🗃️ clean up route segments (#4580)

Author: MrKrisKrisu

app/Http/Controllers/ReRoutingController.php

@@ -189,12 +189,16 @@ private function rerouteBetween(Stopover $start, Stopover $end, OpenRailRoutingP
                 return;
             }
 
+            // TODO: path_type is temporarily set to null because the existing OpenRailRoutingProfile
+            //       values (e.g. tgv_all, all_tracks) are not meaningful for our use.
+            //       Introduce a proper categorisation (e.g. rail, street, water, air) and
+            //       re-enable path_type assignment here once that is in place.
             $segment = $this->tripRepository->createRouteSegment(
                 fromStation: $start->station,
                 toStation: $end->station,
                 encodedPolyline: $encodedPolyline,
                 duration: $duration,
-                pathType: $pathType,
+                pathType: null,
                 distanceInMeters: $route->distanceInMeters
             );
             $this->tripRepository->setRouteSegmentForStop($start, $segment);

database/migrations/2026_03_14_000002_add_polyline_hash_to_route_segments.php

@@ -0,0 +1,39 @@
+<?php
+
+declare(strict_types=1);
+
+use Illuminate\Database\Migrations\Migration;
+use Illuminate\Database\Schema\Blueprint;
+use Illuminate\Support\Facades\DB;
+use Illuminate\Support\Facades\Schema;
+
+// On MySQL/MariaDB: ALGORITHM=INPLACE, LOCK=NONE for a non-blocking DDL.
+// Adding a nullable column without default uses INSTANT on MySQL 8.0+ / MariaDB 10.3+,
+// falling back to INPLACE – no table rebuild, no read/write lock.
+// On SQLite (test env): falls back to standard Schema builder.
+return new class() extends Migration
+{
+    public function up(): void
+    {
+        if (in_array(DB::getDriverName(), ['mysql', 'mariadb'], true)) {
+            DB::statement(
+                'ALTER TABLE route_segments
+                 ADD COLUMN polyline_hash VARCHAR(32) NULL AFTER polyline_precision,
+                 ALGORITHM=INPLACE, LOCK=NONE'
+            );
+
+            return;
+        }
+
+        Schema::table('route_segments', function (Blueprint $table): void {
+            $table->string('polyline_hash', 32)->nullable()->after('polyline_precision');
+        });
+    }
+
+    public function down(): void
+    {
+        Schema::table('route_segments', function (Blueprint $table): void {
+            $table->dropColumn('polyline_hash');
+        });
+    }
+};

database/migrations/2026_03_14_000003_backfill_polyline_hash_in_route_segments.php

@@ -0,0 +1,41 @@
+<?php
+
+declare(strict_types=1);
+
+use Illuminate\Database\Migrations\Migration;
+use Illuminate\Support\Facades\DB;
+
+// On MySQL/MariaDB: batched UPDATE with MD5() and LIMIT 500 keeps row locks short-lived.
+// On SQLite (test env): MD5() and UPDATE...LIMIT are unavailable, so hashes are computed
+// via PHP and applied row-by-row. Not performant, but test databases are tiny.
+return new class() extends Migration
+{
+    private const BATCH_SIZE = 500;
+
+    public function up(): void
+    {
+        if (in_array(DB::getDriverName(), ['mysql', 'mariadb'], true)) {
+            do {
+                $affected = DB::affectingStatement(
+                    'UPDATE route_segments
+                     SET polyline_hash = MD5(polyline)
+                     WHERE polyline_hash IS NULL
+                     LIMIT ' . self::BATCH_SIZE
+                );
+            } while ($affected > 0);
+
+            return;
+        }
+
+        DB::table('route_segments')
+            ->whereNull('polyline_hash')
+            ->orderBy('id')
+            ->chunkById(self::BATCH_SIZE, function (iterable $rows): void {
+                foreach ($rows as $row) {
+                    DB::table('route_segments')
+                        ->where('id', $row->id)
+                        ->update(['polyline_hash' => md5($row->polyline)]);
+                }
+            });
+    }
+};

database/migrations/2026_03_14_000004_nullify_path_type_in_route_segments.php

@@ -0,0 +1,19 @@
+<?php
+
+declare(strict_types=1);
+
+use Illuminate\Database\Migrations\Migration;
+use Illuminate\Support\Facades\DB;
+
+// Resets path_type to NULL on all route_segments in a single statement.
+// The existing path_type values are not meaningful in their current form and are cleared
+// before deduplication so they do not act as a grouping dimension.
+// A proper categorisation (e.g. rail, street, water) may be introduced separately.
+// At ~116k rows this completes in well under a second, so batching is unnecessary.
+return new class() extends Migration
+{
+    public function up(): void
+    {
+        DB::table('route_segments')->whereNotNull('path_type')->update(['path_type' => null]);
+    }
+};

database/migrations/2026_03_14_000005_add_polyline_hash_index_to_route_segments.php

