import datasets
import csv
import json
import os
import csv

# Defining Access
_DATA_URL = "data/BA_Datensatz_V2.tar.gz"

_PROMPTS_URLS = {
    "train": "data/metadata-train.csv.gz",
    "test": "data/metadata-test.csv.gz",
}

class BA_Datensatz_V2(datasets.GeneratorBasedBuilder):

    def _info(self):
        return datasets.DatasetInfo(
            description="German Dataset that focuses on juristical data",
            features=datasets.Features({
                "speaker_id": datasets.Value("string"),
                "path": datasets.Value("string"),
                "audio": datasets.Audio(sampling_rate=16_000),
                "sentence": datasets.Value("string"),
            }),
            supervised_keys=None,
        )

    def _split_generators(self, dl_manager):
        audio_path = dl_manager.download(_DATA_URL)
        prompts_paths = dl_manager.download_and_extract(_PROMPTS_URLS)

        train_dir = os.path.join("BA_Datensatz_V2", "train", "wav/")
        test_dir = os.path.join("BA_Datensatz_V2", "test", "wav/")

        return [
            datasets.SplitGenerator(
                name=datasets.Split.TRAIN,
                gen_kwargs={
                    "audio_files": dl_manager.iter_archive(audio_path),
                    "metadata_path": prompts_paths["train"],
                    "path_to_clips": train_dir,
                },
            ),
            datasets.SplitGenerator(
                name=datasets.Split.TEST,
                gen_kwargs={
                    "audio_files": dl_manager.iter_archive(audio_path),
                    "metadata_path": prompts_paths["test"],
                    "path_to_clips": test_dir,
                },
            ),
        ] 

    def _generate_examples(self, audio_files, metadata_path, path_to_clips):
        metadata = {}
        with open(metadata_path, 'rt', encoding='utf-8') as f:
            reader = csv.reader(f)
            for row in reader:
                filename_with_extension, sentence = row
                speaker_id = filename_with_extension.split('_')[0]
                # Construct path relative to the dataset structure
                relative_path = os.path.join(path_to_clips, speaker_id, filename_with_extension)
                metadata[relative_path] = {
                    "speaker_id": speaker_id,
                    "path": relative_path,
                    "sentence": sentence,
                }
        id_ = 0
        for path, file_content in audio_files:
            if not path.endswith('/') and path.startswith(path_to_clips): # This checks if the file is relevant to the current split by checking its path
                if path in metadata:
                    audio_data = {
                        "path": metadata[path]["path"],
                        "bytes": file_content.read(),
                        }
                    yield id_, {**metadata[path], "audio": audio_data}
                    id_ += 1
                else:
                    print(f"No metadata entry for {path}")
            else: # Skip directories or files not relevant to the current split
                continue