Rework the code building model constructors to leverage `Function` class
and dynamically emit a constructor function named after the model.
Before this change, all model classes were called "ModelConstructor",
which made debugging difficult.
After this change, a model class for model "User" is called "User.
Because not all valid model names are also valid JavaScript identifiers,
we implement a simple sanitization technique (replacing characters like
"-", "." and ":" with underscore "_") and fall back to legacy
"ModelConstructor" if the model name is still not a valid JS identifier.
The uniq function does currently not work when the database is mongodb.
In the case of mongodb, the function will receive an array of bson
object of bson type ObjectID. The indexOf function will return a
different index, even if the mongodb ID is the same, as it is wrapped
in the ObjectID. This commit first transforms any ObjectID in the array
to a string representation. We can then use indexOf to check for
uniqueness.
* updateOnly, forceId changes
* support getUpdateOnlyProperties
* fixup! fix updateOrCreate in forceId mode
The contract of `updateOrCreate` is expecting a full object instance
to be passed to the callback.
The current implementation was creating an empty instance and
calling updateAttributes under the hood. As a result, the callback
was called with the attributes being updated only.
In order to preserve existing behaviour, we have to always build
a full initial instance by calling `findById`.
See the following discussion for more context:
https://github.com/strongloop/loopback-datasource-juggler/issues/966
* fixup! fix tests of upsert validation
* move forceId to model-builder
* remove TODO comment
* revert refactoring and test fixes
* Remove duplicate test
* change testcase names
* change to ModeClass.settingse
* forceId setup from datasource to model-builder
* fix inheritance of auto-generated forceId
* Fixed failing tests for auto change
* fixed a comment
* Fix missing option arguments in scope.js
Added option arguments in find.relatedmodel
Added case test
* Add order filter in verify function
Fix for cloudant test
Differentiate between tests skipped by feature flags, and those
manually marked as skipped.
Tests that use the bdd.describe and bdd.it helper functions
will be marked as [UNSUPPORTED] if they are conditionally disabled.
In #1298, the spec/doc for polymorphic relations was reviewed
**hasX relation**
- `type`: **hasMany**
- `as`: redefines **this** relation's name (optional)
- `model`: **modelTo**
- `polymorphic`:
- typeOf `polymorphic` === `String`
- matching **belongsTo** relation name
- `foreignKey` is generated as `polymorphic + 'Id'`,
- `discriminator` is generated as `polymorphic + 'Type'`
- typeOf `polymorphic` === `Object`
- `as`: **DEPRECATED** should display a warning,
replaced by `selector`
- `selector`: should match **belongsTo** relation name if the
latter is defined with {polymorphic: true}
- (required) if both foreignKey and discriminator
are **NOT** provided
- (extraneous) if both foreignKey and discriminator
are provided
- `foreignKey`: A property of modelTo, representing the fk to
modelFrom's id.
- generated by default as `selector + 'Id'`
- `discriminator`: A property of modelTo, representing the actual
modelFrom to be looked up and defined
dynamically
- generated by default as `selector + 'Type'`
---
**belongsTo relation**
- `type`: **belongsTo**
- `as`: redefines **this** relation's name (optional)
- `model`: **NOT EXPECTED**: should throw an error at
relation validation
- `polymorphic`:
- typeOf `polymorphic` === `Boolean`
- `foreignKey` is generated as `relationName + 'Id'`,
- `discriminator` is generated as `relationName + 'Type'`
- typeOf `polymorphic` === `Object`
- `as`: **DEPRECATED**: should display a warning,
replaced by `selector`
- `selector`:
- (required) if both foreignKey and discriminator
are **NOT** provided
- (extraneous) if both foreignKey and discriminator
are provided
- `foreignKey`: A property of modelTo, representing the fk to
modelFrom's id.
- generated by default as `selector + 'Id'`
- `discriminator`: A property of modelTo, representing the actual
modelFrom to be looked up and defined
dynamically
- generated by default as `selector + 'Type'`
The PR superseeds the existing deepMerge algorithm used to merge
settings of parent and child models with a new algorithm that allows
to specify the way each setting is merged or mixed-in.
This configuration of this algorithm uses a merge policy specification.
The `getMergePolicy()` helper of BaseModelClass can be used to ease
model merge configuration.
Next is presented the expected merge behaviour for each option.
NOTE: This applies to top-level settings properties
- Any
- `{replace: true}` (default): child replaces the value from parent
- assignin `null` on child setting deletes the inherited setting
- Arrays
- `{replace: false}`: unique elements of parent and child cumulate
- `{rank: true}` adds the model inheritance rank to array
elements of type Object {} as internal property `__rank`
- Object {}:
- `{replace: false}`: deep merges parent and child objects
- `{patch: true}`: child replaces inner properties from parent
The recommended merge policy is returned by getMergePolicy()
when calling the method with option `{configureModelMerge: true}`.
