Mercurial > hgrepos > Python2 > PyMuPDF
annotate tests/test_tables.py @ 46:7ee69f120f19 default tip
>>>>> tag v1.26.5+1 for changeset b74429b0f5c4
| author | Franz Glasner <fzglas.hg@dom66.de> |
|---|---|
| date | Sat, 11 Oct 2025 17:17:30 +0200 |
| parents | a6bc019ac0b2 |
| children |
| rev | line source |
|---|---|
|
1
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1 import os |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2 import io |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
3 from pprint import pprint |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
4 import textwrap |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
5 import pickle |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
6 import platform |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
7 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
8 import pymupdf |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
9 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
10 scriptdir = os.path.abspath(os.path.dirname(__file__)) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
11 filename = os.path.join(scriptdir, "resources", "chinese-tables.pdf") |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
12 pickle_file = os.path.join(scriptdir, "resources", "chinese-tables.pickle") |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
13 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
14 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
15 def test_table1(): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
16 """Compare pickled tables with those of the current run.""" |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
17 pickle_in = open(pickle_file, "rb") |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
18 doc = pymupdf.open(filename) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
19 page = doc[0] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
20 tabs = page.find_tables() |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
21 cells = tabs[0].cells + tabs[1].cells # all table cell tuples on page |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
22 extracts = [tabs[0].extract(), tabs[1].extract()] # all table cell content |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
23 old_data = pickle.load(pickle_in) # previously saved data |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
24 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
25 # Compare cell contents |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
26 assert old_data["extracts"] == extracts # same cell contents |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
27 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
28 # Compare cell coordinates. |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
29 # Cell rectangles may get somewhat larger due to more cautious border |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
30 # computations, but any differences must be small. |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
31 old_cells = old_data["cells"][0] + old_data["cells"][1] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
32 assert len(cells) == len(old_cells) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
33 for i in range(len(cells)): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
34 c1 = pymupdf.Rect(cells[i]) # new cell coordinates |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
35 c0 = pymupdf.Rect(old_cells[i]) # old cell coordinates |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
36 assert c0 in c1 # always: old contained in new |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
37 assert abs(c1 - c0) < 0.2 # difference must be small |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
38 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
39 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
40 def test_table2(): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
41 """Confirm header properties.""" |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
42 doc = pymupdf.open(filename) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
43 page = doc[0] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
44 tab1, tab2 = page.find_tables().tables |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
45 # both tables contain their header data |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
46 assert tab1.header.external == False |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
47 assert tab1.header.cells == tab1.rows[0].cells |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
48 assert tab2.header.external == False |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
49 assert tab2.header.cells == tab2.rows[0].cells |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
50 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
51 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
52 def test_2812(): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
53 """Ensure table detection and extraction independent from page rotation. |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
54 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
55 Make 4 pages with rotations 0, 90, 180 and 270 degrees respectively. |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
56 Each page shows the same 8x5 table. |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
57 We will check that each table is detected and delivers the same content. |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
58 """ |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
59 doc = pymupdf.open() |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
60 # Page 0: rotation 0 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
61 page = doc.new_page(width=842, height=595) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
62 rect = page.rect + (72, 72, -72, -72) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
63 cols = 5 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
64 rows = 8 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
65 # define the cells, draw the grid and insert unique text in each cell. |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
66 cells = pymupdf.make_table(rect, rows=rows, cols=cols) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
67 for i in range(rows): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
68 for j in range(cols): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
69 page.draw_rect(cells[i][j]) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
70 for i in range(rows): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
71 for j in range(cols): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
72 page.insert_textbox( |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
73 cells[i][j], |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
74 f"cell[{i}][{j}]", |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
75 align=pymupdf.TEXT_ALIGN_CENTER, |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
76 ) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
77 page.clean_contents() |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
78 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
79 # Page 1: rotation 90 degrees |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
80 page = doc.new_page() |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
81 rect = page.rect + (72, 72, -72, -72) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
82 cols = 8 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
83 rows = 5 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
84 cells = pymupdf.make_table(rect, rows=rows, cols=cols) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
85 for i in range(rows): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
86 for j in range(cols): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
87 page.draw_rect(cells[i][j]) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
88 for i in range(rows): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
89 for j in range(cols): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
