Spaces:
Running
on
Zero
Running
on
Zero
Initial commit running application
Browse files- app.py +544 -7
- utilities.py +72 -0
app.py
CHANGED
@@ -1,7 +1,544 @@
|
|
1 |
-
import gradio as gr
|
2 |
-
|
3 |
-
|
4 |
-
|
5 |
-
|
6 |
-
|
7 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
import pandas as pd
|
3 |
+
import os
|
4 |
+
from qatch.connectors.sqlite_connector import SqliteConnector
|
5 |
+
from qatch.generate_dataset.orchestrator_generator import OrchestratorGenerator
|
6 |
+
from qatch.evaluate_dataset.orchestrator_evaluator import OrchestratorEvaluator
|
7 |
+
from predictor.orchestrator_predictor import OrchestratorPredictor
|
8 |
+
import utilities as us
|
9 |
+
import plotly.express as px
|
10 |
+
import plotly.graph_objects as go
|
11 |
+
|
12 |
+
with open('style.css', 'r') as file:
|
13 |
+
css = file.read()
|
14 |
+
|
15 |
+
# DataFrame di default
|
16 |
+
df_default = pd.DataFrame({
|
17 |
+
'Name': ['Alice', 'Bob', 'Charlie'],
|
18 |
+
'Age': [25, 30, 35],
|
19 |
+
'City': ['New York', 'Los Angeles', 'Chicago']
|
20 |
+
})
|
21 |
+
|
22 |
+
models_path = "models.csv"
|
23 |
+
|
24 |
+
# Variabile globale per tenere traccia dei dati correnti
|
25 |
+
df_current = df_default.copy()
|
26 |
+
|
27 |
+
input_data = {
|
28 |
+
'input_method': "",
|
29 |
+
'data_path': "",
|
30 |
+
'db_name': "",
|
31 |
+
'data': {
|
32 |
+
'data_frames': {}, # dictionary of dataframes
|
33 |
+
'db': None # SQLITE3 database object
|
34 |
+
},
|
35 |
+
'models': []
|
36 |
+
}
|
37 |
+
def load_data(file, path, use_default):
|
38 |
+
"""Carica i dati da un file, un percorso o usa il DataFrame di default."""
|
39 |
+
global df_current
|
40 |
+
if use_default:
|
41 |
+
input_data["input_method"] = 'default'
|
42 |
+
input_data["data_path"] = os.path.join(".", "data", "datainterface", "mytable.sqlite")
|
43 |
+
input_data["db_name"] = os.path.splitext(os.path.basename(input_data["data_path"]))[0]
|
44 |
+
input_data["data"]['data_frames'] = {'MyTable': df_current}
|
45 |
+
|
46 |
+
#TODO assegna il db a input_data["data"]['db']
|
47 |
+
|
48 |
+
df_current = df_default.copy() # Ripristina i dati di default
|
49 |
+
return input_data["data"]['data_frames']
|
50 |
+
|
51 |
+
selected_inputs = sum([file is not None, bool(path), use_default])
|
52 |
+
if selected_inputs > 1:
|
53 |
+
return 'Errore: Selezionare solo un metodo di input alla volta.'
|
54 |
+
|
55 |
+
if file is not None:
|
56 |
+
try:
|
57 |
+
input_data["input_method"] = 'uploaded_file'
|
58 |
+
input_data["db_name"] = os.path.splitext(os.path.basename(file))[0]
|
59 |
+
input_data["data_path"] = os.path.join(".", "data", f"data_interface{input_data['db_name']}.sqlite")
|
60 |
+
input_data["data"] = us.load_data(input_data["data_path"], input_data["db_name"])
|
61 |
+
df_current = input_data["data"]['data_frames'].get('MyTable', df_default) # Carica il DataFrame
|
62 |
+
print(df_current)
|
63 |
+
print(input_data["data"])
|
64 |
+
if( input_data["data"]['data_frames'] and not input_data["data"]['db']):
|
65 |
+
table2primary_key = {}
|
66 |
+
print("ok")
|
67 |
+
for table_name, df in input_data["data"]['data_frames'].items():
|
68 |
+
# Assign primary keys for each table
|
69 |
+
table2primary_key[table_name] = 'id'
|
70 |
+
print("ok2")
|
71 |
+
input_data["data"]["db"] = SqliteConnector(
|
72 |
+
relative_db_path=input_data["data_path"],
|
73 |
+
db_name=input_data["db_name"],
|
74 |
+
tables= input_data["data"]['data_frames'],
|
75 |
+
table2primary_key=table2primary_key
|
76 |
+
)
|
77 |
+
print(input_data["data"]["db"])
|
78 |
+
return input_data["data"]['data_frames']
|
79 |
+
except Exception as e:
|
80 |
+
return f'Errore nel caricamento del file: {e}'
|
81 |
+
|
82 |
+
if path:
|
83 |
+
if not os.path.exists(path):
|
84 |
+
return 'Errore: Il percorso specificato non esiste.'
