dsProject/dsRag/Test/TestReadWordTextAndImage.py

import os

import docx

from Util.WordImageUtil import extract_images_from_docx

# 定义常量
BLOCK_START_KEYWORDS = ('问题', '话题')
OUTPUT_IMAGE_DIR = "D:\\dsWork\\dsProject\\dsRag\\static\\Images"
OUTPUT_TXT_DIR = "D:\\dsWork\\dsProject\\dsRag\\Txt"


def read_docx(file_path):
    """读取docx文件内容，返回包含段落索引和文本的列表"""
    try:
        doc = docx.Document(file_path)
        return [(i, para.text) for i, para in enumerate(doc.paragraphs) if para.text]
    except Exception as e:
        print(f"读取docx文件出错: {str(e)}")
        return []


def is_new_block_start(para_text):
    """判断段落是否是新块的开始"""
    return para_text.startswith(BLOCK_START_KEYWORDS) and any(c.isdigit() for c in para_text[:5])

def split_into_blocks(paragraphs):
    """按段落遍历文本，发现'问题X'或'话题X'时开始分割，保留段落索引和内容"""
    blocks = []
    current_block = []
    current_indices = []
    in_block = False

    for para_idx, para_text in paragraphs:
        if is_new_block_start(para_text):
            if in_block:
                blocks.append((current_indices, '\n'.join(current_block)))
                print(
                    f"问题/话题 {len(blocks)} 管辖段落范围: {current_indices[0]} - {current_indices[-1]} (段落数: {len(current_block)})")
                print(f"当前段落块内容: {''.join(current_block)}")
                print(f"当前段落块索引: {current_indices}")
            current_block = [para_text]
            current_indices = [para_idx]
            in_block = True
        elif not current_block and (para_text.strip() or para_idx == 0):
            # 处理文档开头没有明确标记的问题/话题
            current_block = [para_text]
            current_indices = [para_idx]
        
        if in_block and para_text.strip():  # 只添加非空段落
            # 检查当前段落是否与前一个段落重复，避免重复添加
            if not current_block or para_text.strip() != current_block[-1].strip():
                current_block.append(para_text)
            # 确保 current_indices 包含所有属于当前块的段落索引
            if para_idx not in current_indices:
                current_indices.append(para_idx)

    if current_block:
        blocks.append((current_indices, '\n'.join(current_block)))
        print(
            f"问题/话题 {len(blocks)} 管辖段落范围: {current_indices[0]} - {current_indices[-1]} (段落数: {len(current_block)})")
        print(f"当前段落块内容: {''.join(current_block)}")
        print(f"当前段落块索引: {current_indices}")

    return blocks


def process_images(images, blocks):
    """处理图片与段落块的关联，并将图片信息写入对应的段落块文本中"""
    for img in images:
        img_para_idx = img['location']['paragraph_index']
        img_name = os.path.basename(img['image_path'])
        img_path = img['image_path']

        assigned = False
        for block_idx, (para_indices, block_text) in enumerate(blocks, 1):
            # 检查图片所在的段落索引是否直接包含在段落块的索引列表中
            if img_para_idx in para_indices:
                print(f"图片 {img_name} 属于段落块 {block_idx} (段落索引: {para_indices})")
                # 将图片信息添加到对应的段落块文本中
                blocks[block_idx - 1] = (para_indices, block_text + f"\n[图片: {img_name}, 路径: {img_path}]")
                assigned = True
                break
        
        # 如果图片未被分配到任何特定段落块，则将其添加到最后一个段落块
        if not assigned:
            if blocks:
                last_block_idx = len(blocks) - 1
                last_para_indices, last_block_text = blocks[last_block_idx]
                blocks[last_block_idx] = (last_para_indices, last_block_text + f"\n[图片: {img_name}, 路径: {img_path}]")
                print(f"图片 {img_name} 未找到精确匹配的段落块，已添加到最后一个段落块。")
            else:
                print(f"图片 {img_name} 未找到匹配的段落块，且没有可用的段落块。")

def save_blocks_to_txt(blocks, file_prefix, output_dir):
    """将段落块保存到TXT文件"""
    saved_count = 0
    for block_idx, (para_indices, block_text) in enumerate(blocks, 1):
        block_text = block_text.strip()
        output_file = os.path.join(output_dir, f"{file_prefix}_{block_idx}.txt")
        if save_to_txt(block_text, output_file):
            saved_count += 1
    print(f"处理完成，共保存{saved_count}个文件到目录: {output_dir}")
    return saved_count > 0

def process_document(input_path):
    """处理文档主函数"""
    paragraphs = read_docx(input_path)

    print("段落块信息:" + str(paragraphs))

    if not paragraphs:
        print("无法读取输入文件内容")
        return False

    os.makedirs(OUTPUT_IMAGE_DIR, exist_ok=True)
    print(f"图片将保存到目录: {OUTPUT_IMAGE_DIR}")
    images = extract_images_from_docx(input_path, OUTPUT_IMAGE_DIR)
    print(f"共提取到{len(images)}张图片")
    for img in images:
        print(f"图片保存至: {img['image_path']}")
        loc = img['location']
        print(f"位置信息: 段落 {loc['paragraph_index']}")

    blocks = split_into_blocks(paragraphs)
    print(f"共分割出{len(blocks)}个段落块")

    # 处理图片
    process_images(images, blocks)

    file_prefix = os.path.basename(input_path).split('.')[0].split('_')[-2] + '_' + \
                  os.path.basename(input_path).split('.')[0].split('_')[-1]
    return save_blocks_to_txt(blocks, file_prefix, OUTPUT_TXT_DIR)


# 保留原有的save_to_txt函数
def save_to_txt(content, file_path, mode='w'):
    """将内容保存到文本文件"""
    try:
        with open(file_path, mode, encoding='utf-8') as f:
            f.write(content)
        return True
    except Exception as e:
        print(f"保存文件{file_path}时出错: {str(e)}")
        return False


def process_directory(input_dir):
    """处理指定目录下的所有docx文件"""
    docx_files = [f for f in os.listdir(input_dir) if f.lower().endswith('.docx')]
    if not docx_files:
        print(f"目录中没有找到docx文件: {input_dir}")
        return False

    success_count = 0
    for docx_file in docx_files:
        input_path = os.path.join(input_dir, docx_file)
        print(f"正在处理文件: {docx_file}")
        if process_document(input_path):
            success_count += 1

    print(f"处理完成，共处理{success_count}/{len(docx_files)}个文件")
    return success_count > 0


if __name__ == "__main__":
    input_dir = '../static/Test'
    process_directory(input_dir)