Allow direct save of changes on embedded model to be persisted on
parent document.
Person.embedsOne(Address);
Person.findById(someId)
.then(function(p){
var address = p.addressItem();
address.street = 'new street'
// This will now persist changes on parent document
return address.save();
})
[forward-port of #949]
The setting controls the strict mode used for embedded property types,
for example the type of "address" property in this model definition:
modelBuilder.define('TestEmbedded', {
name: 'string',
address: {
street: 'string',
},
});
It was previously completely undocumented. There are additional methods that add promises but I figure accurately documenting some is better than none. :)
Simplify DataAccessObject.create() and stop returning the
instance/array of instances. Users should always use callback (or
returned promise) to get the instance(s) created.
create() triggers
- before save
- after save
updateById() triggers
- before save
- after save
destroy() triggers
- before delete
- after delete
The implementation here is intentionally left with less features
than the regular DAO methods provide, the goal is to get a partial
(but still useful!) version released soon.
Limitations:
- `before save` & `after save` hooks don't provide `ctx.isNewInstance`
- async validations are not supported yet
- `persist` and `loaded` hooks are not triggered at all
- `before delete` hook does not provide `ctx.where` property and
it's not possible to change the outcome of `destroy()` using this
hook. Note that regular DAO does support this.
- updating embedded instances triggers update of the parent (owning)
model, which is correct and expected. However, the context provided
by `before save` and `after save` hooks on the parent model is sort of
arbitrary and may include wrong/extra data. The same probably applies
to the scenario when deleting embedded instances triggers update of
the parent model.
create() triggers
- before save
- after save
udpate() triggers
- before save
- after save
destroy() triggers
- before delete
- after delete
The implementation here is intentionally left with less features
than the regular DAO methods provide, the goal is to get a partial
(but still useful!) version released soon.
Make automatic validation optional on all CRUD methods in a loopback
model. This can be done in 2 ways
- set `automaticValidation` in the model settings
- set `validate` on the options passed when calling the crud methods
The options take precedence on the model setting.
By default the automatic validation remains true to be backwards
compatible
To replace incorrect location of capturing this condition in find() in
loopback-connector/lib/sql.js proposed earlier in
https://github.com/strongloop/loopback-connector/pull/34.
Being set to 'includeHasMany', findCaller option triggers injection of
PARTITION BY clause in buildColumnNames() in
loopback-connector-mssql/lib/mssql.js only when find() function is
called to process include filter with 'has many' relation.
Add datasource and model setting `normalizeUndefinedInQuery`
to determine how it will handle undefined values. Options:
- nullify : converts undefined to null
- throw : throw an error on undefined value
- ignore : strip the key where undefined value is found
The default operation is to strip the key.
If the PK has a definition like
```
"mysql": {
"dataType": "CHAR",
"dataLength":36,
"nullable":"N"
}
```
dataLength is ignored when the PK is used as a foreignKey in other models
Adds support for the ability to query embedsMany models from the parent.
Fix a memory connector bug that could occur when having an "or" or "and"
clause combined with another property. In that case, the and would revert
true for 'Paul McCartney'
```
{name:'John Lennon',and: [{role:'lead'}, {vip:true}]}}
```
Return `info.count` to the callback to indicate whether the model
instance was deleted or not. When Model's `settings.strictDelete`
is true, return 404 error when the model instance was not found.
Creates a new applyDefaultValues option on the Model constructor
defaulting to true, the current behaviour.
Updates the dao module to pass `{ applyDefaultValues: false }` to the
Model constructor during the updateOrCreate method when we assume an
update is happening.
Changes the after `save` callback in the memory connector to use the
`fromDb` method to deserialize the data passed back to upsert and
updateAttributes methods.
Sub models sometimes need to customize the properties from the base model.
This change allows each sub model has its own copy of the base property
definition to avoid potential conflicts across multiple sub models of the
same base.
- validateUpsert:true reports validation errors back to the callback
- validateUpsert:false does not call `isValid()` at all
- any other value report validation errors via `console.warn`
When a required number property is set to NaN, for example as a result
of coersion (`Number([1,2,3])`), the "presence" validation now correctly
reports an error.
Fix the implementation of updateOrCreate (a.k.a. upsert) to validate
the model before calling the connector.
In order to preserve backwards compatibility, validation errors are
only logged via console.warn by default.
The correct behaviour, where validation errors fail the updateOrCreate
operation, can be enabled via new model setting "validateUpsert".
