hf.py

import glob
import json
import time
from pathlib import Path

import datasets
from datasets.table import embed_table_storage
from tqdm import tqdm


def extract_number(s):
    # 移除 `.png` 后缀部分
    if s.endswith(".png"):
        s = s.rsplit('.', 1)[0]  # 删除文件扩展名

    # 查找最后一个 "_" 后的数字部分
    last_underscore_pos = s.rfind("_")
    if last_underscore_pos == -1:
        return float("inf")  # 没有找到 "_" 时，返回一个较大的值，表示该项排最后

    number_part = s[last_underscore_pos + 1:]

    # 尝试将数字部分转换为整数，如果失败则返回 float('inf')
    try:
        return int(number_part)
    except ValueError:
        return float("inf")

def sort_strings(strings):
    # 使用 sorted 排序，根据数字部分排序
    return sorted(strings, key=extract_number)


def embed_images(dataset: datasets.Dataset) -> datasets.Dataset:
    format = dataset.format
    dataset = dataset.with_format("arrow")
    dataset = dataset.map(embed_table_storage, batched=False)
    dataset = dataset.with_format(**format)
    return dataset

def read_jsonl(file_path):
    with open(file_path, 'r') as f:
        content = f.read()

    # 用来存储完整的 JSON 对象
    json_objects = []

    # 变量用于跟踪大括号的嵌套层级
    brace_count = 0
    start = 0

    # 遍历文件内容
    for idx, char in enumerate(content):
        if char == '{':
            # 增加嵌套层级
            if brace_count == 0:
                start = idx  # 记录开始位置
            brace_count += 1
        elif char == '}':
            # 减少嵌套层级
            brace_count -= 1
            if brace_count == 0:
                # 当嵌套层级归零时，意味着我们找到了一个完整的 JSON 对象
                json_str = content[start:idx+1]
                try:
                    # 将字符串解析为字典
                    json_objects.append(json.loads(json_str))
                except json.JSONDecodeError as e:
                    print(f"Error parsing JSON: {e}")
                    continue
    action_dicts = [d for d in json_objects if 'action' in d]

    return action_dicts


features = datasets.Features(
    {
        "env_id": datasets.Value("string"),
        "traj_id": datasets.Value("string"),
        "image_id": datasets.Value("string"),
        "frame_index": datasets.Value("int32"),
        "image": datasets.Image(),
        "pos": datasets.Sequence(length=3, feature=datasets.Value("float64")),
        "yaw": datasets.Value("float64"),
        "action_type": datasets.Value("string"),
        "action_value": datasets.Value("int32"),
    }
)


json_path = '/cpfs01/shared/optimal/Datasets/Openfly/OpenFly/Annotation/unseen.json'

with open(json_path, 'r') as file:
    data = json.load(file)

for item in tqdm(data, ncols=100, position=0, desc="Progress"):
    path = '/cpfs01/shared/optimal/Datasets/Openfly/OpenFly/Image/'+str(item['image_path']) + '/'
    json_path = path + 'pose.jsonl'
    path += '*.png'  # 假设图片格式为png，可以根据实际情况修改
    print(json_path)
    # 获取所有匹配的文件
    png_files = sort_strings(glob.glob(path))
    action_list = read_jsonl(json_path)  # 假设read_jsonl函数已经定义，用于读取jsonl文件
    action_list.sort(key=lambda x: int(x['action']['imageid']))  # 根据帧号排序action_list
    
    image_path = Path(item['image_path'])
    env_id_item = image_path.parent
    traj_id_item = image_path.stem
    img_id = [Path(item).stem for item in png_files]
    num = len(png_files)
    assert num == len(action_list), f"Number of images and actions do not match: {num} != {len(action_list)}"
    data_dict = {
        "env_id" : [env_id_item] * num,
        "traj_id" : [traj_id_item] * num,
        "image_id" : img_id,
        "frame_index" : [i for i in range(num)],
        "image" : ['/cpfs01/shared/optimal/Datasets/Openfly/OpenFly/Image/'+str(image_path) + '/' + img_id[i] + '.png'  for i in range(num)],
        "pos": [action_list[i]['action']['pos'][:3] for i in range(num)],
        "yaw": [action_list[i]['action']['yaw'] for i in range(num)],
        "action_type": [action_list[i]['action']['type'] for i in range(num)],
        "action_value": [action_list[i]['action']['value'] for i in range(num)]
    }
        

    path = Path("./hf")
    episode_dict = {key: data_dict[key] for key in features}
    ep_dataset = datasets.Dataset.from_dict(episode_dict, features=features, split="train")
    ep_dataset = embed_images(ep_dataset)
    ep_data_path = (path / data_dict["env_id"][0] / data_dict["traj_id"][0]).with_suffix(".parquet")
    ep_data_path.parent.mkdir(parents=True, exist_ok=True)
    ep_dataset.to_parquet(ep_data_path)
#     break

# data = datasets.load_dataset('parquet', data_files=str(ep_data_path), split='train')
# print(data)




# import datasets

# path = '/cpfs01/shared/optimal/Datasets/Openfly/OpenFly/hf/env_airsim_16/astar_data/high_average/2025-1-9_16-50-28_2007905771.parquet'

# data = datasets.load_dataset('parquet', data_files=path, split='train')
# print(data[1])