def test_article_publish_v2_forced(self):
"an unpublished v2 article can be successfully published again, if forced"
# ingest and publish the v1
_, _, av = ajson_ingestor.ingest(self.ajson)
ajson_ingestor.publish(self.msid, self.version)
av = self.freshen(av)
self.assertTrue(av.published())
# modify and ingest+publish a v2
self.ajson['article']['version'] = 2
_, _, av2 = ajson_ingestor.ingest_publish(self.ajson)
av2 = self.freshen(av2)
self.assertTrue(av2.published())
# the v2 should have been published normally.
self.assertEqual(utils.ymd(datetime.now()), utils.ymd(av2.datetime_published))
# give the article-json a 'versionDate' - this won't ordinarily happen until further down the line
# but lets embed this logic while it's still fresh in everybody's heads.
# modify the versionDate of the v2 and ingest+publish again
yesterday = datetime.now() - timedelta(days=1)
self.ajson['article']['versionDate'] = yesterday
_, _, av2v2 = ajson_ingestor.ingest_publish(self.ajson, force=True)
av2v2 = self.freshen(av2v2)
self.assertEqual(utils.ymd(yesterday), utils.ymd(av2v2.datetime_published))
def test_article_publish_v2_forced2(self):
"a PUBLISHED v2 article can be successfully published (again), if forced"
av = ajson_ingestor.ingest(self.ajson)
ajson_ingestor.publish(self.msid, self.version)
av = self.freshen(av)
self.assertTrue(av.published())
# modify and ingest+publish a v2
self.ajson['article']['version'] = 2
# there is a versionDate here, but because we're not forcing it, it doesn't get looked for
# lax is the distributor of non-v1 pub dates. this may find their way into xml later, but
# they will always come from lax.
av2 = ajson_ingestor.ingest_publish(self.ajson)
av2 = self.freshen(av2)
self.assertTrue(av2.published())
self.assertEqual(utils.ymd(datetime.now()),
utils.ymd(av2.datetime_published))
# don't set a versionDate, just force a publish
# we expect the v2.datetime_publish to remain unchanged
del self.ajson['article'][
'versionDate'] # remember, this was copied from a v1 that had a versionDate!
av2v2 = ajson_ingestor.ingest_publish(self.ajson, force=True)
av2v2 = self.freshen(av2v2)
self.assertEqual(av2.datetime_published, av2v2.datetime_published)
def test_article_publish_v2_forced(self):
"an unpublished v2 article can be successfully published again, if forced"
# ingest and publish the v1
av = ajson_ingestor.ingest(self.ajson)
ajson_ingestor.publish(self.msid, self.version)
av = self.freshen(av)
self.assertTrue(av.published())
# modify and ingest+publish a v2
self.ajson['article']['version'] = 2
av2 = ajson_ingestor.ingest_publish(self.ajson)
av2 = self.freshen(av2)
self.assertTrue(av2.published())
# the v2 should have been published normally.
self.assertEqual(utils.ymd(datetime.now()),
utils.ymd(av2.datetime_published))
# give the article-json a 'versionDate' - this won't ordinarily happen until further down the line
# but lets embed this logic while it's still fresh in everybody's heads.
# modify the versionDate of the v2 and ingest+publish again
yesterday = datetime.now() - timedelta(days=1)
self.ajson['article']['versionDate'] = yesterday
av2v2 = ajson_ingestor.ingest_publish(self.ajson, force=True)
av2v2 = self.freshen(av2v2)
self.assertEqual(utils.ymd(yesterday),
utils.ymd(av2v2.datetime_published))
def test_out_of_sequence_publish_fails(self):
"attempting to ingest an article with a version greater than another *published* version fails"
# ingest and publish a v1
_, _, av = ajson_ingestor.ingest(self.ajson) # v1
ajson_ingestor.publish(self.msid, self.version)
# now attempt to ingest a v3
self.ajson['article']['version'] = 3
self.assertRaises(StateError, ajson_ingestor.ingest, self.ajson)
self.assertEqual(models.Article.objects.count(), 1)
self.assertEqual(models.ArticleVersion.objects.count(), 1)
av = self.freshen(av)
self.assertEqual(av.version, 1) # assert the version hasn't changed
def test_out_of_sequence_publish_fails(self):
"attempting to ingest an article with a version greater than another *published* version fails"
# ingest and publish a v1
av = ajson_ingestor.ingest(self.ajson) # v1
ajson_ingestor.publish(self.msid, self.version)
# now attempt to ingest a v3
self.ajson['article']['version'] = 3
self.assertRaises(StateError, ajson_ingestor.ingest, self.ajson)
self.assertEqual(models.Article.objects.count(), 1)
self.assertEqual(models.ArticleVersion.objects.count(), 1)
av = self.freshen(av)
self.assertEqual(av.version, 1) # assert the version hasn't changed
def test_article_publish_v1(self):
"an unpublished v1 article can be successfully published"
av = ajson_ingestor.ingest(self.ajson)
self.assertEqual(models.Journal.objects.count(), 1)
self.assertEqual(models.Article.objects.count(), 1)
self.assertEqual(models.ArticleVersion.objects.count(), 1)
self.assertFalse(av.published())
# publish
av = ajson_ingestor.publish(self.msid, self.version)
# aaand just make sure we still have the expected number of objects
self.assertEqual(models.Journal.objects.count(), 1)
self.assertEqual(models.Article.objects.count(), 1)
self.assertEqual(models.ArticleVersion.objects.count(), 1)
self.assertTrue(av.published())
self.assertTrue(isinstance(av.datetime_published, datetime))