|
85 |
+
try:
|
86 |
+
input_data["input_method"] = 'uploaded_file'
|
87 |
+
input_data["data_path"] = path
|
88 |
+
input_data["db_name"] = os.path.splitext(os.path.basename(path))[0]
|
89 |
+
input_data["data"] = us.load_data(input_data["data_path"], input_data["db_name"])
|
90 |
+
df_current = input_data["data"]['data_frames'].get('MyTable', df_default) # Carica il DataFrame
|
91 |
+
|
92 |
+
return input_data["data"]['data_frames']
|
93 |
+
except Exception as e:
|
94 |
+
return f'Errore nel caricamento del file dal percorso: {e}'
|
95 |
+
|
96 |
+
return input_data["data"]['data_frames']
|
97 |
+
|
98 |
+
def preview_default(use_default):
|
99 |
+
"""Mostra il DataFrame di default se il checkbox è selezionato."""
|
100 |
+
if use_default:
|
101 |
+
return df_default # Mostra il DataFrame di default
|
102 |
+
return df_current # Mostra il DataFrame corrente, che potrebbe essere stato modificato
|
103 |
+
|
104 |
+
def update_df(new_df):
|
105 |
+
"""Aggiorna il DataFrame corrente."""
|
106 |
+
global df_current # Usa la variabile globale per aggiornarla
|
107 |
+
df_current = new_df
|
108 |
+
return df_current
|
109 |
+
|
110 |
+
def open_accordion(target):
|
111 |
+
# Apre uno e chiude l'altro
|
112 |
+
if target == "reset":
|
113 |
+
return gr.update(open=True), gr.update(open=False, visible=False), gr.update(open=False, visible=False), gr.update(open=False, visible=False), gr.update(open=False, visible=False)
|
114 |
+
elif target == "model_selection":
|
115 |
+
return gr.update(open=False), gr.update(open=False), gr.update(open=True, visible=True), gr.update(open=False), gr.update(open=False)
|
116 |
+
|
117 |
+
# Interfaccia Gradio
|
118 |
+
interface = gr.Blocks(theme='d8ahazard/rd_blue', css_paths='style.css')
|
119 |
+
|
120 |
+
with interface:
|
121 |
+
gr.Markdown("# QATCH")
|
122 |
+
data_state = gr.State(None) # Memorizza i dati caricati
|
123 |
+
upload_acc = gr.Accordion("Upload your data section", open=True, visible=True)
|
124 |
+
select_table_acc = gr.Accordion("Select tables", open=False, visible=False)
|
125 |
+
select_model_acc = gr.Accordion("Select models", open=False, visible=False)
|
126 |
+
qatch_acc = gr.Accordion("QATCH execution", open=False, visible=False)
|
127 |
+
metrics_acc = gr.Accordion("Metrics", open=False, visible=False)
|
128 |
+
|
129 |
+
|
130 |
+
|
131 |
+
#################################
|
132 |
+
# PARTE DI INSERIMENTO DEL DB #
|
133 |
+
#################################
|
134 |
+
with upload_acc:
|
135 |
+
gr.Markdown("## Caricamento dei Dati")
|
136 |
+
|
137 |
+
file_input = gr.File(label="Trascina e rilascia un file", file_types=[".csv", ".xlsx", ".sqlite"])
|
138 |
+
path_input = gr.Textbox(label="Oppure inserisci il percorso locale del file")
|
139 |
+
with gr.Row():
|
140 |
+
default_checkbox = gr.Checkbox(label="Usa DataFrame di default")
|
141 |
+
preview_output = gr.DataFrame(interactive=True, visible=True, value=df_default)
|
142 |
+
submit_button = gr.Button("Carica Dati", interactive=False) # Disabilitato di default
|
143 |
+
output = gr.JSON(visible=False) # Output dizionario
|
144 |
+
|
145 |
+
# Funzione per abilitare il bottone se sono presenti dati da caricare
|
146 |
+
def enable_submit(file, path, use_default):
|
147 |
+
return gr.update(interactive=bool(file or path or use_default))
|
148 |
+
|
149 |
+
# Abilita il bottone quando i campi di input sono valorizzati
|
150 |
+
file_input.change(fn=enable_submit, inputs=[file_input, path_input, default_checkbox], outputs=[submit_button])
|
151 |
+
path_input.change(fn=enable_submit, inputs=[file_input, path_input, default_checkbox], outputs=[submit_button])
|
152 |
+
default_checkbox.change(fn=enable_submit, inputs=[file_input, path_input, default_checkbox], outputs=[submit_button])
|
153 |
+
|
154 |
+
# Mostra l'anteprima del DataFrame di default quando il checkbox è selezionato
|
155 |
+
default_checkbox.change(fn=preview_default, inputs=[default_checkbox], outputs=[preview_output])
|
156 |
+
preview_output.change(fn=update_df, inputs=[preview_output], outputs=[preview_output])
|
157 |
+
|
158 |
+
def handle_output(file, path, use_default):
|
159 |
+
"""Gestisce l'output quando si preme il bottone 'Carica Dati'."""
