def post_process_content(self):
self._post_processed_content = self._converted_content
self._post_processed_content = helper_functions.replace_markdown_pseudo_html_href_tag_with_markdown_links(
self._post_processed_content)
self.post_process_obsidian_image_links_if_required()
self.add_meta_data_if_required()
self._current_note_attachment_links = get_attachment_paths(
self._conversion_settings.source_absolute_root,
self._conversion_settings.export_format, self._file,
self._files_to_convert, self._post_processed_content)
self._post_processed_content = update_content_with_new_paths(
self._post_processed_content, self._file,
self._current_note_attachment_links.non_copyable_relative,
self._conversion_settings.make_absolute,
self._conversion_settings.export_folder_absolute)
def post_process_content(self):
self.logger.debug(f'Post process HTML content')
self._post_processed_content = self._converted_content
self._post_processed_content = update_href_link_suffix_in_content(
self._post_processed_content, self._output_extension,
self._files_to_convert)
self.add_meta_data_if_required()
self.update_checklists()
self._current_note_attachment_links = get_attachment_paths(
self._conversion_settings.source_absolute_root,
self._conversion_settings.export_format, self._file,
self._files_to_convert, self._post_processed_content)
self._post_processed_content = update_content_with_new_paths(
self._post_processed_content, self._file,
self._current_note_attachment_links.non_copyable_relative,
self._conversion_settings.make_absolute,
self._conversion_settings.export_folder_absolute)
def post_process_content(self):
self._post_processed_content = self._converted_content
self._post_processed_content = helper_functions.replace_markdown_pseudo_html_href_tag_with_markdown_links(
self._post_processed_content)
self.post_process_obsidian_image_links_if_required()
self.update_checklists()
self.add_meta_data_if_required()
if self._iframes_dict:
self.logger.debug(f'Add iframes to Markdown content')
self._post_processed_content = post_process_iframes_to_markdown(
self._post_processed_content, self._iframes_dict)
self.add_one_last_line_break()
self._current_note_attachment_links = get_attachment_paths(
self._conversion_settings.source_absolute_root,
self._conversion_settings.export_format, self._file,
self._files_to_convert, self._post_processed_content)
self._post_processed_content = update_content_with_new_paths(
self._post_processed_content, self._file,
self._current_note_attachment_links.non_copyable_relative,
self._conversion_settings.make_absolute,
self._conversion_settings.export_folder_absolute)
def test_update_content_with_new_paths_for_non_movable_attachments(tmp_path):
Path(tmp_path,
'some_folder/data/my_notebook/attachments').mkdir(parents=True)
Path(tmp_path, 'some_folder/attachments').mkdir(parents=True)
Path(tmp_path, 'some_folder/data/attachments').mkdir(parents=True)
Path(tmp_path,
'some_folder/data/my_other_notebook/attachments').mkdir(parents=True)
Path(tmp_path, 'some_folder/data/my_notebook/note.md').touch()
Path(tmp_path, 'some_folder/data/my_notebook/attachments/one.png').touch()
Path(tmp_path, 'some_folder/data/attachments/two.csv').touch()
Path(tmp_path, 'some_folder/three.png').touch()
Path(tmp_path, 'some_folder/attachments/four.csv').touch()
Path(tmp_path, 'some_folder/four.csv').touch()
Path(tmp_path,
