def _cleanup_accept_suggestion_subs(apps):
# state/target changes that happened from suggestion had type=3,
# and state change subs did not have their suggestion associated
# They also credited the reviewer not the submitter
# we need to:
# - associate any state change subs with relevant suggestions
# - set the suggester as submitter for all type=3 submissions
# - update type to WEB for all type=3 subs that have suggestions
# any remaining submissions of this type are likely dupes and will be deleted
Submission = apps.get_model("pootle_statistics.Submission")
subs = Submission.objects.all()
suggestions = apps.get_model("pootle_store.Suggestion").objects
# assoc state change subs with suggestions and update the sub
to_update = []
accepted_suggestions = suggestions.filter(state__name="accepted").values(
"review_time", "pk", "unit_id", "reviewer_id", "user_id")
accept_sugg_state_subs = subs.filter(type=SUGG_ACCEPT_TYPE).filter(
field=STATE_FIELD)
accept_subs_for_update = accept_sugg_state_subs.select_related(
"suggestion").only("id", "type", "submitter_id", "suggestion_id")
for suggestion in accepted_suggestions.iterator():
suggestion["submitter_id"] = suggestion["reviewer_id"]
suggestion["creation_time"] = suggestion["review_time"]
suggestion_id = suggestion["pk"]
suggestion_user = suggestion["user_id"]
del suggestion["review_time"]
del suggestion["reviewer_id"]
del suggestion["pk"]
del suggestion["user_id"]
matching_sub = accept_subs_for_update.filter(**suggestion).first()
if matching_sub:
matching_sub.type = NEW_WEB_TYPE
matching_sub.suggestion_id = suggestion_id
matching_sub.submitter_id = suggestion_user
to_update.append(matching_sub)
if to_update:
Batch(Submission.objects, batch_size=5000).update(
to_update,
update_fields=["type", "suggestion_id", "submitter_id"],
reduces=False)
# all remaining valid sugg_accept subs should be field=target and have a suggestion
to_update = []
accept_sugg_subs = subs.filter(type=SUGG_ACCEPT_TYPE).filter(
field=TARGET_FIELD).exclude(suggestion__isnull=True)
accept_subs_for_update = accept_sugg_subs.select_related(
"suggestion").only("id", "type", "submitter_id", "suggestion__user_id")
for sub in accept_subs_for_update.iterator():
sub.submitter_id = sub.suggestion.user_id
sub.type = NEW_WEB_TYPE
to_update.append(sub)
if to_update:
Batch(Submission.objects,
batch_size=5000).update(to_update,
update_fields=["type", "submitter_id"],
reduces=False)
def test_batch_lists(store0, member):
batch = Batch(Suggestion.objects, batch_size=2)
assert batch.target == Suggestion.objects
assert batch.batch_size == 2
last_sugg_pk = Suggestion.objects.order_by("-pk").values_list(
"pk", flat=True).first()
def _create_method(unit, source, mtime):
return dict(unit_id=unit,
creation_time=mtime,
target_f=source,
user_id=member.id)
batch.create(list(store0.units.values_list("id", "source_f", "mtime")),
_create_method,
reduces=False)
new_suggs = Suggestion.objects.filter(pk__gt=last_sugg_pk)
assert new_suggs.count() == store0.units.count()
updated_suggs = []
for suggestion in new_suggs:
suggestion.target_f = "suggestion %s" % suggestion.id
updated_suggs.append(suggestion)
batch.update(updated_suggs, update_fields=["target_f"], reduces=False)
for suggestion in Suggestion.objects.filter(pk__gt=last_sugg_pk):
assert suggestion.target_f == "suggestion %s" % suggestion.id
def test_batch_create_no_reduce(store0, member):
batch = Batch(Suggestion.objects, batch_size=2)
assert batch.target == Suggestion.objects
assert batch.batch_size == 2
last_sugg_pk = Suggestion.objects.order_by(
"-pk").values_list("pk", flat=True).first()
def _create_method(unit, source, mtime):
return dict(
unit_id=unit,
creation_time=mtime,
target_f=source,
user_id=member.id)
batches = batch.batched_create(
store0.units.values_list("id", "source_f", "mtime"),
_create_method,
reduces=False)
new_suggs = Suggestion.objects.filter(pk__gt=last_sugg_pk)
assert new_suggs.count() == 0
for batched in batches:
assert len(batched)
assert len(batched) <= 2
assert all(isinstance(b, Suggestion) for b in batched)
assert all(b.target_f == b.unit.source_f for b in batched)
assert new_suggs.count() == store0.units.count()
new_suggs.delete()
