Sequelize.sync：当表不存在时创建索引

Question

根据documentation sequelize.sync() 没有{force: true} 或{alter:true} 应该忽略已经存在的表并只创建/同步新表。但是至少有两个用例是现有表没有被完全忽略并且引入了错误。

我的设置：

1. sequelize.sync() 用作迁移前步骤以创建架构中不存在的表
2. sequelize.migrate() 用于更改任何现有表。

注意：Sequelize 模型被视为反映数据库模式的单一事实来源。它们总是会更新以反映数据库中存在的所有索引/字段。

复制步骤

第 1 步：创建 User 模型，其中包含两个字段 name 和 email。 Email 具有唯一索引

const users = sequelizeClient.define('users', {
    name: {
      type: DataTypes.STRING,
      allowNull: false,
    },
    email: {
      type: DataTypes.STRING,
      allowNull: false,
    },
  }, {
    indexes: [
      {
        unique: true,
        fields: ['email'],
      },
    ],
  });

没有迁移，因此预计将使用sequelize.sync() 创建表。 一切都按预期工作。这是生成的 SQL 脚本。

Executing (default): CREATE TABLE IF NOT EXISTS "users" ("id"  SERIAL , "name" VARCHAR(255) NOT NULL, "email" VARCHAR(255) NOT NULL, PRIMARY KEY ("id"));
Executing (default): SELECT i.relname AS name, ix.indisprimary AS primary, ix.indisunique AS unique, ix.indkey AS indkey, array_agg(a.attnum) as column_indexes, array_agg(a.attname) AS column_names, pg_get_indexdef(ix.indexrelid) AS definition FROM pg_class t, pg_class i, pg_index ix, pg_attribute a WHERE t.oid = ix.indrelid AND i.oid = ix.indexrelid AND a.attrelid = t.oid AND t.relkind = 'r' and t.relname = 'users' GROUP BY i.relname, ix.indexrelid, ix.indisprimary, ix.indisunique, ix.indkey ORDER BY i.relname;
Executing (default): CREATE UNIQUE INDEX "users_email" ON "users" ("email")

第 2 步 将新字段 phonenumber 添加到用户表并添加唯一索引。添加将更改表结构并创建索引的迁移。 sequelize.sync() 预计会忽略此表，但永远不会执行迁移，因为 sequelize.sync() 会引发以下错误。

Executing (default): CREATE TABLE IF NOT EXISTS "users" ("id"  SERIAL , "name" VARCHAR(255) NOT NULL, "email" VARCHAR(255) NOT NULL, PRIMARY KEY ("id"));
Executing (default): SELECT i.relname AS name, ix.indisprimary AS primary, ix.indisunique AS unique, ix.indkey AS indkey, array_agg(a.attnum) as column_indexes, array_agg(a.attname) AS column_names, pg_get_indexdef(ix.indexrelid) AS definition FROM pg_class t, pg_class i, pg_index ix, pg_attribute a WHERE t.oid = ix.indrelid AND i.oid = ix.indexrelid AND a.attrelid = t.oid AND t.relkind = 'r' and t.relname = 'users' GROUP BY i.relname, ix.indexrelid, ix.indisprimary, ix.indisunique, ix.indkey ORDER BY i.relname;
Executing (default): CREATE UNIQUE INDEX "users_phonenumber" ON "users" ("phonenumber")
{"_bitField":18087936,"_fulfillmentHandler0":{"name":"SequelizeDatabaseError","parent":{"name":"error","length":101,"severity":"ERROR","code":"42703","file":"indexcmds.c","line":"1083","routine":"ComputeIndexAttrs","sql":"CREATE UNIQUE INDEX \"users_phonenumber\" ON \"users\" (\"phonenumber\")"},"original":{"name":"error","length":101,"severity":"ERROR","code":"42703","file":"indexcmds.c","line":"1083","routine":"ComputeIndexAttrs","sql":"CREATE UNIQUE INDEX \"users_phonenumber\" ON \"users\" (\"phonenumber\")"},"sql":"CREATE UNIQUE INDEX \"users_phonenumber\" ON \"users\" (\"phonenumber\")"},"_trace":{"_promisesCreated":0,"_length":1},"level":"error","message":"Unhandled Rejection at: Promise "}

这是最终模型

const users = sequelizeClient.define('users', {
    name: {
      type: DataTypes.STRING,
      allowNull: false,
    },
    email: {
      type: DataTypes.STRING,
      allowNull: false,
    },
    phoneNumber: { // new field
      type: DataTypes.STRING,
      allowNull: false,
    },

  }, {
    indexes: [
      {
        unique: true,
        fields: ['email'],
      },
      { // new index
        unique: true,
        fields: ['phoneNumber'],
      },
    ],
  });

有人可以在这里建议一个解决方法，这样只有在表不存在时才会创建索引

---

另一个用例是当您添加带有评论的新字段时

    fieldWithComment: {
      type: DataTypes.STRING,
      comment: 'my comment goes here',
    },

生成的 SQL 显然会引发错误，因为新列尚不存在。

CREATE TABLE IF NOT EXISTS "users" (
    "id"   SERIAL, 
    "name" VARCHAR(255) NOT NULL, 
    "email" VARCHAR(255) NOT NULL, 
    "phonenumber" VARCHAR(255) NOT NULL, 
    "fieldWithComment" VARCHAR(255) , PRIMARY KEY ("id")); 
        COMMENT ON COLUMN "users"."fieldWithComment" IS 'my comment goes here';

Answer 1

如果有人像我一样因为续集更新等原因仍在苦苦挣扎，对上述答案不满意，这就是我解决问题的方法。

我使用的是underscore:true 选项。
所以，我将fields: ["phone", "countryCode"], 改为fields: ["phone", "country_code"],。

sequelize.define(
  "user",
  {
    phone: {
      type: DataTypes.STRING(20),
      allowNull: false,
    },
    countryCode: {
      type: DataTypes.STRING(4),
      allowNull: false,
    },
    // other attributes ...
  },
  {
    freezeTableName: true,
    timestamps: true,
    underscored: true,
    indexes: [
      {
        unique: true,
        fields: ["phone", "country_code"],
      },
    ],
  }
);

Answer 2

在生产环境中同步数据库的正确且非破坏性的方法是通过迁移。这样可以确保操作的顺序（新字段创建、索引创建等）。

因此，一般来说，同步应该只在开发和测试环境中使用。

引用官方文档：

sync({ force: true }) 和 sync({ alter: true }) 可能是破坏性操作。因此，不建议将它们用于生产级软件。相反，应该在 Sequelize CLI 的帮助下，使用迁移的高级概念来完成同步。

更多信息here（部分：“生产中的同步”）。