laravel-shift / blueprint Goto Github PK

When providing a precision for float or decimal columns, it would be ideal if this precision were passed as the first argument to Faker's randomFloat method

Even better if the max digit were inferred as well.

For example the model-modifiers.bp fixture defines: total: decimal:10,2.

The generated is factory has: 'total' => $faker->randomFloat().

Ideally it would generate: 'total' => $faker->randomFloat(2) or even 'total' => $faker->randomFloat(2, 0, 999999)

Some folks really dislike Laravel's default locations for generated output. A common occurrence is that folks might want to output models to app/Models instead of directly under app/. There are also others (myself included) that attempt to do some Domain separation of the business logic (not entirely DDD, but at least some sort of other structure).

Whilst tools like PhpStorm pack enough firepower to easily refactor the output, I still think it might be a good idea to at least configure some paths for the generator to output to? Love to read your take on this idea.

Passing boolean on

softDeletes: true

produces

$table->string('softDeletes'); on Migration

While it's easy to use a clean working state and version control to manage the generation of new components, it would be better if Blueprint had an "undo" command to revert/remove the generated components of the last built.

Model definitions which include foreign keys (e.g. *_id: id) should also generate for the relationship in the Model and a new, related record in the Factory.

While sufficient to past the current tests, the Rules translator is incomplete.

Given a Column definition, the Rules translator should be have the information it needs to intelligently translate this data into validation rules.

It can do so by:

• Column name (e.g. email or email_address adds an email validation rule)
• Relationship (e.g. user_id adds an exists validation rule with proper attributes)
• Data Type (e.g. integer)
• Attributes (e.g. max lengths, digits precision,in set, etc)
• Modifiers (e.g. gt:0 for unsigned, unique, etc)

I expect this will constantly be tweaked. I plan to finish a base version for v1.0.

Nonetheless, I welcome PRs to begin any portion of the items above. However, large or small. I only ask that you add to the test.

To help get started, you can review the current implementation and review the available validation rules to get ideas of the types of things that can be translated.

I cannot currently see any functionality to add a table without a model, thus when trying to replicate the default laravel auth system, we have to create a full model and factory for the password_resets table rather than just the existing migration.

Error message:" BadMethodCallException : Method Illuminate\Support\Str::beforeLast does not exist."

in yaml refering to user id in relation for instance:

user_id <-- works
user::id <-- Do not work
id <-- Do not work

version 1.2.1

Thank for your efforts! keep it up!

Using the example yaml from the README, when I use php artisan blueprint:build I get an error:

ErrorException : file_put_contents(app/Events/NewPost.php): failed to open stream: No such file or directory

(Laravel v6.6.0, PHP 7.4.0, Mac 10.14.6 (18G1012))

ErrorException  : file_put_contents(app/Events/NewPost.php): failed to open stream: No such file or directory

  at /Users/martin/Sites/blueprint-05/vendor/laravel/framework/src/Illuminate/Filesystem/Filesystem.php:133
    129|      * @return int|bool
    130|      */
    131|     public function put($path, $contents, $lock = false)
    132|     {
  > 133|         return file_put_contents($path, $contents, $lock ? LOCK_EX : 0);
    134|     }
    135|
    136|     /**
    137|      * Write the contents of a file, replacing it atomically if it already exists.

  Exception trace:

  1   file_put_contents("app/Events/NewPost.php", "<?php

namespace App\Events;

use Illuminate\Queue\SerializesModels;

class NewPost
{
    use SerializesModels;

    public $post;

    /**
     * Create a new event instance.
     *
     * @return void
     */
    public function __construct($post)
    {
        $this->post = $post;
    }
}
")
      /Users/martin/Sites/blueprint-05/vendor/laravel/framework/src/Illuminate/Filesystem/Filesystem.php:133

  2   Illuminate\Filesystem\Filesystem::put("app/Events/NewPost.php", "<?php

namespace App\Events;

use Illuminate\Queue\SerializesModels;

class NewPost
{
    use SerializesModels;

    public $post;

    /**
     * Create a new event instance.
     *
     * @return void
     */
    public function __construct($post)
    {
        $this->post = $post;
    }
}
")
      /Users/martin/Sites/blueprint-05/vendor/laravel-shift/blueprint/src/Generators/Statements/EventGenerator.php:46

  Please use the argument -v to see more details.

