def test_build_string():
xml_str = """<String ID="string_3" HPOS="712" VPOS="133" WIDTH="55" HEIGHT="13" WC="0.92" CONTENT="Liberté"/>"""
element = _build_xml(xml_str)
assert String.from_xml(element) == String(
id="string_3",
hpos=712,
vpos=133,
width=55,
height=13,
confidence=0.92,
content="Liberté",
alternatives=[],
)
xml_str = """
<String ID="string_3" HPOS="712" VPOS="133" WIDTH="55" HEIGHT="13" WC="0.92" CONTENT="Liberté">
<Alternative>alt</Alternative>
</String>
"""
element = _build_xml(xml_str)
assert String.from_xml(element) == String(
id="string_3",
hpos=712,
vpos=133,
width=55,
height=13,
confidence=0.92,
content="Liberté",
alternatives=[Alternative("alt")],
)
def test_build_text_block():
element = _build_xml("""
<TextBlock ID="block_1" HPOS="712" VPOS="129" WIDTH="235" HEIGHT="53">
<TextLine ID="line_2" HPOS="712" VPOS="129" WIDTH="235" HEIGHT="21">
<String ID="string_3" HPOS="712" VPOS="133" WIDTH="55" HEIGHT="13" WC="0.92" CONTENT="abc"/>
<SP WIDTH="9" VPOS="133" HPOS="767"/>
</TextLine>
</TextBlock>
""")
strings = [
String(
id="string_3",
hpos=712,
vpos=133,
width=55,
height=13,
confidence=0.92,
content="abc",
alternatives=[],
),
SP(hpos=767, vpos=133, width=9),
]
lines = [
TextLine(id="line_2",
hpos=712,
vpos=129,
width=235,
height=21,
strings=strings)
]
assert TextBlock.from_xml(element) == TextBlock("block_1", 53, 235, 712,
129, lines)
def test_page_extract_strings():
assert Page("page_0", 2339, 1654, 0, None, []).extract_strings() == []
assert Page("page_0", 2339, 1654, 0, None,
[PrintSpace(1, 1, 1, 1, 1, [])]).extract_strings() == []
block = ComposedBlock("", 1, 1, 1, 1, [])
assert Page("page_0", 2339, 1654, 0, None,
[PrintSpace(1, 1, 1, 1, 1, [block])]).extract_strings() == []
tb = TextBlock("", 1, 1, 1, 1, [])
block = ComposedBlock("", 1, 1, 1, 1, [tb])
assert Page("page_0", 2339, 1654, 0, None,
[PrintSpace(1, 1, 1, 1, 1, [block])]).extract_strings() == []
tb = TextBlock(
"", 1, 1, 1, 1,
[TextLine("", 1, 1, 1, 1, [String("", 1, 1, 1, 1, "", 0, [])])])
block = ComposedBlock("", 1, 1, 1, 1, [tb])
page = Page("page_0", 2339, 1654, 0, None,
[PrintSpace(1, 1, 1, 1, 1, [block])])
assert page.extract_strings() == [String("", 1, 1, 1, 1, "", 0, [])]
def _string(word: str = 'abc') -> String:
return String(id="string_3",
hpos=712,
vpos=133,
width=55,
height=13,
confidence=0.92,
content=word,
alternatives=[])
def test_build_text_line():
xml_str = """
<TextLine ID="line_2" HPOS="712" VPOS="129" WIDTH="235" HEIGHT="21">
<String ID="string_3" HPOS="712" VPOS="133" WIDTH="55" HEIGHT="13" WC="0.92" CONTENT="abc"/>
<SP WIDTH="9" VPOS="133" HPOS="767"/>
</TextLine>
"""
element = _build_xml(xml_str)
strings = [
String(
id="string_3",
hpos=712,
vpos=133,
width=55,
height=13,
confidence=0.92,
content="abc",
alternatives=[],
),
SP(hpos=767, vpos=133, width=9),
]
assert TextLine.from_xml(element) == TextLine(id="line_2",
hpos=712,
vpos=129,
width=235,
height=21,
strings=strings)
xml_str = """
<TextLine ID="line_2" HPOS="712" VPOS="129" WIDTH="235" HEIGHT="21">
</TextLine>
"""
element = _build_xml(xml_str)
assert TextLine.from_xml(element) == TextLine(id="line_2",
hpos=712,
vpos=129,
width=235,
height=21,
strings=[])
def test_load_string_or_sp():
xml_str = """<SP WIDTH="9" VPOS="133" HPOS="767"/>"""
element = _build_xml(xml_str)
assert _load_string_or_sp(element) == SP(hpos=767, vpos=133, width=9)
xml_str = """<String ID="string_3" HPOS="712" VPOS="133" WIDTH="55" HEIGHT="13" WC="0.92" CONTENT="abc"/>"""
element = _build_xml(xml_str)
assert _load_string_or_sp(element) == String(
id="string_3",
hpos=712,
vpos=133,
width=55,
height=13,
confidence=0.92,
content="abc",
alternatives=[],
)
with pytest.raises(ValueError):
xml_str = """<Alternative>test</Alternative>"""
element = _build_xml(xml_str)
_load_string_or_sp(element)