Update magic_pdf/__init__.py, magic_pdf/config.ini, magic_pdf/tmp.py,...

Update magic_pdf/__init__.py, magic_pdf/config.ini, magic_pdf/tmp.py, magic_pdf/pdf_parse_by_ocr.py, magic_pdf/pdf_parse_by_txt.py, magic_pdf/user_api.py, magic_pdf/pdf_parse_union_core.py, magic_pdf/pdf_parse_union_core_v2.py, magic_pdf/config/__init__.py, magic_pdf/config/enums.py, magic_pdf/config/exceptions.py, magic_pdf/data/__init__.py, magic_pdf/data/schemas.py, magic_pdf/data/dataset.py, magic_pdf/data/utils.py, magic_pdf/data/read_api.py, magic_pdf/data/data_reader_writer/__init__.py, magic_pdf/data/data_reader_writer/base.py, magic_pdf/data/data_reader_writer/filebase.py, magic_pdf/data/data_reader_writer/s3.py, magic_pdf/data/data_reader_writer/multi_bucket_s3.py, magic_pdf/data/io/__init__.py, magic_pdf/data/io/base.py, magic_pdf/data/io/s3.py, magic_pdf/data/io/http.py, magic_pdf/dict2md/__init__.py, magic_pdf/dict2md/ocr_vllm_client.py, magic_pdf/dict2md/ocr_vllm_server.py, magic_pdf/dict2md/ocr_mkcontent.py, magic_pdf/dict2md/mkcontent.py, magic_pdf/dict2md/ocr_server.py, magic_pdf/dict2md/ocr_client.py, magic_pdf/filter/__init__.py, magic_pdf/filter/pdf_classify_by_type.py, magic_pdf/filter/pdf_meta_scan.py, magic_pdf/integrations/__init__.py, magic_pdf/integrations/rag/__init__.py, magic_pdf/integrations/rag/api.py, magic_pdf/integrations/rag/utils.py, magic_pdf/integrations/rag/type.py, magic_pdf/layout/__init__.py, magic_pdf/layout/layout_det_utils.py, magic_pdf/layout/layout_spiler_recog.py, magic_pdf/layout/mcol_sort.py, magic_pdf/layout/layout_sort.py, magic_pdf/layout/bbox_sort.py, magic_pdf/libs/__init__.py, magic_pdf/libs/boxbase.py, magic_pdf/libs/calc_span_stats.py, magic_pdf/libs/clean_memory.py, magic_pdf/libs/convert_utils.py, magic_pdf/libs/detect_language_from_model.py, magic_pdf/libs/coordinate_transform.py, magic_pdf/libs/drop_tag.py, magic_pdf/libs/nlp_utils.py, magic_pdf/libs/ocr_content_type.py, magic_pdf/libs/hash_utils.py, magic_pdf/libs/path_utils.py, magic_pdf/libs/pdf_check.py, magic_pdf/libs/language.py, magic_pdf/libs/markdown_utils.py, magic_pdf/libs/drop_reason.py, magic_pdf/libs/pdf_image_tools.py, magic_pdf/libs/config_reader.py, magic_pdf/libs/Constants.py, magic_pdf/libs/local_math.py, magic_pdf/libs/ModelBlockTypeEnum.py, magic_pdf/libs/draw_bbox.py, magic_pdf/libs/vis_utils.py, magic_pdf/libs/textbase.py, magic_pdf/libs/safe_filename.py, magic_pdf/libs/MakeContentConfig.py, magic_pdf/libs/version.py, magic_pdf/libs/json_compressor.py, magic_pdf/libs/commons.py, magic_pdf/model/__init__.py, magic_pdf/model/doc_analyze_by_custom_model.py, magic_pdf/model/magic_model.py, magic_pdf/model/model_list.py, magic_pdf/model/pp_structure_v2.py, magic_pdf/model/ppTableModel.py, magic_pdf/model/pdf_extract_kit.py, magic_pdf/model/pek_sub_modules/__init__.py, magic_pdf/model/pek_sub_modules/self_modify.py, magic_pdf/model/pek_sub_modules/post_process.py, magic_pdf/model/pek_sub_modules/layoutlmv3/__init__.py, magic_pdf/model/pek_sub_modules/layoutlmv3/backbone.py, magic_pdf/model/pek_sub_modules/layoutlmv3/beit.py, magic_pdf/model/pek_sub_modules/layoutlmv3/deit.py, magic_pdf/model/pek_sub_modules/layoutlmv3/model_init.py, magic_pdf/model/pek_sub_modules/layoutlmv3/visualizer.py, magic_pdf/model/pek_sub_modules/layoutlmv3/rcnn_vl.py, magic_pdf/model/pek_sub_modules/layoutlmv3/layoutlmft/__init__.py, magic_pdf/model/pek_sub_modules/layoutlmv3/layoutlmft/data/__init__.py, magic_pdf/model/pek_sub_modules/layoutlmv3/layoutlmft/data/funsd.py, magic_pdf/model/pek_sub_modules/layoutlmv3/layoutlmft/data/data_collator.py, magic_pdf/model/pek_sub_modules/layoutlmv3/layoutlmft/data/cord.py, magic_pdf/model/pek_sub_modules/layoutlmv3/layoutlmft/data/image_utils.py, magic_pdf/model/pek_sub_modules/layoutlmv3/layoutlmft/data/xfund.py, magic_pdf/model/pek_sub_modules/layoutlmv3/layoutlmft/models/__init__.py, magic_pdf/model/pek_sub_modules/layoutlmv3/layoutlmft/models/layoutlmv3/__init__.py, magic_pdf/model/pek_sub_modules/layoutlmv3/layoutlmft/models/layoutlmv3/tokenization_layoutlmv3.py, magic_pdf/model/pek_sub_modules/layoutlmv3/layoutlmft/models/layoutlmv3/tokenization_layoutlmv3_fast.py, magic_pdf/model/pek_sub_modules/layoutlmv3/layoutlmft/models/layoutlmv3/configuration_layoutlmv3.py, magic_pdf/model/pek_sub_modules/structeqtable/__init__.py, magic_pdf/model/pek_sub_modules/structeqtable/StructTableModel.py, magic_pdf/para/__init__.py, magic_pdf/para/commons.py, magic_pdf/para/draw.py, magic_pdf/para/layout_match_processor.py, magic_pdf/para/raw_processor.py, magic_pdf/para/title_processor.py, magic_pdf/para/para_split.py, magic_pdf/para/denoise.py, magic_pdf/para/block_continuation_processor.py, magic_pdf/para/block_termination_processor.py, magic_pdf/para/para_pipeline.py, magic_pdf/para/para_split_v2.py, magic_pdf/para/para_split_v3.py, magic_pdf/para/stats.py, magic_pdf/para/exceptions.py, magic_pdf/parse/__init__.py, magic_pdf/parse/excel_parse.py, magic_pdf/parse/common_parse.py, magic_pdf/parse/ofd_parse.py, magic_pdf/parse/pdf_client.py, magic_pdf/pipe/__init__.py, magic_pdf/pipe/AbsPipe.py, magic_pdf/pipe/OCRPipe.py, magic_pdf/pipe/TXTPipe.py, magic_pdf/pipe/UNIPipe.py, magic_pdf/post_proc/__init__.py, magic_pdf/post_proc/pdf_post_filter.py, magic_pdf/post_proc/remove_spaces_html.py, magic_pdf/post_proc/remove_footnote.py, magic_pdf/post_proc/detect_para.py, magic_pdf/pre_proc/__init__.py, magic_pdf/pre_proc/post_layout_split.py, magic_pdf/pre_proc/citationmarker_remove.py, magic_pdf/pre_proc/detect_equation.py, magic_pdf/pre_proc/detect_footer_by_model.py, magic_pdf/pre_proc/cut_image.py, magic_pdf/pre_proc/construct_page_dict.py, magic_pdf/pre_proc/ocr_detect_layout.py, magic_pdf/pre_proc/pdf_pre_filter.py, magic_pdf/pre_proc/ocr_dict_merge.py, magic_pdf/pre_proc/ocr_span_list_modify.py, magic_pdf/pre_proc/remove_bbox_overlap.py, magic_pdf/pre_proc/remove_colored_strip_bbox.py, magic_pdf/pre_proc/remove_footer_header.py, magic_pdf/pre_proc/detect_header.py, magic_pdf/pre_proc/detect_page_number.py, magic_pdf/pre_proc/detect_tables.py, magic_pdf/pre_proc/detect_footer_header_by_statistics.py, magic_pdf/pre_proc/detect_footnote.py, magic_pdf/pre_proc/remove_rotate_bbox.py, magic_pdf/pre_proc/resolve_bbox_conflict.py, magic_pdf/pre_proc/solve_line_alien.py, magic_pdf/pre_proc/statistics.py, magic_pdf/pre_proc/detect_images.py, magic_pdf/pre_proc/equations_replace.py, magic_pdf/pre_proc/fix_image.py, magic_pdf/pre_proc/ocr_detect_all_bboxes.py, magic_pdf/pre_proc/fix_table.py, magic_pdf/pre_proc/main_text_font.py, magic_pdf/resources/fasttext-langdetect/lid.176.ftz, magic_pdf/resources/model_config/model_configs.yaml, magic_pdf/resources/model_config/layoutlmv3/layoutlmv3_base_inference.yaml, magic_pdf/resources/model_config/UniMERNet/demo.yaml, magic_pdf/rw/__init__.py, magic_pdf/rw/AbsReaderWriter.py, magic_pdf/rw/DiskReaderWriter.py, magic_pdf/rw/draw_ofd.py, magic_pdf/rw/ofdtemplate.py, magic_pdf/rw/pdf_parse.py, magic_pdf/rw/draw_pdf.py, magic_pdf/rw/S3ReaderWriter.py, magic_pdf/spark/__init__.py, magic_pdf/spark/spark_api.py, magic_pdf/tools/__init__.py, magic_pdf/tools/cli.py, magic_pdf/tools/cli_dev.py, magic_pdf/tools/common.py, magic_pdf/tools/file_deal.py, magic_pdf/tools/img_deal.py, magic_pdf/tools/find_seal_img.py, magic_pdf/tools/font_tools.py, magic_pdf/tools/file_parser.py, magic_pdf/tools/parameter_parser.py, magic_pdf/tools/ofd.py, magic_pdf/tools/pdf_server.py, magic_pdf/tools/ofd_parser.py, magic_pdf/utils/__init__.py, magic_pdf/utils/annotations.py files

