Collection2

A smart package for Meteor that extends Meteor.Collection to provide support for specifying a schema and then validating against that schema when inserting and updating.

This package requires and automatically installs the simple-schema package, which provides the SimpleSchema object type for defining and validating against schemas.

Installation

Install using Meteorite. When in a Meteorite-managed app directory, enter:

$ mrt add collection2

Why Use Collection2?

While adding allow/deny rules ensures that only authorized users can edit a document from the client, adding a schema ensures that only acceptable properties and values can be set within that document from the client. Thus, client side inserts and updates can be allowed without compromising security or data integrity.
Schema validation for all inserts and updates is reactive, allowing you to easily display customizable validation error messages to the user without any event handling.
Schema validation for all inserts and updates is automatic on both the client and the server, providing both speed and security.
The autoform package can take your collection's schema and automatically create HTML5 forms based on it. AutoForm provides automatic database operations, method calls, validation, and user interface reactivity. You have to write very little markup and no event handling. Refer to the autoform documentation for more information.

Attaching a Schema to a Collection

Let's say we have a normal "books" collection, defined in common.js:

Books = new Meteor.Collection("books");

Let's create a SimpleSchema schema for this collection. We'll do this in common.js, too:

var Schemas = {};

Schemas.Book = new SimpleSchema({
    title: {
        type: String,
        label: "Title",
        max: 200
    },
    author: {
        type: String,
        label: "Author"
    },
    copies: {
        type: Number,
        label: "Number of copies",
        min: 0
    },
    lastCheckedOut: {
        type: Date,
        label: "Last date this book was checked out",
        optional: true
    },
    summary: {
        type: String,
        label: "Brief summary",
        optional: true,
        max: 1000
    }
});

Once we have the SimpleSchema instance, all we need to do is attach it to our collection using the attachSchema method. Again, we will do this in common.js:

Books.attachSchema(Schemas.Book);

Now that our collection has a schema, we can do a validated insert on either the client or the server:

Books.insert({title: "Ulysses", author: "James Joyce"}, function(error, result) {
  //The insert will fail, error will be set,
  //and result will be undefined or false because "copies" is required.
  //
  //The list of errors is available on `error.invalidKeys` or by calling Books.simpleSchema().namedContext().invalidKeys()
});

Or we can do a validated update:

Books.update(book._id, {$unset: {copies: 1}}, function(error, result) {
  //The update will fail, error will be set,
  //and result will be undefined or false because "copies" is required.
  //
  //The list of errors is available on `error.invalidKeys` or by calling Books.simpleSchema().namedContext().invalidKeys()
});

Attach a Schema to Meteor.users

Obviously, when you attach a schema, you must know what the schema should be. For Meteor.users, here is an example schema, which you might have to adjust for your own needs:

Schema = {};

Schema.UserCountry = new SimpleSchema({
    name: {
        type: String
    },
    code: {
        type: String,
        regEx: /^[A-Z]{2}$/
    }
});

Schema.UserProfile = new SimpleSchema({
    firstName: {
        type: String,
        regEx: /^[a-zA-Z-]{2,25}$/,
        optional: true
    },
    lastName: {
        type: String,
        regEx: /^[a-zA-Z]{2,25}$/,
        optional: true
    },
    birthday: {
        type: Date,
        optional: true
    },
    gender: {
        type: String,
        allowedValues: ['Male', 'Female'],
        optional: true
    },
    organization : {
        type: String,
        regEx: /^[a-z0-9A-z .]{3,30}$/,
        optional: true
    },
    website: {
        type: String,
        regEx: SimpleSchema.RegEx.Url,
        optional: true
    },
    bio: {
        type: String,
        optional: true
    },
    country: {
        type: Schema.UserCountry,
        optional: true
    }
});

