From e713d25bf88d8ad2cf4d56eb8f16e0e066cfc4ad Mon Sep 17 00:00:00 2001
From: Ian Chan <ianchan@mobilitydata.org>
Date: Wed, 30 Sep 2026 13:04:19 -0400
Subject: [PATCH] remove mdb-3504 and 3505 feed and datasets

---
 liquibase/changelog.xml                      |  2 +
 liquibase/changes/remove_sao_paulo_feeds.sql | 75 ++++++++++++++++++++
 2 files changed, 77 insertions(+)
 create mode 100644 liquibase/changes/remove_sao_paulo_feeds.sql

diff --git a/liquibase/changelog.xml b/liquibase/changelog.xml
index e3a6d717a..23c5d11a0 100644
--- a/liquibase/changelog.xml
+++ b/liquibase/changelog.xml
@@ -140,6 +140,8 @@
     <!-- Index the unindexed FK gtfsfile.gtfs_dataset_id; the seal criteria scan it per dataset (issue #1831). -->
     <!-- Created CONCURRENTLY, so it needs its own transaction: hence the XML wrapper, as in feat_1432_indexes.xml -->
     <include file="changes/feat_1831_indexes.xml" relativeToChangelogFile="true"/>
+    <!-- Remove Sao Paulo feeds as requested by feed owner -->
+    <include file="changes/remove_sao_paulo_feeds.sql" relativeToChangelogFile="true"/>
     <!-- Keep this after all source table and schema changes so materialized views
          are recreated from the final source schema. -->
     <include file="materialized_views/materialized_views.xml" relativeToChangelogFile="true"/>
diff --git a/liquibase/changes/remove_sao_paulo_feeds.sql b/liquibase/changes/remove_sao_paulo_feeds.sql
new file mode 100644
index 000000000..a0ce74001
--- /dev/null
+++ b/liquibase/changes/remove_sao_paulo_feeds.sql
@@ -0,0 +1,75 @@
+-- Remove feeds mdb-3504 and mdb-3505 (Sao Paulo) and every record related to them.
+--
+-- Deleting the feed row cascades (ON DELETE CASCADE) to:
+--   gtfsfeed / gtfsrealtimefeed, externalid, feed_license_change, feed_reliability_seal,
+--   feedlocationgrouppoint, feedosmlocationgroup, feedrelatedlink, locationfeed,
+--   officialstatushistory, redirectingid (as source or target), seal_criterion,
+--   seal_criterion_snapshot, entitytypefeed, feedreference, gtfs_dataset_changelog,
+--   gtfs_feed_availability_check, gtfsdataset
+-- and from gtfsdataset to:
+--   gtfsfile, notice, location_gtfsdataset, validationreportgtfsdataset
+--
+-- Not covered by cascades, handled explicitly below:
+--   gtfsfeed.bounding_box_dataset_id / visualization_dataset_id (no ON DELETE rule)
+--   config_value_feed (no foreign key to feed)
+--   validationreport and feedinfo rows that would be left orphaned
+--   feedsearch materialized view
+
+CREATE TEMP TABLE sp_feeds_to_delete ON COMMIT DROP AS
+SELECT f.id
+FROM feed f
+WHERE f.stable_id IN ('mdb-3504', 'mdb-3505');
+
+CREATE TEMP TABLE sp_datasets_to_delete ON COMMIT DROP AS
+SELECT g.id, g.feed_info_id
+FROM gtfsdataset g
+WHERE g.feed_id IN (SELECT id FROM sp_feeds_to_delete);
+
+-- Validation reports used exclusively by the datasets being deleted
+CREATE TEMP TABLE sp_validation_reports_to_delete ON COMMIT DROP AS
+SELECT DISTINCT vrd.validation_report_id AS id
+FROM validationreportgtfsdataset vrd
+WHERE vrd.dataset_id IN (SELECT id FROM sp_datasets_to_delete)
+  AND NOT EXISTS (
+    SELECT 1
+    FROM validationreportgtfsdataset other
+    WHERE other.validation_report_id = vrd.validation_report_id
+      AND other.dataset_id NOT IN (SELECT id FROM sp_datasets_to_delete)
+  );
+
+-- Feed info rows used exclusively by the datasets being deleted
+CREATE TEMP TABLE sp_feed_infos_to_delete ON COMMIT DROP AS
+SELECT DISTINCT d.feed_info_id AS id
+FROM sp_datasets_to_delete d
+WHERE d.feed_info_id IS NOT NULL
+  AND NOT EXISTS (
+    SELECT 1
+    FROM gtfsdataset other
+    WHERE other.feed_info_id = d.feed_info_id
+      AND other.id NOT IN (SELECT id FROM sp_datasets_to_delete)
+  );
+
+-- Clear dataset references that have no ON DELETE rule
+UPDATE gtfsfeed
+SET bounding_box_dataset_id = NULL
+WHERE bounding_box_dataset_id IN (SELECT id FROM sp_datasets_to_delete);
+
+UPDATE gtfsfeed
+SET visualization_dataset_id = NULL
+WHERE visualization_dataset_id IN (SELECT id FROM sp_datasets_to_delete);
+
+DELETE FROM config_value_feed
+WHERE feed_id IN (SELECT id FROM sp_feeds_to_delete);
+
+-- Cascades to all feed-level and dataset-level tables listed above
+DELETE FROM feed
+WHERE id IN (SELECT id FROM sp_feeds_to_delete);
+
+-- Cascades to notice and featurevalidationreport
+DELETE FROM validationreport
+WHERE id IN (SELECT id FROM sp_validation_reports_to_delete);
+
+DELETE FROM feedinfo
+WHERE id IN (SELECT id FROM sp_feed_infos_to_delete);
+
+REFRESH MATERIALIZED VIEW feedsearch;