UNIPipe.py 2.65 KB
Newer Older
赵小蒙's avatar
赵小蒙 committed
1
2
import json

3
from loguru import logger
赵小蒙's avatar
赵小蒙 committed
4
5
6
from magic_pdf.io.AbsReaderWriter import AbsReaderWriter
from magic_pdf.io.DiskReaderWriter import DiskReaderWriter
from magic_pdf.libs.commons import join_path
7
from magic_pdf.libs.json_compressor import JsonCompressor
赵小蒙's avatar
赵小蒙 committed
8
from magic_pdf.pipe.AbsPipe import AbsPipe
赵小蒙's avatar
赵小蒙 committed
9
from magic_pdf.user_api import parse_union_pdf, parse_ocr_pdf
10
11


赵小蒙's avatar
赵小蒙 committed
12
class UNIPipe(AbsPipe):
13

赵小蒙's avatar
赵小蒙 committed
14
15
16
17
18
19
20
21
    def __init__(self, pdf_bytes: bytes, model_list: list, image_writer: AbsReaderWriter, img_bucket_path: str):
        self.pdf_type = "ocr"
        self.compressed_pdf_mid_data = None
        self.pdf_mid_data = None
        self.pdf_bytes = pdf_bytes
        self.model_list = model_list
        self.image_writer = image_writer
        self.img_bucket_path = img_bucket_path
22

赵小蒙's avatar
赵小蒙 committed
23
24
    def pipe_classify(self):
        self.pdf_type = UNIPipe.classify(self.pdf_bytes)
25

赵小蒙's avatar
赵小蒙 committed
26
27
28
29
30
31
    def pipe_parse(self):
        if self.pdf_type == "txt":
            self.pdf_mid_data = parse_union_pdf(pdf_bytes, self.model_list, self.image_writer)
        elif self.pdf_type == "ocr":
            self.pdf_mid_data = parse_ocr_pdf(pdf_bytes, self.model_list, self.image_writer)
        self.compressed_pdf_mid_data = JsonCompressor.compress_json(self.pdf_mid_data)
32

赵小蒙's avatar
赵小蒙 committed
33
34
35
    def pipe_mk_uni_format(self):
        content_list = AbsPipe.mk_uni_format(self.compressed_pdf_mid_data, self.img_bucket_path)
        return content_list
赵小蒙's avatar
赵小蒙 committed
36

赵小蒙's avatar
赵小蒙 committed
37
38
39
    def pipe_mk_markdown(self):
        markdown_content = AbsPipe.mk_markdown(self.compressed_pdf_mid_data, self.img_bucket_path)
        return markdown_content
40
41
42

if __name__ == '__main__':
    # 测试
赵小蒙's avatar
赵小蒙 committed
43
44
    drw = DiskReaderWriter(r"D:/project/20231108code-clean")

赵小蒙's avatar
赵小蒙 committed
45
46
    pdf_file_path = r"linshixuqiu\19983-00.pdf"
    model_file_path = r"linshixuqiu\19983-00.json"
赵小蒙's avatar
赵小蒙 committed
47
48
    pdf_bytes = drw.read(pdf_file_path, AbsReaderWriter.MODE_BIN)
    model_json_txt = drw.read(model_file_path, AbsReaderWriter.MODE_TXT)
赵小蒙's avatar
赵小蒙 committed
49
50
51
52
    model_list = json.loads(model_json_txt)
    write_path = r"D:\project\20231108code-clean\linshixuqiu\19983-00"
    img_bucket_path = "imgs"
    img_writer = DiskReaderWriter(join_path(write_path, img_bucket_path))
赵小蒙's avatar
赵小蒙 committed
53

赵小蒙's avatar
赵小蒙 committed
54
55
56
57
58
59
60
61
    pipe = UNIPipe(pdf_bytes, model_list, img_writer, img_bucket_path)
    pipe.pipe_classify()
    pipe.pipe_parse()
    md_content = pipe.pipe_mk_markdown()
    try:
        content_list = pipe.pipe_mk_uni_format()
    except Exception as e:
        logger.exception(e)
赵小蒙's avatar
赵小蒙 committed
62
63

    md_writer = DiskReaderWriter(write_path)
赵小蒙's avatar
赵小蒙 committed
64
65
66
    md_writer.write(md_content, "19983-00.md", AbsReaderWriter.MODE_TXT)
    md_writer.write(json.dumps(pipe.pdf_mid_data, ensure_ascii=False, indent=4), "19983-00.json", AbsReaderWriter.MODE_TXT)
    md_writer.write(str(content_list), "19983-00.txt", AbsReaderWriter.MODE_TXT)