def test_furigana(self):
assert clean_spaces("abc def[ghi]") == "abc def[ghi]"
assert clean_spaces("abc def[ghi]") == "abc def[ghi]"
assert clean_spaces("abc def[ghi]") == "abc def[ghi]"
assert clean_spaces(" abc def[ghi] ") == "abc def[ghi]"
assert clean_spaces(" a[bc] de[fghi] jkl[mnop] qr "
) == "a[bc] de[fghi] jkl[mnop]qr"
def test_html(self):
assert clean_spaces("<b>foo</b>") == "<b>foo</b>"
assert clean_spaces("<b> foo</b>") == "<b>foo</b>"
assert clean_spaces("<b> foo </b>") == "<b>foo</b>"
assert clean_spaces(
"""<a src="foo">bar</b>""") == """<a src="foo">bar</b>"""
assert clean_spaces(
"""<a src="foo">bar baz</a>""") == """<a src="foo">barbaz</a>"""
assert clean_spaces(""" <a src="foo">bar baz</a> """
) == """<a src="foo">barbaz</a>"""
assert clean_spaces("""<a src="foo"> bar baz</a> """
) == """<a src="foo">barbaz</a>"""
assert clean_spaces(""" <a src="foo">bar baz[bee]</a> """
) == """<a src="foo">bar baz[bee]</a>"""
assert clean_spaces("""blah <a src="foo">bar[bee]</a> """
) == """blah <a src="foo">bar[bee]</a>"""
assert clean_spaces("""blah<a src="foo"> bar[bee]</a> """
) == """blah<a src="foo"> bar[bee]</a>"""
assert clean_spaces("""blah <a src="foo"> bar[bee]</a> """
) == """blah<a src="foo"> bar[bee]</a>"""
assert clean_spaces(" abc <a href=\"foo\">def</a>[ghi] \n <b>mn</b>[aa] op[qrs] tuv[wy] ") \
== "abc <a href=\"foo\">def</a>[ghi]\n<b>mn</b>[aa] op[qrs] tuv[wy]"
def test_brackets(self):
with pytest.raises(JapaneseReadingFormattingError):
clean_spaces("abc[def")
with pytest.raises(JapaneseReadingFormattingError):
clean_spaces("abc]def")
clean_spaces("abc[def]")
clean_spaces("abc[def]ghi[jk]")
with pytest.raises(JapaneseReadingFormattingError):
clean_spaces("abc[def]]")
with pytest.raises(JapaneseReadingFormattingError):
clean_spaces("abc[[def]")
def test_multiple_lines(self):
assert clean_spaces(" abc def[ghi] \n mn[aa] op[qrs] tuv[wy] ") \
== "abc def[ghi]\nmn[aa] op[qrs] tuv[wy]"
def test_spaces(self):
assert clean_spaces("abc def") == "abcdef"
assert clean_spaces("abc def") == "abcdef"
assert clean_spaces(" a b c d e f ") == "abcdef"