Schema.User = new SimpleSchema({
    _id: {
        type: String,
        regEx: SimpleSchema.RegEx.Id
    },
    username: {
        type: String,
        regEx: /^[a-z0-9A-Z_]{3,15}$/
    },
    emails: {
        type: [Object]
    },
    "emails.$.address": {
        type: String,
        regEx: SimpleSchema.RegEx.Email
    },
    "emails.$.verified": {
        type: Boolean
    },
    createdAt: {
        type: Date
    },
    profile: {
        type: Schema.UserProfile,
        optional: true
    },
    services: {
        type: Object,
        optional: true,
        blackbox: true
    },
    // Add `roles` to your schema if you use the meteor-roles package.
    // Note that when using this package, you must also specify the
    // `Roles.GLOBAL_GROUP` group whenever you add a user to a role.
    // Roles.addUsersToRoles(userId, ["admin"], Roles.GLOBAL_GROUP);
    // You can't mix and match adding with and without a group since
    // you will fail validation in some cases.
    roles: {
        type: Object,
        optional: true,
        blackbox: true
    }
});

Meteor.users.attachSchema(Schema.User);

This schema has not been thoroughly vetted to ensure that it accounts for all possible properties the accounts packages might try to set. Furthermore, any other packages you add might also try to set additional properties. If you see warnings in the console about keys being removed, that's a good indication that you should add those keys to the schema.

Note also that this schema uses the blackbox: true option for simplicity. You might choose instead to figure out a more specific schema.

(If you figure out a more accurate Meteor.users schema, documentation pull requests are welcome.)

Schema Format

Refer to the simple-schema package documentation for a list of all the available schema rules and validation methods.

Use the MyCollection.simpleSchema() method to access the attached SimpleSchema instance for a Meteor.Collection instance. For example:

check(doc, MyCollection.simpleSchema());

Validation Contexts

In the examples above, note that we called namedContext() with no arguments to access the SimpleSchema reactive validation methods. Contexts let you keep multiple separate lists of invalid keys for a single collection. In practice you might be able to get away with always using the default context. It depends on what you're doing. If you're using the context's reactive methods to update UI elements, you might find the need to use multiple contexts. For example, you might want one context for inserts and one for updates, or you might want a different context for each form on a page.

To use a specific named validation context, use the validationContext option when calling insert or update:

Books.insert({title: "Ulysses", author: "James Joyce"}, { validationContext: "insertForm" }, function(error, result) {
  //The list of errors is available by calling Books.simpleSchema().namedContext("insertForm").invalidKeys()
});

Books.update(book._id, {$unset: {copies: 1}}, { validationContext: "updateForm" }, function(error, result) {
  //The list of errors is available by calling Books.simpleSchema().namedContext("updateForm").invalidKeys()
});

Validating Without Inserting or Updating

It's also possible to validate a document without performing the actual insert or update:

Books.simpleSchema().namedContext().validate({title: "Ulysses", author: "James Joyce"}, {modifier: false});

Set the modifier option to true if the document is a mongo modifier object.

You can also validate just one key in the document:

Books.simpleSchema().namedContext().validateOne({title: "Ulysses", author: "James Joyce"}, "title", {modifier: false});

Or you can specify a certain validation context when calling either method:

Books.simpleSchema().namedContext("insertForm").validate({title: "Ulysses", author: "James Joyce"}, {modifier: false});
Books.simpleSchema().namedContext("insertForm").validateOne({title: "Ulysses", author: "James Joyce"}, "title", {modifier: false});

Refer to the simple-schema package documentation for more information about these methods.

Inserting or Updating Without Validating

To skip validation, use the validate: false option when calling insert or update. On the client (untrusted code), this will skip only client-side validation. On the server (trusted code), it will skip all validation.

