def get_netilt_xml(self, use_spine_as_toc):
self.epub_archive = EpubArchive(self.epub_filename, use_spine_as_toc)
document = etree.Element("document")
add_element_with_text(document, "title", self.epub_archive.title)
add_element_with_text(document, "authors",
", ".join(self.epub_archive.authors))
for page in self.epub_archive.pages:
page_root_elem = self.chapter_elements.get(page.parent_page,
document)
page_elem = etree.Element("page")
if page.children_pages:
add_element_with_text(page_elem, "title", "Overview")
chapter_elem = etree.Element("chapter")
if page.get_page_title() is not None:
add_element_with_text(chapter_elem, "title",
page.get_page_title())
page_root_elem.append(chapter_elem)
self.chapter_elements[page] = chapter_elem
page_root_elem = chapter_elem
else:
if page.get_page_title() is not None:
add_element_with_text(page_elem, "title",
page.get_page_title())
page_root_elem.append(page_elem)
for index, section in enumerate(page.sections):
page_elem.append(
epub_page_section_to_netilt(section, "section", index))
return document
def test_very_last_page_in_doc(self):
archive = EpubArchive("test_data/nested_navpoints.epub", False)
page = archive.pages[18]
self.assertEqual(
page.get_page_title(),
"Side-by-Side Migration"
)
self.assertTrue(
page.page_content_parsed.find(".//body").text_content().strip().endswith(
"after the migration is complete."
))
def test_middle_page(self):
archive = EpubArchive("test_data/nested_navpoints.epub", False)
page = archive.pages[11]
self.assertEqual(
page.get_page_title(),
"Additional SQL Server 2008 R2 Enhancements for DBAs"
)
self.assertEqual(
len(page.page_content_parsed.find(".//body")),
3 # h4, p, ul
)
self.assertTrue(
page.page_content_parsed.find(".//body").text_content().strip().endswith(
"operating systems that support Extended Protection."
))
def test_very_first_page_in_doc(self):
archive = EpubArchive("test_data/nested_navpoints.epub", False)
page = archive.pages[8]
self.assertEqual(
page.get_page_title(),
"CHAPTER 1 SQL Server 2008 R2 Editions and Enhancements"
)
self.assertEqual(
len(page.page_content_parsed.find(".//body")),
3 # h2, p, p
)
self.assertTrue(
page.page_content_parsed.find(".//body").text_content().strip().endswith(
"installation strategies are also identified."
))
def test_nested_pages(self):
archive = EpubArchive("test_data/nested_navpoints.epub", False)
self.assertEqual(len(archive.pages), 29)
self.assertEqual(
archive.pages[11].get_page_title(),
'Additional SQL Server 2008 R2 Enhancements for DBAs'
)
self.assertEqual(
archive.pages[11].parent_page.get_page_title(),
'SQL Server 2008 R2 Enhancements for DBAs'
)
self.assertEqual(
archive.pages[11].parent_page,
archive.pages[9]
)
self.assertEqual(
archive.pages[9].parent_page.get_page_title(),
"CHAPTER 1 SQL Server 2008 R2 Editions and Enhancements"
)
self.assertEqual(
archive.pages[9].parent_page,
archive.pages[8]
)
self.assertEqual(
len(archive.pages[8].sections),
1
)
self.assertEqual(
len(archive.pages[8].sections[0].children_sections),
0
)
self.assertEqual(
archive.pages[8].parent_page.get_page_title(),
"Part I Database Administration"
)
self.assertEqual(
archive.pages[8].parent_page,
archive.pages[7]
)
self.assertEqual(
archive.pages[7].parent_page,
None
)
self.assertEqual(
[p.get_page_title() for p in archive.pages[7].children_pages],
["CHAPTER 1 SQL Server 2008 R2 Editions and Enhancements", "CHAPTER 10 Self-Service Analysis with PowerPivot"]
)
def test_search_headers_by_text(self):
archive = EpubArchive("test_data/sicp.epub", False)
page = archive.pages[5]
self.assertEqual(
page.get_page_title(),
"1.1.2 Naming and the Environment"
)
self.assertTrue(
page.page_content_parsed.find(".//body").text_content().strip().startswith(
"1.1.2 Naming and the Environment A critical aspect of a programming language"
))
self.assertTrue(
page.page_content_parsed.find(".//body").text_content().strip().endswith(
"a number of different environments).9"
))
self.assertEqual(
len(page.sections),
1
)
self.assertEqual(
len(page.sections[0].children_sections),
0
)
def test_nav_alice_short(self):
simple_archive = EpubArchive("test_data/in1.epub", False)
self.assertEqual(len(simple_archive.pages), 4) # all pages besides Cover
def test_alice_short(self):
archive = EpubArchive("test_data/in1.epub")
self.assertEqual(len(archive.pages), 5)