def test_update_fragment(self):
"a fragment of article data can be updated by adding it again with different content"
# ensure we have something that resembles the ingest data
self.assertEqual(models.ArticleFragment.objects.count(), 1)
frag = logic.get(self.av, 'xml->json').fragment
self.assertTrue('title' in frag)
# now update it with some garbage
data = {'title': 'pants-party'}
logic.add(self.av, 'xml->json', data, pos=0, update=True)
# ensure we've just destroyed our very important data
frag = logic.get(self.av, 'xml->json').fragment
self.assertEqual(frag, data)
def setUp(self):
self.nom = 'revalidate'
# publish four valid articles, ensure each is in different states
fixture_list = [
'01968', # SET
'16695', # RESET
'20105', # UNSET
'20125', # NOTSET
]
self.msid1, self.msid2, self.msid3, self.msid4 = fixture_list
self.version = '1'
for msid in fixture_list:
fixture_path = join(self.fixture_dir, 'ajson',
'elife-%s-v1.xml.json' % msid)
ajson_ingestor.ingest_publish(json.load(open(fixture_path, 'r')))
# SET
# remove the valid ajson so it gets set
av = article_logic.article_version(self.msid1, self.version)
av.article_json_v1 = av.article_json_v1_snippet = None
av.save()
# RESET
# already valid, nothing to do
# UNSET
# ensure the fragment is now invalid
av = article_logic.article_version(self.msid3, self.version)
frag = logic.get(av, models.XML2JSON)
frag.fragment['title'] = ''
frag.save()
# NOTSET
# remove the valid article-json
# ensure fragment is invalid and won't be set
av = article_logic.article_version(self.msid4, self.version)
av.article_json_v1 = av.article_json_v1_snippet = None
av.save()
frag = logic.get(av, models.XML2JSON)
frag.fragment['title'] = ''
frag.save()
def _publish(msid, version, force=False):
"""attach a `datetime_published` value to an article version. if none provided, use RIGHT NOW.
you cannot publish an already published article version unless force==True"""
try:
av = models.ArticleVersion.objects.get(article__manuscript_id=msid, version=version)
if av.published():
if not force:
raise StateError("refusing to publish an already published article version")
# NOTE: we don't use any other article fragments for determining the publication date
# except the xml->json fragment.
raw_data = fragments.get(av, XML2JSON)
# the json *will always* have a published date if v1 ...
if version == 1:
# pull that published date from the stored (but unpublished) article-json
# and set the pub-date on the ArticleVersion object
datetime_published = utils.todt(raw_data.get('published'))
if not datetime_published:
raise StateError("found 'published' value in article-json, but it's either null or unparsable as a datetime")
else:
# but *not* if it's > v1. in this case, we generate one.
if av.published() and force:
# this article version is already published and a force publish request has been sent
if False and 'versionDate' in raw_data: # fail this case for now.
# FUTURE CASE: when a 'versionDate' value is present in the article-json, use that.
# as of 2016-10-21 version history IS NOT captured in the xml,
# it won't be parsed by the bot-lax-adaptor and it
# won't find it's way here. this is a future-case only.
datetime_published = utils.todt(raw_data['versionDate'])
if not datetime_published:
raise StateError("found 'versionDate' value in article-json, but it's either null or unparseable as a datetime")
else:
# CURRENT CASE
# preserve the existing pubdate set by lax. ignore anything given in the ajson.
# if the pubdate for an article is to change, it must come from the xml (see above case)
datetime_published = av.datetime_published
else:
# CURRENT CASE
# this article version hasn't been published yet. use a value of RIGHT NOW as the published date.
datetime_published = utils.utcnow()
av.datetime_published = datetime_published
av.save()
# merge the fragments we have available and make them available for serving
# allow errors when the publish operation is being forced
fragments.merge_if_valid(av, quiet=force)
# notify event bus that article change has occurred
transaction.on_commit(partial(events.notify, av.article))
return av
except ValidationError:
raise StateError("refusing to publish an article '%sv%s' with invalid article-json" % (msid, version))
except models.ArticleFragment.DoesNotExist:
raise StateError("no 'xml->json' fragment found. being strict and failing this publish. please INGEST!")
except models.ArticleVersion.DoesNotExist:
# attempted to publish an article that doesn't exist ...
raise StateError("refusing to publish an article '%sv%s' that doesn't exist" % (msid, version))
def _publish(msid, version, force=False) -> models.ArticleVersion:
"""attach a `datetime_published` value to an article version. if none provided, use RIGHT NOW.
you cannot publish an already published article version unless force==True"""
try:
av = models.ArticleVersion.objects.get(article__manuscript_id=msid,
version=version)
if av.published():
if not force:
raise StateError(
codes.ALREADY_PUBLISHED,
"refusing to publish an already published article version")
# NOTE: we don't use any other article fragments for determining the publication date
# except the xml->json fragment.
raw_data = fragments.get(av, XML2JSON).fragment
# the json *will always* have a published date if v1 ...
if version == 1:
# pull that published date from the stored (but unpublished) article-json
# and set the pub-date on the ArticleVersion object
datetime_published = utils.todt(raw_data.get('published'))
if not datetime_published:
raise StateError(
codes.PARSE_ERROR,
"found 'published' value in article-json, but it's either null or unparsable as a date+time"
)
else:
# but *not* if it's > v1. in this case, we generate one.
if av.published() and force:
# this article version is already published and a force publish request has been sent
if False and 'versionDate' in raw_data: # fail this case for now.
# FUTURE CASE: when a 'versionDate' value is present in the article-json, use that.
# as of 2016-10-21 version history IS NOT captured in the xml,
# it won't be parsed by the bot-lax-adaptor and it
# won't find it's way here. this is a future-case only.
datetime_published = utils.todt(raw_data['versionDate'])
if not datetime_published:
raise StateError(
codes.PARSE_ERROR,
"found 'versionDate' value in article-json, but it's either null or unparseable as a datetime"
)
else:
# CURRENT CASE
# preserve the existing pubdate set by lax. ignore anything given in the ajson.
# if the pubdate for an article is to change, it must come from the xml (see above case)
datetime_published = av.datetime_published
else:
# CURRENT CASE
# this article version hasn't been published yet. use a value of RIGHT NOW as the published date.
datetime_published = utils.utcnow()
av.datetime_published = datetime_published
av.save()
events.ajson_publish_events(av, force)
# merge the fragments we have available and make them available for serving.
# allow errors when the publish operation is being forced.
fragments.set_article_json(
av, quiet=False if settings.VALIDATE_FAILS_FORCE else force)
# notify event bus that article change has occurred
transaction.on_commit(partial(aws_events.notify_all, av))
return av
except ValidationError as err:
# the problem isn't that the ajson is invalid, it's that we've allowed invalid ajson into the system
raise StateError(
codes.INVALID,
"refusing to publish an article '%sv%s' with invalid article-json: %s"
% (msid, version, err), err)
except models.ArticleFragment.DoesNotExist:
raise StateError(
codes.NO_RECORD,
"no 'xml->json' fragment found. being strict and failing this publish. please INGEST!"
)
except models.ArticleVersion.DoesNotExist:
# attempted to publish an article that doesn't exist ...
raise StateError(
codes.NO_RECORD,
"refusing to publish an article '%sv%s' that doesn't exist" %
(msid, version))