1. Adding softDeletes in yaml adds 'softDeletes' => $faker->word, inside factory

Steps to reproduce

models:
  Post:
    title: string:400
    content: longtext
    published_at: nullable timestamp
    softDeletes: true

The erase command has two bugs:

1. It errors when previous models were created
2. After running, it still stores the previously built files

It would be helpful to be able to add those columns with the timezone.

1. $table->timestampsTz(); Adds nullable created_at and updated_at TIMESTAMP (with timezone) equivalent columns.
2. $table->softDeletesTz(); Adds a nullable deleted_at TIMESTAMP (with timezone) equivalent column for soft deletes.

In most of the time Every Devs prefers to keep the models and Controllers based on their groups
So For Example i have a app that contains the Blog so i wish to keep the Models inside the app/Models/Blog so this below syntax might be helpful in many situation because i see most of the app that is Built with laravel has the models inside the app/Models Directory

models:
  Models/Blog/Post:
    title: string:400
    content: longtext
    published_at: nullable timestamp

controllers:
  Blog/PostController:
    index:
      query: all posts
      render: post.index with posts

    store:
      validate: title, content
      save: post
      send: ReviewNotification to post.author
      queue: SyncMedia
      flash: post.title
      redirect: post.index

Example input:

  Person:
    additionalName: string
    award: string
    birthDate: date
    birthPlace_id: id:place
    callSign: string
    deathDate: date
    deathPlace_id: id:place

Should render:

// \App\Person::class
// ...
    public function birthPlace()
    {
        return $this->belongsTo(\App\Place::class, 'birthPlace_id');
    }

    public function deathPlace()
    {
        return $this->belongsTo(\App\Place::class, 'deathPlace_id');
    }

Both Laravel and Blueprint follow some documented, as well as community, conventions. Developers new to Laravel may not be familiar with these.

As such, it'd be nice to have a --notes option which would add the Laravel-style block comments with more information about the generated code with links to the documentation where applicable.

With this we can ensure Blueprint users become familiar with Laravel as well as Blueprint, and vice version.

Ideally Blueprint could load existing models to have more context to use as reference when defining controllers.

I like the command name trace for this...

When running php artisan test it complains about not having jasonmccreary/laravel-test-assertions:

php artisan test             
PHP Fatal error:  Trait 'JMac\Testing\Traits\HttpTestAssertions' not found

when SoftDeletes are enabled and relationships exist in the model, then use SoftDeletes; is also added for each function (invalid php code).

As per the Laravel documentation located at https://laravel.com/docs/master/database-testing#relationships, paragraph "Relations & Attribute Closures".

Given the following draft.yaml, modelled after a schema.org/Person:

models:
  Country:
    name: string
    isoAlpha2: string:2
    isoAlpha3: string:3
  ContactPoint:
    contactType: string
    email: string
    faxNumber: string
    telephone: string
  PostalAddress:
    country_id: id:country
    addressLocality: string
    addressRegion: string
    postalOfficeBoxNumber: string
    postalCode: string
    streetAddress: text
    contact_point_id: id:contact_point
  Person:
    additionalName: string
    address_id: id:postal_address
    familyName: string
    givenName: string

The generated factories will currently insert a $faker->randomDigitNotNull, like so:

$factory->define(Person::class, function (Faker $faker) {
    return [
        'additionalName' => $faker->word,
        'address_id' => $faker->randomDigitNotNull,
        'familyName' => $faker->word,
        'givenName' => $faker->word,
    ];
});

As per the documentation, it could be generated like so:

$factory->define(Person::class, function (Faker $faker) {
    return [
        'additionalName' => $faker->word,
        'address_id' => factory(App\PostalAddress::class),
        'familyName' => $faker->word,
        'givenName' => $faker->word,
    ];
});

Subsequently, the other PostalAddressFactory could be like so:

