forked from CGM_Public/pretix_original
Add support for multiple linked_orderpositions on reusable media (#5666)
* change linked orderpositions to many-to-many * Update media views to list ops * return last op as fallback for linked_orderposition * add multi-op to export * update media-API * fix media-view filter * update control media forms * fix API orders * fix API orders matching media * remove cached_property linked_orderposition - keep only in API * fix media-issue signal * adapt checkin API for multiple orderpositions * remove unneeded comment * fix create/update logging * fix tests * fix more tests * fix code style * add label to reusablemedium * fix migration NOT NULL * fix tests * update docs * clarify docs updating multiple linked_orderpositions * clarify docs * no need to prefetch linked_orderpositions * improve readability * select_related order instead prefetch * add filter based on op.valid_from/until * rename secret to claim_token * Update docs for claim_token * unifiy deprecated style * Update reusablemedia.rst * Update reusablemedia.rst * Update reusablemedia.rst * fix missing claim_token in serializer * fix flake8 * add add_to_reusable_medium to order-serializer * fix tests regarding claim_token * fix flake8 * Clarify docs * list ops comma-separated in export * Add test for order-API add_to_reusable_medium * fix linked_orderpositions filter in checkinrpc * add test * Add help-text * fix multi-op media filter * fix flake8 * improve check * Fix sorting of reusable media type in overview * Add copy and qr button to reusable medium detail view * Rebase against origin/master * Add logentrytype reusable_medium.linked_orderposition.removed * add missing label_from_instance for SafeOrderPositionMultipleChoiceField * add tests for create with linked_orderposition * API add test for fallback-values in medium patch * fix flake8 * Fix indentation * fix migrations numbering * fix test * unify qutation marks * fix flake8 * micro-improve linked_op-removal-logging * simplify filter instead of annotate/get * Do not translate API-errors Co-authored-by: Raphael Michel <michel@pretix.eu> * Fix typos in doc Co-authored-by: Raphael Michel <mail@raphaelmichel.de> * Update versionchanged in docs Co-authored-by: Raphael Michel <michel@pretix.eu> * Change log to always added not changed * Add test for checkinrpc for ops out of timerang or canceled * improve tests mixing ops from different organizers * Fix logging of changed order_positions * properly log added/removed when using UI * refactor logging code * unify logging adding/removing ops via API * fix flake8 * remove unnecessary prefetch as already prefetched * optimize fetching ops * combine addon match and time-based validity match * fix combined valid and product check * re-number migrations * Apply suggestion from @raphaelm Co-authored-by: Raphael Michel <michel@pretix.eu> * fix flake8 * New attempt at logic * Improve op_candidate-selection for error message if no op matches check-in * Fix typo * fix valid_from start time being included * use the datetime parameter for the comparison time so that the simulator works too --------- Co-authored-by: Maximilian Richt <richt@pretix.eu> Co-authored-by: Martin Gross <gross@rami.io> Co-authored-by: Raphael Michel <michel@pretix.eu> Co-authored-by: Raphael Michel <mail@raphaelmichel.de>
This commit is contained in:
Richard Schreiber
committed by
GitHub
GitHub
parent
d3151f978d
commit
721b179521
@@ -491,6 +491,7 @@ def _redeem_process(*, checkinlists, raw_barcode, answers_data, datetime, force,