The legacy built-in merge policy is returned by `getMergePolicy()`
when avoiding option `configureModelMerge`.
NOTE: it also delivers ACLs ranking in addition to the legacy
behaviour as well as fixes for settings `description` and `relations`
`getMergePolicy()` can be customized using model's setting
`configureModelMerge` as follows:
```
{
// ..
options: {
configureModelMerge: {
// merge options
}
}
// ..
}
```
`getMergePolicy()` method can also be extended programmatically as
follows:
```
myModel.getMergePolicy = function(options) {
const origin = myModel.base.getMergePolicy(options);
return Object.assign({}, origin, {
// new/overriding options
});
};
```
New type that preserves string input as a string, but ensures that
the string is a valid Date.
Additionally, provides a .toDate function to provide the Date
object representation of the string.
Some databases are storing the value in UTC (not ISO) format,
and truncating the values. Even though this is technically
unavoidable, it does fail strict string equality.
* fix check for null
* add tests
* fix for early return
* Allow check for null and non-existent value
Some connectors uses a non existent prop instead of allowing null
Modified test case to look if null exists or the prop is non existent
* Check for null value with geo near query
* Apply requested changes
* change test to two users and simplify
* check error first
* Fix simple query test case with null value
* BDD for connectors w//o null support
* handle deep geo-near queries (#1216)
a dedicated mongKey is added in geo.nearFilter for mongoDB
fixes geo min distance tests as filter now expects an array
* Fix for string geoPoints
* Add geo point handle for ibmdb connectors
* Handle geo-point type for cassandra connector
Some connector, like arangodb, not support id as Number.
When `forceId` is set to true and id is set `Model.isValid`
report error and the field id is not coerced.
Fix `_targetClass` on scope function when using hasManyThrough
relation with customized relation names and foreignKey/keyThrough.
This bug is cause by `_targetClass` uses the camel-case of
`relationName`(e.g.: if `relationName` is `bbb`, `targetClass`
would be `Bbb`), which is not exists.
This will also suppress "not exposed" warnings when generating
angular sdk, and generate end-points for this scope.
* Fixes#1275 Transform *fields* property into array
`Include` filter takes into consideration string property
'fields' and transforms it into an array containing this string.
* Added error handling for `include` filter.
* ExecTasksWithInterLeave now contains a try-catch block
in order to catch any unexpected errors.
* LinkManyToMany now checks if *modelToIdName* exists on
*target* before continuing.
* Added unit test for *include* with string fields
Before this change, when resolving full connector path, all errors were
ignored. As a result, when the connector was installed but not
correctly built (e.g. loopback-connector-db2 which uses a native addon),
a very confusing message was reported by LoopBack.
In this commit, I am fixing the code handling `require()` errors
to ignore only MODULE_NOT_FOUND errors that contain the name
of the required module.
The query-string parser used by express
https://github.com/ljharb/qs#parsing-arrays
limits the size of arrays that are created from query strings to 20
items. Arrays larger than that are converted to objects using numeric
indices.
This commit fixes the coercion algorithm used by queries to
treat number-indexed objects as arrays. We still maintain a strict
understanding of an "array-like object" to limit the opportunity for
subtle bugs. In particular, the presence of non-index keys is an
indication that the object was not intended to be interpreted as
an array.
- Rename `flush` to `deleteAll`
- Add `delete`
- Detect `delete/deleteAll` before running downstream test suites
- Fall back to unoptimized `deleteAll` when connector does not support
`deleteAll` but supports `delete`
- Return 501 for connectors not supporting `delete` or `deleteAll`
Defining a model relation with the name "trigger" causes the model not
able to insert records. No error is thrown when a model relation with
the name "trigger" is defined. Adding a check for the model relation
name "trigger" will now throw an error.
Miroslav Bajtoš
Peter Bouda
Joost de Bruijn
zbarbuto
Janny
Jürg Lehni
loay
Rashmi Hunt
Tom Kirkpatrick
Raymond Feng
ssh24
dmellonch
biniam
Diana Lau
Simo Moujami
Tetsuo Seto
Kevin Delisle
Raymond Feng
ebarault
Bram Borggreve
Andrew McDonnell
paulussup
loay
Sakib Hasan
Joshua Chaitin-Pollak
Corentin H
Jonathan Sheely
Waldemar Zahn
Thaer Abbas
Matteo Padovano
Vincent Wen
Mikhail
Clark Wang
Nick Oikonomou
Dimitris
Rémi Bèges
jannyHou
Miroslav Bajtoš
Matteo Padovano
Simon Ho
Heath Morrison
Brian Schemp