media_types-serialization 0.8.0 → 1.1.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (47) hide show
  1. checksums.yaml +4 -4
  2. data/.github/workflows/ci.yml +16 -3
  3. data/.prettierrc +1 -0
  4. data/CHANGELOG.md +42 -0
  5. data/CODE_OF_CONDUCT.md +74 -74
  6. data/Gemfile.lock +74 -83
  7. data/README.md +691 -179
  8. data/lib/media_types/problem.rb +64 -0
  9. data/lib/media_types/serialization.rb +497 -173
  10. data/lib/media_types/serialization/base.rb +115 -91
  11. data/lib/media_types/serialization/error.rb +186 -0
  12. data/lib/media_types/serialization/fake_validator.rb +52 -0
  13. data/lib/media_types/serialization/serialization_dsl.rb +117 -0
  14. data/lib/media_types/serialization/serialization_registration.rb +245 -0
  15. data/lib/media_types/serialization/serializers/api_viewer.rb +133 -0
  16. data/lib/media_types/serialization/serializers/common_css.rb +168 -0
  17. data/lib/media_types/serialization/serializers/endpoint_description_serializer.rb +80 -0
  18. data/lib/media_types/serialization/serializers/fallback_not_acceptable_serializer.rb +85 -0
  19. data/lib/media_types/serialization/serializers/fallback_unsupported_media_type_serializer.rb +58 -0
  20. data/lib/media_types/serialization/serializers/input_validation_error_serializer.rb +89 -0
  21. data/lib/media_types/serialization/serializers/problem_serializer.rb +100 -0
  22. data/lib/media_types/serialization/utils/accept_header.rb +77 -0
  23. data/lib/media_types/serialization/utils/accept_language_header.rb +82 -0
  24. data/lib/media_types/serialization/utils/header_list.rb +89 -0
  25. data/lib/media_types/serialization/version.rb +1 -1
  26. data/media_types-serialization.gemspec +48 -50
  27. metadata +48 -79
  28. data/.travis.yml +0 -17
  29. data/lib/generators/media_types/serialization/api_viewer/api_viewer_generator.rb +0 -25
  30. data/lib/generators/media_types/serialization/api_viewer/templates/api_viewer.html.erb +0 -98
  31. data/lib/generators/media_types/serialization/api_viewer/templates/initializer.rb +0 -33
  32. data/lib/generators/media_types/serialization/api_viewer/templates/template_controller.rb +0 -23
  33. data/lib/media_types/serialization/media_type/register.rb +0 -4
  34. data/lib/media_types/serialization/migrations_command.rb +0 -38
  35. data/lib/media_types/serialization/migrations_support.rb +0 -50
  36. data/lib/media_types/serialization/mime_type_support.rb +0 -64
  37. data/lib/media_types/serialization/no_content_type_given.rb +0 -11
  38. data/lib/media_types/serialization/no_media_type_serializers.rb +0 -11
  39. data/lib/media_types/serialization/no_serializer_for_content_type.rb +0 -15
  40. data/lib/media_types/serialization/renderer.rb +0 -41
  41. data/lib/media_types/serialization/renderer/register.rb +0 -4
  42. data/lib/media_types/serialization/wrapper.rb +0 -13
  43. data/lib/media_types/serialization/wrapper/html_wrapper.rb +0 -45
  44. data/lib/media_types/serialization/wrapper/media_collection_wrapper.rb +0 -59
  45. data/lib/media_types/serialization/wrapper/media_index_wrapper.rb +0 -59
  46. data/lib/media_types/serialization/wrapper/media_object_wrapper.rb +0 -55
  47. data/lib/media_types/serialization/wrapper_support.rb +0 -38
data/README.md CHANGED
@@ -4,7 +4,7 @@
4
4
  [![Gem Version](https://badge.fury.io/rb/media_types-serialization.svg)](https://badge.fury.io/rb/media_types-serialization)
5
5
  [![MIT license](http://img.shields.io/badge/license-MIT-brightgreen.svg)](http://opensource.org/licenses/MIT)
6
6
 
