def test_heading(self):
splitter = HtmlSplitter(path=Path(FIXTURE_DIR).joinpath("heading.html"))
html2topics = HtmlToTopics(splitter, render_cover_page=True)
expected = [
NeoTopic(title="Cover Page", content="\n\nIntroduction\n\n"),
NeoTopic(
title="My First Heading",
content="""
<p>My first paragraph.</p>
""",
),
]
self.assertEqual(len(html2topics.resources), 0)
self.assertEqual(html2topics.topics, expected)
def test_empty_title(self):
splitter = HtmlSplitter(path=Path(FIXTURE_DIR).joinpath("empty_title.html"))
html2topics = HtmlToTopics(splitter, render_cover_page=True)
expected = [
NeoTopic(
title="Cover Page",
origin_id=None,
content="\n\nIntroduction\n\n \nText that should be in the introduction.\n\n ",
),
NeoTopic(
title="Installation",
origin_id="_Ref2A4E1AB689A0D2EE52FF15610E2D8283",
content="\n a\n \n b\n\n ",
),
NeoTopic(title="Removal", origin_id="_Re2D8283", content="\n c\n \n d\n "),
]
self.assertEqual(len(html2topics.resources), 0)
self.assertEqual(html2topics.topics, expected)
def test_heading_three_level(self):
splitter = HtmlSplitter(path=Path(FIXTURE_DIR).joinpath("heading_three_levels.html"))
html2topics = HtmlToTopics(splitter, render_cover_page=True)
expected = [
NeoTopic(title="Cover Page", content="\n\nIntroduction\n\n"),
NeoTopic(
title="Heading 1",
content="\na\n",
children=[
NeoTopic(
title="Heading 1-2",
content="\nb\n",
children=[NeoTopic(title="Heading 1-2-3", content="\nc\n")],
)
],
),
]
self.assertEqual(len(html2topics.resources), 0)
self.assertEqual(expected, html2topics.topics)
def test_cover_page(self):
splitter = HtmlSplitter(
content="""<!DOCTYPE html>
<html>
<body>
a
<h1>b</h1>
c
</body>
</html>
"""
)
html2topics = HtmlToTopics(splitter, render_cover_page=True)
self.assertEqual(html2topics.resources, [])
expected = [NeoTopic(title="Cover Page", content="a\n"), NeoTopic(title="b", content="c\n")]
self.assertEqual(html2topics.topics, expected)
html2topics = HtmlToTopics(splitter, render_cover_page=False)
expected = expected[1:]
self.assertEqual(html2topics.resources, [])
self.assertEqual(html2topics.topics, expected)
def test_empty_without_cover_page(self):
splitter = HtmlSplitter(
content="""<!DOCTYPE html>
<html>
<body>
a
b
c
</body>
</html>
"""
)
html2topics = HtmlToTopics(splitter, render_cover_page=True)
self.assertEqual(html2topics.resources, [])
expected_content = "a\nb\nc\n"
expected = [NeoTopic(title="Cover Page", content=expected_content)]
self.assertEqual(html2topics.topics, expected)
html2topics = HtmlToTopics(splitter, render_cover_page=False)
expected = [NeoTopic(title="Flat document", content=expected_content)]
self.assertEqual(html2topics.resources, [])
self.assertEqual(html2topics.topics, expected)
def test_anchor(self):
splitter = HtmlSplitter(path=Path(FIXTURE_DIR).joinpath("anchor.html"))
html2topics = HtmlToTopics(splitter, render_cover_page=True)
expected = [
NeoTopic(title="Cover Page", origin_id=None, content="\n\nIntroduction\n\n"),
NeoTopic(title="Heading 1", origin_id="heading1", content='\n\nHeading 1\n\n<a href="heading2"></a>\n\n'),
NeoTopic(
title="Heading 2",
origin_id="heading2",
content="""
Heading 2
<a href="heading1"></a>
<a href="https://google.com/#Heading2">Clique</a>
""",
),
]
self.assertEqual(len(html2topics.resources), 0)
self.assertEqual(html2topics.topics, expected)
def test_headings(self):
splitter = HtmlSplitter(path=Path(FIXTURE_DIR).joinpath("headings_simple.html"))
html2topics = HtmlToTopics(splitter, render_cover_page=True)
expected = [
NeoTopic(title="Cover Page", content="z"),
NeoTopic(
title="Heading 1",
content="\na\n",
children=[
NeoTopic(
title="Heading 1-2", content="\nb\n", children=[NeoTopic(title="Heading 1-2-3", content="c")]
)
],
),
NeoTopic(title="Heading 1.2", content="\nd\n"),
NeoTopic(title="Heading 1.3", content="\ne\n", children=[NeoTopic(title="Heading 1.3-2", content="\nf\n")]),
]
self.assertEqual(len(html2topics.resources), 0)
for i, part in enumerate(html2topics.topics):
self.assertEqual(
expected[i],
part,
"What we have:\n{}\n\n{}\nWhat we expect:\n{}\n\n".format(part, "-" * 80, expected[i]),
)