'some_folder/data/my_other_notebook/attachments/five.pdf').touch()
Path(tmp_path, 'some_folder/data/my_notebook/six.csv').touch()
Path(tmp_path,
'some_folder/data/my_notebook/attachments/eight.pdf').touch()
Path(tmp_path, 'some_folder/data/my_notebook/nine.md').touch()
Path(tmp_path, 'some_folder/data/my_notebook/attachments/ten.png').touch()
Path(tmp_path,
'some_folder/data/my_notebook/attachments/eleven.pdf').touch()
Path(tmp_path,
'some_folder/data/my_notebook/attachments/file twelve.pdf').touch()
Path(tmp_path,
'some_folder/data/my_notebook/attachments/file fourteen.png').touch()
file_path = Path(tmp_path, 'some_folder/data/my_notebook/note.md')
content = f'}/some_folder/data/my_notebook/attachments/one.png)\n' \
f'}/some_folder/two.png)\n' \
f'}/some_folder/three.png)\n' \
f'\n' \
f'\n' \
f'\n' \
f'\n' \
f'\n' \
f'\n' \
f'\n' \
f'\n' \
f'\n' \
f'[a web link](https://www.google.com)\n' \
f'<img src="attachments/ten.png" />\n' \
f'<a href="attachments/eleven.pdf">example-attachment.pdf</a>\n' \
f'\n' \
f'<a href="attachments/file%20thirteen.pdf">example-attachment.pdf</a>\n' \
f'<img src="attachments/file%20fourteen.png" />\n' \
f'<a href="https://www.google.com">google</a>)\n' \
f'![]()\n' \
f'<img src="" />\n' \
f'<a href="">empty href link</a>'
expected_content = f'}/some_folder/data/my_notebook/attachments/one.png)\n' \
f'}/some_folder/two.png)\n' \
f'}/some_folder/three.png)\n' \
f'}/some_folder/three.png)\n' \
f'\n' \
f'\n' \
f'\n' \
f')}/some_folder/attachments/four.csv)\n' \
f'\n' \
f'\n' \
f'\n' \
f'\n' \
f'[a web link](https://www.google.com)\n' \
f'<img src="attachments/ten.png"/>\n' \
f'<a href="attachments/eleven.pdf">example-attachment.pdf</a>\n' \
f'\n' \
f'<a href="attachments/file%20thirteen.pdf">example-attachment.pdf</a>\n' \
f'<img src="attachments/file%20fourteen.png"/>\n' \
f'<a href="https://www.google.com">google</a>)\n' \
f'![]()\n' \
f'<img src=""/>\n' \
f'<a href="">empty href link</a>'
conversion_settings = ConversionSettings()
conversion_settings.source = Path(tmp_path, 'some_folder/data')
conversion_settings.export_folder = Path(tmp_path, 'some_folder/export')
conversion_settings.export_format = 'obsidian'
conversion_settings.conversion_input = 'markdown'
conversion_settings.make_absolute = True
file_converter = HTMLToMDConverter(conversion_settings, 'files_to_convert')
file_converter._file = file_path
file_converter._files_to_convert = {
Path(tmp_path, 'some_folder/data/my_notebook/nine.md')
}
attachment_links = get_attachment_paths(
file_converter._conversion_settings.source_absolute_root,
file_converter._conversion_settings.export_format,
file_converter._file, file_converter._files_to_convert, content)
result_content = update_content_with_new_paths(
content,
file_converter._file,
attachment_links.non_copyable_relative,
file_converter._conversion_settings.make_absolute,
file_converter._conversion_settings.export_folder_absolute,
)
assert result_content == expected_content
def test_generate_set_of_attachment_paths_html_export_format(tmp_path):
Path(tmp_path,
'some_folder/data/my_notebook/attachments').mkdir(parents=True)
Path(tmp_path, 'some_folder/attachments').mkdir(parents=True)
Path(tmp_path, 'some_folder/data/attachments').mkdir(parents=True)
Path(tmp_path,
'some_folder/data/my_other_notebook/attachments').mkdir(parents=True)
Path(tmp_path,
'some_folder/data/my_other_notebook/attachments/five.pdf').touch()