|
160 |
+
result = load_data(file, path, use_default)
|
161 |
+
|
162 |
+
if isinstance(result, dict): # Se result è un dizionario di DataFrame
|
163 |
+
if len(result) == 1: # Se c'è solo una tabella
|
164 |
+
return (
|
165 |
+
gr.update(visible=False), # Nasconde l'output JSON
|
166 |
+
result, # Salva lo stato dei dati
|
167 |
+
gr.update(visible=False), # Nasconde la selezione tabella
|
168 |
+
result, # Mantiene lo stato dei dati
|
169 |
+
gr.update(interactive=False), # Disabilita il pulsante di submit
|
170 |
+
gr.update(visible=True, open=True), # Passa direttamente a select_model_acc
|
171 |
+
gr.update(visible=True, open=False)
|
172 |
+
)
|
173 |
+
else:
|
174 |
+
return (
|
175 |
+
gr.update(visible=False),
|
176 |
+
result,
|
177 |
+
gr.update(open=True, visible=True),
|
178 |
+
result,
|
179 |
+
gr.update(interactive=False),
|
180 |
+
gr.update(visible=False), # Mantiene il comportamento attuale
|
181 |
+
gr.update(visible=True, open=True)
|
182 |
+
)
|
183 |
+
else:
|
184 |
+
return (
|
185 |
+
gr.update(visible=False),
|
186 |
+
None,
|
187 |
+
gr.update(open=False, visible=True),
|
188 |
+
None,
|
189 |
+
gr.update(interactive=True),
|
190 |
+
gr.update(visible=False),
|
191 |
+
gr.update(visible=True, open=True)
|
192 |
+
)
|
193 |
+
|
194 |
+
submit_button.click(
|
195 |
+
fn=handle_output,
|
196 |
+
inputs=[file_input, path_input, default_checkbox],
|
197 |
+
outputs=[output, output, select_table_acc, data_state, submit_button, select_model_acc, upload_acc]
|
198 |
+
)
|
199 |
+
|
200 |
+
|
201 |
+
|
202 |
+
######################################
|
203 |
+
# PARTE DI SELEZIONE DELLE TABELLE #
|
204 |
+
######################################
|
205 |
+
with select_table_acc:
|
206 |
+
table_selector = gr.CheckboxGroup(choices=[], label="Seleziona le tabelle da visualizzare", value=[])
|
207 |
+
table_outputs = [gr.DataFrame(label=f"Tabella {i+1}", interactive=True, visible=False) for i in range(5)]
|
208 |
+
selected_table_names = gr.Textbox(label="Tabelle selezionate", visible=False, interactive=False)
|
209 |
+
|
210 |
+
# Bottone di selezione modelli (inizialmente disabilitato)
|
211 |
+
open_model_selection = gr.Button("Choose your models", interactive=False)
|
212 |
+
|
213 |
+
def update_table_list(data):
|
214 |
+
"""Aggiorna dinamicamente la lista delle tabelle disponibili."""
|
215 |
+
if isinstance(data, dict) and data:
|
216 |
+
table_names = list(data.keys()) # Ritorna solo i nomi delle tabelle
|
217 |
+
return gr.update(choices=table_names, value=[]) # Reset delle selezioni
|
218 |
+
return gr.update(choices=[], value=[])
|
219 |
+
|
220 |
+
def show_selected_tables(data, selected_tables):
|
221 |
+
"""Mostra solo le tabelle selezionate dall'utente e abilita il bottone."""