# the pubdate of an unpublished v1 article is the same as that found in the
# given json.
av = self.freshen(av)
expected_pubdate = utils.ymd(
utils.todt(self.ajson['article']['published']))
self.assertEqual(expected_pubdate, utils.ymd(av.datetime_published))
def test_article_publish_fails_if_already_published(self):
"a published article CANNOT be published again"
_, _, av = ajson_ingestor.ingest(self.ajson)
av = ajson_ingestor.publish(self.msid, self.version)
av = self.freshen(av)
self.assertTrue(av.published())
# publish again
self.assertRaises(StateError, ajson_ingestor.publish, self.msid, self.version)
def test_publish_dry_run(self):
"specifying a dry run does not commit changes to database"
_, _, saved_av = ajson_ingestor.ingest(self.ajson) # do an actual ingest first
unsaved_av = ajson_ingestor.publish(self.msid, self.version, dry_run=True)
self.assertEqual(models.ArticleVersion.objects.count(), 1)
# ensure the article version stored has no published date
models.ArticleVersion.objects.get(pk=saved_av.pk, datetime_published=None)
# and that the object returned *does* have a datetime published
self.assertTrue(unsaved_av.published())
def test_article_publish_fails_if_already_published(self):
"a published article CANNOT be published again"
av = ajson_ingestor.ingest(self.ajson)
av = ajson_ingestor.publish(self.msid, self.version)
av = self.freshen(av)
self.assertTrue(av.published())
# publish again
self.assertRaises(StateError, ajson_ingestor.publish, self.msid,
self.version)
def test_article_publish_succeeds_for_published_article_if_forced(self):
"publication of an already published article can occur only if forced"
_, _, av = ajson_ingestor.ingest(self.ajson)
av = ajson_ingestor.publish(self.msid, self.version)
av = self.freshen(av)
expected_pubdate = utils.ymd(utils.todt(self.ajson['article']['published']))
self.assertEqual(expected_pubdate, utils.ymd(av.datetime_published))
# publish again, no changes to pubdate expected
av = ajson_ingestor.publish(self.msid, self.version, force=True)
av = self.freshen(av)
self.assertEqual(expected_pubdate, utils.ymd(av.datetime_published))
# ingest new pubdate, force publication
new_pubdate = utils.todt('2016-01-01')
self.ajson['article']['published'] = new_pubdate
ajson_ingestor.ingest_publish(self.ajson, force=True)
av = self.freshen(av)
self.assertEqual(utils.ymd(new_pubdate), utils.ymd(av.datetime_published))
def test_article_publish_v2(self):
"an unpublished v2 article can be successfully published"
_, _, av = ajson_ingestor.ingest(self.ajson)
self.assertEqual(models.Journal.objects.count(), 1)
self.assertEqual(models.Article.objects.count(), 1)
self.assertEqual(models.ArticleVersion.objects.count(), 1)
self.assertFalse(av.published())
ajson_ingestor.publish(self.msid, self.version)
av = self.freshen(av)
self.assertTrue(av.published())
# modify to a v2 and publish
self.ajson['article']['version'] = 2
_, _, av2 = ajson_ingestor.ingest_publish(self.ajson)
av2 = self.freshen(av2)
self.assertEqual(models.ArticleVersion.objects.count(), 2)
self.assertTrue(av2.published())
self.assertEqual(utils.ymd(datetime.now()), utils.ymd(av2.datetime_published))
def test_article_publish_v2(self):
"an unpublished v2 article can be successfully published"
av = ajson_ingestor.ingest(self.ajson)
self.assertEqual(models.Journal.objects.count(), 1)
self.assertEqual(models.Article.objects.count(), 1)
self.assertEqual(models.ArticleVersion.objects.count(), 1)
self.assertFalse(av.published())
ajson_ingestor.publish(self.msid, self.version)
av = self.freshen(av)
self.assertTrue(av.published())
# modify to a v2 and publish
self.ajson['article']['version'] = 2
av2 = ajson_ingestor.ingest_publish(self.ajson)
av2 = self.freshen(av2)
self.assertEqual(models.ArticleVersion.objects.count(), 2)
self.assertTrue(av2.published())
self.assertEqual(utils.ymd(datetime.now()),
utils.ymd(av2.datetime_published))
def test_article_publish_succeeds_for_published_article_if_forced(self):
"publication of an already published article can occur only if forced"
av = ajson_ingestor.ingest(self.ajson)
av = ajson_ingestor.publish(self.msid, self.version)
