Scripts-SAT-OPEC/02_data_generator_hybrid.py at main · Llapra/Scripts-SAT-OPEC · GitHub

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
import os
import json
import numpy as np
import pandas as pd
import tensorflow as tf
from datetime import datetime, timedelta

class HybridDataGenerator(tf.keras.utils.Sequence):
    def __init__(self,
                 csv_path,
                 npy_folder,
                 manifest_path,
                 target_vars,
                 feature_vars,
                 input_length_tab=48,
                 input_length_vis=6,
                 output_length=28,
                 batch_size=32,
                 shuffle=True):

        # 1. Llamada al constructor padre (Corrige el Warning de PyDataset)
        super().__init__()

        self.npy_folder = npy_folder
        self.batch_size = batch_size
        self.shuffle = shuffle
        self.input_length_tab = input_length_tab
        self.input_length_vis = input_length_vis
        self.output_length = output_length

        # 2. Cargar Datos Tabulares
        print(f"Cargando datos tabulares desde {csv_path}...")
        self.df = pd.read_csv(csv_path, index_col='TIMESTAMP', parse_dates=True)

        # Filtrar columnas y convertir a numpy (float32 explícito para evitar errores de tipo)
        self.feature_vars = feature_vars
        self.target_vars = target_vars
        self.data_features = self.df[feature_vars].values.astype(np.float32)
        self.data_targets = self.df[target_vars].values.astype(np.float32)

        # Crear mapa de tiempo
        self.time_to_index = {t: i for i, t in enumerate(self.df.index)}

        # 3. Cargar Manifiesto
        print(f"Cargando manifiesto de secuencias desde {manifest_path}...")
        with open(manifest_path, 'r') as f:
            manifest = json.load(f)

        self.indices_validos = []
        self.preparar_indices(manifest['secuencias'])
        print(f"Generador inicializado con {len(self.indices_validos)} muestras.")

    def preparar_indices(self, secuencias):
        max_lookback = max(self.input_length_tab, self.input_length_vis)

        for seq_archivos in secuencias:
            fechas_seq = [self.parse_filename(f) for f in seq_archivos]

            for i in range(max_lookback, len(fechas_seq) - self.output_length):
                fecha_presente = fechas_seq[i]

                if fecha_presente in self.time_to_index:
                    idx_csv = self.time_to_index[fecha_presente]

                    if idx_csv < self.input_length_tab:
                        continue

                    if (idx_csv + self.output_length) < len(self.df):
                        self.indices_validos.append({
                            'csv_idx': idx_csv,
                            'img_names': seq_archivos[i - self.input_length_vis + 1 : i + 1]
                        })

    def parse_filename(self, filename):
        try:
            str_date = filename.split('_')[0]
            return datetime.strptime(str_date, "%Y-%m-%d-%H%M")
        except Exception:
            return None

    def __len__(self):
        return int(np.floor(len(self.indices_validos) / self.batch_size))

    def __getitem__(self, index):
        indices = self.indices_validos[index * self.batch_size:(index + 1) * self.batch_size]

        batch_x_vis = []
        batch_x_tab = []
        batch_y = []

        for sample in indices:
            # --- VISUAL ---
            seq_imgs = []
            for fname in sample['img_names']:
                npy_name = fname.replace('.png', '.npy')
                npy_path = os.path.join(self.npy_folder, npy_name)

                try:
                    img_temp = np.load(npy_path).astype(np.float32)
                    # Normalización (-90 a 40 -> 0 a 1)
                    img_norm = (img_temp + 90.0) / 130.0
                    img_norm = np.clip(img_norm, 0.0, 1.0)
                except (FileNotFoundError, OSError):
                    img_norm = np.zeros((128, 128), dtype=np.float32)

                seq_imgs.append(img_norm)

            # (Time, H, W, 1)
            x_vis = np.array(seq_imgs)[..., np.newaxis]
            batch_x_vis.append(x_vis)

            # --- TABULAR ---
            idx = sample['csv_idx']
            x_tab = self.data_features[idx - self.input_length_tab + 1 : idx + 1]
            batch_x_tab.append(x_tab)

            # --- TARGET ---
            y = self.data_targets[idx + 1 : idx + 1 + self.output_length]
            batch_y.append(y)

        # CORRECCIÓN CRÍTICA AQUÍ:
        # Devolvemos una TUPLA para inputs: (x_visual, x_tabular)
        # Y un array para outputs: y
        # Convertimos todo explícitamente a tensores o arrays de float32
        inputs = (np.array(batch_x_vis, dtype=np.float32), np.array(batch_x_tab, dtype=np.float32))
        targets = np.array(batch_y, dtype=np.float32)

        return inputs, targets

    def on_epoch_end(self):
        if self.shuffle:
            np.random.shuffle(self.indices_validos)