Path(tmp_path, 'some_folder/data/my_notebook/nine.md').touch()
Path(tmp_path, 'some_folder/data/my_notebook/attachments/ten.png').touch()
Path(tmp_path,
'some_folder/data/my_notebook/attachments/eleven.pdf').touch()
Path(tmp_path,
'some_folder/data/my_notebook/attachments/file twelve.pdf').touch()
Path(tmp_path,
'some_folder/data/my_notebook/attachments/file fourteen.png').touch()
file_path = Path(tmp_path, 'some_folder/data/my_notebook/note.md')
content = f'\n' \
f'\n' \
f'[a web link](https://www.google.com "google")\n' \
f'<img src="attachments/ten.png" />\n' \
f'<a href="attachments/eleven.pdf">example-attachment.pdf</a>\n' \
f'\n' \
f'<a href="attachments/file%20thirteen.pdf">example-attachment.pdf</a>\n' \
f'<img src="attachments/file%20fourteen.png" />'
# expected_content = f'\n' \
# f'\n' \
# f'[a web link](https:\\www.google.com "google")\n' \
# f'<img src="attachments/ten.png" />\n' \
# f'<a href="attachments/eleven.pdf">example-attachment.pdf</a>\n' \
# f'\n' \
# f'<a href="attachments/file%20thirteen.pdf">example-attachment.pdf</a>\n' \
# f'<img src="attachments/file%20fourteen.png" />'
conversion_settings = ConversionSettings()
conversion_settings.source = Path(tmp_path, 'some_folder/data')
conversion_settings.export_folder = Path(tmp_path, 'some_folder/export')
conversion_settings.export_format = 'html'
file_converter = MDToHTMLConverter(conversion_settings, 'files_to_convert')
file_converter._file = file_path
file_converter._files_to_convert = {
Path(tmp_path, 'some_folder/data/my_notebook/nine.md')
}
attachment_links = get_attachment_paths(
file_converter._conversion_settings.source_absolute_root,
file_converter._conversion_settings.conversion_input,
file_converter._file, file_converter._files_to_convert, content)
assert Path(tmp_path, 'some_folder/data/my_notebook/attachments/ten.png') \
in attachment_links.copyable_absolute
assert Path(tmp_path, 'some_folder/data/my_notebook/attachments/eleven.pdf') \
in attachment_links.copyable_absolute
assert Path(tmp_path, 'some_folder/data/my_notebook/attachments/file fourteen.png') \
in attachment_links.copyable_absolute
assert Path(tmp_path, 'some_folder/data/my_other_notebook/attachments/five.pdf') \
in attachment_links.copyable_absolute
assert Path(tmp_path, 'some_folder/data/my_notebook/attachments/file twelve.pdf') \
in attachment_links.copyable_absolute
assert len(attachment_links.copyable_absolute) == 5
assert f'attachments/file thirteen.pdf' in attachment_links.non_existing
assert len(attachment_links.non_existing) == 1
def test_get_attachment_paths_where_make_absolute_for_non_copyable_files(tmp_path):
Path(tmp_path, 'some_folder/data/my_notebook/attachments').mkdir(parents=True)
Path(tmp_path, 'some_folder/attachments').mkdir(parents=True)
Path(tmp_path, 'some_folder/data/attachments').mkdir(parents=True)
Path(tmp_path, 'some_folder/data/my_other_notebook/attachments').mkdir(parents=True)
Path(tmp_path, 'some_folder/data/my_notebook/note.md').touch()
Path(tmp_path, 'some_folder/data/my_notebook/attachments/one.png').touch()
Path(tmp_path, 'some_folder/data/attachments/two.csv').touch()
Path(tmp_path, 'some_folder/three.png').touch()
Path(tmp_path, 'some_folder/attachments/four.csv').touch()
Path(tmp_path, 'some_folder/four.csv').touch()
Path(tmp_path, 'some_folder/data/my_other_notebook/attachments/five.pdf').touch()
Path(tmp_path, 'some_folder/data/my_notebook/six.csv').touch()