Additional SimpleSchema Options

In addition to all the other schema validation options documented in the simple-schema package, the collection2 package adds additional options explained in this section.

denyInsert and denyUpdate

If you set denyUpdate: true, any collection update that modifies the field will fail. For instance:

Posts = new Meteor.Collection('posts', {
  schema: new SimpleSchema({
    title: {
      type: String
    },
    content: {
      type: String
    },
    createdAt: {
      type: Date,
      denyUpdate: true
    }
  })
});

var postId = Posts.insert({title: 'Hello', content: 'World', createdAt: new Date});

The denyInsert option works the same way, but for inserts. If you set denyInsert to true, you will need to set optional: true as well.

autoValue

The autoValue option is provided by the SimpleSchema package and is documented there. Collection2 adds the following properties to this for any autoValue function that is called as part of a C2 database operation:

isInsert: True if it's an insert operation
isUpdate: True if it's an update operation
isUpsert: True if it's an upsert operation (either upsert() or upsert: true)
userId: The ID of the currently logged in user. (Always null for server-initiated actions.)
isFromTrustedCode: True if the insert, update, or upsert was initiated from trusted (server) code

Note that autoValue functions are run on the client only for validation purposes, but the actual value saved will always be generated on the server, regardless of whether the insert/update is initiated from the client or from the server.

There are many possible use cases for autoValue. It's probably easiest to explain by way of several examples:

{
  // Force value to be current date (on server) upon insert
  // and prevent updates thereafter.
  createdAt: {
    type: Date,
      autoValue: function() {
        if (this.isInsert) {
          return new Date;
        } else if (this.isUpsert) {
          return {$setOnInsert: new Date};
        } else {
          this.unset();
        }
      }
  },
  // Force value to be current date (on server) upon update
  // and don't allow it to be set upon insert.
  updatedAt: {
    type: Date,
    autoValue: function() {
      if (this.isUpdate) {
        return new Date();
      }
    },
    denyInsert: true,
    optional: true
  },
  // Whenever the "content" field is updated, automatically set
  // the first word of the content into firstWord field.
  firstWord: {
    type: String,
    optional: true,
    autoValue: function() {
      var content = this.field("content");
      if (content.isSet) {
        return content.value.split(' ')[0];
      } else {
        this.unset(); // Prevent user from supplying her own value
      }
    }
  },
  // Whenever the "content" field is updated, automatically
  // update a history array.
  updatesHistory: {
    type: [Object],
    optional: true,
    autoValue: function() {
      var content = this.field("content");
      if (content.isSet) {
        if (this.isInsert) {
          return [{
              date: new Date,
              content: content.value
            }];
        } else {
          return {
            $push: {
              date: new Date,
              content: content.value
            }
          };
        }
      } else {
        this.unset();
      }
    }
  },
  'updatesHistory.$.date': {
    type: Date,
    optional: true
  },
  'updatesHistory.$.content': {
    type: String,
    optional: true
  },
  // Automatically set HTML content based on markdown content
  // whenever the markdown content is set.
  htmlContent: {
    type: String,
    optional: true,
    autoValue: function(doc) {
      var markdownContent = this.field("markdownContent");
      if (Meteor.isServer && markdownContent.isSet) {
        return MarkdownToHTML(markdownContent.value);
      }
    }
  }
}

index and unique

Use the index option to ensure a MongoDB index for a specific field:

{
  title: {
    type: String,
    index: 1
  }
}

Set to 1 or true for an ascending index. Set to -1 for a descending index. Or you may set this to another type of specific MongoDB index, such as "2d". Indexes works on embedded sub-documents as well.

If you have created an index for a field by mistake and you want to remove it, set index to false:

{
  "address.street": {
    type: String,
    index: false
  }
}

If a field has the unique option set to true, the MongoDB index will be a unique index as well. Then on the server, Collection2 will rely on MongoDB to check uniqueness of your field, which is more efficient than our custom checking.

{
  "pseudo": {
    type: String,
    index: true,
    unique: true
  }
}

For the unique option to work, index must be true, 1, or -1. The error message for uniqueness is very generic. It's best to define your own using MyCollection.simpleSchema().messages(). The error type string is "notUnique".

Indexes are built in the background so indexing does not block other database queries.

custom

The custom option is provided by the SimpleSchema package and is documented there. Collection2 adds the following properties to this for any custom function that is called as part of a C2 database operation:

isInsert: True if it's an insert operation
isUpdate: True if it's an update operation
isUpsert: True if it's an upsert operation (either upsert() or upsert: true)
userId: The ID of the currently logged in user. (Always null for server-initiated actions.)
isFromTrustedCode: True if the insert, update, or upsert was initiated from trusted (server) code

What Happens When The Document Is Invalid?

