用pdfminer把PDF文件转化为文本文件

安装pdfminer

python3安装pdfminer3k, python2安装pdfminer2k 

具体实现

def extract_layout_by_page(pdf_path):
    # 提取页面布局
    # 　创建pdf设备对象
    laparams = LAParams()

    fp = open(pdf_path, 'rb')  # 以二進制的模式打開
    # 用文件对象来创建一个pdf文档分析器
    parser = PDFParser(fp)
    # 创建一个pdf文档
    document = PDFDocument(parser)

    # 连接分析器 与文档对象
    parser.set_document(document)
    document.set_parser(parser)
    document.initialize()
    # 检查文档是否提供txt装换，不提供就忽略
    if not document.is_extractable:
        raise PDFTextExtractionNotAllowed
    # 创建PDF资源管理器　来共享资源
    rsrcmgr = PDFResourceManager()

    device = PDFPageAggregator(rsrcmgr, laparams=laparams)
    # 创建一个pdf解释器对象
    interpreter = PDFPageInterpreter(rsrcmgr, device)

    layouts = []
    # 循环遍历列表，每次处理page的内容
    for page in document.get_pages():
        interpreter.process_page(page)
        layouts.append(device.get_result())
    return layouts


def pdf_to_txt(pdf_path, newname1):
    page_layouts = extract_layout_by_page(pdf_path)
    with open(newname1, 'w', encoding='utf-8') as f:
        for current_page in page_layouts:
            # 获取文本
            for x in current_page:
                if hasattr(x, "get_text"):
                    # result.append(x.get_text())
                    results = x.get_text()
                    # print(type(results))
                    if results == ' \n':
                        continue
                    # results=list(results)
                    # with open(newname1, 'a', encoding='utf-8') as f:
                    f.write(results)
                    # f.close()

    # 恢复文本原来换行
    with open(newname1, 'r', encoding='UTF-8-sig') as f:
        content = f.readlines()
        print(content)
    pattern = re.compile(r'\s{1,2}\n')
    newcontent = []
    for i in content:
        if bool(re.search(pattern, i)) is False:
            j = i.replace('\n', '').replace(' \n', '')
            newcontent.append(j)
        else:
            newcontent.append(i)
    f.close()
    print(newcontent)
    with open(newname1, 'w', encoding='UTF-8') as f:
        for i in newcontent:
            f.write(i)
    f.close()
    return newname1