|
222 |
+
updates = []
|
223 |
+
if isinstance(data, dict) and data:
|
224 |
+
available_tables = list(data.keys()) # Nomi effettivamente disponibili
|
225 |
+
selected_tables = [t for t in selected_tables if t in available_tables] # Filtra selezioni valide
|
226 |
+
|
227 |
+
tables = {name: data[name] for name in selected_tables} # Filtra i DataFrame
|
228 |
+
|
229 |
+
for i, (name, df) in enumerate(tables.items()):
|
230 |
+
updates.append(gr.update(value=df, label=f"Tabella: {name}", visible=True))
|
231 |
+
|
232 |
+
# Se ci sono meno di 5 tabelle, nascondi gli altri DataFrame
|
233 |
+
for _ in range(len(tables), 5):
|
234 |
+
updates.append(gr.update(visible=False))
|
235 |
+
else:
|
236 |
+
updates = [gr.update(value=pd.DataFrame(), visible=False) for _ in range(5)]
|
237 |
+
|
238 |
+
# Abilitare/disabilitare il bottone in base alle selezioni
|
239 |
+
button_state = bool(selected_tables) # True se almeno una tabella è selezionata, False altrimenti
|
240 |
+
updates.append(gr.update(interactive=button_state)) # Aggiorna stato bottone
|
241 |
+
|
242 |
+
return updates
|
243 |
+
|
244 |
+
def show_selected_table_names(selected_tables):
|
245 |
+
"""Mostra i nomi delle tabelle selezionate quando si preme il bottone."""
|
246 |
+
if selected_tables:
|
247 |
+
return gr.update(value=", ".join(selected_tables), visible=False)
|
248 |
+
return gr.update(value="", visible=False)
|
249 |
+
|
250 |
+
# Aggiorna automaticamente la lista delle checkbox quando `data_state` cambia
|
251 |
+
data_state.change(fn=update_table_list, inputs=[data_state], outputs=[table_selector])
|
252 |
+
|
253 |
+
# Aggiorna le tabelle visibili e lo stato del bottone in base alle selezioni dell'utente
|
254 |
+
table_selector.change(fn=show_selected_tables, inputs=[data_state, table_selector], outputs=table_outputs + [open_model_selection])
|
255 |
+
|
256 |
+
# Mostra la lista delle tabelle selezionate quando si preme "Choose your models"
|
257 |
+
open_model_selection.click(fn=show_selected_table_names, inputs=[table_selector], outputs=[selected_table_names])
|
258 |
+
open_model_selection.click(open_accordion, inputs=gr.State("model_selection"), outputs=[upload_acc, select_table_acc, select_model_acc, qatch_acc, metrics_acc])
|
259 |
+
|
260 |
+
|
261 |
+
|
262 |
+
####################################
|
263 |
+
# PARTE DI SELEZIONE DEL MODELLO #
|
264 |
+
####################################
|
265 |
+
with select_model_acc:
|
266 |
+
gr.Markdown("**Model Selection**")
|
267 |
+
|
268 |
+
# Supponiamo che `us.read_models_csv` restituisca anche il percorso dell'immagine
|
269 |
+
model_list_dict = us.read_models_csv(models_path)
|
270 |
+
model_list = [model["name"] for model in model_list_dict]
|
271 |
+
model_images = [model["image_path"] for model in model_list_dict]
|
272 |
+
|
273 |
+
# Creazione dinamica di checkbox con immagini
|
274 |
+
model_checkboxes = []
|
275 |
+
for model, image_path in zip(model_list, model_images):
|
276 |
+
with gr.Row():
|
277 |
+
with gr.Column(scale=1):
|
278 |
+
|
279 |
+
gr.Image(image_path, show_label=False)
|
280 |
+
with gr.Column(scale=2):
|
281 |
+
model_checkboxes.append(gr.Checkbox(label=model, value=False))
|
282 |
+
|
283 |
+
selected_models_output = gr.JSON(visible = False)
|
284 |
+
|
285 |
+
# Funzione per ottenere i modelli selezionati
|
286 |
+
def get_selected_models(*model_selections):
|
287 |
+
selected_models = [model for model, selected in zip(model_list, model_selections) if selected]
|
