import os
from Bio import SeqIO
from tqdm.auto import tqdm
import pandas as pd
from huggingface_hub import HfApi


aav_files = [
    "des_mut",
    "low_vs_high",
    "mut_des",
    "one_vs_many",
    "sampled",
    "seven_vs_many",
    "two_vs_many",
]


meltome_files = [
    "human",
    "human_cell",
    "mixed_split",
]


sav_files = [
    "human",
    "only_savs",
    "mixed",
]


scl_files = [
    "balanced",
    "human_hard",
    "human_soft",
    "mixed_hard",
    "mixed_soft",
]


gb1_files = [
    "low_vs_high",
    "one_vs_rest",
    "sampled",
    "three_vs_rest",
    "two_vs_rest",
]


def download_wget(filename, repo):
    if os.path.exists(f"{filename}.fasta"):
        return
    url = f"http://data.bioembeddings.com/public/FLIP/fasta/{repo}/{filename}.fasta"
    os.system(f"wget {url}")


def upload_aav():
    repo = "aav"
    for filename in aav_files:
        download_wget(filename, repo)
        fasta_file = f"{filename}.fasta"
        csv_file = f"{filename}.csv"
        db = SeqIO.index(fasta_file, "fasta")
        output = []

        for _, record in tqdm(db.items()):
            description = record.description
            description = description.split()
            sequence = str(record.seq)
            seqid = description[0]
            label = float(description[1].split("=")[1])
            split = description[2].split("=")[1]
            validation = description[3].split("=")[1]

            output.append({
                "seqid": seqid,
                "label": label,
                "sequence": sequence,
                "split": split.lower(),
                "validation": validation.lower(),
            })
        pd.DataFrame(output).to_csv(csv_file, index=False)

        api = HfApi()
        api.create_repo(
            repo_id=f"hazemessam/{repo}",
            repo_type="dataset",
            exist_ok=True,
        )
        api.upload_file(
            path_or_fileobj=csv_file,
            path_in_repo=csv_file,
            repo_id=f"hazemessam/{repo}",
            repo_type="dataset",
        )
        os.system(f"rm -rf {fasta_file}")
        os.system(f"rm -rf {csv_file}")


def upload_meltome():
    repo = "meltome"
    for filename in meltome_files:
        download_wget(filename, repo)
        fasta_file = f"{filename}.fasta"
        csv_file = f"{filename}.csv"
        db = SeqIO.index(fasta_file, "fasta")
        output = []
        for _, record in tqdm(db.items()):
            description = record.description
            description = description.split()
            sequence = str(record.seq)
            seqid = description[0]
            label = float(description[1].split("=")[1])
            split = description[2].split("=")[1]
            validation = description[3].split("=")[1]

            output.append({
                "seqid": seqid,
                "label": label,
                "sequence": sequence,
                "split": split.lower(),
                "validation": validation.lower(),
            })
        pd.DataFrame(output).to_csv(csv_file, index=False)

        api = HfApi()
        api.create_repo(
            repo_id=f"hazemessam/{repo}",
            repo_type="dataset",
            exist_ok=True,
        )
        api.upload_file(
            path_or_fileobj=csv_file,
            path_in_repo=csv_file,
            repo_id=f"hazemessam/{repo}",
            repo_type="dataset",
        )
        os.system(f"rm -rf {fasta_file}")
        os.system(f"rm -rf {csv_file}")


def upload_sav():
    repo = "sav"
    for filename in sav_files:
        download_wget(filename, repo)
        fasta_file = f"{filename}.fasta"
        csv_file = f"{filename}.csv"
        db = SeqIO.index(fasta_file, "fasta")
        output = []
        for _, record in tqdm(db.items()):
            description = record.description
            description = description.split()
            sequence = str(record.seq)
            seqid = description[0]
            # the label is a string in sav and scl datasets
            label = description[1].split("=")[1]
            split = description[2].split("=")[1]
            validation = description[3].split("=")[1]

            output.append({
                "seqid": seqid,
                "label": label,
                "sequence": sequence,
                "split": split.lower(),
                "validation": validation.lower(),
            })
        pd.DataFrame(output).to_csv(csv_file, index=False)
        api = HfApi()
        api.create_repo(
            repo_id=f"hazemessam/{repo}",
            repo_type="dataset",
            exist_ok=True,
        )
        api.upload_file(
            path_or_fileobj=csv_file,
            path_in_repo=csv_file,
            repo_id=f"hazemessam/{repo}",
            repo_type="dataset",
        )
        os.system(f"rm -rf {fasta_file}")
        os.system(f"rm -rf {csv_file}")


def upload_scl():
    repo = "scl"
    for filename in scl_files:
        download_wget(filename, repo)
        fasta_file = f"{filename}.fasta"
        csv_file = f"{filename}.csv"
        db = SeqIO.index(fasta_file, "fasta")
        output = []
        for _, record in tqdm(db.items()):
            description = record.description
            description = description.split()
            sequence = str(record.seq)
            seqid = description[0]
            # the label is a string in sav and scl datasets
            label = description[1].split("=")[1]
            split = description[2].split("=")[1]
            validation = description[3].split("=")[1]

            output.append({
                "seqid": seqid,
                "label": label,
                "sequence": sequence,
                "split": split.lower(),
                "validation": validation.lower(),
            })
        pd.DataFrame(output).to_csv(csv_file, index=False)
        api = HfApi()
        api.create_repo(
            repo_id=f"hazemessam/{repo}",
            repo_type="dataset",
            exist_ok=True,
        )
        api.upload_file(
            path_or_fileobj=csv_file,
            path_in_repo=csv_file,
            repo_id=f"hazemessam/{repo}",
            repo_type="dataset",
        )
        os.system(f"rm -rf {fasta_file}")
        os.system(f"rm -rf {csv_file}")


def upload_gb1():
    repo = "gb1"
    for filename in gb1_files:
        download_wget(filename, repo)
        fasta_file = f"{filename}.fasta"
        csv_file = f"{filename}.csv"
        db = SeqIO.index(fasta_file, "fasta")
        output = []
        for _, record in tqdm(db.items()):
            description = record.description
            description = description.split()
            sequence = str(record.seq)
            seqid = description[0]
            label = float(description[1].split("=")[1])
            split = description[2].split("=")[1]
            validation = description[3].split("=")[1]

            output.append({
                "seqid": seqid,
                "label": label,
                "sequence": sequence,
                "split": split.lower(),
                "validation": validation.lower(),
            })
        pd.DataFrame(output).to_csv(csv_file, index=False)
        api = HfApi()
        api.create_repo(
            repo_id=f"hazemessam/{repo}",
            repo_type="dataset",
            exist_ok=True,
        )
        api.upload_file(
            path_or_fileobj=csv_file,
            path_in_repo=csv_file,
            repo_id=f"hazemessam/{repo}",
            repo_type="dataset",
        )
        os.system(f"rm -rf {fasta_file}")
        os.system(f"rm -rf {csv_file}")


if __name__ == "__main__":
    upload_gb1()
    upload_meltome()
    upload_sav()
    upload_scl()
    upload_aav()