magic_pdf/tools/cli_dev.py

+import json as json_parse
+import os
+from pathlib import Path
+
+import click
+
+import magic_pdf.model as model_config
+from magic_pdf.libs.config_reader import get_s3_config
+from magic_pdf.libs.path_utils import (parse_s3_range_params, parse_s3path,
+                                       remove_non_official_s3_args)
+from magic_pdf.libs.version import __version__
+from magic_pdf.rw.AbsReaderWriter import AbsReaderWriter
+from magic_pdf.rw.DiskReaderWriter import DiskReaderWriter
+from magic_pdf.rw.S3ReaderWriter import S3ReaderWriter
+from magic_pdf.tools.common import do_parse, parse_pdf_methods
+
+
+def read_s3_path(s3path):
+    bucket, key = parse_s3path(s3path)
+
+    s3_ak, s3_sk, s3_endpoint = get_s3_config(bucket)
+    s3_rw = S3ReaderWriter(s3_ak, s3_sk, s3_endpoint, 'auto',
+                           remove_non_official_s3_args(s3path))
+    may_range_params = parse_s3_range_params(s3path)
+    if may_range_params is None or 2 != len(may_range_params):
+        byte_start, byte_end = 0, None
+    else:
+        byte_start, byte_end = int(may_range_params[0]), int(
+            may_range_params[1])
+    return s3_rw.read_offset(
+        remove_non_official_s3_args(s3path),
+        byte_start,
+        byte_end,
+    )
+
+
+@click.group()
+@click.version_option(__version__, '--version', '-v', help='显示版本信息')
+def cli():
+    pass
+
+
+@cli.command()
+@click.option(
+    '-j',
+    '--jsonl',
+    'jsonl',
+    type=str,
+    help='输入 jsonl 路径，本地或者 s3 上的文件',
+    required=True,
+)
+@click.option(
+    '-m',
+    '--method',
+    'method',
+    type=parse_pdf_methods,
+    help='指定解析方法。txt: 文本型 pdf 解析方法， ocr: 光学识别解析 pdf, auto: 程序智能选择解析方法',
+    default='auto',
+)
+@click.option(
+    '-o',
+    '--output-dir',
+    'output_dir',
+    type=click.Path(),
+    required=True,
+    help='输出到本地目录',
+)
+def jsonl(jsonl, method, output_dir):
+    model_config.__use_inside_model__ = False
+    if jsonl.startswith('s3://'):
+        jso = json_parse.loads(read_s3_path(jsonl).decode('utf-8'))
+    else:
+        with open(jsonl) as f:
+            jso = json_parse.loads(f.readline())
+    os.makedirs(output_dir, exist_ok=True)
+    s3_file_path = jso.get('file_location')
+    if s3_file_path is None:
+        s3_file_path = jso.get('path')
+    pdf_file_name = Path(s3_file_path).stem
+    pdf_data = read_s3_path(s3_file_path)
+
+    print(pdf_file_name, jso, method)
+    do_parse(
+        output_dir,
+        pdf_file_name,
+        pdf_data,
+        jso['doc_layout_result'],
+        method,
+        False,
+        f_dump_content_list=True,
+        f_draw_model_bbox=True,
+    )
+
+
+@cli.command()
+@click.option(
+    '-p',
+    '--pdf',
+    'pdf',
+    type=click.Path(exists=True),
+    required=True,
+    help='本地 PDF 文件',
+)
+@click.option(
+    '-j',
+    '--json',
+    'json_data',
+    type=click.Path(exists=True),
+    required=True,
+    help='本地模型推理出的 json 数据',
+)
+@click.option('-o',
+              '--output-dir',
+              'output_dir',
+              type=click.Path(),
+              required=True,
+              help='本地输出目录')
+@click.option(
+    '-m',
+    '--method',
+    'method',
+    type=parse_pdf_methods,
+    help='指定解析方法。txt: 文本型 pdf 解析方法， ocr: 光学识别解析 pdf, auto: 程序智能选择解析方法',
+    default='auto',
+)
+def pdf(pdf, json_data, output_dir, method):
+    model_config.__use_inside_model__ = False
+    full_pdf_path = os.path.realpath(pdf)
+    os.makedirs(output_dir, exist_ok=True)
+
+    def read_fn(path):
+        disk_rw = DiskReaderWriter(os.path.dirname(path))
+        return disk_rw.read(os.path.basename(path), AbsReaderWriter.MODE_BIN)
+
+    model_json_list = json_parse.loads(read_fn(json_data).decode('utf-8'))
+
+    file_name = str(Path(full_pdf_path).stem)
+    pdf_data = read_fn(full_pdf_path)
+    do_parse(
+        output_dir,
+        file_name,
+        pdf_data,
+        model_json_list,
+        method,
+        False,
+        f_dump_content_list=True,
+        f_draw_model_bbox=True,
+    )
+
+
+if __name__ == '__main__':
+    cli()

magic_pdf/tools/common.py

+import copy
+import json as json_parse
+import os
+
+import click
+from loguru import logger
+
+import magic_pdf.model as model_config
+from magic_pdf.libs.draw_bbox import (draw_layout_bbox, draw_line_sort_bbox,
+                                      draw_model_bbox, draw_span_bbox)
+from magic_pdf.libs.MakeContentConfig import DropMode, MakeMode
+from magic_pdf.pipe.OCRPipe import OCRPipe
+from magic_pdf.pipe.TXTPipe import TXTPipe
+from magic_pdf.pipe.UNIPipe import UNIPipe
+from magic_pdf.rw.AbsReaderWriter import AbsReaderWriter
+from magic_pdf.rw.DiskReaderWriter import DiskReaderWriter
+from magic_pdf.post_proc.remove_spaces_html import remove_extra_spaces_html_txt
+
+
+def prepare_env(output_dir, pdf_file_name, method):
+    local_parent_dir = os.path.join(output_dir, pdf_file_name, method)
+
+    local_image_dir = os.path.join(str(local_parent_dir), 'images')
+    local_md_dir = local_parent_dir
+    os.makedirs(local_image_dir, exist_ok=True)
+    os.makedirs(local_md_dir, exist_ok=True)
+    return local_image_dir, local_md_dir
+
+def remove_empty_lines_from_file(file_path):
+    # 读取文件内容
+    with open(file_path, 'r', encoding='utf-8') as file:
+        lines = file.readlines()
+
+    # 删除空行
+    non_empty_lines = [line for line in lines if line.strip()]
+
+    # 将非空行写回原文件
+    with open(file_path, 'w', encoding='utf-8') as file:
+        file.writelines(non_empty_lines)
+
+
+def do_parse(
+    ocr_status,
+    config_path,
+    output_dir,
+    pdf_file_name,
+    pdf_bytes,
+    model_list,
+    parse_method,
+    debug_able,
+    model,
+    f_draw_span_bbox=True,
+    f_draw_layout_bbox=True,
+    f_dump_md=True,
+    f_dump_middle_json=True,
+    f_dump_model_json=True,
+    f_dump_orig_pdf=True,
+    f_dump_content_list=True,
+    f_make_md_mode=MakeMode.MM_MD,
+    f_draw_model_bbox=False,
+    f_draw_line_sort_bbox=False,
+    start_page_id=0,
+    end_page_id=None,
+    lang=None,
+    layout_model=None,
+    formula_enable=None,
+    table_enable=None,
+):
+    if debug_able:
+        logger.warning('debug mode is on')
+        f_draw_model_bbox = True
+        f_draw_line_sort_bbox = True
+
+    orig_model_list = copy.deepcopy(model_list)
+    local_image_dir, local_md_dir = prepare_env(output_dir, pdf_file_name,
+                                                parse_method)
+
+    image_writer, md_writer = DiskReaderWriter(
+        local_image_dir), DiskReaderWriter(local_md_dir)
+    image_dir = str(os.path.basename(local_image_dir))
+
+    if parse_method == 'auto':
+        jso_useful_key = {'_pdf_type': '', 'model_list': model_list}
+        pipe = UNIPipe(pdf_bytes, jso_useful_key, image_writer, is_debug=True,
+                       start_page_id=start_page_id, end_page_id=end_page_id, lang=lang,
+                       layout_model=layout_model, formula_enable=formula_enable, table_enable=table_enable)
+    elif parse_method == 'txt':
+        pipe = TXTPipe(pdf_bytes, model_list, image_writer, is_debug=True,
+                       start_page_id=start_page_id, end_page_id=end_page_id, lang=lang,
+                       layout_model=layout_model, formula_enable=formula_enable, table_enable=table_enable)
+    elif parse_method == 'ocr':
+        pipe = OCRPipe(pdf_bytes, model_list, image_writer, is_debug=True,
+                       start_page_id=start_page_id, end_page_id=end_page_id, lang=lang,
+                       layout_model=layout_model, formula_enable=formula_enable, table_enable=table_enable)
+    else:
+        logger.error('unknown parse method')
+        exit(1)
+
+    pipe.pipe_classify()
+
+    if len(model_list) == 0:
+        if model_config.__use_inside_model__:
+            pipe.pipe_analyze(model)
+            orig_model_list = copy.deepcopy(pipe.model_list)
+        else:
+            logger.error('need model list input')
+            exit(2)
+
+    pipe.pipe_parse(ocr_status,config_path,local_image_dir)
+    pdf_info = pipe.pdf_mid_data['pdf_info']
+    if f_draw_layout_bbox:
+        draw_layout_bbox(pdf_info, pdf_bytes, local_md_dir, pdf_file_name)
+
+    md_content = pipe.pipe_mk_markdown(image_dir,
+                                       drop_mode=DropMode.NONE,
+                                       md_make_mode=f_make_md_mode)
+
+    try:
+        txt_file = f'{pdf_file_name}.txt'
+        md_writer.write(
+            content=md_content,
+            path=txt_file,
+            mode=AbsReaderWriter.MODE_TXT,
+        )
+        filepath = os.path.join(str(local_md_dir), f'{pdf_file_name}.txt')
+        # logger.info(f'txt文件保存在filepath:{filepath}')
+        remove_empty_lines_from_file(filepath)
+
+        if f_dump_orig_pdf:
+            md_writer.write(
+                content=pdf_bytes,
+                path=f'{pdf_file_name}_origin.pdf',
+                mode=AbsReaderWriter.MODE_BIN,
+            )
+        remove_extra_spaces_html_txt(filepath)
+
+
+        return filepath
+    except Exception as e:
+        logger.error(f'{pdf_file_name}导出txt文件失败，具体原因为：\n{e}')
+        return None
+
+parse_pdf_methods = click.Choice(['ocr', 'txt', 'auto'])