7
- Add media types supported serialization using your favourite serializer
7
+ `respond_to` on steroids. Add [HATEOAS](https://docs.delftsolutions.nl/wiki/HATEOAS_API) compatible serialization and deserialization to your Rails projects.
8
8
 
9
9
  ## Installation
10
10
 
@@ -21,275 +21,787 @@ And then execute:
21
21
  Or install it yourself as:
22
22
 
23
23
  $ gem install media_types-serialization
24
-
25
- If you have not done this before, and you're using `rails`, install the necessary parts using:
26
24
 
27
- ```bash
28
- rails g media_types:serialization:api_viewer
25
+ ## Usage
26
+
27
+ Serializers help you in converting a ruby object to a representation matching a specified [Media Type validator](https://github.com/SleeplessByte/media-types-ruby) and the other way around.
28
+
29
+ ### Creating a serializer
30
+
31
+ ```ruby
32
+ class BookSerializer < MediaTypes::Serialization::Base
33
+ unvalidated 'application/vnd.acme.book'
34
+
35
+ # outputs with a Content-Type of application/vnd.acme.book.v1+json
36
+ output version: 1 do |obj, version, context|
37
+ {
38
+ book: {
39
+ title: obj.title
40
+ }
41
+ }
42
+ end
43
+ end
29
44
  ```
30
45
 
31
- This will:
46
+ To convert a ruby object to a json representation:
32
47
 
33
- - Add the default `html_wrapper` layout which is an API Viewer used as fallback or the `.api_viewer` format
34
- - Add the default `template_controller` which allows the API Viewer to post templated links
35
- - Add the `route` for these templated link forms
36
- - Add an initializer that registers the `media` renderer and `api_viewer` media type
48
+ ```ruby
49
+ class Book
50
+ attr_accessor :title
51
+ end
37
52
 
38
- ## Usage
53
+ book = Book.new
54
+ book.title = 'Everything, abridged'
55
+
56
+ BookSerializer.serialize(book, 'vnd.acme.book.v1+json', context: nil)
57
+ # => { "book": { "title": "Everything, abridged" } }
58
+ ```
39
59
 
40
- In order to use media type serialization you only need to do 2 things:
60
+ ### Controller integration
41
61
 
42
- ### Serializer
62
+ You can integrate the serialization system in rails, giving you automatic [Content-Type negotiation](https://en.wikipedia.org/wiki/Content_negotiation) using the `Accept` header:
43
63
 
44
- Add a serializer that can serialize a certain media type. The `to_hash` function will be called _explicitly_ in your
45
- controller, so you can always use your own, favourite serializer here to do the hefty work. This gem does provide some
46
- easy tools, usually enough to do most serialization.
47
-
48
64
  ```ruby
49
- class Book < ApplicationRecord
50
- class Serializer < MediaTypes::Serialization::Base
51
- serializes_media_type MyNamespace::MediaTypes::Book
52
-
53
- def fields
54
- if current_media_type.create?
55
- return %i[name author]
56
- end
57
-
58
- %i[name author updated_at views]
59
- end
65
+ require 'media_types/serialization'
66
+
67
+ class BookController < ActionController::API
68
+ include MediaTypes::Serialization
69
+
70
+ allow_output_serializer(BookSerializer, only: %i[show])
71
+ freeze_io!
72
+
73
+ def show
74
+ book = Book.new
75
+ book.title = 'Everything, abridged'
76
+
77
+ render_media book
78
+ end
79
+ end
80
+ ```
81
+
82
+ While using the controller integration the context will always be set to the current controller. This allows you to construct urls.
83
+
84
+ ### Adding HATEOAS responses to existing routes
85
+
86
+ When creating a mobile application it's often useful to allow the app to request a non-html representation of a specific url. If you have an existing route:
87
+
88
+ ```ruby
89
+ class BookController < ApplicationController
90
+ def show
91
+ @book = Book.new
92
+
93
+ # Use view corresponding to the controller
94
+ end
95
+ end
96
+ ```
97
+
98
+ You can add a json representation as follows:
99
+
100
+ ```ruby
101
+ class BookController < ApplicationController
102
+ allow_output_serializer(BookSerializer, only: %i[show])
103
+ allow_output_html
104
+ freeze_io!
105
+
106
+ def show
107
+ @book = Book.new
60
108
 