$factory->define(PostalAddress::class, function (Faker $faker) {
    return [
        'country_id' => factory(App\Country::class),
        'addressLocality' => $faker->city,
        'addressRegion' => $faker->state,
        'postalOfficeBoxNumber' => $faker->randomNumber(4),
        'postalCode' => $faker->postcode,
        'streetAddress' => $faker->streetAddress,
        'contact_point_id' => factory(App\ContactPoint::class),
    ];
});

The benefit of this, is that when using the PersonFactory in a test, you will get a Person with a full set of related data:

factory(App\Person::class, 100)->create();
App\Person::with('address.country', 'address.contactPoint')->first();

could yield:

> App\Person {#1062
     id: 1,
     additionalName: "Serena",
     address_id: 1,
     familyName: "Medhurst",
     givenName: "Andreane",
     created_at: "2019-12-06 15:13:14",
     updated_at: "2019-12-06 15:13:14",
     address: App\PostalAddress {#1042
       id: 1,
       country_id: 1,
       addressLocality: "Daniellaville",
       addressRegion: "Texas",
       postalOfficeBoxNumber: "6424",
       postalCode: "68138",
       streetAddress: "65044 Gutkowski Heights Apt. 741",
       contact_point_id: 1,
       created_at: "2019-12-06 15:13:14",
       updated_at: "2019-12-06 15:13:14",
       country: App\Country {#1018
         id: 1,
         name: "Canada",
         isoAlpha2: "PF",
         isoAlpha3: "GRL",
         created_at: "2019-12-06 15:13:14",
         updated_at: "2019-12-06 15:13:14",
       },
       contactPoint: App\ContactPoint {#1026
         id: 1,
         contactType: "Logistics",
         email: "[email protected]",
         faxNumber: "941-910-7739 x572",
         telephone: "1-368-817-4568",
         created_at: "2019-12-06 15:13:14",
         updated_at: "2019-12-06 15:13:14",
       },
     },
   }

Everything works fine for me, except when I add softDeletes to yaml file, then it shows the error as displayed in the screenshot.

I don't know if I not following the syntax of blueprint or what?

Here's an idea to 'type hint' the fakes in the draft.yaml

models:
  Country:
    name: string fake:country
    isoAlpha2: string:2 fake:countryCode
    isoAlpha3: string:3 fake:countryISOAlpha3
  ContactPoint:
    contactType: string fake:randomElement['Procurement', 'Sales', 'Logistics', 'Planning', 'HR', 'Helpdesk', 'PR']
    email: string fake:email
    faxNumber: string fake:phoneNumber
    telephone: string fake:phoneNumber
  PostalAddress:
    country_id: id:country
    addressLocality: string fake:city
    addressRegion: string fake:state
    postalOfficeBoxNumber: string fake:randomNumber(4)
    postalCode: string fake:postcode
    streetAddress: text fake:streetAddress
    contact_point_id: id:contact_point
  Person:
    additionalName: string fake:firstName
    address_id: id:postal_address
    familyName: string fake:lastName
    givenName: string fake:firstName

Just had a thought about this while i was using it -

Would it be worth having draft.yaml able to generated with the basic auth framework for a user (Users Table and password_resets table) as an example of how to use Blueprint, this could be a good reference for users just starting to use blueprint.

This could be an option flag i.e. -auth on the command.

First thanks for this amazing tool!!!

Idea:
Sometimes we have projects that are 80% similar (data / workflow) to something else we built in the past, it would be really cool if we can do the following:

• have a blueprint file reverse-engineered from an existing project
• import it to the new project
• change the 20% that is different and bootup the new project that way.

It would be awesome to be able to indicate a model is soft deletable, which would add the trait and the deleted_at table column.

It would be great if blueprints generate Nova resources as will

Just two rough logo ideas for Blueprint.
Probably not more than just some inspiration or base for discussion as they're just the obvious and super simple.

It would be great if policy generation was possible, maybe as config associated to the model definition? Perhaps even just calling php artisan make:policy {Model}Policy --model={Model}

When defining models it would be good to optimize any column data types and modifiers.

For example, the definition column: integer unsigned currently generates $table->integer('column')->unsigned(), but could be optimized to generate $table->unsignedInteger('column').