magic_pdf/tools/file_deal.py

+#!/usr/bin/env python
+#-*- coding: utf-8 -*-
+#PROJECT_NAME: D:\code\easyofd\easyofd\parser
+#CREATE_TIME: 2023-07-27 
+#E_MAIL: renoyuan@foxmail.com
+#AUTHOR: reno 
+#NOTE:  文件处理
+import os
+import base64
+import shutil
+from typing import Any
+from uuid import uuid1
+
+import xmltodict
+import zipfile
+
+    
+class FileRead(object):
+    """
+    文件读取，清除
+    'root': OFD.xml 
+    "root_doc" Doc_0/Document.xml
+    xml_path : xml_obj
+    other_path : b64string
+    """
+    def __init__(self, ofdb64:str):
+        self.ofdbyte = base64.b64decode(ofdb64) 
+        pid=os.getpid()
+        self.name = f"{pid}_{str(uuid1())}.ofd"
+        self.pdf_name = self.name.replace(".ofd",".pdf")
+        self.zip_path = f"{os.getcwd()}/{self.name}"
+        self.unzip_path = ""
+        self.file_tree = {}
+    
+    def unzip_file(self):
+        """
+        :param zip_path: ofd格式文件路径
+        :param unzip_path: 解压后的文件存放目录
+        :return: unzip_path
+        """
+        with open(self.zip_path,"wb") as f:
+            f.write(self.ofdbyte)
+        self.unzip_path = self.zip_path.split('.')[0]
+        with zipfile.ZipFile(self.zip_path, 'r') as f:
+            for file in f.namelist():
+                f.extract(file, path=self.unzip_path)
+        if self.save_xml:
+            print("saving xml {}".format(self.xml_name))
+            with zipfile.ZipFile(self.zip_path, 'r') as f:
+                for file in f.namelist():
+                    f.extract(file, path=self.xml_name)
+       
+    def buld_file_tree(self):
+        "xml读取对象其他b64"
+        self.file_tree["root"] = self.unzip_path
+        self.file_tree["pdf_name"] = self.pdf_name
+        for root, dirs, files in os.walk(self.unzip_path):
+            for file in files:
+                
+                abs_path = os.path.join(root,file)
+                # 资源文件 则 b64 xml 则  xml——obj
+                self.file_tree[abs_path] = str(base64.b64encode(open(f"{abs_path}","rb").read()),"utf-8")  \
+                    if "xml" not in file else xmltodict.parse(open(f"{abs_path}" , "r", encoding="utf-8").read())
+        self.file_tree["root_doc"] = os.path.join(self.unzip_path,"OFD.xml") if os.path.join(self.unzip_path,"OFD.xml") in self.file_tree else ""
+  
+        if os.path.exists(self.unzip_path):
+            shutil.rmtree(self.unzip_path)
+       
+        if os.path.exists(self.zip_path):
+            os.remove(self.zip_path)
+                   
+    def __call__(self, *args: Any, **kwds: Any) -> Any:
+        self.save_xml=kwds.get("save_xml",False)
+        self.xml_name=kwds.get("xml_name")
+    
+        self.unzip_file()
+        self.buld_file_tree()
+        return self.file_tree 
+
+if __name__ == "__main__":
+    with open(r"D:/code/easyofd/test/增值税电子专票5.ofd","rb") as f:
+        ofdb64 = str(base64.b64encode(f.read()),"utf-8")
+    a = FileRead(ofdb64)()
+    print(list(a.keys()))
\ No newline at end of file