|
||||
)
|
||||
raw_barcode_for_checkin = None
|
||||
from_revoked_secret = False
|
||||
reusable_medium_used = None
|
||||
if simulate:
|
||||
common_checkin_args['__fake_arg_to_prevent_this_from_being_saved'] = True
|
||||
|
||||
@@ -521,11 +522,12 @@ def _redeem_process(*, checkinlists, raw_barcode, answers_data, datetime, force,
|
||||
# with respecting the force option), or it's a reusable medium (-> proceed with that)
|
||||
if not op_candidates:
|
||||
try:
|
||||
media = ReusableMedium.objects.select_related('linked_orderposition').active().get(
|
||||
media = ReusableMedium.objects.active().filter(
|
||||
Exists(ReusableMedium.linked_orderpositions.through.objects.filter(reusablemedium_id=OuterRef('pk')))
|
||||
).get(
|
||||
organizer_id=checkinlists[0].event.organizer_id,
|
||||
type=source_type,
|
||||
identifier=raw_barcode,
|
||||
linked_orderposition__isnull=False,
|
||||
)
|
||||
raw_barcode_for_checkin = raw_barcode
|
||||
except ReusableMedium.DoesNotExist:
|
||||
@@ -628,7 +630,9 @@ def _redeem_process(*, checkinlists, raw_barcode, answers_data, datetime, force,
|
||||
'list': MiniCheckinListSerializer(list_by_event[revoked_matches[0].event_id]).data,
|
||||
}, status=400)
|
||||
else:
|
||||
if media.linked_orderposition.order.event_id not in list_by_event:
|
||||
linked_ops = media.linked_orderpositions.all().select_related("order").prefetch_related("addons")
|
||||
linked_event_ids = {op.order.event_id for op in linked_ops}
|
||||
if not any(event_id in list_by_event for event_id in linked_event_ids):
|
||||
# Medium exists but connected ticket is for the wrong event
|
||||
if not simulate:
|
||||
checkinlists[0].event.log_action('pretix.event.checkin.unknown', data={
|
||||
@@ -654,28 +658,91 @@ def _redeem_process(*, checkinlists, raw_barcode, answers_data, datetime, force,
|
||||
'checkin_texts': [],
|
||||
'list': MiniCheckinListSerializer(checkinlists[0]).data,
|
||||
}, status=404)
|
||||
op_candidates = [media.linked_orderposition]
|
||||
if list_by_event[media.linked_orderposition.order.event_id].addon_match:
|
||||
op_candidates += list(media.linked_orderposition.addons.all())
|
||||
op_candidates = []
|
||||
for op in linked_ops:
|
||||
if op.order.event_id in list_by_event:
|
||||
reusable_medium_used = media
|
||||
op_candidates.append(op)
|
||||
if list_by_event[op.order.event_id].addon_match:
|
||||
op_candidates += list(op.addons.all())
|
||||
|
||||
# 3. Handle the "multiple options found" case: Except for the unlikely case of a secret being also a valid primary
|
||||
# key on the same list, we're probably dealing with the ``addon_match`` case here and need to figure out
|
||||
# which add-on has the right product.
|
||||
# key on the same list, we're probably dealing with multiple linked_orderpositions or the ``addon_match`` case
|
||||
# here and need to figure out which op has the right product. This basically is a valid-for-checkin-test on every op.
|
||||
if len(op_candidates) > 1:
|
||||
op_candidates_matching_product = [
|
||||
op for op in op_candidates
|
||||
if (
|
||||
(list_by_event[op.order.event_id].addon_match or op.secret == raw_barcode or legacy_url_support) and
|
||||
(list_by_event[op.order.event_id].all_products or op.item_id in {i.pk for i in list_by_event[op.order.event_id].limit_products.all()})
|
||||
)
|
||||
]
|
||||
|
||||
if len(op_candidates_matching_product) == 0:
|
||||
# None of the found add-ons has the correct product, too bad! We could just error out here, but
|
||||
if not reusable_medium_used:
|
||||
# 3a. First, we clean up that we made an imprecise query above. If a scan is made for multiple check-in lists,
|
||||
# we have queried ``addon_to__secret=raw_barcode``, even if some of the lists in question do not allow addon
|
||||
# matching. So we accept all candidates that match one of these cases:
|
||||
# - Exactly the ticket secret we scanned (because that's always a possible result)
|
||||
# - Exactly the ticket pk we scanned (on legacy endpoints)
|
||||
# - An add-on on a list that allows add-on matching
|
||||
# This is not necessary when a reusable media was used, since in that case we already obeyed list.addon_match
|
||||
# correctly above.
|
||||
op_candidates_filtered = [
|
||||
op for op in op_candidates
|
||||
if (
|
||||
op.secret == raw_barcode or
|
||||
list_by_event[op.order.event_id].addon_match or
|
||||
(str(op.pk) == raw_barcode and legacy_url_support and not untrusted_input)
|
||||
)
|
||||
]
|
||||
else:
|
||||
op_candidates_filtered = op_candidates
|
||||
|
||||
if len(op_candidates_filtered) > 1:
|
||||
# 3b. If we still have multiple candidates, we filter by product based on the check-in list configuration.
|
||||
# This is relevant for the addon_match scenario where the scanned ticket has multiple add-ons, but only
|
||||
# one is contained in the check-in list used to scan. It makes sense to filter this first, since it is a
|
||||
# "static" check, i.e. scanning the same QR code on the same check-in list will always do the same, no matter
|
||||
# when I scan it, and it is "intentional" filtering in the sense that the admin configured this behaviour
|
||||
# into the check-in list.
|
||||
op_candidates_filtered = [
|
||||
op for op in op_candidates_filtered
|
||||
if list_by_event[op.order.event_id].all_products or op.item_id in {i.pk for i in list_by_event[op.order.event_id].limit_products.all()}
|
||||
]
|
||||
|
||||
if len(op_candidates_filtered) > 1:
|
||||
# 3c. If we still have multiple candidates, we filter by validity date. This was introduced for the case where
|
||||
# a reusable media refers to two tickets, one currently valid and one expired or in the future. Howeer,
|
||||
# it could in theory also happen with two add-ons being on the same check-in list but without overlapping
|
||||
# validity. It makes sense to filter this "after" the previous checks since it is not "intentional" filtering
|
||||
# configured by the admin but "accidental" filtering that depends on the time of execution.
|
||||
op_candidates_filtered = [
|
||||
op for op in op_candidates_filtered
|
||||
if (
|
||||
(not op.valid_from or op.valid_from <= datetime) and
|
||||
(not op.valid_until or op.valid_until > datetime)
|
||||
)
|
||||
]
|
||||
|
||||
if len(op_candidates_filtered) == 0:
|
||||
# None of the ops is valid today or has the correct product, too bad! We could just error out here, but
|
||||
# instead we just continue with *any* product and have it rejected by the check in perform_checkin.
|
||||
# This has the advantage of a better error message.
|
||||
op_candidates = [op_candidates[0]]
|
||||
elif len(op_candidates_matching_product) > 1:
|
||||
# To improve the error message, we select the op that will "work next" or - if none matches - "worked last".
|
||||
op_candidate = None
|
||||
for op in op_candidates:
|
||||
if (
|
||||
op.valid_from and op.valid_from > datetime and
|
||||
(not op_candidate or op.valid_from < op_candidate.valid_from)
|
||||
):
|
||||
op_candidate = op
|
||||
|
||||
if not op_candidate:
|
||||
# no candidate in the future, get closest in the past
|
||||
for op in op_candidates:
|
||||
if (
|
||||
op.valid_until and op.valid_until < datetime and
|
||||
(not op_candidate or op.valid_until > op_candidate.valid_until)
|
||||
):
|
||||
op_candidate = op
|
||||
|
||||
if not op_candidate:
|
||||
op_candidate = op_candidates[0]
|
||||
|
||||
op_candidates = [op_candidate]
|
||||
elif len(op_candidates_filtered) > 1:
|
||||
# It's still ambiguous, we'll error out.
|
||||
# We choose the first match (regardless of product) for the logging since it's most likely to be the
|
||||
# base product according to our order_by above.
|
||||
@@ -709,7 +776,7 @@ def _redeem_process(*, checkinlists, raw_barcode, answers_data, datetime, force,
|
||||
'list': MiniCheckinListSerializer(list_by_event[op.order.event_id]).data,
|
||||
}, status=400)
|
||||
else:
|
||||
op_candidates = op_candidates_matching_product
|
||||
op_candidates = op_candidates_filtered
|
||||
|
||||
op = op_candidates[0]
|
||||
common_checkin_args['list'] = list_by_event[op.order.event_id]
|
||||
|
||||
Reference in New Issue
Block a user