def test_extract_strings(self):
template, strings = extract_strings("""
<p><b>Bread</b>\xa0is a\xa0<a href="https://en.wikipedia.org/wiki/Staple_food">staple food</a>\xa0prepared from a\xa0<a href="https://en.wikipedia.org/wiki/Dough">dough</a>\xa0of\xa0<a href="https://en.wikipedia.org/wiki/Flour">flour</a>\xa0and\xa0<a href="https://en.wikipedia.org/wiki/Water">water</a>, usually by\xa0<a href="https://en.wikipedia.org/wiki/Baking">baking</a>. Throughout recorded history it has been popular around the world and is one of the oldest artificial foods, having been of importance since the dawn of\xa0<a href="https://en.wikipedia.org/wiki/Agriculture#History">agriculture</a>.</p>
<p>Proportions of types of flour and other ingredients vary widely, as do modes of preparation. As a result, types, shapes, sizes, and textures of breads differ around the world. Bread may be\xa0<a href="https://en.wikipedia.org/wiki/Leaven">leavened</a>\xa0by processes such as reliance on naturally occurring\xa0<a href="https://en.wikipedia.org/wiki/Sourdough">sourdough</a>\xa0microbes, chemicals, industrially produced yeast, or high-pressure aeration. Some bread is cooked before it can leaven, including for traditional or religious reasons. Non-cereal ingredients such as fruits, nuts and fats may be included. Commercial bread commonly contains additives to improve flavor, texture, color, shelf life, and ease of manufacturing.</p>
""")
self.assertHTMLEqual(
template,
"""
<p><text position="0"></text></p>
<p><text position="1"></text></p>
""",
)
self.assertEqual(
strings,
[
StringValue.from_html(
'<b>Bread</b>\xa0is a\xa0<a href="https://en.wikipedia.org/wiki/Staple_food">staple food</a>\xa0prepared from a\xa0<a href="https://en.wikipedia.org/wiki/Dough">dough</a>\xa0of\xa0<a href="https://en.wikipedia.org/wiki/Flour">flour</a>\xa0and\xa0<a href="https://en.wikipedia.org/wiki/Water">water</a>, usually by\xa0<a href="https://en.wikipedia.org/wiki/Baking">baking</a>. Throughout recorded history it has been popular around the world and is one of the oldest artificial foods, having been of importance since the dawn of\xa0<a href="https://en.wikipedia.org/wiki/Agriculture#History">agriculture</a>.'
),
StringValue.from_html(
'Proportions of types of flour and other ingredients vary widely, as do modes of preparation. As a result, types, shapes, sizes, and textures of breads differ around the world. Bread may be\xa0<a href="https://en.wikipedia.org/wiki/Leaven">leavened</a>\xa0by processes such as reliance on naturally occurring\xa0<a href="https://en.wikipedia.org/wiki/Sourdough">sourdough</a>\xa0microbes, chemicals, industrially produced yeast, or high-pressure aeration. Some bread is cooked before it can leaven, including for traditional or religious reasons. Non-cereal ingredients such as fruits, nuts and fats may be included. Commercial bread commonly contains additives to improve flavor, texture, color, shelf life, and ease of manufacturing.'
),
],
)
def test_block_tag_in_inline_tag(self):
# If an inline tag contains a block tag. The inline tag must be in the template.
# Testing for issue https://github.com/mozilla/donate-wagtail/issues/586
template, strings = extract_strings("<p><i>Foo <p>Bar</p></i></p>")
self.assertHTMLEqual(
template,
'<p><i><text position="0"></text> <p><text position="1"></text></p></i></p>',
)
self.assertEqual(
strings,
[StringValue.from_html("Foo"),
StringValue.from_html("Bar")])
def test_string_from_html(self):
string, attrs = StringValue.from_html(
'<b>Bread</b>\xa0is a\xa0<a href="https://en.wikipedia.org/wiki/Staple_food">staple food</a>\xa0prepared from a\xa0<a href="https://en.wikipedia.org/wiki/Dough">dough</a>\xa0of\xa0<a href="https://en.wikipedia.org/wiki/Flour">flour</a>\xa0and\xa0<a href="https://en.wikipedia.org/wiki/Water">water</a>'
)
self.assertEqual(
string.data,
'<b>Bread</b> is a <a id="a1">staple food</a> prepared from a <a id="a2">dough</a> of <a id="a3">flour</a> and <a id="a4">water</a>',
)
self.assertEqual(
attrs, {
'a1': {
'href': 'https://en.wikipedia.org/wiki/Staple_food'
},
'a2': {
'href': 'https://en.wikipedia.org/wiki/Dough'
},
'a3': {
'href': 'https://en.wikipedia.org/wiki/Flour'
},
'a4': {
'href': 'https://en.wikipedia.org/wiki/Water'
},
})
def test_empty_inline_tag(self):
template, strings = extract_strings("<p><i></i>Foo</p>")
self.assertHTMLEqual(template,
'<p><i></i><text position="0"></text></p>')
self.assertEqual(strings, [StringValue.from_html("Foo")])
def test_br_tag_is_removed_when_it_appears_at_end_of_segment(self):
template, strings = extract_strings("<p><i>Foo</i><br/></p>")
self.assertHTMLEqual(template,
'<p><i><text position="0"></text></i><br/></p>')
self.assertEqual(strings, [StringValue.from_html("Foo")])
def test_br_tag_is_treated_as_inline_tag(self):
template, strings = extract_strings(
"<p><b>Foo <i>Bar<br/>Baz</i></b></p>")
self.assertHTMLEqual(template,
'<p><b><text position="0"></text></b></p>')
self.assertEqual(strings,
[StringValue.from_html("Foo <i>Bar<br/>Baz</i>")])
def test_extract_strings_2(self):
template, strings = extract_strings("""
<h1>Foo bar baz</h1>
<p>This is a paragraph. <b>This is some bold <i>and now italic</i></b> text</p>
<p><script> this should be interpreted as text.</p>
<ul>
<li>List item one</li>
<li><b>List item two</li>
</ul>
<img src="foo" alt="This bit isn't translatable">
""")
self.assertHTMLEqual(
template,
"""
<h1><text position="0"></text></h1>
<p><text position="1"></text></p>
<p><text position="2"></text></p>
<ul>
<li><text position="3"></text></li>
<li><b><text position="4"></text></b></li>
</ul>
<img alt="This bit isn\'t translatable" src="foo">
""",
)
self.assertEqual(
strings,
[
StringValue.from_html("Foo bar baz"),
StringValue.from_html(
"This is a paragraph. <b>This is some bold <i>and now italic</i></b> text"
),
StringValue.from_html(
"<script> this should be interpreted as text."),
StringValue.from_html("List item one"),
StringValue.from_html("List item two"),
],
)