288 |
+
input_data['models'] = selected_models
|
289 |
+
button_state = bool(selected_models) # True se almeno un modello è selezionato, False altrimenti
|
290 |
+
return selected_models, gr.update(open=True, visible=True), gr.update(interactive=button_state)
|
291 |
+
|
292 |
+
# Bottone di submit (inizialmente disabilitato)
|
293 |
+
submit_models_button = gr.Button("Submit Models", interactive=False)
|
294 |
+
|
295 |
+
# Collegamento dei checkbox agli eventi di selezione
|
296 |
+
for checkbox in model_checkboxes:
|
297 |
+
checkbox.change(
|
298 |
+
fn=get_selected_models,
|
299 |
+
inputs=model_checkboxes,
|
300 |
+
outputs=[selected_models_output, select_model_acc, submit_models_button]
|
301 |
+
)
|
302 |
+
|
303 |
+
submit_models_button.click(
|
304 |
+
fn=lambda *args: (get_selected_models(*args), gr.update(open=False, visible=True), gr.update(open=True, visible=True)),
|
305 |
+
inputs=model_checkboxes,
|
306 |
+
outputs=[selected_models_output, select_model_acc, qatch_acc]
|
307 |
+
)
|
308 |
+
|
309 |
+
reset_data = gr.Button("Open upload data section")
|
310 |
+
reset_data.click(open_accordion, inputs=gr.State("reset"), outputs=[upload_acc, select_table_acc, select_model_acc, qatch_acc, metrics_acc])
|
311 |
+
|
312 |
+
|
313 |
+
|
314 |
+
###############################
|
315 |
+
# PARTE DI ESECUZIONE QATCH #
|
316 |
+
###############################
|
317 |
+
with qatch_acc:
|
318 |
+
selected_models_display = gr.JSON(label="Modelli selezionati")
|
319 |
+
submit_models_button.click(
|
320 |
+
fn=lambda: gr.update(value=input_data),
|
321 |
+
outputs=[selected_models_display]
|
322 |
+
)
|
323 |
+
|
324 |
+
proceed_to_metrics_button = gr.Button("Proceed to Metrics")
|
325 |
+
proceed_to_metrics_button.click(
|
326 |
+
fn=lambda: (gr.update(open=False, visible=True), gr.update(open=True, visible=True)),
|
327 |
+
outputs=[qatch_acc, metrics_acc]
|
328 |
+
)
|
329 |
+
|
330 |
+
reset_data = gr.Button("Open upload data section")
|
331 |
+
reset_data.click(open_accordion, inputs=gr.State("reset"), outputs=[upload_acc, select_table_acc, select_model_acc, qatch_acc, metrics_acc])
|
332 |
+
|
333 |
+
|
334 |
+
#######################################
|
335 |
+
# PARTE DI VISUALIZZAZIONE METRICHE #
|
336 |
+
#######################################
|
337 |
+
with metrics_acc:
|
338 |
+
confirmation_text = gr.Markdown("## Metrics successfully loaded")
|
339 |
+
|
340 |
+
data_path = 'metrics_random2.csv'
|
341 |
+
|
342 |
+
def load_data_csv_es():
|
343 |
+
return pd.read_csv(data_path)
|
344 |
+
|
345 |
+
def calculate_average_metrics(df, selected_metrics):
|
346 |
+
df['avg_metric'] = df[selected_metrics].mean(axis=1)
|
347 |
+
return df
|
348 |
+
|
349 |
+
def plot_metric(df, selected_metrics, group_by, selected_models):
|
350 |
+
df = df[df['model'].isin(selected_models)]
|
351 |
+
df = calculate_average_metrics(df, selected_metrics)
|
352 |
+
avg_metrics = df.groupby(group_by)['avg_metric'].mean().reset_index()
|
353 |
+
fig = px.bar(
|
354 |
+
avg_metrics, x=group_by[0], y='avg_metric', color=group_by[-1], barmode='group',
|
355 |
+
title=f'Media metrica per {group_by[0]}',
|
356 |
+
labels={group_by[0]: group_by[0].capitalize(), 'avg_metric': 'Media Metrica'},
|
357 |
+
template='plotly_dark'
|
358 |
+
)
|
359 |
+
return fig
|
360 |
+
|
361 |
+
def plot_radar(df, selected_models):
|
362 |
+
radar_data = []
|
363 |
+
for model in selected_models:
|
364 |
+
model_df = df[df['model'] == model]
|
365 |
+
