- 提供DocMind文档结构化输出的doc-json结果反序列化对象，以及辅助功能函数SDK
- 源码安装
#poetry 准备环境
poetry install
#使用虚拟环境
poetry shell
# 构建
poetry build
twine check $pkg_path
# 上传
twine upload -r aliyun-pypi pkg_path --verbose- python 3.7以上 环境
集团环境
pip3 install -i http://yum.tbsite.net/aliyun-pypi/simple/ --extra-index-url https://mirrors.aliyun.com/pypi/simple/ --trusted-host=yum.tbsite.net doc_json_sdk云上环境
pip install https://docmind-api-cn-hangzhou.oss-cn-hangzhou.aliyuncs.com/sdk/doc_json_sdk-1.0.11-py3-none-any.whlrelease:
- 1.0.11 : 当前版本，修复llm_enhancement参数支持，优化代码结构，兼容Python 3.12和macOS环境，修复字体文件包含问题，增强文档解析兼容性，全面增强错误处理
- 1.0.10 : 已废除，存在其他字段缺失问题
- 1.0.9 : 已废除，存在styleId缺失问题
- 1.0.8 : 已废除，存在文档解析兼容性问题
- 1.0.4 : 已废除，不兼容Python 3.12
- 1.0.3 : 已废除，请勿使用
- 1.0.0 : 正式版本
- 0.1.9.0: 新调用和接口方式
- 0.1.8.0：修复
- 设置DocMind文档智能解析环境变量
export ALIBABA_CLOUD_ACCESS_KEY_ID=<access_key_id>
export ALIBABA_CLOUD_ACCESS_KEY_SECRET=<access_key_secret>
#调用服务- 调用文档智能解析 阿里云官网SDK调用API
加载对象可以是：
- doc-json 字符串对象
from doc_json_sdk.loader.document_model_loader import DocumentModelLoader
def test_local_json_document():
file_path = "gongshi.json"
loader = DocumentModelLoader()
document = loader.load(doc_json_fp=open(file_path,"r"))- 公有云环境调用（配置ALIBABA_CLOUD_ACCESS_KEY_ID，ALIBABA_CLOUD_ACCESS_KEY_SECRET）
from doc_json_sdk.loader.document_model_loader import DocumentModelLoader
from doc_json_sdk.handler.document_handler import DocumentExtractHandler, DocumentDigitalExtractHandler
from doc_json_sdk.handler.document_parser_handler import DocumentParserHandler, DocumentParserWithCallbackHandler
def test_document_hander():
file_path = "gongshi.png"
file_url = None
# DocumentExtractHandler:文档智能解析，DocumentDigitalExtractHandler:文档电子解析
loader = DocumentModelLoader(handler=DocumentExtractHandler())
document = loader.load(file_path=file_path,file_url=file_url)- 公式参数调用/markdown输出/json保存
from doc_json_sdk.loader.document_model_loader import DocumentModelLoader
from doc_json_sdk.handler.document_handler import DocumentExtractHandler
def test_render_formula_markdown():
file_path = "gongshi.png"
file_url = None
handler = DocumentExtractHandler()
loader = DocumentModelLoader(handler=handler)
document = loader.load(file_path=file_path,file_url=file_url,
formula_enhancement=True,
markdown_result=True,
save_json_path="/Users/sanchuan/Downloads/docmind.json")- 私有化服务调用（配置PRIVATE_DOCMIND_HOST或显式传入）
from doc_json_sdk.loader.document_model_loader import DocumentModelLoader
from doc_json_sdk.handler.document_private_handler import PrivateDocumentExtractHandler,PrivateDigitalDocumentExtractHandler
def test_private_document_hander():
file_path = "gongshi.png"
file_url = None
loader = DocumentModelLoader(handler=PrivateDocumentExtractHandler(host="127.0.0.1:7001"))
document = loader.load(file_path=file_path,file_url=file_url)使用内置函数处理为markdown
from doc_json_sdk.loader.document_model_loader import DocumentModelLoader
from doc_json_sdk.handler.document_handler import DocumentExtractHandler, DocumentDigitalExtractHandler
from doc_json_sdk.handler.document_parser_handler import DocumentParserHandler, DocumentParserWithCallbackHandler
from doc_json_sdk.render.document_model_render import DocumentModelRender
def test_render_markdown():
file_path = "gongshi.png"
file_url = None
loader = DocumentModelLoader(handler=DocumentExtractHandler())
document = loader.load(file_path=file_path,file_url=file_url,markdown_result=True)
render = DocumentModelRender(document_model=document)
with open("/Users/sanchuan/Downloads/docmind.md","w") as f:
f.write(render.render_markdown_result())可视化查看处理效果
from doc_json_sdk.loader.document_model_loader import DocumentModelLoader
from doc_json_sdk.handler.document_handler import DocumentExtractHandler, DocumentDigitalExtractHandler
from doc_json_sdk.handler.document_parser_handler import DocumentParserHandler, DocumentParserWithCallbackHandler
from doc_json_sdk.render.document_model_render import DocumentModelRender
def test_document_hander():
file_path = "gongshi.png"
file_url = None
loader = DocumentModelLoader(handler=DocumentExtractHandler())
document = loader.load(file_path=file_path,file_url=file_url)
render = DocumentModelRender(document_model=document)
render.render_image_result("/Users/sanchuan/Downloads")LayoutModel 对象分为内容信息（来源电子解析/OCR）、版面类型信息（来源OCR/NLP）、逻辑关系信息（来源NLP）
from doc_json_sdk.model.enums.layout_type_enum import LayoutTypeEnum
for layout in document:
type_enum = layout.get_layout_type_enum()
if (type_enum == LayoutTypeEnum.Elements.FOOTER or
type_enum == LayoutTypeEnum.Elements.HEADER or
type_enum == LayoutTypeEnum.Elements.NOTE):
# header and footer notes
pass
elif type_enum == LayoutTypeEnum.Elements.IMAGE:
# image with head_line or split_line
if layout.type.find("_line")!=-1:
continue
elif type_enum == LayoutTypeEnum.Elements.TABLE:
#table
pass
else:
# paragraph or note(table or figure)
pass