magic_pdf/tools/file_parser.py

+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+# PROJECT_NAME: D:\code\easyofd\easyofd\parser
+# CREATE_TIME: 2023-07-27
+# E_MAIL: renoyuan@foxmail.com
+# AUTHOR: reno
+# NOTE: 每种类型的文件定义一个解析器
+
+import sys
+
+sys.path.insert(0, "..")
+import logging
+import os
+import traceback
+import base64
+import re
+from typing import Any
+from magic_pdf.tools.parameter_parser import ParameterParser
+logger = logging.getLogger("root")
+
+
+class FileParserBase(object):
+    """xml解析"""
+
+    def __init__(self, xml_obj):
+        assert xml_obj
+        self.ofd_param = ParameterParser()
+        self.xml_obj = xml_obj
+        # print(xml_obj)
+
+    def recursion_ext(self, need_ext_obj, ext_list, key):
+        """
+        抽取需要xml要素
+        need_ext_obj : xmltree
+        ext_list: data container
+        key: key
+        """
+
+        if isinstance(need_ext_obj, dict):
+
+            for k, v in need_ext_obj.items():
+                if k == key:
+
+                    if isinstance(v, (dict, str)):
+                        ext_list.append(v)
+                    elif isinstance(v, list):
+                        ext_list.extend(v)
+
+
+                else:
+
+                    if isinstance(v, dict):
+                        self.recursion_ext(v, ext_list, key)
+                    elif isinstance(v, list):
+                        for cell in v:
+                            self.recursion_ext(cell, ext_list, key)
+                    else:
+
+                        pass
+        else:
+
+            print(type(need_ext_obj))
+
+
+class OFDFileParser(FileParserBase):
+    def __call__(self):
+        info = {}
+        # DocRoot 
+        doc_root: list = []
+        doc_root_key = "ofd:DocRoot"
+        # print(self.xml_obj,doc_root)
+        self.recursion_ext(self.xml_obj, doc_root, doc_root_key)
+        info["doc_root"] = doc_root
+
+        signatures: list = []
+        signatures_key = "ofd:Signatures"
+        self.recursion_ext(self.xml_obj, signatures, signatures_key)
+        info["signatures"] = signatures
+
+        # ofd:Creator 
+        creator: list = []
+        creator_key = "ofd:Creator"
+        self.recursion_ext(self.xml_obj, creator, creator_key)
+        info["creator"] = creator
+
+        # ofd:CreationDate 
+        reation_date: list = []
+        creation_date_key = "ofd:CreationDate"
+        self.recursion_ext(self.xml_obj, reation_date, creation_date_key)
+        info["creationDate"] = reation_date
+
+        return info
+
+
+class DocumentFileParser(FileParserBase):
+    """
+    Document 为doc内的根节点 包含：
+    1 文件的路径 2 doc的size 
+    """
+
+    def loc2page_no(self, loc, idx):
+        pg_no = re.search(r"\d+", loc)
+        if pg_no:
+            pg_no = int(pg_no.group())
+        else:
+            pg_no = idx
+        return pg_no
+
+    def __call__(self):
+        document_info = {}
+
+        # size 
+        physical_box: list = []
+        physical_box_key = "ofd:PhysicalBox"
+        self.recursion_ext(self.xml_obj, physical_box, physical_box_key)
+        document_info["size"] = physical_box[0] if physical_box else ""
+
+        # ofd:PublicRes路径 包含字体路径信息 
+        public_res: list = []
+        public_res_key = "ofd:PublicRes"
+        self.recursion_ext(self.xml_obj, public_res, public_res_key)
+        document_info["public_res"] = public_res
+
+        # ofd:DocumentRes路径  包含静态资源图片
+        document_res: list = []
+        document_res_key = "ofd:DocumentRes"
+        self.recursion_ext(self.xml_obj, document_res, document_res_key)
+        document_info["document_res"] = document_res
+
+        # ofd:Page 正文
+        page: list = []
+        page_id_map = {}
+        apage_key = "ofd:Page"
+        self.recursion_ext(self.xml_obj, page, apage_key)
+        if page:
+            page_id_map = {
+                i.get("@ID"): self.loc2page_no(i.get("@BaseLoc"), idx)
+                for idx, i in enumerate(page)
+            }
+            page = [i.get("@BaseLoc") if isinstance(i, dict) else i for i in page]
+
+        document_info["page"] = page
+        document_info["page_id_map"] = page_id_map
+
+        tpls: list = []
+        template_page_key = "ofd:TemplatePage"
+        self.recursion_ext(self.xml_obj, tpls, template_page_key)
+        if tpls:
+            tpls = [i.get("@BaseLoc") if isinstance(i, dict) else i for i in tpls]
+        document_info["tpls"] = tpls
+
+        return document_info
+
+
+class ContentFileParser(FileParserBase):
+    """
+    Parser Contents&tpls
+
+    """
+
+    def fetch_cell_info(self, row, TextObject):
+        """fetch_cell_info"""
+        cell_d = {}
+        cell_d = {}
+        cell_d["ID"] = row['@ID']  # 字体
+        # 字体字形信息
+        if row.get("ofd:CGTransform"):
+            Glyphs_d = {
+                "Glyphs": row.get("ofd:CGTransform").get("ofd:Glyphs"),
+                "GlyphCount": row.get("ofd:CGTransform").get("@GlyphCount"),
+                "CodeCount": row.get("ofd:CGTransform").get("@CodeCount"),
+                "CodePosition": row.get("ofd:CGTransform").get("@CodePosition")
+            }
+            cell_d["Glyphs_d"] = Glyphs_d
+
+        cell_d["pos"] = [float(pos_i) for pos_i in row['@Boundary'].split(" ")]  # 文本框
+        if row.get('ofd:Clips', {}).get('ofd:Clip', {}).get('ofd:Area', {}).get('ofd:Path', {}):
+            cell_d["clips_pos"] = [float(pos_i) for pos_i in
+                                   row.get('ofd:Clips', {}).get('ofd:Clip', {}).get('ofd:Area', {}).get('ofd:Path',
+                                                                                                        {}).get(
+                                       '@Boundary', "").split(" ")]
+        cell_d["text"] = str(TextObject.get('#text'))
+        cell_d["font"] = row['@Font']  # 字体
+        cell_d["size"] = float(row['@Size'])  # 字号
+        # print("row", row)
+
+        color =self.ofd_param("ofd:FillColor", row).get("@Value", "0 0 0")
+
+        cell_d["color"] = tuple(color.split(" "))  # 颜色
+        cell_d["DeltaY"] = TextObject.get("@DeltaY", "")  # y 轴偏移量 竖版文字表示方法之一
+        cell_d["DeltaX"] = TextObject.get("@DeltaX", "")  # x 轴偏移量
+        cell_d["CTM"] = row.get("@CTM", "")  # 平移矩阵换
+
+        cell_d["X"] = TextObject.get("@X", "")  # X 文本之与文本框距离
+        cell_d["Y"] = TextObject.get("@Y", "")  # Y 文本之与文本框距离
+        return cell_d
+
+    def __call__(self) -> list:
+        """
+       
+        输出主体坐标和文字信息 cell_list
+        [{"pos":row['@Boundary'].split(" "),
+                    "text":row['ofd:TextCode'].get('#text'),
+                    "font":row['@Font'],
+                    "size":row['@Size'],}]
+        """
+        text_list = []
+        img_list = []
+        line_list = []
+
+        content_d = {
+            "text_list": text_list,
+            "img_list": img_list,
+            "line_list": line_list,
+        }
+
+        text: list = []  # 正文
+        text_key = "ofd:TextObject"
+        self.recursion_ext(self.xml_obj, text, text_key)
+
+        if text:
+            for row in text:
+                # print("row", row.get('ofd:TextCode', {}))
+                if isinstance(row.get('ofd:TextCode', {}), list):
+                    for _i in row.get('ofd:TextCode', {}):
+                        if not _i.get('#text'):
+                            continue
+                        cell_d = self.fetch_cell_info(row, _i)
+                        text_list.append(cell_d)
+
+                elif isinstance(row.get('ofd:TextCode', {}), dict):
+                    if not row.get('ofd:TextCode', {}).get('#text'):
+                        continue
+                    cell_d = self.fetch_cell_info(row, row.get('ofd:TextCode', {}))
+                    text_list.append(cell_d)
+
+                else:
+                    logger.error(f"'ofd:TextCode' format nonsupport  {row.get('ofd:TextCode', {})}")
+                    continue
+
+        line: list = []  # 路径线条
+        line_key = "ofd:PathObject"
+        self.recursion_ext(self.xml_obj, line, line_key)
+
+        if line:
+            # print(line)
+            for _i in line:
+                line_d = {}
+                # print("line",_i)
+                line_d["ID"] = _i.get("@ID", "")  # 图片id
+                line_d["pos"] = [float(pos_i) for pos_i in _i['@Boundary'].split(" ")]  # 平移矩阵换
+                line_d["LineWidth"] = _i.get("@LineWidth", "")  # 图片id
+                line_d["AbbreviatedData"] = _i.get("ofd:AbbreviatedData", "")  # 路径指令
+                line_d["FillColor"] = self.ofd_param("ofd:FillColor", _i).get('@Value', "0 0 0").split(" ")  # 颜色
+                line_d["StrokeColor"] = self.ofd_param("ofd:StrokeColor", _i).get('@Value', "0 0 0")  # 颜色
+
+                line_list.append(line_d)
+
+        img: list = []  # 图片
+        img_key = "ofd:ImageObject"
+        self.recursion_ext(self.xml_obj, img, img_key)
+
+        if img:
+            for _i in img:
+                img_d = {}
+                img_d["CTM"] = _i.get("@CTM", "")  # 平移矩阵换
+                img_d["ID"] = _i.get("ID", "")  # 图片id
+                img_d["ResourceID"] = _i.get("@ResourceID", "")  # 图片id
+                img_d["pos"] = [float(pos_i) for pos_i in _i['@Boundary'].split(" ")]  # 平移矩阵换
+                img_list.append(img_d)
+
+        return content_d
+
+
+class DocumentResFileParser(FileParserBase):
+    """
+    Parser DocumentRes 抽取里面图片信息
+
+    """
+
+    def __call__(self):
+        info = {}
+        muti_media: list = []
+        muti_media_key = "ofd:MultiMedia"
+        self.recursion_ext(self.xml_obj, muti_media, muti_media_key)
+        if muti_media:
+            for media in muti_media:
+                name = media.get("ofd:MediaFile", "")
+                info[media.get("@ID")] = {
+                    "format": media.get("@Format", ""),
+                    "wrap_pos": media.get("@wrap_pos", ""),
+                    "type": media.get("@Type", ""),
+                    "suffix": os.path.splitext(name)[-1].replace(".", ""),  # 文件后缀名
+                    "fileName": name,
+                }
+        return info
+
+
+class PublicResFileParser(FileParserBase):
+    """
+    Parser PublicRes 抽取里面 获取公共信息 字体信息
+
+    """
+
+    def __call__(self):
+        info = {}
+        public_res: list = []
+        public_res_key = "ofd:Font"
+        self.recursion_ext(self.xml_obj, public_res, public_res_key)
+
+        if public_res:
+            for i in public_res:
+                info[i.get("@ID")] = {
+                    "FontName": i.get("@FontName"),
+                    "FamilyName": i.get("@FamilyName"),
+                    "Bold": i.get("@Bold"),
+                    "Serif": i.get("@Serif"),
+                    "FixedWidth": i.get("@FixedWidth"),
+                    "FontFile": i.get("ofd:FontFile"),
+                }
+        return info
+
+
+class AnnotationFileParser(FileParserBase):
+    """
+    Parser Annotation
+    签名信息 暂不用
+    """
+    pass
+
+
+class SignaturesFileParser(FileParserBase):
+    """
+    Parser Signatures
+    签章信息-总
+    """
+
+    def __call__(self):
+        info = {}
+        signature_res: list = []
+        signature_res_key = "ofd:Signature"
+        self.recursion_ext(self.xml_obj, signature_res, signature_res_key)
+
+        if signature_res:
+            for i in signature_res:
+                info[i.get("@ID")] = {
+                    "BaseLoc": i.get("@BaseLoc"),
+                    "Type": i.get("@Type"),
+                    "ID": i.get("@ID"),
+
+                }
+        return info
+
+
+class SignatureFileParser(FileParserBase):
+    """
+    Parser Signature
+    签章信息
+    """
+
+    def __call__(self, prefix=""):
+        info = {}
+        StampAnnot_res: list = []
+        StampAnnot_res_key = "ofd:StampAnnot"
+
+        self.recursion_ext(self.xml_obj, StampAnnot_res, StampAnnot_res_key)
+
+        SignedValue_res: list = []
+        SignedValue_res_key = "ofd:SignedValue"
+        self.recursion_ext(self.xml_obj, SignedValue_res, SignedValue_res_key)
+
+        # print("SignedValue_res", SignedValue_res)
+        # print("prefix", prefix)
+        if StampAnnot_res:
+            for i in StampAnnot_res:
+                info = {
+                    "PageRef": i.get("@PageRef"),  # page id
+                    "Boundary": i.get("@Boundary"),
+                    "ID": i.get("@ID"),
+                    "SignedValue": f"{prefix}/{SignedValue_res[0]}" if SignedValue_res else f"{prefix}/SignedValue.dat",
+                }
+
+        return info
+
+
+if __name__ == "__main__":
+    FileParserBase("")()