61
- def to_hash
62
- extract(serializable, fields).tap do |result|
63
- result[:_links] = extract_links unless current_media_type.create?
109
+ render_media @book
110
+ end
111
+ end
112
+ ```
113
+
114
+ ### Validations
115
+
116
+ Right now the serializer does not validate incoming or outgoing information. This can cause issues when you accidentally emit non-conforming data that people start to depend on. To make sure you don't do that you can specify a [Media Type validator](https://github.com/SleeplessByte/media-types-ruby):
117
+
118
+ ```ruby
119
+ require 'media_types'
120
+
121
+ class BookValidator
122
+ include MediaTypes::Dsl
123
+
124
+ def self.organisation
125
+ 'acme'
126
+ end
127
+
128
+ use_name 'book'
129
+
130
+ validations do
131
+ version 1 do
132
+ attribute :book do
133
+ attribute :title, String
64
134
  end
65
135
  end
136
+ end
137
+ end
66
138
 
67
- alias to_h to_hash
68
-
69
- protected
70
-
71
- def extract_self
72
- # A serializer gets the controller as context
73
- { href: context.api_book_url(serializable) }
74
- end
139
+ class BookSerializer < MediaTypes::Serialization::Base
140
+ validator BookValidator
75
141
 
76
- def extract_links
77
- {
78
- 'self': extract_self,
79
- 'signatures': { href: context.api_book_signatures_url(serializable) }
142
+ # outputs with a Content-Type of application/vnd.acme.book.v1+json
143
+ output version: 1 do |obj, version, context|
144
+ {
145
+ book: {
146
+ title: obj.title
80
147
  }
148
+ }
149
+ end
150
+ end
151
+ ```
152
+
153
+ For more information, see the [Media Types docs](https://github.com/SleeplessByte/media-types-ruby).
154
+
155
+ ### Versioning
156
+
157
+ To help with supporting older versions, serializers have a [DSL](https://en.wikipedia.org/wiki/Domain-specific_language) to construct json objects:
158
+
159
+ ```ruby
160
+ class BookSerializer < MediaTypes::Serialization::Base
161
+ validator BookValidator
162
+
163
+ output versions: [1, 2] do |obj, version, context|
164
+ attribute :book do
165
+ attribute :title, obj.title
166
+ attribute :description, obj.description if version >= 2
81
167
  end
82
168
  end
83
169
  end
84
170
  ```
85
- By default, The passed in `MediaType` gets converted into a constructable (via `to_constructable`) and invoked with the
86
- current `view` (e.g. `create`, `index`, `collection` or ` `). This means that by default it will be able to serialize
87
- the latest version you `MediaType` is reporting. The best way to supply your media type is via the [`media_types`](https://github.com/SleeplessByte/media-types-ruby) gem.
88
171
 
89
- #### Multiple suffixes, one serializer
172
+ ```ruby
173
+ BookSerializer.serialize(book, BookValidator.version(1), context: nil)
174
+ # => { "book": { "title": "Everything, abridged" } }
175
+
176
+ BookSerializer.serialize(book, BookValidator.version(2), context: nil)
177
+ # => { "book": { "title": "Everything, abridged", "description": "Mu" } }
178
+ ```
179
+
180
+ ### Links
90
181
 
91
- By default, the media renderer will automatically detect and inject the following:
92
- - suffix `+json` if you define `to_json`
93
- - suffix `+xml` if you define `to_xml`
94
- - type `text/html` if you define `to_html`
182
+ When making [HATEOAS](https://docs.delftsolutions.nl/wiki/HATEOAS_API) compliant applications it's very useful to include `Link` headers in your response so clients can use a `HEAD` request instead of having to fetch the entire resource. Serializers have convenience methods to help with this:
95
183
 
96
- If you do _not_ define these methods, only the `default` suffix / type will be used, `accepts_html` for the `text/html`
97
- content-type.
184
+ ```ruby
185
+ class BookSerializer < MediaTypes::Serialization::Base
186
+ validator BookValidator
98
187
 