90 page.insert_textbox( |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
91 cells[i][j], |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
92 f"cell[{j}][{rows-i-1}]", |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
93 rotate=90, |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
94 align=pymupdf.TEXT_ALIGN_CENTER, |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
95 ) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
96 page.set_rotation(90) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
97 page.clean_contents() |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
98 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
99 # Page 2: rotation 180 degrees |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
100 page = doc.new_page(width=842, height=595) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
101 rect = page.rect + (72, 72, -72, -72) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
102 cols = 5 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
103 rows = 8 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
104 cells = pymupdf.make_table(rect, rows=rows, cols=cols) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
105 for i in range(rows): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
106 for j in range(cols): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
107 page.draw_rect(cells[i][j]) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
108 for i in range(rows): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
109 for j in range(cols): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
110 page.insert_textbox( |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
111 cells[i][j], |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
112 f"cell[{rows-i-1}][{cols-j-1}]", |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
113 rotate=180, |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
114 align=pymupdf.TEXT_ALIGN_CENTER, |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
115 ) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
116 page.set_rotation(180) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
117 page.clean_contents() |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
118 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
119 # Page 3: rotation 270 degrees |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
120 page = doc.new_page() |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
121 rect = page.rect + (72, 72, -72, -72) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
122 cols = 8 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
123 rows = 5 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
124 cells = pymupdf.make_table(rect, rows=rows, cols=cols) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
125 for i in range(rows): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
126 for j in range(cols): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
127 page.draw_rect(cells[i][j]) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
128 for i in range(rows): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
129 for j in range(cols): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
130 page.insert_textbox( |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
131 cells[i][j], |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
132 f"cell[{cols-j-1}][{i}]", |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
133 rotate=270, |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
134 align=pymupdf.TEXT_ALIGN_CENTER, |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
135 ) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
136 page.set_rotation(270) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
137 page.clean_contents() |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
138 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
139 pdfdata = doc.tobytes() |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
140 # doc.ez_save("test-2812.pdf") |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
141 doc.close() |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
142 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
143 # ------------------------------------------------------------------------- |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
144 # Test PDF prepared. Extract table on each page and |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
145 # ensure identical extracted table data. |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
146 # ------------------------------------------------------------------------- |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
147 doc = pymupdf.open("pdf", pdfdata) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
148 extracts = [] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
149 for page in doc: |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
150 tabs = page.find_tables() |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
151 assert len(tabs.tables) == 1 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
152 tab = tabs[0] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
153 fp = io.StringIO() |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
154 pprint(tab.extract(), stream=fp) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
155 extracts.append(fp.getvalue()) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
156 fp = None |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
157 assert tab.row_count == 8 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
158 assert tab.col_count == 5 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
159 e0 = extracts[0] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
160 for e in extracts[1:]: |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
161 assert e == e0 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
162 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
163 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
164 def test_2979(): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
165 """This tests fix #2979 and #3001. |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
166 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
167 2979: identical cell count for each row |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
168 3001: no change of global glyph heights |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
169 """ |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
170 filename = os.path.join(scriptdir, "resources", "test_2979.pdf") |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
171 doc = pymupdf.open(filename) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
172 page = doc[0] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
173 tab = page.find_tables()[0] # extract the table |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
174 lengths = set() # stores all row cell counts |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
175 for e in tab.extract(): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
176 lengths.add(len(e)) # store number of cells for row |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
177 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
178 # test 2979 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
179 assert len(lengths) == 1 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
180 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
181 # test 3001 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
182 assert ( |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
183 pymupdf.TOOLS.set_small_glyph_heights() is False |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
184 ), f"{pymupdf.TOOLS.set_small_glyph_heights()=}" |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
185 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
186 wt = pymupdf.TOOLS.mupdf_warnings() |
|
39
a6bc019ac0b2
ADD: PyMuPDF v1.26.5: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
1
diff
changeset
|
187 if pymupdf.mupdf_version_tuple >= (1, 26, 8): |
|
a6bc019ac0b2
ADD: PyMuPDF v1.26.5: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