av = self.freshen(av)
expected_pubdate = utils.ymd(
utils.todt(self.ajson['article']['published']))
self.assertEqual(expected_pubdate, utils.ymd(av.datetime_published))
# publish again, no changes to pubdate expected
av = ajson_ingestor.publish(self.msid, self.version, force=True)
av = self.freshen(av)
self.assertEqual(expected_pubdate, utils.ymd(av.datetime_published))
# ingest new pubdate, force publication
new_pubdate = utils.todt('2016-01-01')
self.ajson['article']['published'] = new_pubdate
ajson_ingestor.ingest_publish(self.ajson, force=True)
av = self.freshen(av)
self.assertEqual(utils.ymd(new_pubdate),
utils.ymd(av.datetime_published))
def test_publish_dry_run(self):
"specifying a dry run does not commit changes to database"
saved_av = ajson_ingestor.ingest(
self.ajson) # do an actual ingest first
unsaved_av = ajson_ingestor.publish(self.msid,
self.version,
dry_run=True)
self.assertEqual(models.ArticleVersion.objects.count(), 1)
# ensure the article version stored has no published date
models.ArticleVersion.objects.get(pk=saved_av.pk,
datetime_published=None)
# and that the object returned *does* have a datetime published
self.assertTrue(unsaved_av.published())
def handle_single(print_queue, action, infile, msid, version, force, dry_run):
data = None
log_context = {"msid": msid, "version": version}
LOG.info("attempting to %s article %s", action, msid, extra=log_context)
# read and check the article-json given, if necessary
try:
if action in [INGEST, BOTH]:
raw_data = infile.read()
log_context["data"] = str(raw_data[:25]) + "... (truncated)" if raw_data else ""
data = json.loads(raw_data)
# vagary of the CLI interface: article id and version are required
# these may not match the data given
data_version = data["article"].get("version")
if not data_version == version:
raise StateError(
"version in the data (%s) does not match version passed to script (%s)" % (data_version, version)
)
data_msid = int(data["article"]["id"])
if not data_msid == msid:
raise StateError(
"manuscript-id in the data (%s) does not match id passed to script (%s)" % (data_msid, msid)
)
except StateError as err:
error(print_queue, INVALID, err.message, log_context)
except ValueError as err:
msg = "could not decode the json you gave me: %r for data: %r" % (err.message, raw_data)
error(print_queue, INVALID, msg, log_context)
choices = {
# all these return a models.ArticleVersion object
INGEST: lambda msid, ver, force, data, dry: ajson_ingestor.ingest(data, force, dry_run=dry)[-1],
PUBLISH: lambda msid, ver, force, data, dry: ajson_ingestor.publish(msid, ver, force, dry_run=dry),
BOTH: lambda msid, ver, force, data, dry: ajson_ingestor.ingest_publish(data, force, dry_run=dry)[-1],
}
try:
av = choices[action](msid, version, force, data, dry_run)
success(print_queue, action, av, log_context, dry_run)
except StateError as err:
error(print_queue, INVALID, "failed to call action %r: %s" % (action, err.message), log_context)
except Exception as err:
msg = "unhandled exception attempting to %r article: %s" % (action, err)
LOG.exception(msg, extra=log_context)
error(print_queue, ERROR, msg, log_context)
def test_article_publish_v2_forced2(self):
"a PUBLISHED v2 article can be successfully published (again), if forced"
_, _, av = ajson_ingestor.ingest(self.ajson)
ajson_ingestor.publish(self.msid, self.version)
av = self.freshen(av)
self.assertTrue(av.published())
# modify and ingest+publish a v2
self.ajson['article']['version'] = 2
# there is a versionDate here, but because we're not forcing it, it doesn't get looked for
# lax is the distributor of non-v1 pub dates. this may find their way into xml later, but
# they will always come from lax.
_, _, av2 = ajson_ingestor.ingest_publish(self.ajson)
av2 = self.freshen(av2)
self.assertTrue(av2.published())
self.assertEqual(utils.ymd(datetime.now()), utils.ymd(av2.datetime_published))
# don't set a versionDate, just force a publish
# we expect the v2.datetime_publish to remain unchanged
del self.ajson['article']['versionDate'] # remember, this was copied from a v1 that had a versionDate!