Path(tmp_path, 'some_folder/data/my_notebook/attachments/eight.pdf').touch()
Path(tmp_path, 'some_folder/data/my_notebook/nine.md').touch()
Path(tmp_path, 'some_folder/data/my_notebook/attachments/ten.png').touch()
Path(tmp_path, 'some_folder/data/my_notebook/attachments/eleven.pdf').touch()
Path(tmp_path, 'some_folder/data/my_notebook/attachments/file twelve.pdf').touch()
Path(tmp_path, 'some_folder/data/my_notebook/attachments/file fourteen.png').touch()
file_path = Path(tmp_path, 'some_folder/data/my_notebook/note.md')
content = f'}/some_folder/data/my_notebook/attachments/one.png)\n' \
f'}/some_folder/two.png)\n' \
f'}/some_folder/three.png)\n' \
f'\n' \
f'\n' \
f'\n' \
f'\n' \
f'\n' \
f'\n' \
f'\n' \
f'\n' \
f'\n' \
f'[a web link](https://www.google.com "google")\n' \
f'<img src="attachments/ten.png" />\n' \
f'<a href="attachments/eleven.pdf">example-attachment.pdf</a>\n' \
f'\n' \
f'<a href="attachments/file%20thirteen.pdf">example-attachment.pdf</a>\n' \
f'<img src="attachments/file%20fourteen.png" />\n' \
f'<a href="https://www.google.com "google">)'
# expected_content = f'}/some_folder/data/my_notebook/attachments/one.png)\n' \
# f'}/some_folder/two.png)\n' \
# f'}/some_folder/three.png)\n' \
# f'}/some_folder/three.png)\n' \
# f'\n' \
# f'\n' \
# f'\n' \
# f'}/some_folder/attachments/four.csv)\n' \
# f'\n' \
# f'\n' \
# f'\n' \
# f'\n' \
# f'[a web link](https://www.google.com "google")\n' \
# f'<img src="attachments/ten.png" />\n' \
# f'<a href="attachments/eleven.pdf">example-attachment.pdf</a>\n' \
# f'\n' \
# f'<a href="attachments/file%20thirteen.pdf">example-attachment.pdf</a>\n' \
# f'<img src="attachments/file%20fourteen.png" />\n' \
# f'<a href="https://www.google.com "google">)'
conversion_settings = ConversionSettings()
conversion_settings.source = Path(tmp_path, 'some_folder/data')
conversion_settings.export_folder = Path(tmp_path, 'some_folder/export')
conversion_settings.export_format = 'obsidian'
# conversion_settings.conversion_input = 'markdown'
file_converter = HTMLToMDConverter(conversion_settings, 'files_to_convert')
file_converter._file = file_path
file_converter._files_to_convert = {Path(tmp_path, 'some_folder/data/my_notebook/nine.md')}
attachment_links = get_attachment_paths(file_converter._conversion_settings.source_absolute_root,
file_converter._conversion_settings.conversion_input,
file_converter._file,
file_converter._files_to_convert, content)
assert len(attachment_links.all) == 16
assert len(attachment_links.copyable) == 9
assert len(attachment_links.copyable_absolute) == 9
assert len(attachment_links.existing) == 12
assert len(attachment_links.non_copyable_absolute) == 1
assert len(attachment_links.non_copyable_relative) == 2
assert len(attachment_links.non_existing) == 4
assert Path(tmp_path, 'some_folder/data/my_other_notebook/attachments/five.pdf') \
in attachment_links.copyable_absolute
assert Path(tmp_path, 'some_folder/data/my_notebook/attachments/one.png') \
in attachment_links.copyable_absolute
assert Path(tmp_path, 'some_folder/data/my_notebook/six.csv') \
in attachment_links.copyable_absolute
assert Path(tmp_path, 'some_folder/data/attachments/two.csv') \
in attachment_links.copyable_absolute
assert Path(tmp_path, 'some_folder/data/my_notebook/attachments/eight.pdf') \
in attachment_links.copyable_absolute
assert Path(tmp_path, 'some_folder/data/my_notebook/attachments/ten.png') \