1
diff
changeset
|
188 assert ( |
|
a6bc019ac0b2
ADD: PyMuPDF v1.26.5: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
1
diff
changeset
|
189 wt |
|
a6bc019ac0b2
ADD: PyMuPDF v1.26.5: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
1
diff
changeset
|
190 == "bogus font ascent/descent values (3117 / -2463)\n... repeated 2 times...\nActualtext with no position. Text may be lost or mispositioned.\n... repeated 96 times..." |
|
a6bc019ac0b2
ADD: PyMuPDF v1.26.5: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
1
diff
changeset
|
191 ) |
|
a6bc019ac0b2
ADD: PyMuPDF v1.26.5: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
1
diff
changeset
|
192 elif pymupdf.mupdf_version_tuple >= (1, 26, 0): |
|
1
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
193 assert ( |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
194 wt |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
195 == "bogus font ascent/descent values (3117 / -2463)\n... repeated 2 times..." |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
196 ) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
197 else: |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
198 assert not wt |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
199 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
200 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
201 def test_3062(): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
202 """Tests the fix for #3062. |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
203 After table extraction, a rotated page should behave and look |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
204 like as before.""" |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
205 if platform.python_implementation() == 'GraalVM': |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
206 print(f'test_3062(): Not running because slow on GraalVM.') |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
207 return |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
208 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
209 filename = os.path.join(scriptdir, "resources", "test_3062.pdf") |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
210 doc = pymupdf.open(filename) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
211 page = doc[0] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
212 tab0 = page.find_tables()[0] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
213 cells0 = tab0.cells |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
214 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
215 page = None |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
216 page = doc[0] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
217 tab1 = page.find_tables()[0] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
218 cells1 = tab1.cells |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
219 assert cells1 == cells0 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
220 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
221 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
222 def test_strict_lines(): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
223 """Confirm that ignoring borderless rectangles improves table detection.""" |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
224 filename = os.path.join(scriptdir, "resources", "strict-yes-no.pdf") |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
225 doc = pymupdf.open(filename) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
226 page = doc[0] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
227 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
228 tab1 = page.find_tables()[0] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
229 tab2 = page.find_tables(strategy="lines_strict")[0] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
230 assert tab2.row_count < tab1.row_count |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
231 assert tab2.col_count < tab1.col_count |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
232 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
233 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
234 def test_add_lines(): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
235 """Test new parameter add_lines for table recognition.""" |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
236 if platform.python_implementation() == 'GraalVM': |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
237 print(f'test_add_lines(): Not running because breaks later tests on GraalVM.') |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
238 return |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
239 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
240 filename = os.path.join(scriptdir, "resources", "small-table.pdf") |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
241 doc = pymupdf.open(filename) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
242 page = doc[0] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
243 assert page.find_tables().tables == [] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
244 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
245 more_lines = [ |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
246 ((238.9949951171875, 200.0), (238.9949951171875, 300.0)), |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
247 ((334.5559997558594, 200.0), (334.5559997558594, 300.0)), |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
248 ((433.1809997558594, 200.0), (433.1809997558594, 300.0)), |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
249 ] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
250 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
251 # these 3 additional vertical lines should additional 3 columns |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
252 tab2 = page.find_tables(add_lines=more_lines)[0] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
253 assert tab2.col_count == 4 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
254 assert tab2.row_count == 5 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
255 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
256 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
257 def test_3148(): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
258 """Ensure correct extraction text of rotated text.""" |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
259 doc = pymupdf.open() |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
260 page = doc.new_page() |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
261 rect = pymupdf.Rect(100, 100, 300, 300) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
262 text = ( |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
263 "rotation 0 degrees", |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
264 "rotation 90 degrees", |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
265 "rotation 180 degrees", |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
266 "rotation 270 degrees", |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
267 ) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
268 degrees = (0, 90, 180, 270) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
269 delta = (2, 2, -2, -2) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
270 cells = pymupdf.make_table(rect, cols=3, rows=4) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
271 for i in range(3): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
272 for j in range(4): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
273 page.draw_rect(cells[j][i]) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
274 k = (i + j) % 4 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
275 page.insert_textbox(cells[j][i] + delta, text[k], rotate=degrees[k]) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
276 # doc.save("multi-degree.pdf") |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
277 tabs = page.find_tables() |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
278 tab = tabs[0] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