99
- If you don't define `to_html`, but try to make a serializer output `html`, it will be rendered in the layout at:
100
- `serializers/wrapper/html_wrapper.html.erb` (or any other templating extension).
188
+ output versions: [1, 2, 3] do |obj, version, context|
189
+ attribute :book do
190
+ link :self, href: context.book_url(obj) if version >= 3
101
191
 
102
- #### Migrations (versions)
192
+ attribute :title, obj.title
193
+ attribute :description, obj.description if version >= 2
194
+ end
195
+ end
196
+ end
197
+ ```
103
198
 
104
- If the serializer can serialize multiple _versions_, you can supply them through `additional_versions: [2, 3]`. A way to
105
- handle this is via backward migrations, meaning you'll migrate from the current version back to an older version.
199
+ This returns the following response:
106
200
 
107
201
  ```ruby
108
- class Book < ApplicationRecord
109
- class BasicSerializer < MediaTypes::Serialization::Base
110
-
111
- # Maybe it's currently at version 2, so tell the base that this also serializes version 1
112
- # You can also use a range to_a: (1...4).to_a
113
- #
114
- serializes_media_type MyNamespace::MediaTypes::Book, additional_versions: [1]
115
-
116
- def to_hash
117
- # This enables migrations right when it's being serialized
118
- #
119
- migrate do
120
- extract(serializable, fields).tap do |result|
121
- result[:_links] = extract_links unless current_media_type.create?
122
- end
123
- end
202
+ BookSerializer.serialize(book, BookValidator.version(3), context: controller)
203
+ # header = Link: <https://example.org/>; rel="self"
204
+ # => {
205
+ # "book": {
206
+ # "_links": {
207
+ # "self": { "href": "https://example.org" }
208
+ # },
209
+ # "title": "Everything, abridged",
210
+ # "description": "Mu"
211
+ # }
212
+ # }
213
+ ```
214
+
215
+ ### Collections
216
+
217
+ There are convenience methods for serializing arrays of objects based on a template.
218
+
219
+ #### Indexes
220
+
221
+ An index is a collection of urls that point to members of the array. The index method automatically generates it based on the self links defined in the default view of the same version.
222
+
223
+ ```ruby
224
+ class BookSerializer < MediaTypes::Serialization::Base
225
+ validator BookValidator
226
+
227
+ output versions: [1, 2, 3] do |obj, version, context|
228
+ attribute :book do
229
+ link :self, href: context.book_url(obj) if version >= 3
230
+
231
+ attribute :title, obj.title
232
+ attribute :description, obj.description if version >= 2
124
233
  end
125
-
126
- # This defines migrations. You can use classes, commands or anything else to execute this code
127
- # but inline migrations work fine if you don't have a lot of them.
128
- backward_migrations do
129
-
130
- # This is called if the version requested is 1 _or_ lower. This means you can compose your migrations. The
131
- # migrations with a _lower_ version than the requested version are NOT executed.
132
- version 1 do |result|
133
- result.tap do |r|
134
- if r.key?(:views)
135
- r[:views_count] = r.delete(:views)
136
- end
137
- end
138
- end
234
+ end
235
+
236
+ output view: :index, version: 3 do |arr, version, context|
237
+ attribute :books do
238
+ link :self, href: context.book_index_url
239
+
240
+ index arr, version: version
139
241
  end
140
242
  end
141
243
  end
142
244
  ```
143
245
 
144
- ### Controller
246
+ ```ruby
247
+ BookSerializer.serialize([book], BookValidator.view(:index).version(3), context: controller)
248
+ # header = Link: <https://example.org/index>; rel="self"
249
+ # => {
250
+ # "books": {
251
+ # "_links": {
252
+ # "self": { "href": "https://example.org" }
253
+ # },
254
+ # "_index": [
255
+ # { "href": "https://example.org" }
256
+ # ]
257
+ # }
258
+ # }
259
+ ```
260
+
261
+ #### Collections
145
262
 
146
- In your base controller, or wherever you'd like, include the `MediaTypes::Serialization` concern. In the controller that
147
- uses the serialization, you need to explicitly `accept` it if you want to use the built-in lookups.
263
+ A collection inlines the member objects. The collection method automatically generates it based on the default view of the same version.
148
264
 
149
265
  ```ruby
150
- require 'media_types/serialization'
151
- require 'media_types/serialization/renderer/register'
266
+ class BookSerializer < MediaTypes::Serialization::Base
267
+ validator BookValidator
152
268
 
