pygexml.strategies

 1# mypy: disallow_untyped_defs=False
 2# mypy: disallow_untyped_calls=False
 3
 4import string
 5
 6import hypothesis.strategies as st
 7
 8from pygexml.geometry import Point, Box, Polygon
 9from pygexml.page import Coords, Page, TextLine, TextRegion
10
11st_points = st.builds(Point, x=st.integers(min_value=0), y=st.integers(min_value=0))
12
13
14@st.composite
15def st_box_points(draw):
16    tl = draw(st_points)
17    br_x = draw(st.integers(min_value=tl.x + 1))
18    br_y = draw(st.integers(min_value=tl.y + 1))
19    br = Point(x=br_x, y=br_y)
20    return tl, br
21
22
23st_boxes = st.builds(
24    lambda pp: Box(top_left=pp[0], bottom_right=pp[1]), st_box_points()
25)
26
27st_polygons = st.builds(Polygon, st.lists(st_points, min_size=1))
28st_polygons2 = st.builds(Polygon, st.lists(st_points, min_size=2))
29
30st_coords = st.builds(Coords, polygon=st_polygons2)
31
32st_coords_strings = st.builds(str, st_coords)
33
34
35def st_xml_text(**kwargs):
36    xml_chars = (
37        st.characters(min_codepoint=0x20, max_codepoint=0xD7FF)
38        | st.characters(min_codepoint=0xE000, max_codepoint=0xFFFD)
39        | st.characters(min_codepoint=0x10000, max_codepoint=0x10FFFF)
40        | st.sampled_from(["\t", "\n"])
41    )
42    return st.text(alphabet=xml_chars, **kwargs)
43
44
45def st_simple_text(**kwargs):
46    simple = string.ascii_letters + string.digits + " _-"
47    return st.text(alphabet=simple, max_size=20, **kwargs)
48
49
50st_text_lines = st.builds(
51    TextLine, id=st_simple_text(), coords=st_coords, text=st_xml_text()
52)
53
54st_text_regions = st.builds(
55    TextRegion,
56    id=st_simple_text(),
57    coords=st_coords,
58    textlines=st.builds(
59        lambda lines: {l.id: l for l in lines}, st.lists(st_text_lines)
60    ),
61)
62
63
64@st.composite
65def st_pages(draw):
66    image_filename = draw(st_simple_text())
67    regions = {tr.id: tr for tr in draw(st.lists(st_text_regions))}
68    page = Page(image_filename=image_filename, regions=regions)
69    return page