用Python将hOCR解析为JSON

import json
from bs4 import BeautifulSoup

def hocr_to_json(hocr_file):
    with open(hocr_file, 'r', encoding='utf-8') as f:
        soup = BeautifulSoup(f, 'html.parser')
    
    json_data = {}
    json_data['pages'] = []
    
    # 解析每个页面
    for page in soup.find_all('div', class_='ocr_page'):
        page_data = {}
        page_data['width'] = int(page['data-image-width'])
        page_data['height'] = int(page['data-image-height'])
        page_data['text'] = []
        
        # 解析每个识别的文本块
        for block in page.find_all('div', class_='ocr_carea'):
            block_data = {}
            block_data['bbox'] = [int(x) for x in block['title'].split()[1:]]
            block_data['text'] = []
            
            # 解析每个识别的文本行
            for line in block.find_all('span', class_='ocr_line'):
                line_data = {}
                line_data['bbox'] = [int(x) for x in line['title'].split()[1:]]
                line_data['text'] = line.get_text()
                
                block_data['text'].append(line_data)
            
            page_data['text'].append(block_data)
        
        json_data['pages'].append(page_data)
    
    return json.dumps(json_data)

# 示例用法
hocr_file = 'example.hocr'
json_data = hocr_to_json(hocr_file)
print(json_data)