Additional optimizations can be inferred from the list of available data types.

This adds support of the id shorthand column type added in Laravel 7.

Since Blueprint will continue to support back to Laravel 6 (LTS) for a while, this should only do so for apps running Laravel 7 or higher.

Given the following draft.yaml which is modeled after a schema.org/Person:

models:
  Country:
    name: string
    isoAlpha2: string:2
    isoAlpha3: string:3
  ContactPoint:
    contactType: string
    email: string
    faxNumber: string
    telephone: string
  PostalAddress:
    country_id: id:country
    addressLocality: string
    addressRegion: string
    postalOfficeBoxNumber: string
    postalCode: string
    streetAddress: text
    contact_point_id: id:contact_point
  Person:
    additionalName: string
    address_id: id:postal_address
    familyName: string
    givenName: string

When running blueprint:build, it will yield migrations that do not have foreign key constraints for the related items.

Additionally: if the migrations would have had foreign key constraints, there would have been issues: the generated migrations all share the same timestamp in the filename, which means that they will be processed in alphabetical order, which will create a people table, before the postal_addresses table and therefore referencing a table that does not yet exists.

Example:

$ ls -1 database/migrations
...
2019_12_06_142421_create_contact_points_table.php
2019_12_06_142421_create_countries_table.php
2019_12_06_142421_create_people_table.php
2019_12_06_142421_create_postal_addresses_table.php

The order of the models: section was deliberate to make sure that the blueprint:build would generate in the correct order. However, since all of the generation occurred within the same second, the generated files fall back into alphabetical ordering.

So, implementing this would also require adding some kind of sequence numbering in the filename.

The generator that attaches relationships to models us cutting off the last 3 letters of the class. Traced it to this PR: #36

Issue is on this line:

It results in code that looks something like this (you can see each is missing characters at the end):

        'department' => factory(\App\Departm::class),
        'supervisor' => factory(\App\Supervi::class),
        'comments' => $faker->,
        'organization' => factory(\App\Organizat::class),
        'campus' => factory(\App\Cam::class),
        'building' => factory(\App\Build::class),

I wanted to post it as an issue to make sure I wasn't doing something wrong. I can't see why the column name is being cut here.

In retrospect filling fake data for nullable columns seems to be too explicit. Often these columns should be left undefined in the default factory and instead defined in a factory state.

This is open for discussion as it would be a breaking change based on original behavior.

While the Blueprint class has methods to register additional lexers and generators it is created within the command and as such not readily extensible.

Potential options:

• resolved as a singleton to share the same object with other components
• allow publishing of a configuration file to register additional functionality

The translation used by Blueprint is also used by the Model Factory Generator. I envision other packages benefiting from this data and as such would like to separate it to its own package and include it as a dependency.

i'm using this pacakge from awhile and it's excellent but after adding testing generators support it gives me an error about not found tests/Feature/Http/Controllers beucase it's not exist by default

Hey, thank you for your awesome work on blueprint it does wanders.

I've made some uuid and API resources work in my fork and was wandering do you want those changes in upstream?
(https://github.com/tectiv3/blueprint/tree/add-api-resources-support)
I can separate uuids, resources generation and foreign keys and indexes support into different PRs if you are interested.
Currently I've added support for uuid and json columns, index modifiers, api controllers, policies, resources generation and foreign keys in migrations.

models:
  Order:
    id: uuid primary
    user_id: id foreign:user
    external_id: string nullable index
    subscription_id: uuid foreign:subscription
    sku: string nullable
    status: enum:created,pending,paid,deleted,refunded default:'created'
    expires_at: timestamp nullable index
    meta: json default:'[]'
controllers:
  Order:
    index:
      query: user.all
      resource: order collection:paginate
    store:
      validate: type, name
      save: order
      fire: NewOrder with:order
      resource: order
    show:
      authorize: order
      resource: order
    destroy:
      authorize: order
      delete: order
      resource: empty
  Subscription:
    resource: api
    index:
      query: user.all
      resource: subscription collection:paginate

Once again, thank you for creating this tool!

Hi,

Great package love it, It would be great if it could also generate form boilerplate too.

Has anyone got any interest in this?