_, _, av2v2 = ajson_ingestor.ingest_publish(self.ajson, force=True)
av2v2 = self.freshen(av2v2)
self.assertEqual(av2.datetime_published, av2v2.datetime_published)
def test_article_publish_v1(self):
"an unpublished v1 article can be successfully published"
_, _, av = ajson_ingestor.ingest(self.ajson)
self.assertEqual(models.Journal.objects.count(), 1)
self.assertEqual(models.Article.objects.count(), 1)
self.assertEqual(models.ArticleVersion.objects.count(), 1)
self.assertFalse(av.published())
# publish
av = ajson_ingestor.publish(self.msid, self.version)
# aaand just make sure we still have the expected number of objects
self.assertEqual(models.Journal.objects.count(), 1)
self.assertEqual(models.Article.objects.count(), 1)
self.assertEqual(models.ArticleVersion.objects.count(), 1)
self.assertTrue(av.published())
self.assertTrue(isinstance(av.datetime_published, datetime))
# the pubdate of an unpublished v1 article is the same as that found in the
# given json.
av = self.freshen(av)
expected_pubdate = utils.ymd(utils.todt(self.ajson['article']['published']))
self.assertEqual(expected_pubdate, utils.ymd(av.datetime_published))
def handle_single(print_queue, action, infile, msid, version, force, dry_run):
data = None
log_context = {'msid': msid, 'version': version}
LOG.info('attempting to %s article %s v%s',
action,
msid,
version,
extra=log_context)
# read and check the article-json given, if necessary
try:
if action not in [PUBLISH]:
raw_data = infile.read()
log_context['data'] = str(
raw_data[:25]) + "... (truncated)" if raw_data else ''
try:
data = json.loads(raw_data)
except ValueError as err:
msg = "could not decode the json you gave me: %r for data: %r" % (
err.msg, raw_data)
raise StateError(codes.BAD_REQUEST, msg)
# vagary of the CLI interface: article id and version are required
# these may not match the data given
data_version = data['article'].get('version')
if not data_version == version:
raise StateError(
codes.BAD_REQUEST,
"'version' in the data (%s) does not match 'version' passed to script (%s)"
% (data_version, version))
data_msid = int(data['article']['id'])
if not data_msid == msid:
raise StateError(
codes.BAD_REQUEST,
"'id' in the data (%s) does not match 'msid' passed to script (%s)"
% (data_msid, msid))
except KeyboardInterrupt as err:
LOG.warn("ctrl-c caught during data load")
raise
except StateError as err:
error_from_err(print_queue, INVALID, err, force, dry_run, log_context)
except BaseException as err:
LOG.exception("unhandled exception attempting to ingest article-json",
extra=log_context)
error(print_queue,
ERROR,
codes.UNKNOWN,
str(err),
force,
dry_run,
log_context,
trace=ftb(err))
choices = {
# all these return a models.ArticleVersion object
INGEST:
lambda msid, ver, force, data, dry: ajson_ingestor.ingest(
data, force, dry_run=dry),
PUBLISH:
lambda msid, ver, force, data, dry: ajson_ingestor.publish(
msid, ver, force, dry_run=dry),
INGEST_PUBLISH:
lambda msid, ver, force, data, dry: ajson_ingestor.ingest_publish(
data, force, dry_run=dry),
}
try:
av = choices[action](msid, version, force, data, dry_run)
success(print_queue, action, av, force, dry_run, log_context)
except KeyboardInterrupt as err:
LOG.warn("ctrl-c caught during ingest/publish")
raise
except SystemExit:
# `success` and `error` use `exit` to indicate success or failure with their return code.
# this is handled in the `job` function, so re-raise it here and handle it there
raise
except StateError as err:
# handled error
error_from_err(print_queue, INVALID, err, force, dry_run, log_context)
except BaseException as err:
# unhandled error
msg = "unhandled exception attempting to %r article: %r" % (action,
err)
LOG.exception(msg, extra=log_context)
error(print_queue,
ERROR,
codes.UNKNOWN,
msg,
force,
dry_run,
log_context,
trace=ftb(err))