magic_pdf/tools/find_seal_img.py

+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+# PROJECT_NAME: easyofd read_seal_img
+# CREATE_TIME: 2024/5/28 14:13
+# E_MAIL: renoyuan@foxmail.com
+# AUTHOR: renoyuan
+# note: 根据 ASN.1 解析签章 拿到 签章图片
+import io
+import base64
+
+from PIL import Image, UnidentifiedImageError
+from loguru import logger
+from pyasn1.codec.der.decoder import decode
+from pyasn1.type import univ
+from pyasn1.error import PyAsn1Error
+
+
+
+class SealExtract(object):
+    def __init__(self,):
+        pass
+    def read_signed_value(self, path="", b64=""):
+        # 读取二进制文件
+        if b64:
+            binary_data = base64.b64decode(b64)
+        elif path:
+
+            with open(path, 'rb') as file:
+                binary_data = file.read()
+        else:
+            return
+        # 尝试解码为通用的 ASN.1 结构
+        try:
+            decoded_data, _ = decode(binary_data)
+        except PyAsn1Error as e:
+            # print(f"Decoding failed: {e}")
+            decoded_data = None
+        finally:
+           return  decoded_data
+
+
+    def find_octet_strings(self, asn1_data,octet_strings:list):
+
+        # 递归查找所有的 OctetString 实例
+
+        if isinstance(asn1_data, univ.OctetString):
+
+            octet_strings.append(asn1_data)
+        elif isinstance(asn1_data, univ.Sequence) or isinstance(asn1_data, univ.Set):
+            for component in asn1_data:
+                self.find_octet_strings(asn1_data[f"{component}"], octet_strings)
+        elif isinstance(asn1_data, univ.Choice):
+            self.find_octet_strings(asn1_data.getComponent(), octet_strings)
+        elif isinstance(asn1_data, univ.Any):
+            try:
+                sub_data, _ = decode(asn1_data.asOctets())
+                self.find_octet_strings(sub_data, octet_strings)
+            except PyAsn1Error:
+                pass
+
+
+    def hex_to_image(self, hex_data, image_format='PNG',inx=0):
+        """
+        将16进制数据转换为图片并保存。
+        :param hex_data: 图片的16进制数据字符串
+        :param image_format: 图片的格式，默认为'PNG'
+        """
+        # 将16进制数据转换为二进制数据
+
+        binary_data = bytes.fromhex(hex_data)
+
+        # 创建BytesIO对象以读取二进制数据
+        image_stream = io.BytesIO(binary_data)
+
+        # 使用Pillow打开图像数据并保存
+        try:
+            image = Image.open(image_stream)
+            # image.save(f'{inx}_image.{image_format}', format=image_format)
+            # print(f"图片已保存为'image.{image_format}'")
+            return image
+        except UnidentifiedImageError:
+            # logger.info("not img ")
+            pass
+
+    def __call__(self, path="", b64=""):
+
+        decoded_data = self.read_signed_value(path=path, b64=b64)
+        octet_strings = []
+        img_list = []  # 目前是只有一个的，若存在多个的话关联后面考虑
+        if decoded_data:
+            self.find_octet_strings(decoded_data, octet_strings)
+
+            for i, octet_string in enumerate(octet_strings):
+                # logger.info(f"octet_string{octet_string}")
+                if str(octet_string.prettyPrint()).startswith("0x"):
+
+                    img = self.hex_to_image(str(octet_string.prettyPrint())[2:],inx= i)
+                    if img:
+                        logger.info("ASN.1 data found.")
+                        img_list.append(img)
+        else:
+            logger.info("No valid ASN.1 data found.")
+
+        return  img_list
+
+if __name__=="__main__":
+    print(SealExtract()(r"F:\code\easyofd\test\1111_xml\Doc_0\Signs\Sign_0\SignedValue.dat" ))
+

magic_pdf/tools/font_tools.py

+#!/usr/bin/env python
+#-*- coding: utf-8 -*-
+#PROJECT_NAME: D:\code\easyofd\easyofd
+#CREATE_TIME: 2023-07-27 
+#E_MAIL: renoyuan@foxmail.com
+#AUTHOR: reno 
+#NOTE: 字体处理  
+import time
+import re
+import json
+import base64
+import zipfile
+import os
+import shutil
+import logging
+from io import BytesIO, StringIO
+import string
+from uuid import uuid1
+import random
+import traceback
+import logging
+
+
+import tempfile
+import xmltodict
+from fontTools.ttLib import TTFont as ttLib_TTFont
+from fontTools.pens.basePen import BasePen
+from reportlab.graphics.shapes import Path
+from reportlab.lib import colors
+from reportlab.graphics import renderPM
+from reportlab.graphics.shapes import Group, Drawing, scale
+from reportlab import platypus
+from reportlab.lib.pagesizes import letter, A4
+from reportlab.lib.units import mm,inch
+from reportlab.platypus import SimpleDocTemplate, Image
+from reportlab.lib.utils import ImageReader
+from reportlab.pdfgen import canvas
+from reportlab.pdfbase import pdfmetrics
+from reportlab.pdfbase.cidfonts import UnicodeCIDFont
+from reportlab.pdfbase.ttfonts import TTFont
+from concurrent.futures import ThreadPoolExecutor
+import threading
+import multiprocessing
+import PIL
+
+
+from reportlab.lib.fonts import _tt2ps_map 
+from reportlab.lib.fonts import _family_alias 
+
+from loguru import logger
+
+FONTS = ['宋体',"SWPMEH+SimSun",'SimSun','KaiTi','楷体',"STKAITI","SWLCQE+KaiTi",
+         'Courier New','STSong-Light',"CourierNew","SWANVV+CourierNewPSMT",
+         "CourierNewPSMT","BWSimKai","hei","黑体","SimHei","SWDKON+SimSun",
+         "SWCRMF+CourierNewPSMT","SWHGME+KaiTi"]
+
+
+FONTS_ttf = [{'宋体':'simsun.ttc'},{'SimSun':'simsun.ttc'},{'KaiTi':'simkai.ttf'},{'楷体':'simkai.ttf'},
+{'STKAITI':'STKAITI.TTF'},{'hei':'simhei.ttf'},{'STXihei':'simhei.ttf'},{'黑体':'simhei.ttf'},
+{'华文细黑':'STXIHEI.TTF'},{'SimHei':'simhei.ttf'},{'Courier New':'/usr/share/fonts/truetype/msttcorefonts/cour.ttf'}]
+
+path_prefix = "/usr/local/share/fonts/zit/"
+
+# 批量注册字体
+for font in FONTS_ttf:
+    for font_name, file_name in font.items():
+        try:
+            if 'usr' in file_name:
+                font_path = file_name
+            else:
+                font_path = os.path.join(path_prefix, file_name)
+            pdfmetrics.registerFont(TTFont(font_name, font_path))
+        except Exception as e:
+            print(f"无法注册字体 {font_name}: {e}")
+
+class FontTool(object):
+    FONTS = FONTS
+    
+    def __init__(self):
+        # 初始支持字体
+        # 字体检测
+        pass
+    
+    def font_check(self):
+        logger.info("f{_tt2ps_map}")
+        logger.info("f{_family_alias}")
+        
+        for font in self.FONTS:
+            if font in _tt2ps_map.values():
+                logger.info(f"已注册{font}")
+            else:
+                logger.warning(f"-{font}-未注册可能导致写入失败")
+                 
+        
+        
+    def register_font(self,file_name,FontName,font_b64):
+        
+        if font_b64:
+            
+            file_name = os.path.split(file_name)
+            # logger.error(f"file_name:{file_name}")
+            # logger.info(f"file_name:{file_name}")
+            if isinstance(file_name, (tuple, list)):
+                    file_name = file_name[1]
+            if not FontName:
+                FontName = file_name.split(".")[0]
+            try:
+                with open(file_name, "wb") as f:
+                    f.write(base64.b64decode(font_b64))
+                # print("FontName", FontName, "file_name", file_name)
+                pdfmetrics.registerFont(TTFont(FontName, file_name))
+                self.FONTS.append(FontName)
+                
+            except Exception as e:
+                traceback.print_exc()
+                logger.error(f"register_font_error:\n{e} \n 包含不支持解析字体格式")
+            
+            finally:
+                
+                if os.path.exists(file_name):
+                        os.remove(file_name)

magic_pdf/tools/img_deal.py

+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+# PROJECT_NAME: easyofd img_deal
+# CREATE_TIME: 2024/7/18 11:20
+# E_MAIL: renoyuan@foxmail.com
+# AUTHOR: renoyuan
+# note: img 操作
+from io import BytesIO
+class DealImg(object):
+    def __init__(self):
+        pass
+    def resize(self):
+        """resize img"""
+        pass
+    def pil2bytes(self, image):
+        """pil2bytes"""
+        # 创建一个 BytesIO 对象
+        img_bytesio = BytesIO()
+        # 将图像保存到 BytesIO 对象
+        image.save(img_bytesio, format='PNG')  # 你可以根据需要选择其他图像格式
+        # 获取 BytesIO 对象中的字节
+        img_bytes = img_bytesio.getvalue()
+        # 关闭 BytesIO 对象
+        img_bytesio.close()
+        return img_bytes
+    def pil2bytes_io(self, image):
+        """pil2bytes_io"""
+        # 创建一个 BytesIO 对象
+        img_bytesio = BytesIO()
+        # 将图像保存到 BytesIO 对象
+        image.save(img_bytesio, format='PNG')  # 你可以根据需要选择其他图像格式
+        return img_bytesio
+
+
+

