def test_anonymous_transcription_submission(self):
asset = create_asset()
anon = get_anonymous_user()
transcription = Transcription(asset=asset,
user=anon,
text="previous entry")
transcription.full_clean()
transcription.save()
resp = self.client.post(
reverse("submit-transcription", args=(transcription.pk, )))
data = self.assertValidJSON(resp, expected_status=401)
self.assertIn("key", data)
self.assertIn("image", data)
self.assertFalse(
Transcription.objects.filter(submitted__isnull=False).exists())
self.completeCaptcha(data["key"])
self.client.post(
reverse("submit-transcription", args=(transcription.pk, )))
self.assertTrue(
Transcription.objects.filter(submitted__isnull=False).exists())
def test_transcription_review_asset_status_updates(self):
"""
Confirm that the Asset.transcription_status field is correctly updated
throughout the review process
"""
asset = create_asset()
anon = get_anonymous_user()
# We should see NOT_STARTED only when no transcription records exist:
self.assertEqual(asset.transcription_set.count(), 0)
self.assertEqual(
Asset.objects.get(pk=asset.pk).transcription_status,
TranscriptionStatus.NOT_STARTED,
)
t1 = Transcription(asset=asset,
user=anon,
text="test",
submitted=now())
t1.full_clean()
t1.save()
self.assertEqual(
Asset.objects.get(pk=asset.pk).transcription_status,
TranscriptionStatus.SUBMITTED,
)
# “Login” so we can review the anonymous transcription:
self.login_user()
self.assertEqual(
1,
Transcription.objects.filter(pk=t1.pk,
accepted__isnull=True).count())
resp = self.client.post(reverse("review-transcription",
args=(t1.pk, )),
data={"action": "reject"})
self.assertValidJSON(resp, expected_status=200)
# After rejecting a transcription, the asset status should be reset to
# in-progress:
self.assertEqual(
1,
Transcription.objects.filter(pk=t1.pk,
accepted__isnull=True,
rejected__isnull=False).count(),
)
self.assertEqual(
Asset.objects.get(pk=asset.pk).transcription_status,
TranscriptionStatus.IN_PROGRESS,
)
# We'll simulate a second attempt:
t2 = Transcription(asset=asset,
user=anon,
text="test",
submitted=now(),
supersedes=t1)
t2.full_clean()
t2.save()
self.assertEqual(
Asset.objects.get(pk=asset.pk).transcription_status,
TranscriptionStatus.SUBMITTED,
)
resp = self.client.post(reverse("review-transcription",
args=(t2.pk, )),
data={"action": "accept"})
self.assertValidJSON(resp, expected_status=200)
self.assertEqual(
1,
Transcription.objects.filter(pk=t2.pk,
accepted__isnull=False).count())
self.assertEqual(
Asset.objects.get(pk=asset.pk).transcription_status,
TranscriptionStatus.COMPLETED,
)
def site_report():
report = {
"assets_not_started": 0,
"assets_in_progress": 0,
"assets_submitted": 0,
"assets_completed": 0,
}
asset_count_qs = Asset.objects.values_list("transcription_status").annotate(
Count("transcription_status")
)
for status, count in asset_count_qs:
logger.debug("Assets %s: %d", status, count)
report[f"assets_{status}"] = count
assets_total = Asset.objects.count()
assets_published = Asset.objects.published().count()
assets_unpublished = Asset.objects.unpublished().count()
items_published = Item.objects.published().count()
items_unpublished = Item.objects.unpublished().count()
projects_published = Project.objects.published().count()
projects_unpublished = Project.objects.unpublished().count()
campaigns_published = Campaign.objects.published().count()
campaigns_unpublished = Campaign.objects.unpublished().count()
users_registered = User.objects.all().count()
users_activated = User.objects.filter(is_active=True).count()
anonymous_transcriptions = Transcription.objects.filter(
user=get_anonymous_user()
).count()
transcriptions_saved = Transcription.objects.all().count()
stats = UserAssetTagCollection.objects.aggregate(Count("tags"))
tag_count = stats["tags__count"]
distinct_tag_count = Tag.objects.all().count()
site_report = SiteReport()
site_report.assets_total = assets_total
site_report.assets_published = assets_published
site_report.assets_not_started = report["assets_not_started"]
site_report.assets_in_progress = report["assets_in_progress"]
site_report.assets_waiting_review = report["assets_submitted"]
site_report.assets_completed = report["assets_completed"]
site_report.assets_unpublished = assets_unpublished
site_report.items_published = items_published
site_report.items_unpublished = items_unpublished
site_report.projects_published = projects_published
site_report.projects_unpublished = projects_unpublished
site_report.anonymous_transcriptions = anonymous_transcriptions
site_report.transcriptions_saved = transcriptions_saved
site_report.distinct_tags = distinct_tag_count
site_report.tag_uses = tag_count
site_report.campaigns_published = campaigns_published
site_report.campaigns_unpublished = campaigns_unpublished
site_report.users_registered = users_registered
site_report.users_activated = users_activated
site_report.save()
for campaign in Campaign.objects.all():
campaign_report(campaign)
for topic in Topic.objects.all():
topic_report(topic)
def campaign_report(campaign):
report = {
"assets_not_started": 0,
"assets_in_progress": 0,
"assets_submitted": 0,
"assets_completed": 0,
}
asset_count_qs = (
Asset.objects.filter(item__project__campaign=campaign)
.values_list("transcription_status")
.annotate(Count("transcription_status"))
)
for status, count in asset_count_qs:
logger.debug("Campaign %s assets %s: %d", campaign.slug, status, count)
report[f"assets_{status}"] = count
assets_total = Asset.objects.filter(item__project__campaign=campaign).count()
assets_published = (
Asset.objects.published().filter(item__project__campaign=campaign).count()
)
assets_unpublished = (
Asset.objects.unpublished().filter(item__project__campaign=campaign).count()
)
items_published = (
Item.objects.published().filter(project__campaign=campaign).count()
)
items_unpublished = (
Item.objects.published().filter(project__campaign=campaign).count()
)
projects_published = Project.objects.published().filter(campaign=campaign).count()
projects_unpublished = (
Project.objects.unpublished().filter(campaign=campaign).count()
)
anonymous_transcriptions = Transcription.objects.filter(
asset__item__project__campaign=campaign, user=get_anonymous_user()
).count()
transcriptions_saved = Transcription.objects.filter(
asset__item__project__campaign=campaign
).count()
asset_tag_collections = UserAssetTagCollection.objects.filter(
asset__item__project__campaign=campaign
)
stats = asset_tag_collections.order_by().aggregate(tag_count=Count("tags"))
tag_count = stats["tag_count"]
distinct_tag_list = set()
for tag_collection in asset_tag_collections:
distinct_tag_list.update(tag_collection.tags.values_list("pk", flat=True))
distinct_tag_count = len(distinct_tag_list)
site_report = SiteReport()
site_report.campaign = campaign
site_report.assets_total = assets_total
site_report.assets_published = assets_published
site_report.assets_not_started = report["assets_not_started"]
site_report.assets_in_progress = report["assets_in_progress"]
site_report.assets_waiting_review = report["assets_submitted"]
site_report.assets_completed = report["assets_completed"]
site_report.assets_unpublished = assets_unpublished
site_report.items_published = items_published
site_report.items_unpublished = items_unpublished
site_report.projects_published = projects_published
site_report.projects_unpublished = projects_unpublished
site_report.anonymous_transcriptions = anonymous_transcriptions
site_report.transcriptions_saved = transcriptions_saved
site_report.distinct_tags = distinct_tag_count
site_report.tag_uses = tag_count
site_report.save()