nsq_convert2txt/file_processing/file_swf.py

"""
.swf文件的处理
"""

import subprocess, os
from file_processing.file_picture import ocr_patch
from loguru import logger
from docs.config import error_number


def read_swf(file_path: str):
    swf_picture_paths, status = extract_swf(file_path)
    image_list = []
    img_str = ""
    for position, img_path in swf_picture_paths:
        if os.path.exists(img_path):
            image_list.append((position, img_path))
    print(f"image_list-->{image_list}")
    if image_list:
        ocr_text = ocr_patch(image_list)
        list_data = sorted(ocr_text.items(), key=lambda x: x[0], reverse=False)
        img_str = '\n'.join([v[1] for v in list_data])
    return img_str, status


def extract_swf(file_path):
    """
    解压到.swf文件夹中,返回文件路径列表(有序)
    :param file_path:
    :return:
    """
    swf_picture_paths = []
    swf_to_picture_dir = os.path.dirname(file_path)

    try:
        # 抽取swf，获取图片路径
        output = subprocess.check_output('./extract.sh %s %s' % (file_path, swf_to_picture_dir), shell=True)
        output_string = output.decode()
        picture_num = output_string.split('\n')
        if len(picture_num) > 5:
            return str(output_string)
        else:
            if picture_num[0] != '':
                for i in picture_num[0].split(','):
                    swf_to_picture_path = os.path.join(swf_to_picture_dir, '%s.jpg' % (i.strip()))
                    swf_picture_paths.append(swf_to_picture_path)

            if picture_num[1] != '':
                for i in picture_num[1].split(','):
                    swf_to_picture_path = os.path.join(swf_to_picture_dir, '%s.png' % (i.strip()))
                    swf_picture_paths.append(swf_to_picture_path)

    except subprocess.CalledProcessError as e:
        logger.error('swf文件解析出错')
        logger.error(e)
        return [], error_number["解析错误"]

    # 对图片进行排序
    try:
        swf_picture_dict = dict([(int(p.split("/")[-1].split(".")[0]), p) for p in swf_picture_paths])
        print(f"swf_picture_dict-->{swf_picture_dict}")
        swf_picture_paths = sorted(swf_picture_dict.items(), key=lambda x: x[0])
    except Exception as e:
        logger.warning(f"{e}")
    return swf_picture_paths, error_number["成功"]