magic_pdf/tools/ofd.py

+# -*- coding: utf-8 -*-
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+# PROJECT_NAME: F:\code\easyofd\easyofd
+# CREATE_TIME: 2023-10-07
+# E_MAIL: renoyuan@foxmail.com
+# AUTHOR: reno
+# note:  ofd 基础类
+import base64
+import os
+import sys
+from io import BytesIO
+from typing import Union
+
+sys.path.insert(0, os.getcwd())
+sys.path.insert(0, "..")
+
+import fitz
+
+from PIL import Image
+from loguru import logger
+
+from magic_pdf.tools.ofd_parser import OFDParser
+from magic_pdf.rw.draw_pdf import DrawPDF
+from magic_pdf.rw.draw_ofd import OFDWrite
+
+
+class OFD(object):
+    """ofd对象"""
+
+    def __init__(self, ):
+        self.data = None
+
+    def read(self, ofd_f: Union[str, bytes, BytesIO], fmt="b64", save_xml=False, xml_name="testxml"):
+        """_summary_
+        Args:
+            file (_type_): _description_
+            fomat (str, optional): _description_. Defaults to "path".
+            fomat in ("path","b64","binary")
+        """
+        if fmt == "path":
+            with open(ofd_f, "rb") as f:
+                ofd_f = str(base64.b64encode(f.read()), encoding="utf-8")
+        elif fmt == "b64":
+            pass
+        elif fmt == "binary":
+            ofd_f = str(base64.b64encode(ofd_f), encoding="utf-8")
+        elif fmt == "io":
+            ofd_f = str(base64.b64encode(ofd_f.getvalue()), encoding="utf-8")
+        else:
+            raise "fomat Error: %s" % fmt
+
+        self.data = OFDParser(ofd_f)(save_xml=save_xml, xml_name=xml_name)
+
+    def save(self, ):
+        """
+        draw ofd xml
+        初始化一个xml 文件
+        self.data > file
+        """
+        assert self.data, f"data is None"
+
+    def pdf2ofd(self, pdfbyte, optional_text=False):
+        """pdf转ofd"""
+        assert pdfbyte, f"pdfbyte is None"
+        # logger.info(f"pdf2ofd")
+        ofd_byte = OFDWrite()(pdfbyte, optional_text=optional_text)
+        return ofd_byte
+
+    def to_pdf(self, ):
+        """return ofdbytes"""
+
+        assert self.data, f"data is None"
+        # logger.info(f"to_pdf")
+        return DrawPDF(self.data)()
+
+    def pdf2img(self, pdfbytes):
+
+        image_list = []
+
+        doc = fitz.open(stream=pdfbytes, filetype="pdf")
+
+        for page in doc:
+            rotate = int(0)
+            zoom_x, zoom_y = 1.6, 1.6
+            zoom_x, zoom_y = 2, 2
+            mat = fitz.Matrix(zoom_x, zoom_y).prerotate(rotate)
+            pix = page.get_pixmap(matrix=mat, alpha=False)
+            pil_image = Image.frombytes("RGB", [pix.width, pix.height], pix.samples)
+            # image = np.ndarray((pix.height, pix.width, 3), dtype=np.uint8, buffer=pix.samples)
+            # print(image.shape)
+            # print(image[2])
+            image_list.append(pil_image)
+        logger.info(f"pdf2img")
+        return image_list
+
+    def jpg2ofd(self, imglist: list):
+        """
+        imglist: pil image list
+        """
+        ofd_byte = OFDWrite()(pil_img_list=imglist)
+        return ofd_byte
+
+    def jpg2pfd(self, imglist: list):
+        """
+        imglist: PIL image list
+        1 构建data
+        2 DrawPDF(self.data)()
+        """
+
+        data = OFDParser(None).img2data(imglist)
+        return DrawPDF(data)()
+
+    def to_jpg(self, format="jpg"):
+        """
+        return pil list
+        """
+        assert self.data, f"data is None"
+        image_list = []
+        pdfbytes = self.to_pdf()
+        image_list = self.pdf2img(pdfbytes)
+        return image_list,pdfbytes
+
+    def del_data(self, ):
+        """销毁self.data"""
+        self.data = None
+
+    def __del__(self):
+        del self
+
+    def disposal(self, ):
+        """销毁对象"""
+        self.__del__()
+
+
+if __name__ == "__main__":
+    ofd = OFD()

