File size: 9,297 Bytes
2b4a9d2
 
 
 
 
 
 
 
d76808b
2b4a9d2
62c4367
2b4a9d2
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
62c4367
2b4a9d2
62c4367
2b4a9d2
 
 
62c4367
 
 
2b4a9d2
 
 
 
 
 
 
 
 
 
62c4367
 
 
 
 
2b4a9d2
 
62c4367
2b4a9d2
62c4367
2b4a9d2
62c4367
2b4a9d2
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
d76808b
 
 
 
 
 
2b4a9d2
 
 
 
 
 
 
 
 
 
d76808b
2b4a9d2
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
d76808b
 
 
 
 
 
2b4a9d2
 
 
 
 
 
 
 
 
d76808b
2b4a9d2
d76808b
 
62c4367
 
 
2b4a9d2
 
d76808b
 
2b4a9d2
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
from utils import (
    container_emphasis_css,
    container_regular_css,
    question_font,
    modify_sql_guid_injection
)

import streamlit as st
import json
import pandas as pd
from datetime import datetime
from streamlit_extras.stylable_container import stylable_container

@st.cache_data
def load_data(file):
    return pd.read_csv(file)

st.set_page_config(layout="wide", initial_sidebar_state="expanded")
st.title("Relevancy Compare App")
st.divider()
st.markdown(question_font, unsafe_allow_html=True)


def edit_df_data_change():
    edited_rows: dict = st.session_state.edit_df_data['edited_rows']
    st.session_state.previous_row_index = st.session_state.previous_row_index
    st.session_state.current_row_index = next(iter(edited_rows))
    st.session_state.df = st.session_state.df.assign(selected=False)
    update_dict = {idx: values for idx, values in edited_rows.items()}
    st.session_state.df.update(pd.DataFrame.from_dict(update_dict, orient='index'))


st.need_rerun = False
if "df" not in st.session_state:
    role_pick = st.selectbox("You are?", options=["Kyle", "Shadi", "Daniel", "Xin"], index=None, key='role_pick')
    uploaded_file = st.file_uploader("Choose a CSV file", type="csv")
    st.session_state.role = role_pick
    st.need_rerun = True
else:
    uploaded_file = st.session_state.df
if uploaded_file is not None and 'role' in st.session_state:
    "You are modifying as :", st.session_state.role
    with st.container():
        if "df" not in st.session_state:
            df = load_data(uploaded_file)
            st.session_state.df_shape = df.shape
            # selected load
            select_values = [False] * len(df)
            
            if "Comment" not in df.columns:
                comments = [''] * len(df)
                df.insert(0, 'Comment', comments)
                st.session_state.default_row_index = 0
            
            if "LastModified" not in df.columns:
                modifies = [''] * len(df)
                df.insert(0, 'LastModified', modifies)

            else:
                df['Comment'] = df['Comment'].astype(str)
                df['LastModified'] = df['LastModified'].astype(str)
                comments = [c if c != 'nan' else '' for c in df['Comment'].tolist()]
                modifies = [c if c != 'nan' else '' for c in df['LastModified'].tolist()]
                df['Comment'] = comments
                df['LastModified'] = modifies
                first_nan = 0
                for i, c in enumerate(comments):
                    if not c:
                        first_nan = i
                        break
                st.session_state.default_row_index = first_nan
            
            st.session_state.current_row_index = st.session_state.default_row_index
            st.session_state.previous_row_index = st.session_state.current_row_index

            df.insert(0, 'Select', select_values)
            df.at[st.session_state.current_row_index, 'Select'] = True
            st.session_state['df'] = df
        
        df_col, = st.columns(1)

    st.session_state.previous_row_index = st.session_state.current_row_index
    with st.container():
        bcol1, bcol2, _ = st.columns([1, 1, 5])
        with bcol1:
            previous_row = st.button("Prev Row")
        with bcol2:
            next_row = st.button("Next Row")
            

    if "current_row_index" in st.session_state and previous_row:
        if st.session_state.current_row_index > 0:
            st.session_state.previous_row_index = st.session_state.current_row_index
            st.session_state.current_row_index -= 1

    if "current_row_index" in st.session_state and next_row:
        if st.session_state.current_row_index < st.session_state.df_shape[0] - 1:
            st.session_state.previous_row_index = st.session_state.current_row_index
            st.session_state.current_row_index += 1