valid_efficiency = model_df['valid_efficiency_score'].mean()
|
366 |
+
avg_time = model_df['time'].mean()
|
367 |
+
avg_tuple_order = model_df['tuple_order'].dropna().mean()
|
368 |
+
|
369 |
+
radar_data.append({
|
370 |
+
'model': model,
|
371 |
+
'valid_efficiency_score': valid_efficiency,
|
372 |
+
'time': avg_time,
|
373 |
+
'tuple_order': avg_tuple_order
|
374 |
+
})
|
375 |
+
|
376 |
+
radar_df = pd.DataFrame(radar_data)
|
377 |
+
categories = ['valid_efficiency_score', 'time', 'tuple_order']
|
378 |
+
|
379 |
+
# Calcola il range dinamico per il grafico
|
380 |
+
min_val = radar_df[categories].min().min()
|
381 |
+
max_val = radar_df[categories].max().max()
|
382 |
+
radar_df[categories] = (radar_df[categories] - min_val) / (max_val - min_val)
|
383 |
+
|
384 |
+
fig = go.Figure()
|
385 |
+
for _, row in radar_df.iterrows():
|
386 |
+
fig.add_trace(go.Scatterpolar(
|
387 |
+
r=[row[cat] for cat in categories],
|
388 |
+
theta=categories,
|
389 |
+
fill='toself',
|
390 |
+
name=row['model']
|
391 |
+
))
|
392 |
+
|
393 |
+
fig.update_layout(
|
394 |
+
polar=dict(radialaxis=dict(visible=True, range=[min_val, max_val])),
|
395 |
+
title='Radar Plot delle Metriche per Modello',
|
396 |
+
template='plotly_dark',
|
397 |
+
width=700, height=700
|
398 |
+
)
|
399 |
+
|
400 |
+
return fig
|
401 |
+
|
402 |
+
def plot_query_rate(df, selected_models, show_labels):
|
403 |
+
df = df[df['model'].isin(selected_models)]
|
404 |
+
|
405 |
+
fig = go.Figure()
|
406 |
+
|
407 |
+
for model in selected_models:
|
408 |
+
model_df = df[df['model'] == model].copy()
|
409 |
+
|
410 |
+
model_df['cumulative_time'] = model_df['time'].cumsum()
|
411 |
+
model_df['query_rate'] = 1 / model_df['time']
|
412 |
+
|
413 |
+
fig.add_trace(go.Scatter(
|
414 |
+
x=model_df['cumulative_time'],
|
415 |
+
y=model_df['query_rate'],
|
416 |
+
mode='lines+markers',
|
417 |
+
name=model,
|
418 |
+
line=dict(width=2)
|
419 |
+
))
|
420 |
+
|
421 |
+
if show_labels:
|
422 |
+
prev_category = None
|
423 |
+
prev_time = -float('inf')
|
424 |
+
y_positions = [1.1, 1.3]
|
425 |
+
y_idx = 0
|
426 |
+
|
427 |
+
for i, row in model_df.iterrows():
|
428 |
+
current_category = row['test_category']
|
429 |
+
if current_category != prev_category and row['cumulative_time'] - prev_time > 5:
|
430 |
+
fig.add_vline(x=row['cumulative_time'], line_width=1, line_dash="dash", line_color="gray")
|
431 |
+
fig.add_annotation(
|
432 |
+
x=row['cumulative_time'],
|
433 |
+
y=max(model_df['query_rate']) * y_positions[y_idx % 2],
|
434 |
+
text=current_category,
|
435 |
+
showarrow=False,
|
436 |
+
font=dict(size=10, color="white"),
|
437 |
+
textangle=45,
|
438 |
+
yshift=10,
|
439 |
+
bgcolor="rgba(0,0,0,0.6)"
|
440 |
+
)
|
441 |
+
prev_category = current_category
|
442 |
+
prev_time = row['cumulative_time']
|
443 |
+
y_idx += 1
|
444 |
+
|
445 |
+
fig.update_layout(
|
446 |
+
title="Rate di Generazione delle Query per Modello",
|
447 |
+
xaxis_title="Tempo Cumulativo (s)",
|
448 |
+
yaxis_title="Query al Secondo",
|
449 |
+
template='plotly_dark',
|
450 |
+
legend_title="Modelli"
|
451 |
+
)
|
452 |
+
|
453 |
+
return fig
|
454 |
+
|
455 |
+
def update_plot(selected_metrics, group_by, selected_models):
|
456 |
+
df = load_data_csv_es()
|
457 |
+
return plot_metric(df, selected_metrics, group_by, selected_models)
|
458 |
+
|
459 |
+
def update_radar(selected_models):
|
460 |
+
df = load_data_csv_es()
|
461 |
+
return plot_radar(df, selected_models)
|
462 |
+
|
463 |
+
def update_query_rate(selected_models, show_labels):
