worklenz/worklenz-backend/database/pg-migrations/20250128000000_fix-window-function-error.js

/* eslint-disable camelcase */

exports.shorthands = undefined;

exports.up = pgm => {
  // Replace the optimized sort functions to avoid CTE usage in UPDATE statements
  pgm.createFunction(
    'handle_task_list_sort_between_groups_optimized',
    [
      { name: '_from_index', type: 'integer' },
      { name: '_to_index', type: 'integer' },
      { name: '_task_id', type: 'uuid' },
      { name: '_project_id', type: 'uuid' },
      { name: '_batch_size', type: 'integer', default: 100 }
    ],
    { returns: 'void', language: 'plpgsql', replace: true },
    `
DECLARE
    _offset INT := 0;
    _affected_rows INT;
BEGIN
    -- PERFORMANCE OPTIMIZATION: Use direct updates without CTE in UPDATE
    IF (_to_index = -1)
    THEN
        _to_index = COALESCE((SELECT MAX(sort_order) + 1 FROM tasks WHERE project_id = _project_id), 0);
    END IF;

    -- PERFORMANCE OPTIMIZATION: Batch updates for large datasets
    IF _to_index > _from_index
    THEN
        LOOP
            UPDATE tasks
            SET sort_order = sort_order - 1
            WHERE project_id = _project_id
              AND sort_order > _from_index
              AND sort_order < _to_index
              AND sort_order > _offset
              AND sort_order <= _offset + _batch_size;
            
            GET DIAGNOSTICS _affected_rows = ROW_COUNT;
            EXIT WHEN _affected_rows = 0;
            _offset := _offset + _batch_size;
        END LOOP;

        UPDATE tasks SET sort_order = _to_index - 1 WHERE id = _task_id AND project_id = _project_id;
    END IF;

    IF _to_index < _from_index
    THEN
        _offset := 0;
        LOOP
            UPDATE tasks
            SET sort_order = sort_order + 1
            WHERE project_id = _project_id
              AND sort_order > _to_index
              AND sort_order < _from_index
              AND sort_order > _offset
              AND sort_order <= _offset + _batch_size;
            
            GET DIAGNOSTICS _affected_rows = ROW_COUNT;
            EXIT WHEN _affected_rows = 0;
            _offset := _offset + _batch_size;
        END LOOP;

        UPDATE tasks SET sort_order = _to_index + 1 WHERE id = _task_id AND project_id = _project_id;
    END IF;
END
    `
  );

  // Replace the second optimized sort function
  pgm.createFunction(
    'handle_task_list_sort_inside_group_optimized',
    [
      { name: '_from_index', type: 'integer' },
      { name: '_to_index', type: 'integer' },
      { name: '_task_id', type: 'uuid' },
      { name: '_project_id', type: 'uuid' },
      { name: '_batch_size', type: 'integer', default: 100 }
    ],
    { returns: 'void', language: 'plpgsql', replace: true },
    `
DECLARE
    _offset INT := 0;
    _affected_rows INT;
BEGIN
    -- PERFORMANCE OPTIMIZATION: Batch updates for large datasets without CTE in UPDATE
    IF _to_index > _from_index
    THEN
        LOOP
            UPDATE tasks
            SET sort_order = sort_order - 1
            WHERE project_id = _project_id
              AND sort_order > _from_index
              AND sort_order <= _to_index
              AND sort_order > _offset
              AND sort_order <= _offset + _batch_size;
            
            GET DIAGNOSTICS _affected_rows = ROW_COUNT;
            EXIT WHEN _affected_rows = 0;
            _offset := _offset + _batch_size;
        END LOOP;
    END IF;

    IF _to_index < _from_index
    THEN
        _offset := 0;
        LOOP
            UPDATE tasks
            SET sort_order = sort_order + 1
            WHERE project_id = _project_id
              AND sort_order >= _to_index
              AND sort_order < _from_index
              AND sort_order > _offset
              AND sort_order <= _offset + _batch_size;
            
            GET DIAGNOSTICS _affected_rows = ROW_COUNT;
            EXIT WHEN _affected_rows = 0;
            _offset := _offset + _batch_size;
        END LOOP;
    END IF;

    UPDATE tasks SET sort_order = _to_index WHERE id = _task_id AND project_id = _project_id;
END
    `
  );

  // Add simple bulk update function as alternative
  pgm.createFunction(
    'update_task_sort_orders_bulk',
    [{ name: '_updates', type: 'json' }],
    { returns: 'void', language: 'plpgsql', replace: true },
    `
DECLARE
    _update_record RECORD;
BEGIN
    -- Simple approach: update each task's sort_order from the provided array
    FOR _update_record IN 
        SELECT 
            (item->>'task_id')::uuid as task_id,
            (item->>'sort_order')::int as sort_order,
            (item->>'status_id')::uuid as status_id,
            (item->>'priority_id')::uuid as priority_id,
            (item->>'phase_id')::uuid as phase_id
        FROM json_array_elements(_updates) as item
    LOOP
        UPDATE tasks 
        SET 
            sort_order = _update_record.sort_order,
            status_id = COALESCE(_update_record.status_id, status_id),
            priority_id = COALESCE(_update_record.priority_id, priority_id)
        WHERE id = _update_record.task_id;
        
        -- Handle phase updates separately since it's in a different table
        IF _update_record.phase_id IS NOT NULL THEN
            INSERT INTO task_phase (task_id, phase_id)
            VALUES (_update_record.task_id, _update_record.phase_id)
            ON CONFLICT (task_id) DO UPDATE SET phase_id = _update_record.phase_id;
        END IF;
    END LOOP;
END
    `
  );
};

exports.down = pgm => {
  // Drop the functions if needed to rollback
  pgm.dropFunction('update_task_sort_orders_bulk', [{ name: '_updates', type: 'json' }], { ifExists: true });
  pgm.dropFunction('handle_task_list_sort_inside_group_optimized', [
    { name: '_from_index', type: 'integer' },
    { name: '_to_index', type: 'integer' },
    { name: '_task_id', type: 'uuid' },
    { name: '_project_id', type: 'uuid' },
    { name: '_batch_size', type: 'integer' }
  ], { ifExists: true });
  pgm.dropFunction('handle_task_list_sort_between_groups_optimized', [
    { name: '_from_index', type: 'integer' },
    { name: '_to_index', type: 'integer' },
    { name: '_task_id', type: 'uuid' },
    { name: '_project_id', type: 'uuid' },
    { name: '_batch_size', type: 'integer' }
  ], { ifExists: true });
};