import csv
import pandas as pd
import random
import datasets
from sklearn.model_selection import train_test_split

_DOWNLOAD_URL = "https://huggingface.co/datasets/cfigueroa/glosas_etiquetadas/resolve/main/datasetNew.csv"

class GlosasEtiquetadas(datasets.GeneratorBasedBuilder):
    """Glosas Etiquetadas classification dataset."""

    def _info(self):
        return datasets.DatasetInfo(
            features=datasets.Features(
                {
                    "text": datasets.Value("string"),
                    "label": datasets.ClassLabel(names = ["LIBRERIA_ACCESORIOS","ALIMENTOS","ASEOPERSONAL","VEHICULOS",
"BANO","BEBE","BEBIDAS","CARNE","CERVEZA","CONSTRUCCION",
"DECORACION","DEPORTE","DORMITORIO","ELECTRONICA","EQUIPAJE","FERRETERIA",
"GRIFERIA","ILUMINACION","JARDIN","JOYERIA","JUGUETERIA","LACTEO","LAVANDERIA",
"LICORES","LIMPIEZA","MASCOTAS","MENAJE","MOBILIARIO",
"ORGANIZACION","OUTDOOR","PANADERIA","PISOSMUROS","PUERTASVENTANAS",
"SERVICIOS","TECNOLOGIA","TERRAZAS","VESTUARIO","VINOS","ZAPATERIA", "ILEGIBLE"] ),
                }
            )
        )

    def _split_generators(self, dl_manager):
        path = dl_manager.download_and_extract(_DOWNLOAD_URL)
        # Cargar el archivo .csv en un dataframe de pandas
        with open(path, 'r', encoding='latin-1', errors='ignore') as archivo:
          data_df = pd.read_csv(archivo ,sep=";")

        data_df.info()
        
        # Estratificación manual con sklearn
        train_data, temp_data = train_test_split(
            data_df, test_size=0.2, stratify=data_df[" categoria"]
        )
        test_data, val_data = train_test_split(
            temp_data, test_size=0.5, stratify=temp_data[" categoria"]
        )
        train_data.info()
        test_data.info()
        val_data.info()
        
        # Guardar divisiones en archivos temporales
        train_data.to_csv("train_data.csv", index=False)
        test_data.to_csv("test_data.csv", index=False)
        val_data.to_csv("val_data.csv", index=False)

        return [
            datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"filepath": "train_data.csv"}),
            datasets.SplitGenerator(name=datasets.Split.TEST, gen_kwargs={"filepath": "test_data.csv"}),
            datasets.SplitGenerator(name=datasets.Split.VALIDATION, gen_kwargs={"filepath": "val_data.csv"}),
        ]

    def _generate_examples(self, filepath):
        # Generar ejemplos sin duplicación de claves
        with open(filepath, 'r', encoding='latin-1', errors='ignore') as archivo:
          data = pd.read_csv(archivo ,sep=",")  
        data.info()
        for idx, row in data.iterrows():
            yield idx, {
                "text": row["glosa"],
                "label": row[" categoria"]
            }