created = batch.create(
store0.units.values_list("id", "source_f", "mtime"),
_create_method,
reduces=False)
new_suggs = Suggestion.objects.filter(pk__gt=last_sugg_pk)
assert created == new_suggs.count() == store0.units.count()
assert (
list(new_suggs.values_list("unit"))
== list(store0.units.values_list("id")))
def set_unit_source_data(apps, schema_editor):
from pootle_store.models import Unit
UnitSource = apps.get_model("pootle_store.UnitSource")
counter = wordcount.get(Unit)
unit_sources = (UnitSource.objects.select_related('unit').filter(
source_wordcount=0).only("unit__source_f", "id").order_by("id"))
Batch(unit_sources, batch_size=BATCH_SIZE).update(
unit_sources,
update_method=functools.partial(unit_source_update, counter),
update_fields=['source_hash', 'source_length', 'source_wordcount'])
def update_sources_with_orm(apps, creators):
UnitSource = apps.get_model("pootle_store.UnitSource")
def _set_created_by(unit_source):
unit_source.created_by_id = creators.get(unit_source.unit_id)
return unit_source
Batch(UnitSource.objects, batch_size=500).update(
list(UnitSource.objects.filter(unit_id__in=creators.keys())),
update_method=_set_created_by,
update_fields=["created_by"],
reduces=False)
def _add_submitted(qs, unit_changes):
values = qs.values_list("id", "submitted_on", "submitted_by")
def _create_method(unit, timestamp, user):
return dict(
unit_id=unit,
changed_with=(SubmissionTypes.SYSTEM if user
== get_system_user_id() else SubmissionTypes.WEB),
submitted_by_id=user,
submitted_on=timestamp)
Batch(unit_changes).create(values, _create_method)
def create_changes_with_orm(apps):
UnitChange = apps.get_model("pootle_store.UnitChange")
missing = list(_missing_changes(apps).iterator())
def _unit_change_create(unit):
return dict(unit_id=unit.pk,
changed_with=SubmissionTypes.WEB,
commented_by_id=unit.commented_by_id,
commented_on=unit.commented_on,
reviewed_by_id=unit.reviewed_by_id,
reviewed_on=unit.reviewed_on,
submitted_by_id=unit.submitted_by_id,
submitted_on=unit.submitted_on)
Batch(UnitChange.objects, batch_size=500).create(missing,
_unit_change_create,
reduces=False)
def create_sources_with_orm(apps, creators):
sysuser = get_system_user_id()
UnitSource = apps.get_model("pootle_store.UnitSource")
units = apps.get_model("pootle_store.Unit").objects.all()
_units = list(
units.values_list("id", "source_hash", "source_length",
"source_wordcount").iterator())
def _unit_source_create(pk, source_hash, source_length, source_wordcount):
return dict(unit_id=pk,
source_hash=source_hash,
source_wordcount=source_wordcount,
source_length=source_length,
created_with=SubmissionTypes.WEB,
created_by_id=creators.get(pk, sysuser))
Batch(UnitSource.objects, batch_size=500).create(_units,
_unit_source_create,
reduces=False)
def clean_abs_file_paths(apps, schema_editor):
"""Replace wrong absolute store file paths by proper relative paths
built based on store.pootle_path values.
"""
store_model = apps.get_model("pootle_store.Store")
stores = store_model.objects.filter(
translation_project__project__treestyle="nongnu")
stores = stores.filter(file__startswith="/").only("file", "pootle_path")
to_update = []
for store in stores.iterator():
lang, prj, d, fn = split_pootle_path(store.pootle_path)
store.file = os.path.join(prj, lang, d, fn)
to_update.append(store)
if to_update:
result = Batch(store_model,
batch_size=500).update(to_update,
update_fields=["file"],
reduces=False)
logger.debug("Cleaned %s store paths" % result)
def test_batch_update_reduce(store0, member):
"""The source queryset reduces as batches are created"""
batch = Batch(Unit, batch_size=2)
store0.units.exclude(pk=store0.units.first().pk).update(target_f="FOO")
def _update_method(unit):
unit.target_f = "BAR"
return unit
count = batch.update(store0.units.filter(target_f="FOO"), _update_method)
assert count == store0.units.count() - 1
assert count == store0.units.filter(target_f="BAR").count()
store0.units.update(target_f="BAZ")
newcount = batch.update(store0.units.filter(target_f="FOO"),
_update_method,
update_fields=["source_f"])
# units not updated, only source in update_fields
assert newcount == 0
assert (store0.units.count() == store0.units.filter(
target_f="BAZ").count())