['sheet_labels'] = np.array([3, 6, 4, 2, 9, 8, 7, 9, 1, 6, 3, 7, 6, 2, 5, 8, 4, 1, 9, 5,
                                  2, 9, 6, 8, 1, 7, 5, 3, 4, 7, 2, 5, 3, 1, 9, 6, 7, 8, 2, 4,
                                  6, 3, 7, 9, 5, 4, 8, 1, 2, 9, 8, 4, 1, 8, 7, 5, 9, 2, 6, 3,
                                  1, 8, 5, 2, 4, 3, 9, 6, 7, 1, 5, 6, 7, 5, 1, 3, 4, 9, 8, 2,
                                  9, 7, 4, 6, 8, 2, 1, 5, 3, 2, 9, 8, 4, 3, 6, 7, 2, 5, 9, 1,
                                  4, 6, 1, 5, 3, 9, 2, 7, 8, 5, 6, 3, 9, 4, 5, 2, 6, 7, 1, 8,
                                  5, 2, 8, 1, 7, 6, 3, 4, 9, 8, 1, 2, 6, 9, 8, 1, 5, 3, 4, 7,
                                  3, 5, 2, 4, 9, 1, 7, 8, 6, 4, 3, 1, 8, 7, 2, 4, 3, 6, 5, 9,
                                  7, 1, 9, 3, 6, 8, 4, 2, 5, 6, 7, 9, 2, 6, 3, 8, 1, 4, 7, 5,
                                  8, 4, 3, 7, 2, 5, 6, 9, 1, 3, 4, 7, 5, 2, 4, 9, 8, 1, 3, 6])
    def set_labels(self):
        st.session_state['sheet_labels'] = self.labels['sheet_labels']
        st.session_state['blank_labels'] = self.labels['blank_labels']
        if 'annot_labels' not in st.session_state:
            st.session_state['annot_labels'] = st.session_state['sheet_labels'].copy()
        if 'correct_labels' not in st.session_state:
            st.session_state['correct_labels'] = st.session_state['sheet_labels'].copy()
        if 'predicted_labels' not in st.session_state:
            st.session_state['predicted_labels'] = st.session_state['blank_labels'].copy()

def sidebar_progress():
    '''
    Sidebar function, used to display progress in workflow.
    '''
    if 'uploaded_file_name' in st.session_state:
        st.sidebar.success(f"Fichier analysé : {repr(st.session_state['uploaded_file_name'])}")
    else:
        st.sidebar.success("Aucun fichier uploadé")
    if 'ex_roi_symbols' in st.session_state:
        st.sidebar.success(f"Symboles extraits : {len(st.session_state['ex_roi_symbols'])}")

def sidebar_legend():
    '''
    Sidebar function, used to display legend (number - symbol match) for correction and annotation.
    '''
    legend = np.loadtxt("data_resultsheets/legend_pictures.csv", delimiter=',')
    leg_pict, leg_label = np.split(legend, [19740], axis=1)
    leg_pict = leg_pict.reshape((len(leg_pict), 141, 140))
    leg_pict = leg_pict / 255
    col1, col2, col3 = st.sidebar.columns([0.33, 0.33, 0.33], gap="small")
    pattern = [col1, col2, col3]
    col_list= [pattern[i % len(pattern)] for i in range(len(leg_pict))]
    for i, col in zip(range(len(leg_pict)), col_list):
        with col:
            col.image(leg_pict[i])
            # st.header(int(leg_label[i]))
            st.markdown(
                f"<p style='position:relative; top:-20px; color:red; font-size:1.5em;padding:10px;margin:0;text-align:center;'>{leg_label[i]}</p>",
                unsafe_allow_html=True)
def detect_empty(picts):
    '''
    Method to detect empty boxes among an image array
    :param picts: array of X images (X, height, width)
    :return boolean mask array of dimension (X,)
    '''
    fill_mask = picts.reshape((len(picts), picts.shape[1]*picts.shape[2])).std(axis=1) > 10
    return fill_mask
def exclude_example(prefix):
    '''
    Method to display streamlit controllers allowing to exclude example boxes (first 9 boxes), and empty boxes (as
    detected).
    :param prefix: string, used to specify the context in which function is called
    '''
    st.checkbox("Exclure les cases exemple", key=f'{prefix}_exclude-exemple', value=True)
    if st.session_state[f'{prefix}_exclude-exemple'] == True:
        excluded_index = range(9)
    else:
        excluded_index = []
    st.session_state[f'{prefix}_keeper_mask'] = [False if i in excluded_index else True for i in (range(200))]
    st.checkbox("Exclure les cases vides", key=f'{prefix}_exclude_empty', value=True)
    if st.session_state[f'{prefix}_exclude_empty'] == True:
        st.session_state[f'{prefix}_fill_mask'] = detect_empty(st.session_state['ex_roi_symbols'])
    else:
        st.session_state[f'{prefix}_fill_mask'] = np.ones(200)
    st.session_state[f'{prefix}_mask'] = [True if f and k else False for f, k in zip(st.session_state[f'{prefix}_keeper_mask'],
                                                       st.session_state[f'{prefix}_fill_mask'])]