When a callback is omitted from a method on a model relation that
supports promises, return that promise. This includes all the standard
DAO methods, as well as any user-defined methods that return promises.
e.g.:
mylist.todos.create({name: 'Item 1'}) // returns Promise
This API will use native ES6 promises if available. If not available,
or to force the use of another Promise library, you must assign the
global.Promise object.
e.g.:
global.Promise = require('bluebird')
Relations affected:
- BelongsTo
- HasOne
- HasMany
- HasManyThrough
- HasAndBelongsToMany
- ReferencesMany
- EmbedsOne
Exceptions:
The EmbedsMany relation has not been promisified, because most of the
methods return synchronous values.
The base relation getter method [e.g.: mylist.todos()] has not been
promisified, due to its default caching behavior.
New Methods:
- getAsync(condition, cb)
A new method "getAsync()" has been added to all relations except
EmbedsMany, which always fetches from the datasource rather than from
the cache. It takes an optional "where" condition (except for HasOne
and BelongsTo) and an optional callback. If the callback is omitted,
a Promise is returned.
When the setting "persistUndefinedAsNull" is true,
the model will use `null` instead of `undefined` in
all property values.
- Known optional model properties are set to `null` when no value
was provided.
- When setting model properties, `undefined` is always converted
to `null`. This applies to both known (model-defined) properties
and additional (custom, dynamic) properties.
- The instance method `toObject()` converts `undefined` to `null` too.
Because `toJSON()` calls `toObject()` under the hood, the change
applies to `toJSON()` too.
Rename the second callback argument to `info` to make its purpose more
clear. Fix jsdoc comments, note that these comments are NOT shown
on http://apidocs.strongloop.com/.
This commit is dropping lodash in favour of hand-written implementation
based on ES5 Array methods. As a result, the size of the (unminified)
loopback browser bundle is decreased by approx 360KB.
ffcaa4e7 added a "data" argument to the callback function, which
shadowed the original data with the data returned by a connector.
Not all connectors are returning a data object though, in which case
the model instance ("this" object) is updated with wrong values.
This commit fixes the problem by renaming the second callback argument
to "unusedData".
"before save" hooks provide "ctx.isNewInstance" whenever "ctx.instance"
is set. Possible values:
- true for all CREATE operations
- false for all UPDATE operations
- undefined for "prototype.save"
"after save" hooks provide "ctx.isNewInstance" whenever "ctx.instance"
is set. Possible values:
- true after all CREATE operations
- false after all UPDATE operations
- undefined after "updateOrCreate" and "save"
Note: both "updateOrCreate" and "prototype.updateAttributes"
don't provide `ctx.instance` to "before save" hooks, therefore
`ctx.isNewInstance` it not provided either.
Before this commit, the following code would not work:
Change.updateOrCreate({...}, function(err, ch) {
// somewhere later, modify "ch" and save the changes
ch.save(cb);
});
"before delete" and "after delete" hooks receive `ctx.instance`
when a single model is being deleted.
"before save" hook receives `ctx.currentInstance` when triggered
by `prototype.updateAttributes()`.
Note that "after save" hook triggered by `prototype.updateAttributes()`
already provides `ctx.instance`.
List of deprecated hooks:
- beforeValidate
- afterValidate
- beforeCreate
- afterCreate
- beforeSave
- afterSave
- beforeUpdate
- afterUpdate
- beforeDestroy
- afterDestroy
Also add a lightweight browser version of "depd", because the "depd"
does not support browser and it is not trivial to fix that.
This commits adds a lightweight implementation of depd's "deprecate"
function.
When a callback is omitted from a DAO method, return a Promise that
resolves to the value normally passed to the callback of that method.
If a callback is provided, behave normally.
This API will use native ES6 promises if available. If not available,
or to force the use of another Promise library, you must assign the
global.Promise object.
e.g.:
global.Promise = require('bluebird')
Class methods affected:
- create
- updateOrCreate / upsert
- findOrCreate
- exists
- find
- findOne
- findById
- findByIds
- remove / deleteAll / destroyAll
- removeById / deleteById / destroyById
- count
- update / updateAll
Prototype methods affected:
- save
- delete / remove / destroy
- updateAttribute
- updateAttributes
- reload
Exceptions / edge cases:
- create() used to return the data object that was passed in, even if
no callback was provided. Now, if a callback is provided, it will
return the data object, otherwise it will return a Promise.
- If create() is provided an array of data objects for creation, it
will continue to always return the array. This batch creation mode
does not support promises.
- findOrCreate() has a callback of the form: cb(err, instance, created),
with the extra parameter indicating whether the instance was created
or not. When called with its promise variant, the resolver will
receive a single array parameter: [instance, created]