PRENT/PRENT-Codebook

1import os2import sys3 4import streamlit as st5from st_aggrid import AgGrid, DataReturnMode6 7current = os.path.dirname(os.path.realpath(__file__))8parent = os.path.dirname(current)9sys.path.append(parent)10from helpers import apply_style, get_idx_column, read_csv_from_web, read_json_from_web11 12apply_style()13 14codebook = {}15 16st.markdown(17    """18    # Codebook Creation/Edition Tool based on the PR-ENT Approach.19    ### *Rethinking the Event Coding Pipeline with Prompt Entailment*20    ### Author: Anonymized for submission21    ##### Version: 1.022"""23)24st.markdown("***********")25 26st.markdown(27    """28## Data Loading29"""30)31 32 33st.markdown(34    """35    ### Upload a CSV of event descriptions.36"""37)38uploaded_file = st.file_uploader("Upload a csv file containing event descriptions")39if uploaded_file is not None:40    st.session_state.data = read_csv_from_web(uploaded_file)41 42 43if "data" in st.session_state:44    # Filter will be reset if the page is left and then used again45    loading_df = st.text("Loading data display...")46    st.write(47        """48        The below display of the data can be used to filter the data. Click on the *3 bars logo* when hovering over a column name and the filtering49        tool will appear. Filters are kept in memory on the whole dashboard as long as the `Reset Filters` button is not clicked.50 51        Current limitation: If a filter is set and the user change page. Then it can not be modified anymore and needs to be reset.52    """53    )54    if "filtered_df" not in st.session_state:55        st.session_state.filtered_df = st.session_state.data56    if st.button("Reset Filters"):57        st.session_state.filtered_df = st.session_state.data58 59    st.session_state.filtered_df = AgGrid(60        st.session_state.filtered_df,61        height=400,62        data_return_mode=DataReturnMode.FILTERED,63        update_mode="MANUAL",64    )["data"]65 66    if "text_column_design_perm" not in st.session_state:67        st.session_state[68            "text_column_design_perm"69        ] = st.session_state.filtered_df.columns[0]70 71    def callback_function(mod, key):72        st.session_state[mod] = st.session_state[key]73 74    st.write("Select the column which contains the event descriptions.")75    st.selectbox(76        "Select the event description column:",77        st.session_state.filtered_df.columns,78        key="text_column_design",79        on_change=callback_function,80        args=("text_column_design_perm", "text_column_design"),81        index=get_idx_column(82            st.session_state["text_column_design_perm"],83            list(st.session_state.filtered_df.columns),84        ),85    )86    loading_df.text("")87 88    # Remove NaN Texts89    if st.button("Remove Empty Event Descriptions"):90        st.session_state.filtered_df = st.session_state.filtered_df.dropna(91            subset=[st.session_state["text_column_design_perm"]]92        )93 94 95st.write("********")96st.markdown("## Optional Upload")97 98 99st.markdown(100    """101    ### Upload a codebook if available. It needs to be in the format used in this dashboard.102"""103)104uploaded_codebook = st.file_uploader("Upload a codebook if available (OPTIONAL)")105if uploaded_codebook is not None:106    codebook = read_json_from_web(uploaded_codebook)107    st.session_state.codebook = codebook108 109st.markdown(110    """111    ### Upload a validated dataset (accept, reject, ignored) in the format of this dashboard.112"""113)114 115uploaded_validated_data = st.file_uploader(116    "Upload a json file containing validated data (OPTIONAL)"117)118if uploaded_validated_data is not None:119    st.session_state.validated_data = read_json_from_web(uploaded_validated_data)120