279 for extract in tab.extract(): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
280 for item in extract: |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
281 item = item.replace("\n", " ") |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
282 assert item in text |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
283 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
284 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
285 def test_3179(): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
286 """Test correct separation of multiple tables on page.""" |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
287 filename = os.path.join(scriptdir, "resources", "test_3179.pdf") |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
288 doc = pymupdf.open(filename) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
289 page = doc[0] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
290 tabs = page.find_tables() |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
291 assert len(tabs.tables) == 3 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
292 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
293 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
294 def test_battery_file(): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
295 """Tests correctly ignoring non-table suspects. |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
296 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
297 Earlier versions erroneously tried to identify table headers |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
298 where there existed no table at all. |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
299 """ |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
300 filename = os.path.join(scriptdir, "resources", "battery-file-22.pdf") |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
301 doc = pymupdf.open(filename) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
302 page = doc[0] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
303 tabs = page.find_tables() |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
304 assert len(tabs.tables) == 0 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
305 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
306 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
307 def test_markdown(): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
308 """Confirm correct markdown output.""" |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
309 filename = os.path.join(scriptdir, "resources", "strict-yes-no.pdf") |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
310 doc = pymupdf.open(filename) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
311 page = doc[0] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
312 tab = page.find_tables(strategy="lines_strict")[0] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
313 if pymupdf.mupdf_version_tuple < (1, 26, 3): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
314 md_expected = textwrap.dedent(''' |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
315 |Header1|Header2|Header3| |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
316 |---|---|---| |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
317 |Col11<br>Col12|~~Col21~~<br>~~Col22~~|Col31<br>Col32<br>Col33| |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
318 |Col13|~~Col23~~|Col34<br>Col35| |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
319 |Col14|~~Col24~~|Col36| |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
320 |Col15|~~Col25~~<br>~~Col26~~|| |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
321 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
322 ''').lstrip() |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
323 else: |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
324 md_expected = ( |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
325 "|Header1|Header2|Header3|\n" |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
326 "|---|---|---|\n" |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
327 "|Col11<br>Col12|Col21<br>Col22|Col31<br>Col32<br>Col33|\n" |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
328 "|Col13|Col23|Col34<br>Col35|\n" |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
329 "|Col14|Col24|Col36|\n" |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
330 "|Col15|Col25<br>Col26||\n\n" |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
331 ) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
332 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
333 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
334 md = tab.to_markdown() |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
335 assert md == md_expected, f'Incorrect md:\n{textwrap.indent(md, " ")}' |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
336 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
337 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
338 def test_paths_param(): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
339 """Confirm acceptance of supplied vector graphics list.""" |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
340 filename = os.path.join(scriptdir, "resources", "strict-yes-no.pdf") |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
341 doc = pymupdf.open(filename) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
342 page = doc[0] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
343 tabs = page.find_tables(paths=[]) # will cause all tables are missed |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
344 assert tabs.tables == [] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
345 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
346 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
347 def test_boxes_param(): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
348 """Confirm acceptance of supplied boxes list.""" |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
349 filename = os.path.join(scriptdir, "resources", "small-table.pdf") |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
350 doc = pymupdf.open(filename) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
351 page = doc[0] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
352 paths = page.get_drawings() |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
353 box0 = page.cluster_drawings(drawings=paths)[0] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
354 boxes = [box0] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
355 words = page.get_text("words") |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
356 x_vals = [w[0] - 5 for w in words if w[4] in ("min", "max", "avg")] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
357 for x in x_vals: |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
358 r = +box0 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
359 r.x1 = x |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
360 boxes.append(r) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
361 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
362 y_vals = sorted(set([round(w[3]) for w in words])) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
363 for y in y_vals[:-1]: # skip last one to avoid empty row |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
364 r = +box0 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
365 r.y1 = y |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
366 boxes.append(r) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
367 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
368 tabs = page.find_tables(paths=[], add_boxes=boxes) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
369 tab = tabs.tables[0] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
370 assert tab.extract() == [ |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
371 ["Boiling Points °C", "min", "max", "avg"], |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
372 ["Noble gases", "-269", "-62", "-170.5"], |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
373 ["Nonmetals", "-253", "4827", "414.1"], |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