in attachment_links.copyable_absolute
assert Path(tmp_path, 'some_folder/data/my_notebook/attachments/eleven.pdf') \
in attachment_links.copyable_absolute
assert Path(tmp_path, 'some_folder/data/my_notebook/attachments/file twelve.pdf') \
in attachment_links.copyable_absolute
assert Path(tmp_path, 'some_folder/data/my_notebook/attachments/file fourteen.png') \
in attachment_links.copyable_absolute
assert len(attachment_links.copyable_absolute) == 9
assert f'{tmp_path}/some_folder/two.png' in attachment_links.non_existing
assert f'attachments/three.pdf' in attachment_links.non_existing
assert f'../my_notebook/seven.csv' in attachment_links.non_existing
assert f'attachments/file thirteen.pdf' in attachment_links.non_existing
assert len(attachment_links.non_existing) == 4
assert '../../attachments/four.csv' in attachment_links.non_copyable_relative
assert '../../three.png' in attachment_links.non_copyable_relative
assert len(attachment_links.non_copyable_relative) == 2
# three.png is used twice, once with relative path and once with absolute path as we are
# searching post processed content the content will already have had make absolute applied if it was turned on
# so we still get one path to three in the relative list
assert f'{tmp_path}/some_folder/three.png' in attachment_links.non_copyable_absolute
assert len(attachment_links.non_copyable_absolute) == 1
def test_get_attachment_paths_markdown_export_format(tmp_path):
Path(tmp_path, 'some_folder/data/my_notebook/attachments').mkdir(parents=True)
Path(tmp_path, 'some_folder/attachments').mkdir(parents=True)
Path(tmp_path, 'some_folder/data/attachments').mkdir(parents=True)
Path(tmp_path, 'some_folder/data/my_other_notebook/attachments').mkdir(parents=True)
Path(tmp_path, 'some_folder/data/my_notebook/note.md').touch()
Path(tmp_path, 'some_folder/data/my_notebook/attachments/one.png').touch()
Path(tmp_path, 'some_folder/data/attachments/two.csv').touch()
Path(tmp_path, 'some_folder/three.png').touch()
Path(tmp_path, 'some_folder/attachments/four.csv').touch()
Path(tmp_path, 'some_folder/four.csv').touch()
Path(tmp_path, 'some_folder/data/my_other_notebook/attachments/five.pdf').touch()
Path(tmp_path, 'some_folder/data/my_notebook/six.csv').touch()
Path(tmp_path, 'some_folder/data/my_notebook/attachments/eight.pdf').touch()
Path(tmp_path, 'some_folder/data/my_notebook/nine.md').touch()
Path(tmp_path, 'some_folder/data/my_notebook/attachments/ten.png').touch()
Path(tmp_path, 'some_folder/data/my_notebook/attachments/eleven.pdf').touch()
Path(tmp_path, 'some_folder/data/my_notebook/attachments/file twelve.pdf').touch()
Path(tmp_path, 'some_folder/data/my_notebook/attachments/file fourteen.png').touch()
file_path = Path(tmp_path, 'some_folder/data/my_notebook/note.md')
content = f'}/some_folder/data/my_notebook/attachments/one.png)\n' \
f'}/some_folder/two.png)\n' \
f'}/some_folder/three.png)\n' \
f'\n' \
f'\n' \
f'\n' \
f'\n' \
f'\n' \
f'\n' \
f'\n' \
f'\n' \
f'[a web link](https://www.google.com "google")\n' \
f'<img src="attachments/ten.png" />\n' \
f'<a href="attachments/eleven.pdf">example-attachment.pdf</a>\n' \
f'\n' \
f'<a href="attachments/file%20thirteen.pdf">example-attachment.pdf</a>\n' \
f'<img src="attachments/file%20fourteen.png" />'