I see a stament

send: ReviewNotification to post.author

Which means that it need to Create a Notification class

send: ReviewMail to post.author

Which means that it need to Create a Mail class

If you create a User model and a Messages model, if "user_id" is a field in the SendMessagesRequest validator, it should be able to detect that "User" is an existing model, and require that the ID already exists in the "users" table.

Also, for the "validate:" line, there should be some kind of mark (I've used "(*)" in the example below, but it could be anything) that will tell the FormRequestGenerator to make the field unique.

For example:

models:
  Post:
    title: string:400
    content: longtext
    published_at: nullable timestamp

controllers:
  Post:
    index:
      query: all
      render: post.index with:posts

    store:
      validate: title(*), content

Thoughts?

When a "decimal" column is defined with precision and decimal attributes it would be possible to add a between rule.

For example, given the definition:

total: decimal:10,2

Blueprint could add between to generate the following rule:

'decimal' => 'required|numeric|between:0,999999.99'

Based on the controller definition, routes should also be generated. It would be nice if these routes were properly optimized as a resource where applicable.

While uuid column types are support, they are currently a bit verbose to define.

It would be nice instead of the current definition of id: uuid primary, to be able to simply use a shorthand of uuid that would expand to id: uuid

This implies the use of the primary modifier if a user were to define id: uuid.

I have a fresh installation of Laravel with blueprint installation after running the command

php artisan blueprint:build

Illuminate\Contracts\Filesystem\FileNotFoundException  : File does not exist at path draft.yaml

at D:\xampp\htdocs\websockets\vendor\laravel\framework\src\Illuminate\Filesystem\Filesystem.php:52
  48|         if ($this->isFile($path)) {
  49|             return $lock ? $this->sharedGet($path) : file_get_contents($path);
  50|         }
  51| 
> 52|         throw new FileNotFoundException("File does not exist at path {$path}");
  53|     }
  54| 
  55|     /**
  56|      * Get contents of a file with shared access.

Exception trace:

1   Illuminate\Filesystem\Filesystem::get("draft.yaml")
    D:\xampp\htdocs\websockets\vendor\laravel-shift\blueprint\src\BlueprintCommand.php:47

2   Blueprint\BlueprintCommand::handle()
    D:\xampp\htdocs\websockets\vendor\laravel\framework\src\Illuminate\Container\BoundMethod.php:32

Please use the argument -v to see more details.

I just want to know where to create the draft.yaml inside the Laravel Application

please include jasonmccreary/laravel-test-assertions package in composer.json of this package because it gives me an error

Before marking a v1.0, Blueprint needs to generate controllers supporting the following actions:

• validate
• query
• save
• update
• send
• queue
• flash
• store
• redirect
• display

I think it would be amazing if a kind of manifest file would be generated in the app store where the generated files of the blueprint are documented.
Later, if we regenerate the blueprint, we override all the files which are having the same hash sums like before, also we should scan the web.php for the occurrence of the lines we want to generate and then we should skip the lines in a given case.

Are pivot tables supported yet?

Blueprint only generates new Model components. As such, assumptions have been made about file naming, migration schema, and class existence.

Ideally, Blueprint would also append existing components with new, updated, or removed columns.

I create my .yaml file

models:
  Category:
    category_uuid: string:200
    name: string:200
    status: enum:ACTIVE,NONACTIVE
controllers:
  Category:
    index:
      query: all
      render: category.index with:categories
    create:
      render: category.create
    store:
      validate: category_uuid, name, status
      save: category
      flash: category
      redirect: category.index
    update:
      find: category
      redirect: category.index
    show:
      find: category
      render: category.show with:category
    edit:
      find: category
      render: category.edit with:category
    destroy:
      delete: category.id
      redirect: category.index

But when I run, factory file is fail

<?php

/** @var \Illuminate\Database\Eloquent\Factory $factory */

use App\Category;
use Faker\Generator as Faker;

$factory->define(Category::class, function (Faker $faker) {
    return [
        'category_uuid' => $faker->word,
        'name' => $faker->name,
        'status' => $faker->,
    ];
});

As you see, faker for enum is fail. Am I missing something?