374 ["Metalloids", "335", "3900", "741.5"], |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
375 ["Metals", "357", ">5000", "2755.9"], |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
376 ] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
377 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
378 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
379 def test_dotted_grid(): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
380 """Confirm dotted lines are detected as gridlines.""" |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
381 filename = os.path.join(scriptdir, "resources", "dotted-gridlines.pdf") |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
382 doc = pymupdf.open(filename) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
383 page = doc[0] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
384 tabs = page.find_tables() |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
385 assert len(tabs.tables) == 3 # must be 3 tables |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
386 t0, t1, t2 = tabs # extract them |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
387 # check that they have expected dimensions |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
388 assert t0.row_count, t0.col_count == (11, 12) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
389 assert t1.row_count, t1.col_count == (25, 11) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
390 assert t2.row_count, t2.col_count == (1, 10) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
391 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
392 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
393 def test_4017(): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
394 path = os.path.normpath(f"{__file__}/../../tests/resources/test_4017.pdf") |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
395 with pymupdf.open(path) as document: |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
396 page = document[0] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
397 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
398 tables = page.find_tables(add_lines=None) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
399 print(f"{len(tables.tables)=}.") |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
400 tables_text = list() |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
401 for i, table in enumerate(tables): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
402 print(f"## {i=}.") |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
403 t = table.extract() |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
404 for tt in t: |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
405 print(f" {tt}") |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
406 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
407 # 2024-11-29: expect current incorrect output for last two tables. |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
408 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
409 expected_a = [ |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
410 ["Class A/B Overcollateralization", "131.44%", ">=", "122.60%", "", "PASS"], |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
411 [None, None, None, None, None, "PASS"], |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
412 ["Class D Overcollateralization", "112.24%", ">=", "106.40%", "", "PASS"], |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
413 [None, None, None, None, None, "PASS"], |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
414 ["Event of Default", "156.08%", ">=", "102.50%", "", "PASS"], |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
415 [None, None, None, None, None, "PASS"], |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
416 ["Class A/B Interest Coverage", "N/A", ">=", "120.00%", "", "N/A"], |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
417 [None, None, None, None, None, "N/A"], |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
418 ["Class D Interest Coverage", "N/A", ">=", "105.00%", "", "N/A"], |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
419 ] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
420 assert tables[-2].extract() == expected_a |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
421 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
422 expected_b = [ |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
423 [ |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
424 "Moody's Maximum Rating Factor Test", |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
425 "2,577", |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
426 "<=", |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
427 "3,250", |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
428 "", |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
429 "PASS", |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
430 "2,581", |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
431 ], |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
432 [None, None, None, None, None, "PASS", None], |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
433 [ |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
434 "Minimum Floating Spread", |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
435 "3.5006%", |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
436 ">=", |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
437 "2.0000%", |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
438 "", |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
439 "PASS", |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
440 "3.4871%", |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
441 ], |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
442 [None, None, None, None, None, "PASS", None], |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
443 [ |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
444 "Minimum Weighted Average S&P Recovery\nRate Test", |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
445 "40.50%", |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
446 ">=", |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
447 "40.00%", |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
448 "", |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
449 "PASS", |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
450 "40.40%", |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
451 ], |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
452 [None, None, None, None, None, "PASS", None], |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
453 ["Weighted Average Life", "4.83", "<=", "9.00", "", "PASS", "4.92"], |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
454 ] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
455 assert tables[-1].extract() == expected_b |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
456 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
457 |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
458 def test_md_styles(): |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
459 """Test output of table with MD-styled cells.""" |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
460 filename = os.path.join(scriptdir, "resources", "test-styled-table.pdf") |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
461 doc = pymupdf.open(filename) |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
462 page = doc[0] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
463 tabs = page.find_tables()[0] |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
464 text = """|Column 1|Column 2|Column 3|\n|---|---|---|\n|Zelle (0,0)|**Bold (0,1)**|Zelle (0,2)|\n|~~Strikeout (1,0), Zeile 1~~<br>~~Hier kommt Zeile 2.~~|Zelle (1,1)|~~Strikeout (1,2)~~|\n|**`Bold-monospaced`**<br>**`(2,0)`**|_Italic (2,1)_|**_Bold-italic_**<br>**_(2,2)_**|\n|Zelle (3,0)|~~**Bold-strikeout**~~<br>~~**(3,1)**~~|Zelle (3,2)|\n\n""" |
|
1d09e1dec1d9
ADD: PyMuPDF v1.26.4: the original sdist.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
465 assert tabs.to_markdown() == text |
