def treat_metadatas(name, metadatas):
found_origin_id = False
found_script = False
script_name = "{}-{}".format(
"antidot-html-connector",
pkg_resources.get_distribution("antidot-html-connector").version)
new_metadatas = []
for metadata in metadatas:
if metadata.key == "ft:forcedOriginId":
LOGGER.debug("Forcing the origin ID to '%s'.",
metadata.first_value)
name = metadata.first_value
found_origin_id = True
else:
if metadata.key == METADATA_SCRIPT:
found_script = True
metadata = Metadata.string(
METADATA_SCRIPT,
["{}-{}".format(metadata.first_value, script_name)])
new_metadatas.append(metadata)
if not found_script:
new_metadatas.append(Metadata.string(METADATA_SCRIPT, [script_name]))
if logging.WARNING and not found_origin_id:
LOGGER.warning(
"For <%s>, we used a default origin_id based on the file name and its metadatas."
" Sending the same file with the same metadata will replace it.",
name,
)
return name, new_metadatas
def test_excluded_metadata_name(self):
"""We can change metadata without affecting the default name created from metadata"""
ignored_metadatas = [
Metadata.string(METADATA_SCRIPT, ["this test script"]),
Metadata.string("style-map-hash", [hash("STYLE_MAP")]),
]
_, name = get_html_from_path(Path(__file__), ignored_metadatas)
_, other_name = get_html_from_path(Path(__file__), [])
self.assertEqual(name, other_name,
"The ignored metadata should not affect the name!")
def test_treat_metadatas(self):
"""We change the title when we use the foredOriginId metdata. """
forced_name = "This is the forced origin_id"
self.metadatas.append(
Metadata.string("ft:forcedOriginId", [forced_name]))
new_name, new_metadatas = treat_metadatas("name", self.metadatas)
expected_metadatas = [
Metadata.string(
METADATA_SCRIPT,
{"this test script-{}".format(self.connector_version)})
]
self.assertEqual(forced_name, new_name)
self.assertEqual(new_metadatas, expected_metadatas)
def test_no_metadata_script(self):
"""We add script metadata when there is nothing."""
new_name, new_metadatas = treat_metadatas("name", [])
expected_metadatas = [
Metadata.string(METADATA_SCRIPT, {self.connector_version})
]
self.assertEqual("name", new_name)
self.assertEqual(new_metadatas, expected_metadatas)
def test_publish_with_meta(self):
client = RemoteClient(url="url",
authentication=LoginAuthentication(
"login", "password"),
source_id="source_id")
origin_id = "Forced joke origin id"
title = "THE LITTLE FORCED TITLE IN THE PRAIRIE"
metadatas = [
Metadata.string("ft:forcedTitle", [title]),
Metadata.string("ft:forcedOriginId", [origin_id])
]
response = publish_html_with_client(HTML_PATHS[0],
client,
metadatas=metadatas)
self.assertIsNotNone(response.publications)
self.assertEqual(response.publications.id, origin_id)
self.assertEqual(response.publications.title, title)
self.assertEqual(len(response.publications.metadata), 1)
def setUp(self) -> None:
self.temp_sys = sys.argv
self.script_name = "script_name"
self.connector_version = "antidot-html-connector-{}".format(
pkg_resources.get_distribution("antidot-html-connector").version)
self.metadatas = [
Metadata.string(METADATA_SCRIPT, ["this test script"])
]
sys.argv = [self.script_name, "myhtml.html"]
def test_preexiting_metadata_script(self):
"""We add script metadata when we already have one"""
new_name, new_metadatas = treat_metadatas("name", self.metadatas)
expected_metadatas = [
Metadata.string(
METADATA_SCRIPT,
{"this test script-{}".format(self.connector_version)})
]
self.assertEqual("name", new_name)
self.assertEqual(new_metadatas, expected_metadatas)
def test_get_html_from_path(self):
"""Origin id is not gigantic even with a lot of metadata"""
metadatas = []
for i in range(1000):
metadatas.append(Metadata.string("i", [-i, "long"]))
_, name = get_html_from_path(Path(__file__), metadatas)
self.assertLess(len(name), ORIGIN_ID_MAX_SIZE)
_, other_name = get_html_from_path(Path(__file__), metadatas[:-1])
self.assertNotEqual(name, other_name,
"Different metadatas, same origin id !")
def ft_content_from_html_content(html_content, title, use_ftml,
render_cover_page, path):
if use_ftml and not FTML_AVAILABLE:
raise ModuleNotFoundError(
"Please install the FTML connector in order to use FTML.")
if use_ftml:
topic = TopicBuilder().title(Metadata.title(title)).content(
html_content).origin_id("0").build()
topics = [TopicsSplitter().split(topic)]
nodes = PublicationConverter().convert_toc(topics)
content = StructuredContent(toc=nodes,
editorial_type=EditorialType.DEFAULT)
resources = []
else:
if path:
splitter = HtmlSplitter(path=path)
else:
splitter = HtmlSplitter(content=html_content)
html2topics = HtmlToTopics(splitter,
render_cover_page=render_cover_page)
content = StructuredContent(toc=(html2topics.topics),
editorial_type=EditorialType.DEFAULT)
resources = html2topics.resources
return content, resources