    row_index = st.session_state.current_row_index
    row_data = st.session_state.df.loc[row_index]
    question = row_data['question']
    pred_relevant_choice = row_data['pred_relevant_choice']
    correct_response = row_data['correct_response']
    relative_relevancy = json.loads(row_data['relative_relevancy'])
    reason = relative_relevancy['reason']
    most_relevant = relative_relevancy['most_relevant']
    options=['first', 'second', 'both']
    default_index = options.index(most_relevant)
    default_comment = row_data['Comment']
    first_css = container_regular_css
    second_css = container_regular_css
    first_color = ':black'
    second_color = ':black'
    
        
    st.write(f"##### Row {row_index}/{st.session_state.df_shape[0]-1}")
    with st.container(border=True):
        # st.write(f"Question: {question}")
        st.markdown(f'<p class="question-font">[Q]: {question}</p>', unsafe_allow_html=True)
        select_by = st.selectbox("Mark By: ", ["pred_relevant_choice", "correct_response"], key='mark_by_dropdown')
        col1_res, col2_res, _ = st.columns([0.2, 0.2, 0.6])
        if select_by == 'pred_relevant_choice':
            col1_res.write(f":blue[pred_relevant_choice: {pred_relevant_choice}]")
            col2_res.write(f"correct_response: {correct_response}")
        else:
            col1_res.write(f"pred_relevant_choice: {pred_relevant_choice}")
            col2_res.write(f":blue[correct_response: {correct_response}]")


        if row_data[select_by].lower() == 'first' or row_data[select_by].lower() == 'both':
            first_css = container_emphasis_css
            first_color = ':blue'

            if row_data[select_by].lower() == 'both':
                second_css = container_emphasis_css
                second_color = ':blue'
        else:
            second_css = container_emphasis_css
            second_color = ':blue'

        col1, col2 = st.columns(2)
        with col1:
            st.write("Target SQL:", use_container_width=True)
            with stylable_container(
                key="first_sql_container",
                css_styles=first_css
                ):
                target_sql_text = modify_sql_guid_injection(row_data['target_sql'].strip())
                #st.write(f"{first_color}[{target_sql_text}]", use_container_width=True)
                col1, _ = st.columns([0.99, 0.01])
                with col1:
                    st.code(
                        target_sql_text,
                        language="sql"
                    )
                    # st.code(
                    #     "\n".join(
                    #         tw.wrap(
                    #         target_sql_text,
                    #         width=60,
                    #         )
                    #     ),
                    #     language="sql"
                    # )

        with col2:
            st.write("Predicted SQL:", use_container_width=True)
            with stylable_container(
                key="second_sql_container",
                css_styles=second_css
                ):
                predicted_sql_text = modify_sql_guid_injection(row_data['predicted_sql'].strip())
                # st.write(f"{second_color}[{predicted_sql_text}]")
                col1, _ = st.columns([0.99, 0.01])
                with col1:
                     st.code(
                        predicted_sql_text,
                        language="sql"
                    )
        col1_comment, col2_comment = st.columns([1, 1])
        with col1_comment:
            select_most_relevant = st.selectbox("most_relevant", options=options, index=default_index, key='most_relevant_drop_down')
            reason_box = st.text_area("reason", value=reason, key='reason_text')
        with col2_comment:
            comment = st.text_area("Comment", value=default_comment, key='comment_text')
    
    with st.container():
        bcol1, bcol2, _, _, _, _, _, _, _, _ = st.columns(10)
        with bcol1:
            save = st.button("Save")
        with bcol2:
            download = st.button("Download")

    if save:
        updated_relative_relevancy = json.dumps({'most_relevant': select_most_relevant, "reason": reason_box})
        st.session_state.df.at[row_index, 'Comment'] = comment
        st.session_state.df.at[row_index, 'relative_relevancy'] = updated_relative_relevancy
        st.session_state.df.at[row_index, 'correct_response'] = select_most_relevant
        current_time = datetime.now()
        current_time_str = current_time.strftime("%Y-%m-%d %H:%M:%S")
        st.session_state.df.at[row_index, 'LastModified'] = f"{st.session_state.role}@{current_time_str}"
        df = st.session_state.df.drop('Select', axis=1)
        df.to_csv('modified.csv', index=False)
        st.rerun()

    if download:
        with open('modified.csv') as f:
            st.download_button('Download modified CSV', f, file_name='modified.csv')
    
    st.session_state.df['Select'] = False
    st.session_state.df.at[st.session_state.current_row_index, 'Select'] = True

    edited_df = df_col.data_editor(st.session_state.df, num_rows="fixed", use_container_width=True, key='edit_df_data', on_change=edit_df_data_change)

    #st.session_state.mannul_select = edited_df[edited_df['Select']].index
    st.session_state.df = edited_df

    if st.need_rerun:
        st.need_rerun = False
        st.rerun()