153
- class ApiController < ActionController::API
154
- include MediaTypes::Serialization
269
+ output versions: [1, 2, 3] do |obj, version, context|
270
+ attribute :book do
271
+ link :self, href: context.book_url(obj) if version >= 3
272
+
273
+ attribute :title, obj.title
274
+ attribute :description, obj.description if version >= 2
275
+ end
276
+ end
277
+
278
+ output view: :index, version: 3 do |arr, version, context|
279
+ attribute :books do
280
+ link :self, href: context.book_index_url
281
+
282
+ index arr, version: version
283
+ end
284
+ end
285
+
286
+ output view: :collection, version: 3 do |arr, version, context|
287
+ attribute :books do
288
+ link :self, href: context.book_collection_url
289
+
290
+ collection arr, version: version
291
+ end
292
+ end
293
+ end
294
+ ```
295
+
296
+ ```ruby
297
+ BookSerializer.serialize([book], BookValidator.view(:collection).version(3), context: controller)
298
+ # header = Link: <https://example.org/collection>; rel="self"
299
+ # => {
300
+ # "books": {
301
+ # "_links": {
302
+ # "self": { "href": "https://example.org" }
303
+ # },
304
+ # "_embedded": [
305
+ # {
306
+ # "_links": {
307
+ # "self": { "href": "https://example.org" }
308
+ # },
309
+ # "title": "Everything, abridged",
310
+ # "description": "Mu"
311
+ # }
312
+ # ]
313
+ # }
314
+ # }
315
+ ```
316
+
317
+ ### Input deserialization
318
+
319
+ You can mark a media type as something that's allowed to be sent along with a PUT request as follows:
320
+
321
+ ```ruby
322
+ class BookSerializer < MediaTypes::Serialization::Base
323
+ validator BookValidator
324
+
325
+ output versions: [1, 2, 3] do |obj, version, context|
326
+ attribute :book do
327
+ link :self, href: context.book_url(obj) if version >= 3
328
+
329
+ attribute :title, obj.title
330
+ attribute :description, obj.description if version >= 2
331
+ end
332
+
333
+ input version: 3
155
334
  end
156
335
 