magic_pdf/tools/ofd_parser.py

+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+# PROJECT_NAME: D:\code\easyofd\easyofd\parser
+# CREATE_TIME: 2023-07-27
+# E_MAIL: renoyuan@foxmail.com
+# AUTHOR: reno
+# NOTE: ofd解析主流程
+
+import os
+import sys
+
+sys.path.insert(0, "..")
+
+import traceback
+import base64
+import re
+
+from typing import Any, List
+from PIL import Image
+from PIL.Image import Image as ImageClass
+from loguru import logger
+
+from magic_pdf.tools.img_deal import DealImg
+from magic_pdf.tools.file_deal import FileRead
+from magic_pdf.tools.file_parser import (OFDFileParser, DocumentFileParser, ContentFileParser, DocumentResFileParser,
+                          PublicResFileParser,
+                          SignaturesFileParser, SignatureFileParser)
+
+
+class OFDParser(object):
+    """
+    OFDParser 解析
+    1 解压文件 创建文件映射表 释放文件
+    2 解析 xml 逐级去 收集需要信息  结构文本 以及 资源
+    2 调用font 注册 字体
+    """
+
+    def __init__(self, ofdb64):
+        self.img_deal = DealImg()
+        self.ofdb64 = ofdb64
+        self.file_tree = None
+        self.jbig2dec_path = r"C:/msys64/mingw64/bin/jbig2dec.exe"
+
+    def img2data(self, imglist: List[ImageClass]):
+        """
+        imglist to ofd data
+        
+        """
+        OP = 200 / 25.4
+        doc_list = []
+        img_info = {}
+        page_size = []
+        font_info = {}
+        page_info_d = {}
+
+        for idx, img_pil in enumerate(imglist):
+            w, h = img_pil.size
+            img_bytes = self.img_deal.pil2bytes(img_pil)
+            imgb64 = str(base64.b64encode(img_bytes), encoding="utf-8")
+            img_info[str(idx)] = {
+                "format": "jpg",
+                "wrap_pos": "",
+                "type": "IMG",
+                "suffix": "jpg",
+                "fileName": f"{idx}.jpg",
+                "imgb64": imgb64,
+
+            }
+            text_list = []
+            img_list = []
+            img_d = {}
+            img_d["CTM"] = ""  # 平移矩阵换 平移 缩放 旋转
+            img_d["ID"] = str(idx)  # 图片id
+            img_d["ResourceID"] = str(idx)  # 图片id
+            img_d["pos"] = [0, 0, w / OP, h / OP]  # 平移矩阵换
+            page_size = [0, 0, w / OP, h / OP]
+            # print(page_size)
+            img_list.append(img_d)
+
+            content_d = {
+                "text_list": text_list,
+                "img_list": img_list,
+            }
+            page_info_d[idx] = content_d
+        doc_list.append({
+            "pdf_name": "demo.pdf",
+            "doc_no": "0",
+            "images": img_info,
+            "page_size": page_size,
+            "fonts": font_info,
+            "page_info": page_info_d
+        })
+
+        return doc_list
+
+    # 获得xml 对象
+    def get_xml_obj(self, label):
+        assert label
+        # print(self.file_tree.keys())
+        for abs_p in self.file_tree:
+            # 统一符号，避免win linux 路径冲突
+
+            abs_p_compare = abs_p.replace("\\\\", "-").replace("//", "-").replace("\\", "-").replace("/", "-")
+            label_compare = label.replace("\\\\", "-").replace("//", "-").replace("\\", "-").replace("/", "-")
+            if label_compare in abs_p_compare:
+                # logger.info(f"{label} {abs_p}")
+                return self.file_tree[abs_p]
+        # logger.info(f"{label} ofd file path is not")
+        return ""
+
+    def jb22png(self, img_d: dict):
+        """
+        jb22png
+        没有安装 jbig2dec 无法操作 
+        """
+        if not os.path.exists(self.jbig2dec_path):
+            logger.warning(f"未安装jbig2dec，无法处理jb2文件")
+            return
+
+        # todo ib2 转png C:/msys64/mingw64/bin/jbig2dec.exe -o F:\code\easyofd\test\image_80.png F:\code\easyofd\test\image_80.jb2
+        fileName = img_d["fileName"]
+        new_fileName = img_d['fileName'].replace(".jb2", ".png")
+        with open(fileName, "wb") as f:
+            f.write(base64.b64decode(img_d["imgb64"]))
+        command = "{} -o {} {}"
+        res = os.system(command.format(self.jbig2dec_path, new_fileName, fileName))
+        if res != 0:
+            logger.warning(f"jbig2dec处理失败")
+        if os.path.exists(fileName):
+            os.remove(fileName)
+        if os.path.exists(new_fileName):
+            logger.info(f"jbig2dec处理成功{fileName}>>{new_fileName}")
+            img_d["fileName"] = new_fileName
+            img_d["suffix"] = "png"
+            img_d["format"] = "png"
+            with open(new_fileName, "rb") as f:
+                data = f.read()
+                img_d["imgb64"] = str(base64.b64encode(data), encoding="utf-8")
+
+            os.remove(new_fileName)
+
+    def bmp2jpg(self, img_d: dict):
+
+        fileName = img_d["fileName"]
+        new_fileName = img_d['fileName'].replace(".bmp", ".jpg")
+        with open(fileName, "wb") as f:
+            f.write(base64.b64decode(img_d["imgb64"]))
+
+        # 打开 BMP 图像
+        bmp_image = Image.open(fileName)
+
+        # 将 BMP 图像保存为 JPG 格式
+        bmp_image.convert("RGB").save(new_fileName, "JPEG")
+
+        # 关闭图像
+        bmp_image.close()
+        if os.path.exists(new_fileName):
+            logger.info(f"jbig2dec处理成功{fileName}>>{new_fileName}")
+            img_d["fileName"] = new_fileName
+            img_d["suffix"] = "jpg"
+            img_d["format"] = "jpg"
+            with open(new_fileName, "rb") as f:
+                data = f.read()
+                img_d["imgb64"] = str(base64.b64encode(data), encoding="utf-8")
+            os.remove(new_fileName)
+
+    def parser(self, ):
+        """
+        解析流程
+        """
+        # 默认只有 doc_0 一层有多层后面改
+        page_size_details = []
+        default_page_size = []
+        doc_list = []
+        ofd_xml_obj = self.get_xml_obj(self.file_tree["root_doc"])  # OFD.xml xml 对象 
+
+        if ofd_xml_obj:
+            ofd_obj_res = OFDFileParser(ofd_xml_obj)()
+            doc_root_name = ofd_obj_res.get("doc_root")
+            signatures = ofd_obj_res.get("signatures")
+        else:
+            # 考虑根节点丢失情况
+            doc_root_name = ["Doc_0/Document.xml"]
+            signatures = ["Doc_0/Signs/Signatures.xml"]
+
+        doc_root_xml_obj = self.get_xml_obj(doc_root_name[0])
+        doc_root_info = DocumentFileParser(doc_root_xml_obj)()
+        doc_size = doc_root_info.get("size")
+
+        if doc_size:
+            try:
+                default_page_size = [float(pos_i) for pos_i in doc_size.split(" ") if re.match("[\d\.]", pos_i)]
+            except:
+                traceback.print_exc()
+
+        # 字体信息
+        font_info = {}
+        public_res_name: list = doc_root_info.get("public_res")
+        if public_res_name:
+            public_xml_obj = self.get_xml_obj(public_res_name[0])
+            font_info = PublicResFileParser(public_xml_obj)()
+
+            # 注册字体
+            for font_id, font_v in font_info.items():
+                file_name = font_v.get("FontFile")
+                if file_name:
+                    font_b64 = self.get_xml_obj(file_name)
+                    if font_b64:
+                        font_v["font_b64"] = font_b64
+
+        # 图片资源
+        img_info: dict = dict()
+        document_res_name: list = doc_root_info.get("document_res")
+        if document_res_name:
+            document_res_xml_obj = self.get_xml_obj(document_res_name[0])
+
+            img_info = DocumentResFileParser(document_res_xml_obj)()
+            # 找到图片b64
+            for img_id, img_v in img_info.items():
+                img_v["imgb64"] = self.get_xml_obj(img_v.get("fileName"))
+                if img_v[
+                    "suffix"] == 'jb2':  # todo ib2 转png C:/msys64/mingw64/bin/jbig2dec.exe -o F:\code\easyofd\test\image_80.png F:\code\easyofd\test\image_80.jb2
+                    self.jb22png(img_v)
+                if img_v["suffix"] == 'bmp':
+                    self.bmp2jpg(img_v)
+
+        page_id_map: list = doc_root_info.get("page_id_map")
+        signatures_page_id = {}
+
+        # 签章信息
+        if signatures and (signatures_xml_obj := self.get_xml_obj(signatures[0])):
+            print("signatures_xml_obj", signatures,signatures_xml_obj)
+            signatures_info = SignaturesFileParser(signatures_xml_obj)()
+            if signatures_info:  # 获取签章具体信息
+                for _, signatures_cell in signatures_info.items():
+                    # print(signatures_info)
+                    BaseLoc = signatures_cell.get("BaseLoc")
+                    signature_xml_obj = self.get_xml_obj(BaseLoc)
+                    # print(BaseLoc)
+                    prefix = BaseLoc.split("/")[0]
+                    signatures_info = SignatureFileParser(signature_xml_obj)(prefix=prefix)
+                    # print(signatures_info)
+                    print("signatures_info", signatures_info)
+                    PageRef = signatures_info.get("PageRef")
+                    Boundary = signatures_info.get("Boundary")
+                    SignedValue = signatures_info.get("SignedValue")
+                    sing_page_no = page_id_map.get(PageRef)
+                    # print("self.file_tree",self.file_tree.keys)
+                    # print(page_id_map,PageRef)
+                    # print(SignedValue, self.get_xml_obj(SignedValue))
+                    # with open("b64.txt","w") as f:
+                    #     f.write(self.get_xml_obj(SignedValue))
+                    if signatures_page_id.get(sing_page_no):
+                        signatures_page_id[sing_page_no].append(
+                            {
+                                "sing_page_no": sing_page_no,
+                                "PageRef": PageRef,
+                                "Boundary": Boundary,
+                                "SignedValue": self.get_xml_obj(SignedValue),
+                            }
+                        )
+                    else:
+                        signatures_page_id[sing_page_no] = [
+                            {
+                                "sing_page_no": sing_page_no,
+                                "PageRef": PageRef,
+                                "Boundary": Boundary,
+                                "SignedValue": self.get_xml_obj(SignedValue),
+                            }
+                        ]
+
+        # 正文信息 会有多页 情况
+        page_name: list = doc_root_info.get("page")
+
+        page_info_d = {}
+        if page_name:
+            for index, _page in enumerate(page_name):
+                page_xml_obj = self.get_xml_obj(_page)
+                # 重新获取页面size
+                try:
+                    page_size = [float(pos_i) for pos_i in
+                                     page_xml_obj.get('ofd:Page', {}).get("ofd:Area", {}).get("ofd:PhysicalBox",
+                                                                                              "").split(" ")
+                                     if re.match("[\d\.]", pos_i)]
+                    if page_size and len(page_size) >= 2:
+                        page_size_details.append(page_size)
+                    else:
+                        page_size_details.append([])
+                except Exception as e:
+                    traceback.print_exc()
+                    page_size.append([])
+                page_info = ContentFileParser(page_xml_obj)()
+                pg_no = re.search(r"\d+", _page)
+                if pg_no:
+                    pg_no = int(pg_no.group())
+                else:
+                    pg_no = index
+                page_info_d[pg_no] = page_info
+
+        # 模板信息
+        tpls_name: list = doc_root_info.get("tpls")
+        if tpls_name:
+            for index, _tpl in enumerate(tpls_name):
+                tpl_xml_obj = self.get_xml_obj(_tpl)
+                tpl_info = ContentFileParser(tpl_xml_obj)()
+                tpl_no = re.search(r"\d+", _tpl)
+
+                if tpl_no:
+                    tpl_no = int(tpl_no.group())
+                else:
+                    tpl_no = index
+
+                if tpl_no in page_info_d:
+                    page_info_d[pg_no]["text_list"].extend(tpl_info["text_list"])
+                    page_info_d[pg_no]["text_list"].sort(
+                        key=lambda pos_text: (float(pos_text.get("pos")[1]), float(pos_text.get("pos")[0])))
+                    page_info_d[pg_no]["img_list"].extend(tpl_info["img_list"])
+                    page_info_d[pg_no]["img_list"].sort(
+                        key=lambda pos_text: (float(pos_text.get("pos")[1]), float(pos_text.get("pos")[0])))
+                    page_info_d[pg_no]["line_list"].extend(tpl_info["line_list"])
+                    page_info_d[pg_no]["line_list"].sort(
+                        key=lambda pos_text: (float(pos_text.get("pos")[1]), float(pos_text.get("pos")[0])))
+                else:
+                    page_info_d[tpl_no] = tpl_info
+                    page_info_d[tpl_no].sort(
+                        key=lambda pos_text: (float(pos_text.get("pos")[1]), float(pos_text.get("pos")[0])))
+
+        page_ID = 0  # 没遇到过doc多个的情况
+        # print("page_info",len(page_info))
+        doc_list.append({
+            "default_page_size": default_page_size,
+            "page_size": page_size_details,
+            "pdf_name": self.file_tree["pdf_name"],
+            "doc_no": page_ID,
+            "images": img_info,
+            "signatures_page_id": signatures_page_id,
+            "page_id_map": page_id_map,
+            "fonts": font_info,
+            "page_info": page_info_d
+        })
+        return doc_list
+
+    def __call__(self, *args: Any, **kwds: Any) -> Any:
+        """
+        输出ofd解析结果
+        
+        """
+        save_xml = kwds.get("save_xml", False)
+        xml_name = kwds.get("xml_name")
+        self.file_tree = FileRead(self.ofdb64)(save_xml=save_xml, xml_name=xml_name)
+        # logger.info(self.file_tree)
+        return self.parser()
+
+
+if __name__ == "__main__":
+    with open(r"E:\code\easyofd\test\增值税电子专票5.ofd", "rb") as f:
+        ofdb64 = str(base64.b64encode(f.read()), "utf-8")
+    print(OFDParser(ofdb64)())

magic_pdf/tools/parameter_parser.py

+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+# PROJECT_NAME: easyofd
+# CREATE_TIME: 
+# E_MAIL: renoyuan@foxmail.com
+# AUTHOR: renoyuan
+# note:参数解析器
+from loguru import logger
+from typing import List, Dict, Any, Union, Tuple, Optional
+
+
+class ParameterParser(object):
+    parameter = {
+        "ofd:FillColor": (dict, dict),
+        "ofd:StrokeColor": (dict, dict),
+        "ofd:Test": ((str, int), str),
+        "ofd:Font": (str, str),
+        "@Value": (str, str)
+    }
+
+    def __call__(self, key, container):
+        if key in ParameterParser.parameter:
+            v = container.get(key, None)
+            t = ParameterParser.parameter[key]
+            if isinstance(v, t[0]):
+                return v
+            else:
+                return t[1]()
+        else:
+            logger.warning(f"{key} not in ParameterParser")
+            return None

