Migrate sector taggings onto the canonical sector list

lib/tasks/consolidate_sectors.rake

@@ -0,0 +1,150 @@
+# frozen_string_literal: true
+
+# Migrates sector TAGGINGS from the legacy service-area names onto the canonical
+# SECTOR_TYPES list. PR #1703 already refreshed SECTOR_TYPES, seeded descriptions,
+# and made seeds create the new names while *unpublishing* (not destroying) the
+# old ones — but it deliberately did not move taggings, so every existing
+# workshop/story/etc. tagging is stranded on a now-unpublished legacy sector while
+# the new published sectors are empty. This task closes that gap.
+#
+# It is deploy-order-agnostic by design:
+#   - Run before #1703 seeds (DB still on the old names): legacy names are renamed
+#     in place, preserving taggings, and Student is merged into Education.
+#   - Run after #1703 seeds (new names already exist): each rename hits the
+#     collision check and falls back to a merge, re-pointing taggings onto the
+#     seeded published record and destroying the empty legacy one.
+namespace :data do
+  desc "Migrate sector taggings from legacy names onto the canonical SECTOR_TYPES (rename in place or merge via ModelDeduper). DRY_RUN=false to apply."
+  task consolidate_sectors: :environment do
+    dry_run = ENV.fetch("DRY_RUN", "true") != "false"
+
+    # Current name => new name. Renames are done in place with update!, which
+    # preserves every tagging because sectorable_items reference sector_id, not
+    # the name string.
+    renames = {
+      "Child Abuse" => "Child Abuse/Neglect",
+      "Community Oppression/Violence" => "Community Violence",
+      "Criminal/Legal" => "Court/Legal System",
+      "Disability" => "Disability Services",
+      "Education/Schools" => "Education",
+      "Homeless" => "Homelessness",
+      "Reproductive" => "Reproductive Services",
+      "Substance Use" => "Substance Use/Recovery",
+      "Veterans & Military" => "Military/Veterans"
+    }
+
+    # [ source_to_delete, target_to_keep ]. ModelDeduper re-points all taggings
+    # from the source onto the target, then destroys the source. Runs after the
+    # renames, so the target ("Education") already exists by this point.
+    merges = [
+      [ "Student", "Education" ]
+    ]
+
+    # Brand-new sectors with no current equivalent. Created published to match
+    # the seed behavior in db/seeds.rb.
+    new_sectors = [
+      "Batterers Intervention",
+      "Climate/Environmental",
+      "Fundraising/Donor Engagement",
+      "Grief/Loss",
+      "Health/Medical",
+      "Private Practice/Sole Proprietor",
+      "Racial/Social Justice",
+      "Religious/Faith Based",
+      "Self-Care/Personal Growth",
+      "Staff/Organizational Development",
+      "Systems/Policy Change"
+    ]
+
+    # "Other" is intentionally left untouched — it has special handling in the
+    # model (OTHER_SECTOR_NAME, the excluding_other scope) and is not part of
+    # the proposed list.
+
+    find_by_name = ->(name) { name.present? ? Sector.where("LOWER(name) = LOWER(?)", name).first : nil }
+
+    puts "=" * 80
+    puts "Sector consolidation  (DRY_RUN=#{dry_run})"
+    puts "=" * 80
+
+    renamed = 0
+    merged = 0
+    created = 0
+    skipped = []
+
+    # --- 1. Renames -------------------------------------------------------
+    puts "\nRenaming sectors…"
+    renames.each do |from, to|
+      source = find_by_name.call(from)
+      unless source
+        skipped << "rename #{from.inspect} -> #{to.inspect}: source not found (already renamed?)"
+        next
+      end
+
+      existing_target = find_by_name.call(to)
+      if existing_target && existing_target.id != source.id
+        # The target name is already taken by a different record, so an in-place
+        # rename would violate the uniqueness validation. Fall back to a merge.
+        puts "  COLLISION #{from.inspect} -> #{to.inspect}: target ##{existing_target.id} exists, merging instead"
+        merges << [ from, to ]
+        next
+      end
+
+      puts "  #{from.inspect} -> #{to.inspect} (sector ##{source.id})"
+      source.update!(name: to) unless dry_run
+      renamed += 1
+    end
+
+    # --- 2. Merges --------------------------------------------------------
+    puts "\nMerging sectors…"
+    deduper = ModelDeduper.new(model_class: Sector, dry_run: dry_run)
+    merges.each do |source_name, target_name|
+      source = find_by_name.call(source_name)
+      target = find_by_name.call(target_name)
+      # In a dry run the prerequisite rename hasn't been persisted, so resolve
+      # the target back through its pre-rename name to preview accurately.
+      target ||= find_by_name.call(renames.key(target_name))
+
+      unless source
+        skipped << "merge #{source_name.inspect} -> #{target_name.inspect}: source not found"
+        next
+      end
+      unless target
+        skipped << "merge #{source_name.inspect} -> #{target_name.inspect}: target not found"
+        next
+      end
+
+      puts "  #{source_name.inspect} (##{source.id}) -> #{target_name.inspect} (##{target.id})"
+      deduper.merge(target, source)
+      merged += 1
+    end
+
+    # --- 3. New sectors ---------------------------------------------------
+    puts "\nAdding new sectors…"
+    new_sectors.each do |name|
+      if find_by_name.call(name)
+        skipped << "create #{name.inspect}: already exists"
+        next
+      end
+
+      puts "  + #{name.inspect}"
+      Sector.create!(name: name, published: true) unless dry_run
+      created += 1
+    end
+
+    # --- Summary ----------------------------------------------------------
+    puts "\n#{'=' * 80}"
+    puts "Renamed: #{renamed}  Merged: #{merged}  Created: #{created}"
+    if skipped.any?
+      puts "\nSkipped (#{skipped.size}):"
+      skipped.each { |s| puts "  - #{s}" }
+    end
+
+    if dry_run
+      puts "\nDRY RUN — nothing was written. Re-run with DRY_RUN=false to apply."
+    else
+      puts "\nResulting sectors (#{Sector.count}):"
+      Sector.order(:name).pluck(:name).each { |n| puts "  - #{n}" }
+    end
+    puts "=" * 80
+  end
+end