157
- class BookController < ApiController
158
-
159
- accept_serialization(Book::BasicSerializer, accept_html: false, only: %i[show])
160
- accept_html(Book::CoverHtmlSerializer, only: %i[show])
161
- freeze_accepted_media!
162
-
163
- def show
164
- # If you do NOT pass in the content_type, it will re-use the current content_type of the response if set or
165
- # use the default content type of the serializer. This is fine if you only output one Content-Type in the
166
- # action, but not if you are relying on content-negotiation.
167
-
168
- render media: serialize_media(@book), content_type: request.format.to_s
336
+ class BookController < ActionController::API
337
+ include MediaTypes::Serialization
338
+
339
+ allow_output_serializer(BookSerializer, only: %i[show])
340
+ allow_input_serializer(BookSerializer, only: %i[create])
341
+ freeze_io!
342
+
343
+ def show
344
+ book = Book.new
345
+ book.title = 'Everything, abridged'
346
+
347
+ render media: serialize_media(book), content_type: request.format.to_s
348
+ end
349
+
350
+ def create
351
+ json = deserialize(request, context: self) # does validation for us
352
+ puts json
169
353
  end
170
354
  end
171
355
  ```
172
356
 
173
- If you have normalized your resources (e.g. into `@resource`), you can add a `render_media` method to your
174
- `BaseController` and render resources like so:
357
+ If you use [ActiveRecord](https://guides.rubyonrails.org/active_record_basics.html) you might want to convert the verified json data during deserialization:
175
358
 
176
359
  ```ruby
177
- class ApiController < ActionController::API
178
- def render_media(**opts)
179
- render media: serialize_media(@resource), content_type: request.format.to_s, **opts
360
+ class BookSerializer < MediaTypes::Serialization::Base
361
+ validator BookValidator
362
+
363
+ output versions: [1, 2, 3] do |obj, version, context|
364
+ attribute :book do
365
+ link :self, href: context.book_url(obj) if version >= 3
366
+
367
+ attribute :title, obj.title
368
+ attribute :description, obj.description if version >= 2
369
+ end
370
+
371
+ input versions: [1, 2, 3] do |json, version, context|
372
+ book = Book.new
373
+ book.title = json['book']['title']
374
+ book.description = 'Not available'
375
+ book.description = json['book']['description'] if version >= 2
376
+
377
+ # Best practise is to only save in the controller.
378
+ book
379
+ end
380
+ end
381
+
382
+ class BookController < ActionController::API
383
+ include MediaTypes::Serialization
384
+
385
+ allow_output_serializer(BookSerializer, only: %i[show])
386
+ allow_input_serializer(BookSerializer, only: %i[create])
387
+ freeze_io!
388
+
389
+ def show
390
+ book = Book.new
391
+ book.title = 'Everything, abridged'
392
+
393
+ render media: serialize_media(book), content_type: request.format.to_s
394
+ end
395
+
396
+ def create
397
+ book = deserialize(request, context: self)
398
+ book.save!
399
+
400
+ render media: serialize_media(book), content_type request.format.to_s
180
401
  end
181
402
  end
182
403
  ```
183
404
 
184
- And then call `render_media` whenever you're ready to render.
405
+ If you don't want to apply any input validation or deserialization you can use the `allow_all_input` method instead of `allow_input_serialization`.
185
406
 
186
- ### HTML output
407
+ ### Raw output
187
408
 
188
- You can define HTML outputs for example by creating a serializer that accepts `text/html`. At this moment, there may
189
- only be one (1) active `text/html` serializer for each action; a single controller can have multiple registered, but
190
- never for the same preconditions in `before_action` (because how else would it know which one to pick?).
409
+ Sometimes you need to output raw data. This cannot be validated. You do this as follows:
191
410
 
192
- Use the `render` method to generate your HTML:
193
411
  ```ruby
194
- class Book::CoverHtmlSerializer < MediaTypes::Serialization::Base
195
- # Tell the serializer that this accepts HTML, but this is also signaled by `to_html`
196
- serializes_html
197
-
198
- def to_html
199
- ApplicationController.render(
200
- 'serializers/book/cover',
201
- assigns: {
202
- title: extract_title,