# expected_content = f'}/some_folder/data/my_notebook/attachments/one.png)\n' \
# f'}/some_folder/two.png)\n' \
# f'}/some_folder/three.png)\n' \
# f'\n' \
# f'\n' \
# f'\n' \
# f'\n' \
# f'\n' \
# f'\n' \
# f'\n' \
# f'\n' \
# f'[a web link](https://www.google.com "google")\n' \
# f'<img src="attachments/ten.png" />\n' \
# f'<a href="attachments/eleven.pdf">example-attachment.pdf</a>\n' \
# f'\n' \
# f'<a href="attachments/file%20thirteen.pdf">example-attachment.pdf</a>\n' \
# f'<img src="attachments/file%20fourteen.png" />'
conversion_settings = ConversionSettings()
conversion_settings.source = Path(tmp_path, 'some_folder/data')
conversion_settings.export_folder = Path(tmp_path, 'some_folder/export')
conversion_settings.export_format = 'obsidian'
conversion_settings.make_absolute = False
# conversion_settings.conversion_input = 'markdown'
file_converter = HTMLToMDConverter(conversion_settings, 'files_to_convert')
file_converter._file = file_path
file_converter._files_to_convert = {Path(tmp_path, 'some_folder/data/my_notebook/nine.md')}
attachment_links = get_attachment_paths(file_converter._conversion_settings.source_absolute_root,
file_converter._conversion_settings.conversion_input,
file_converter._file,
file_converter._files_to_convert, content)
assert len(attachment_links.all) == 15
assert len(attachment_links.copyable) == 9
assert len(attachment_links.copyable_absolute) == 9
assert len(attachment_links.existing) == 11
assert len(attachment_links.non_copyable_relative) == 1
assert len(attachment_links.non_copyable_absolute) == 1
assert len(attachment_links.non_existing) == 4
assert Path(tmp_path,
'some_folder/data/my_other_notebook/attachments/five.pdf') in attachment_links.copyable_absolute
assert Path(tmp_path,
'some_folder/data/my_notebook/attachments/one.png') in attachment_links.copyable_absolute
assert Path(tmp_path,
'some_folder/data/my_notebook/six.csv') in attachment_links.copyable_absolute
assert Path(tmp_path,
'some_folder/data/attachments/two.csv') in attachment_links.copyable_absolute
assert Path(tmp_path,
'some_folder/data/my_notebook/attachments/eight.pdf') in attachment_links.copyable_absolute
assert Path(tmp_path,
'some_folder/data/my_notebook/attachments/ten.png') in attachment_links.copyable_absolute
assert Path(tmp_path,
'some_folder/data/my_notebook/attachments/eleven.pdf') in attachment_links.copyable_absolute
assert Path(tmp_path,
'some_folder/data/my_notebook/attachments/file twelve.pdf') in attachment_links.copyable_absolute
assert Path(tmp_path,
'some_folder/data/my_notebook/attachments/file fourteen.png') in attachment_links.copyable_absolute
assert len(attachment_links.copyable_absolute) == 9
assert f'{tmp_path}/some_folder/two.png' in attachment_links.non_existing
assert f'attachments/three.pdf' in attachment_links.non_existing
assert f'../my_notebook/seven.csv' in attachment_links.non_existing
assert f'attachments/file thirteen.pdf' in attachment_links.non_existing
assert len(attachment_links.non_existing) == 4
# NOTE for the "some_folder/attachments/four.csv" attachment the content should be updated to a new relative link
# assert Path(tmp_path, 'some_folder/attachments/four.csv') in file_converter._non_copyable_attachment_path_set
assert '../../attachments/four.csv' in attachment_links.non_copyable_relative
assert len(attachment_links.non_copyable_relative) == 1
assert f'{tmp_path}/some_folder/three.png' in attachment_links.non_copyable_absolute
assert len(attachment_links.non_copyable_absolute) == 1