The callback you specify as the last argument of your insert() or update() call will have the first argument (error) set to an Error instance. The error message for the first invalid key is set in the error.message, and the full invalidKeys array is available on error.invalidKeys. This is true on both client and server, even if validation for a client-initiated operation does not fail until checked on the server.

If you attempt a synchronous operation in server code, the same validation error is thrown since there is no callback to pass it to. If this happens in a server method (defined with Meteor.methods), a more generic Meteor.Error is passed to your callback back on the client. This error does not have an invalidKeys property, but it does have the error message for the first invalid key set in error.reason.

Generally speaking, you would probably not use the Error for displaying to the user. You can instead use the reactive methods provided by the SimpleSchema validation context to display the specific error messages to the user somewhere in the UI. The autoform package provides some UI components and helpers for this purpose.

More Details

For the curious, this is exactly what Collection2 does before every insert or update:

Removes properties from your document or mongo modifier object if they are not explicitly listed in the schema. (To skip this, set the filter option to false when you call insert or update.)
Automatically converts some properties to match what the schema expects, if possible. (To skip this, set the autoConvert option to false when you call insert or update.)
Optimizes your operation so that empty string values will not be stored. (To skip this, set the removeEmptyStrings option to false when you call insert or update.)
Adds automatic (forced or default) values based on your schema. (Values are added only on the server and will make their way back to your client when your subscription is updated.)
Validates your document or mongo modifier object.
Performs the insert or update like normal, only if it was valid.

Collection2 is simply calling SimpleSchema methods to do these things. The validation happens on both the client and the server for client-initiated actions, giving you the speed of client-side validation along with the security of server-side validation.

Problems?

You might find yourself in a situation where it seems as though validation is not working correctly. First, you should enable SimpleSchema debug mode by setting SimpleSchema.debug = true, which may log some additional information. If you're still confused, read through the following tricky, confusing situations.

SubObjects and Arrays of Objects

One critical thing to know about Collection2 and SimpleSchema is that they don't validate the saved document but rather the proposed insert doc or the update modifier. In the case of updates, this means there is some information unknown to SimpleSchema, such as whether the array object you're attempting to modify already exists or not. If it doesn't exist, MongoDB would create it, so SimpleSchema will validate conservatively. It will assume that any properties not set by the modifier will not exist after the update. This means that the modifier will be deemed invalid if any required keys in the same object are not explicitly set in the update modifier.

For example, say we add the following keys to our "books" schema:

{
    borrowedBy: {
        type: [Object]
    },
    "borrowedBy.$.name": {
        type: String
    },
    "borrowedBy.$.email": {
        type: String,
        regEx: SimpleSchema.RegEx.Email
    },
}

Every object in the borrowedBy array must have a name and email property.

Now we discover that the name is incorrect in item 1, although the email address is correct. So we will just set the name to the correct value:

Books.update(id, {$set: {"borrowedBy.1.name": "Frank"}});

However, this will not pass validation. Why? Because we don't know whether item 1 in the borrowedBy array already exists, so we don't know whether it will have the required email property after the update finishes.

There are three ways to make this work:

$set the entire object
$set all required keys in the object
Perform the update on the server, and pass the validate: false option to skip validation.

When this situation occurs on the client with an autoForm, it generally does not cause any problems because AutoForm is smart enough to $set the entire object; it's aware of this potential issue. However, this means that you need to ensure that all required properties are represented by an input on the form. In our example, if you want an autoForm that only shows a field for changing the borrowedBy name and not the email, you should include both fields but make the email field hidden. Alternatively, you can submit the autoForm to a server method and then do a server update without validation.

Although these examples focused on an array of objects, sub-objects are treated basically the same way.

Contributing

Anyone is welcome to contribute. Fork, make and test your changes (mrt test-packages ./), and then submit a pull request.

Major Contributors

@mquandalle

(Add yourself if you should be listed here.)

CulturalMe/meteor-collection2