test_common.py 1.7 KB
Newer Older
1
2
import os
import shutil
3
import tempfile
4
5
6
7
8
9

import pytest

from magic_pdf.tools.common import do_parse


10
@pytest.mark.parametrize('method', ['auto', 'txt', 'ocr'])
11
def test_common_do_parse(method):
12
13
    import magic_pdf.model as model_config
    model_config.__use_inside_model__ = True
14
    # setup
15
16
    unitest_dir = '/tmp/magic_pdf/unittest/tools'
    filename = 'fake'
17
18
    os.makedirs(unitest_dir, exist_ok=True)

19
    temp_output_dir = tempfile.mkdtemp(dir='/tmp/magic_pdf/unittest/tools')
20
21

    # run
22
    with open('tests/unittest/test_tools/assets/common/cli_test_01.pdf', 'rb') as f:
23
24
25
26
27
28
29
30
31
        bits = f.read()
    do_parse(temp_output_dir,
             filename,
             bits, [],
             method,
             False,
             f_dump_content_list=True)

    # check
32
    base_output_dir = os.path.join(temp_output_dir, f'fake/{method}')
33

34
    r = os.stat(os.path.join(base_output_dir, f'{filename}_content_list.json'))
35
36
    assert r.st_size > 5000

37
    r = os.stat(os.path.join(base_output_dir, f'{filename}.md'))
38
39
    assert r.st_size > 7000

40
    r = os.stat(os.path.join(base_output_dir, f'{filename}_middle.json'))
41
42
    assert r.st_size > 200000

43
    r = os.stat(os.path.join(base_output_dir, f'{filename}_model.json'))
44
45
    assert r.st_size > 15000

46
47
    r = os.stat(os.path.join(base_output_dir, f'{filename}_origin.pdf'))
    assert r.st_size > 400000
48

49
50
    r = os.stat(os.path.join(base_output_dir, f'{filename}_layout.pdf'))
    assert r.st_size > 400000
51

52
53
    r = os.stat(os.path.join(base_output_dir, f'{filename}_spans.pdf'))
    assert r.st_size > 400000
54

55
56
    os.path.exists(os.path.join(base_output_dir, 'images'))
    os.path.isdir(os.path.join(base_output_dir, 'images'))
57
58
59

    # teardown
    shutil.rmtree(temp_output_dir)