magic_pdf/tools/pdf_server.py

+import os
+from pathlib import Path
+
+import click
+from loguru import logger
+from typing import List
+from fastapi import FastAPI, HTTPException, Request
+import magic_pdf.model as model_config
+# from magic_pdf.dict2md.ocr_client import PredictClient
+from magic_pdf.dict2md.ocr_vllm_client import PredictClient
+from magic_pdf.libs.version import __version__
+from magic_pdf.model.doc_analyze_by_custom_model import ModelSingleton
+from magic_pdf.rw.AbsReaderWriter import AbsReaderWriter
+from magic_pdf.rw.DiskReaderWriter import DiskReaderWriter
+from magic_pdf.tools.common import do_parse, parse_pdf_methods
+from argparse import ArgumentParser
+from pydantic import BaseModel
+import uvicorn
+import time
+import configparser
+#from magic_pdf.tools.config import update_config
+
+app = FastAPI()
+method = 'auto'
+
+logger.add("parse.log", rotation="10 MB", level="INFO",
+           format="{time} {level} {message}", encoding='utf-8', enqueue=True)
+config_path = None
+
+ocr_status = None
+custom_model = None
+
+class ocrRequest(BaseModel):
+    path: str
+    output_dir: str
+
+class ocrResponse(BaseModel):
+    status_code: int
+    output_path: str
+
+
+def parse_args():
+    parser = ArgumentParser()
+    parser.add_argument(
+        '--dcu_id',
+        default='0',
+        help='设置DCU')
+    parser.add_argument(
+        '--method',
+        type=parse_pdf_methods,
+        help = """the method for parsing pdf.
+        ocr: using ocr technique to extract information from pdf.
+        txt: suitable for the text-based pdf only and outperform ocr.
+        auto: automatically choose the best method for parsing pdf from ocr and txt.
+        without method specified, auto will be used by default.""",
+        default = 'auto',
+        )
+    parser.add_argument(
+        '--debug',
+        type=bool,
+        help='Enables detailed debugging information during the execution of the CLI commands.',
+        default=False,
+    )
+    parser.add_argument(
+        '--config_path',
+        default='/home/practice/magic_pdf-main/magic_pdf/config.ini')
+
+    args = parser.parse_args()
+    return args
+
+def ocr_pdf_serve(args: str):
+    os.environ["CUDA_VISIBLE_DEVICES"] = args.dcu_id
+    config = configparser.ConfigParser()
+    config.read(args.config_path)
+    # host = config.get('server', 'pdf_host')
+    # port = int(config.get('server', 'pdf_port'))
+    pdf_server = config.get('server', 'pdf_server').split('://')[1]
+    host, port = pdf_server.split(':')[0], int(pdf_server.split(':')[1])
+    global config_path
+    config_path = args.config_path
+    ocr_server = config.get('server', 'ocr_server')
+    ocr_client = PredictClient(ocr_server)
+    global ocr_status
+    ocr_status = ocr_client.check_health()
+
+    ocr = True
+    show_log = False
+    model_manager = ModelSingleton()
+    global custom_model
+    custom_model = model_manager.get_model(ocr, show_log)
+
+
+    uvicorn.run(app, host=host, port=port)
+
+@app.get("/health")
+async def health_check():
+    return {"status": "healthy"}
+
+@app.post("/pdf_ocr")
+# def cli(path, output_dir, method, debug_able, start_page_id, end_page_id):
+async def pdf_ocr(request: ocrRequest):
+    model_config.__use_inside_model__ = True
+    model_config.__model_mode__ = 'full'
+    output_dir = request.output_dir
+    path = request.path
+    #config_path = request.config_path
+    os.makedirs(output_dir, exist_ok=True)
+    debug_able = False
+    start_page_id = 0
+    end_page_id = None
+    logger.info(f"正在处理文件: {path}")
+
+    def read_fn(path):
+        disk_rw = DiskReaderWriter(os.path.dirname(path))
+        return disk_rw.read(os.path.basename(path), AbsReaderWriter.MODE_BIN)
+
+    def parse_doc(doc_path: str, config_path: str):
+        try:
+            file_name = str(Path(doc_path).stem)
+            pdf_data = read_fn(doc_path)
+            output_path = do_parse(
+                ocr_status,
+                config_path,
+                output_dir,
+                file_name,
+                pdf_data,
+                [],
+                method,
+                debug_able,
+                model=custom_model,
+                start_page_id=start_page_id,
+                end_page_id=end_page_id,
+            )
+
+            # logger.info(f'文件解析成功：{output_path}')
+            return output_path
+
+        except Exception as e:
+            logger.exception(e)
+
+    # logger.info(f'config_path：{config_path}')
+
+    output_path = parse_doc(path,config_path)
+    if output_path:
+        logger.info(f'文件解析成功：{output_path}')
+        return {"status_code": 200, "output_path": output_path}
+    else:
+        logger.error(f'文件解析失败，文件为：{path}')
+        raise HTTPException(status_code=500)
+
+def main():
+    args = parse_args()
+    ocr_pdf_serve(args)
+
+
+
+if __name__ == '__main__':
+    main()
+
+
+
+
+
+

magic_pdf/user_api.py

+"""
+用户输入：
+    model数组，每个元素代表一个页面
+    pdf在s3的路径
+    截图保存的s3位置
+
+然后：
+    1）根据s3路径，调用spark集群的api,拿到ak,sk,endpoint，构造出s3PDFReader
+    2）根据用户输入的s3地址，调用spark集群的api,拿到ak,sk,endpoint，构造出s3ImageWriter
+
+其余部分至于构造s3cli, 获取ak,sk都在code-clean里写代码完成。不要反向依赖！！！
+
+"""
+import re
+
+from loguru import logger
+
+from magic_pdf.libs.version import __version__
+from magic_pdf.model.doc_analyze_by_custom_model import doc_analyze
+from magic_pdf.rw import AbsReaderWriter
+from magic_pdf.pdf_parse_by_ocr import parse_pdf_by_ocr
+from magic_pdf.pdf_parse_by_txt import parse_pdf_by_txt
+
+PARSE_TYPE_TXT = "txt"
+PARSE_TYPE_OCR = "ocr"
+
+
+def parse_txt_pdf(pdf_bytes: bytes, pdf_models: list, imageWriter: AbsReaderWriter, is_debug=False,
+                  start_page_id=0, end_page_id=None, lang=None,
+                  *args, **kwargs):
+    """
+    解析文本类pdf
+    """
+    pdf_info_dict = parse_pdf_by_txt(
+        pdf_bytes,
+        pdf_models,
+        imageWriter,
+        start_page_id=start_page_id,
+        end_page_id=end_page_id,
+        debug_mode=is_debug,
+    )
+
+    pdf_info_dict["_parse_type"] = PARSE_TYPE_TXT
+
+    pdf_info_dict["_version_name"] = __version__
+
+    if lang is not None:
+        pdf_info_dict["_lang"] = lang
+
+    return pdf_info_dict
+
+
+def parse_ocr_pdf(ocr_status,config_path,local_image_dir,pdf_bytes: bytes, pdf_models: list, imageWriter: AbsReaderWriter, is_debug=False,
+                  start_page_id=0, end_page_id=None, lang=None,
+                  *args, **kwargs):
+    """
+    解析ocr类pdf
+    """
+    pdf_info_dict = parse_pdf_by_ocr(ocr_status,config_path,local_image_dir,
+        pdf_bytes,
+        pdf_models,
+        imageWriter,
+        start_page_id=start_page_id,
+        end_page_id=end_page_id,
+        debug_mode=is_debug,
+    )
+
+    pdf_info_dict["_parse_type"] = PARSE_TYPE_OCR
+
+    pdf_info_dict["_version_name"] = __version__
+
+    if lang is not None:
+        pdf_info_dict["_lang"] = lang
+
+    return pdf_info_dict
+
+
+def parse_union_pdf(pdf_bytes: bytes, pdf_models: list, imageWriter: AbsReaderWriter, is_debug=False,
+                    input_model_is_empty: bool = False,
+                    start_page_id=0, end_page_id=None, lang=None,
+                    *args, **kwargs):
+    """
+    ocr和文本混合的pdf，全部解析出来
+    """
+
+    def parse_pdf(method):
+        try:
+            return method(
+                pdf_bytes,
+                pdf_models,
+                imageWriter,
+                start_page_id=start_page_id,
+                end_page_id=end_page_id,
+                debug_mode=is_debug,
+            )
+        except Exception as e:
+            logger.exception(e)
+            return None
+
+    pdf_info_dict = parse_pdf(parse_pdf_by_txt)
+    if pdf_info_dict is None or pdf_info_dict.get("_need_drop", False):
+        logger.warning(f"parse_pdf_by_txt drop or error, switch to parse_pdf_by_ocr")
+        if input_model_is_empty:
+            layout_model = kwargs.get("layout_model", None)
+            formula_enable = kwargs.get("formula_enable", None)
+            table_enable = kwargs.get("table_enable", None)
+            pdf_models = doc_analyze(
+                pdf_bytes,
+                ocr=True,
+                start_page_id=start_page_id,
+                end_page_id=end_page_id,
+                lang=lang,
+                layout_model=layout_model,
+                formula_enable=formula_enable,
+                table_enable=table_enable,
+            )
+        pdf_info_dict = parse_pdf(parse_pdf_by_ocr)
+        if pdf_info_dict is None:
+            raise Exception("Both parse_pdf_by_txt and parse_pdf_by_ocr failed.")
+        else:
+            pdf_info_dict["_parse_type"] = PARSE_TYPE_OCR
+    else:
+        pdf_info_dict["_parse_type"] = PARSE_TYPE_TXT
+
+    pdf_info_dict["_version_name"] = __version__
+
+    if lang is not None:
+        pdf_info_dict["_lang"] = lang
+
+    return pdf_info_dict

magic_pdf/utils/annotations.py

+
+from loguru import logger
+
+
+def ImportPIL(f):
+    try:
+        import PIL  # noqa: F401
+    except ImportError:
+        logger.error('Pillow not installed, please install by pip.')
+        exit(1)
+    return f