@@ -0,0 +1,27 @@
+<?php
+
+declare(strict_types=1);
+
+use Illuminate\Database\Migrations\Migration;
+use Illuminate\Database\Schema\Blueprint;
+use Illuminate\Support\Facades\Schema;
+
+return new class() extends Migration
+{
+    public function up(): void
+    {
+        Schema::table('route_segments', function (Blueprint $table): void {
+            $table->index(
+                ['from_station_id', 'to_station_id', 'polyline_hash'],
+                'route_segments_polyline_lookup'
+            );
+        });
+    }
+
+    public function down(): void
+    {
+        Schema::table('route_segments', function (Blueprint $table): void {
+            $table->dropIndex('route_segments_polyline_lookup');
+        });
+    }
+};

database/migrations/2026_03_14_000006_deduplicate_route_segments.php

@@ -0,0 +1,85 @@
+<?php
+
+declare(strict_types=1);
+
+use Illuminate\Database\Migrations\Migration;
+use Illuminate\Support\Facades\DB;
+
+// Deduplicates route_segments where (from_station_id, to_station_id, polyline_hash)
+// has more than one row. path_type is ignored as it has been nullified in 000004.
+// Winner selection priority:
+//   1. Has custom_waypoints (manually curated)
+//   2. Most references in train_stopovers
+//   3. Oldest record (lowest UUID = earliest UUIDv7 timestamp)
+//
+// Each duplicate group is processed in its own small transaction to keep lock
+// duration minimal. The table remains usable throughout.
+return new class() extends Migration
+{
+    private const GROUP_BATCH_SIZE = 100;
+
+    public function up(): void
+    {
+        do {
+            $groups = DB::table('route_segments')
+                ->selectRaw('from_station_id, to_station_id, polyline_hash')
+                ->whereNotNull('polyline_hash')
+                ->groupBy(['from_station_id', 'to_station_id', 'polyline_hash'])
+                ->havingRaw('COUNT(*) > 1')
+                ->orderBy('from_station_id')
+                ->orderBy('to_station_id')
+                ->limit(self::GROUP_BATCH_SIZE)
+                ->get();
+
+            foreach ($groups as $group) {
+                $this->deduplicateGroup($group);
+            }
+
+            // Re-query from offset 0 after each batch: deletions shrink the result set
+            // so a fixed offset would skip groups.
+        } while ($groups->isNotEmpty());
+    }
+
+    private function deduplicateGroup(object $group): void
+    {
+        DB::transaction(function () use ($group): void {
+            $segments = DB::table('route_segments')
+                ->leftJoin(
+                    'train_stopovers',
+                    'route_segments.id',
+                    '=',
+                    'train_stopovers.route_segment_id'
+                )
+                ->select([
+                    'route_segments.id',
+                    DB::raw('COUNT(train_stopovers.id) AS ref_count'),
+                ])
+                ->where('route_segments.from_station_id', $group->from_station_id)
+                ->where('route_segments.to_station_id', $group->to_station_id)
+                ->where('route_segments.polyline_hash', $group->polyline_hash)
+                ->groupBy(['route_segments.id', 'route_segments.custom_waypoints'])
+                // Winner priority: custom_waypoints → most refs → oldest UUID
+                ->orderByRaw('route_segments.custom_waypoints IS NOT NULL DESC')
+                ->orderByDesc('ref_count')
+                ->orderBy('route_segments.id')
+                ->get();
+
+            if ($segments->count() < 2) {
+                return;
+            }
+
+            $winnerId = $segments->first()->id;
+            $loserIds = $segments->skip(1)->pluck('id')->all();
+
+            // Reassign all stopover references from losers to winner
+            DB::table('train_stopovers')
+                ->whereIn('route_segment_id', $loserIds)
+                ->update(['route_segment_id' => $winnerId]);
+
+            // Delete the now-unreferenced duplicate segments
+            DB::table('route_segments')
+                ->whereIn('id', $loserIds)
+                ->delete();
+        });
+    }
+};