|
464 |
+
df = load_data_csv_es()
|
465 |
+
return plot_query_rate(df, selected_models, show_labels)
|
466 |
+
|
467 |
+
def plot_query_time_evolution(df, selected_models):
|
468 |
+
# Filtriamo i dati per i modelli selezionati
|
469 |
+
df = df[df['model'].isin(selected_models)]
|
470 |
+
|
471 |
+
# Ordinare per modello e tempo per tracciare l'evoluzione
|
472 |
+
df_sorted = df.sort_values(by=['model', 'time'])
|
473 |
+
|
474 |
+
fig = go.Figure()
|
475 |
+
|
476 |
+
# Aggiungiamo una traccia per ogni modello
|
477 |
+
for model in selected_models:
|
478 |
+
model_df = df_sorted[df_sorted['model'] == model]
|
479 |
+
fig.add_trace(go.Scatter(
|
480 |
+
x=model_df.index, y=model_df['time'], mode='lines+markers', name=model,
|
481 |
+
line=dict(shape='linear'),
|
482 |
+
text=model_df['model']
|
483 |
+
))
|
484 |
+
|
485 |
+
fig.update_layout(
|
486 |
+
title="Evoluzione del Tempo di Generazione per Modello",
|
487 |
+
xaxis_title="Indice della Query",
|
488 |
+
yaxis_title="Tempo (s)",
|
489 |
+
template='plotly_dark'
|
490 |
+
)
|
491 |
+
|
492 |
+
return fig
|
493 |
+
|
494 |
+
|
495 |
+
metrics = ["cell_precision", "cell_recall", "execution_accuracy", "tuple_cardinality", "tuple_constraint"]
|
496 |
+
group_options = {
|
497 |
+
"SQL Category": ["test_category", "model"],
|
498 |
+
"Tabella": ["tbl_name", "model"],
|
499 |
+
"Modello": ["model"]
|
500 |
+
}
|
501 |
+
|
502 |
+
df_initial = load_data_csv_es()
|
503 |
+
models = df_initial['model'].unique().tolist()
|
504 |
+
|
505 |
+
#with gr.Blocks(theme=gr.themes.Default(primary_hue='blue')) as demo:
|
506 |
+
gr.Markdown("""## Analisi delle prestazioni dei modelli
|
507 |
+
Seleziona una o più metriche per calcolare la media e visualizzare gli istogrammi e radar plots.
|
508 |
+
""")
|
509 |
+
|
510 |
+
# Sezione di selezione delle opzioni
|
511 |
+
with gr.Row():
|
512 |
+
metric_multiselect = gr.CheckboxGroup(choices=metrics, label="Seleziona le metriche")
|
513 |
+
model_multiselect = gr.CheckboxGroup(choices=models, label="Seleziona i modelli", value=models)
|
514 |
+
group_radio = gr.Radio(choices=list(group_options.keys()), label="Seleziona il raggruppamento", value="SQL Category")
|
515 |
+
#show_labels_checkbox = gr.Checkbox(label="Mostra etichette test category", value=True)
|
516 |
+
|
517 |
+
with gr.Row():
|
518 |
+
output_plot = gr.Plot()
|
519 |
+
# Dividi la pagina in due colonne
|
520 |
+
with gr.Row():
|
521 |
+
with gr.Column(scale=1): # Imposta la colonna a occupare metà della larghezza
|
522 |
+
radar_plot = gr.Plot(value=update_radar(models))
|
523 |
+
with gr.Column(scale=2): # Imposta la seconda colonna a occupare l'altra metà
|
524 |
+
show_labels_checkbox = gr.Checkbox(label="Mostra etichette test category", value=True)
|
525 |
+
query_rate_plot = gr.Plot(value=update_query_rate(models, True))
|
526 |
+
|
527 |
+
# Funzioni di callback per il cambiamento dei grafici
|
528 |
+
def on_change(selected_metrics, selected_group, selected_models):
|
529 |
+
return update_plot(selected_metrics, group_options[selected_group], selected_models)
|
530 |
+
|
531 |
+
def on_radar_change(selected_models):
|
532 |
+
return update_radar(selected_models)
|
533 |
+
|
534 |
+
show_labels_checkbox.change(update_query_rate, inputs=[model_multiselect, show_labels_checkbox], outputs=query_rate_plot)
|
535 |
+
metric_multiselect.change(on_change, inputs=[metric_multiselect, group_radio, model_multiselect], outputs=output_plot)
|
536 |
+
group_radio.change(on_change, inputs=[metric_multiselect, group_radio, model_multiselect], outputs=output_plot)
|
537 |
+
model_multiselect.change(on_change, inputs=[metric_multiselect, group_radio, model_multiselect], outputs=output_plot)
|
538 |
+
model_multiselect.change(on_radar_change, inputs=model_multiselect, outputs=radar_plot)
|
539 |
+
model_multiselect.change(update_query_rate, inputs=[model_multiselect, show_labels_checkbox], outputs=query_rate_plot)
|
540 |
+
|
541 |
+
reset_data = gr.Button("Open upload data section")
|
542 |
+
reset_data.click(open_accordion, inputs=gr.State("reset"), outputs=[upload_acc, select_table_acc, select_model_acc, qatch_acc, metrics_acc])
|
543 |
+
|
544 |
+
interface.launch()
|
utilities.py
ADDED
@@ -0,0 +1,72 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import csv
|
2 |
+
import pandas as pd
|
3 |
+
import sqlite3
|
4 |
+
import gradio as gr
|
5 |
+
import os
|
6 |
+
|
7 |
+
def carica_sqlite(file_path):
|
8 |
+
conn = sqlite3.connect(file_path)
|
9 |
+
cursor = conn.cursor()
|
10 |
+
cursor.execute("SELECT name FROM sqlite_master WHERE type='table';")
|
11 |
+
tabelle = cursor.fetchall()
|
12 |
+
tabelle = [tabella for tabella in tabelle if tabella[0] != 'sqlite_sequence']
|
13 |
+
|
14 |
+
dfs = {}
|
15 |
+
for tabella in tabelle:
|
16 |
+
nome_tabella = tabella[0]
|
17 |
+
df = pd.read_sql_query(f"SELECT * FROM {nome_tabella}", conn)
|
18 |
+
dfs[nome_tabella] = df
|
19 |
+
conn.close()
|
20 |
+
data_output = {'data_frames': dfs,'db': conn}
|
21 |
+
return data_output
|
22 |
+
|
23 |
+
# Funzione per leggere un file CSV
|
24 |
+
def carica_csv(file):
|
25 |
+
df = pd.read_csv(file)
|
26 |
+
return df
|
27 |
+
|
28 |
+
# Funzione per leggere un file Excel
|
29 |
+
def carica_excel(file):
|
30 |
+
xls = pd.ExcelFile(file)
|
31 |
+
dfs = {}
|
32 |
+
for sheet_name in xls.sheet_names:
|
33 |
+
dfs[sheet_name] = xls.parse(sheet_name)
|
34 |
+
return dfs
|
35 |
+
|
36 |
+
def load_data(data_path : str, db_name : str):
|
37 |
+
data_output = {'data_frames': {} ,'db': None}
|
38 |
+
table_name = os.path.splitext(os.path.basename(data_path))[0]
|
39 |
+
if data_path.endswith(".sqlite") :
|
40 |
+
data_output = carica_sqlite(data_path)
|
41 |
+
elif data_path.endswith(".csv"):
|
42 |
+
data_output['data_frames'] = {f"{table_name}_table" : carica_csv(data_path)}
|
43 |
+
elif data_path.endswith(".xlsx"):
|
44 |
+
data_output['data_frames'] = carica_excel(data_path)
|
45 |
+
else:
|
46 |
+
raise gr.Error("Formato file non supportato. Carica un file SQLite, CSV o Excel.")
|
47 |
+
return data_output
|
48 |
+
|
49 |
+
def read_api(api_key_path):
|
50 |
+
with open(api_key_path, "r", encoding="utf-8") as file:
|
51 |
+
api_key = file.read()
|
52 |
+
return api_key
|
53 |
+
|
54 |
+
def read_models_csv(file_path):
|
55 |
+
# Reads a CSV file and returns a list of dictionaries
|
56 |
+
models = [] # Change {} to []
|
57 |
+
with open(file_path, mode="r", newline="") as file:
|
58 |
+
reader = csv.DictReader(file)
|
59 |
+
for row in reader:
|
60 |
+
row["price"] = float(row["price"]) # Convert price to float
|
61 |
+
models.append(row) # Append to the list
|
62 |
+
return models
|
63 |
+
|
64 |
+
def csv_to_dict(file_path):
|
65 |
+
with open(file_path, mode='r', encoding='utf-8') as file:
|
66 |
+
reader = csv.DictReader(file)
|
67 |
+
data = []
|
68 |
+
for row in reader:
|
69 |
+
if "price" in row:
|
70 |
+
row["price"] = float(row["price"])
|
71 |
+
data.append(row)
|
72 |
+
return data
|