203
- image: resolve_file_url(covers.first&.version_url('small')),
204
- description: extract_description,
205
- language_links: language_links,
206
- },
207
- layout: false
208
- )
412
+ class BookSerializer < MediaTypes::Serialization::Base
413
+ validator BookValidator
414
+
415
+ output_raw view: :raw, version: 3 do |obj, version, context|
416
+ hidden do
417
+ # Make sure links are only set in the headers, not in the body.
418
+
419
+ link :self, href: context.book_url(obj)
420
+ end
421
+
422
+ "I'm a non-json output"
209
423
  end
210
-
211
- # Naturally you have to define extract_title, etc etc
212
424
  end
213
425
  ```
214
426
 
215
- You can change the default `wrapper` / `to_html` implementation by setting:
427
+ ### Raw input
428
+
429
+ You can do the same with input:
216
430
 
217
431
  ```ruby
218
- ::MediaTypes::Serialization.html_wrapper_layout = '/path/to/wrapper/layout'
432
+ class BookSerializer < MediaTypes::Serialization::Base
433
+ validator BookValidator
434
+
435
+ input_raw view: raw, version: 3 do |bytes, version, context|
436
+ book = Book.new
437
+ book.description = bytes
438
+
439
+ book
440
+ end
441
+ end
219
442
  ```
220
443
 
221
- ### API viewer
444
+ ### Remapping media type identifiers
222
445
 
223
- There is a special media type exposed by this gem at `::MediaTypes::Serialization::MEDIA_TYPE_API_VIEWER`. If you're
224
- using `rails` you'll want to register it. You can do so manually, or by `require`ing:
446
+ Sometimes you already have old clients using an `application/json` media type identifier when they do requests. While this is not a good practise as this makes it hard to add new fields or remove old ones, this library has support for migrating away:
225
447
 
226
448
  ```ruby
227
- require 'media_types/serialization/media_type/register'
449
+ class BookSerializer < MediaTypes::Serialization::Base
450
+ validator BookValidator
451
+
452
+ output versions: [1, 2, 3] do |obj, version, context|
453
+ attribute :book do
454
+ link :self, href: context.book_url(obj) if version >= 3
455
+
456
+ attribute :title, obj.title
457
+ attribute :description, obj.description if version >= 2
458
+ end
459
+ end
460
+ output_alias 'application/json' # maps application/json to to applicaton/vnd.acme.book.v1+json
461
+
462
+ input view: :create, versions: [1, 2, 3] do |json, version, context|
463
+ book = Book.new
464
+ book.title = json['book']['title']
465
+ book.description = 'Not available'
466
+ book.description = json['book']['description'] if version >= 2
467
+
468
+ # Make sure not to save here but only save in the controller
469
+ book
470
+ end
471
+ input_alias 'application/json', view: :create # maps application/json to to applicaton/vnd.acme.book.v1+json
228
472
  ```
229
473
 
230
- If you do so, the `.api_viewer` format becomes available for all actions that call into `render media:`.
474
+ Validation will be done using the remapped validator. Aliasses map to version `nil` if that is available or `1` otherwise. It is not possible to configure this version.
231
475
 
232
- You can change the default `wrapper` implementation by setting:
476
+ ### HTML
477
+
478
+ This library has a built in API viewer. The viewer can be accessed by by appending a `?api_viewer=last` query parameter to the URL.
479
+
480
+ To enable the API viewer, use: `allow_api_viewer` in the controller.
233
481
 
234
482
  ```ruby
235
- ::MediaTypes::Serialization.api_viewer_layout = '/path/to/wrapper/layout'
236
- ```
483
+ class BookController < ActionController::API
484
+ include MediaTypes::Serialization
237
485
 
238
- ### Wrapping output
486
+ allow_api_viewer
239
487
 
240
- By convention, `index` views are wrapped in `_index: [items]`, `collection` views are wrapped in `_embedded: [items]`
241
- and `create` / no views are wrapped in `[ROOT_KEY]: item`. This is currently only enabled for `to_json` serialization
242
- but planned for `xml` as well.
488
+ allow_output_serializer(BookSerializer, only: %i[show])
489
+ allow_input_serializer(BookSerializer, only: %i[create])
490
+ freeze_io!
243
491
 
244
- This behaviour can not be turned of as of writing. However, you may _overwrite_ this behaviour via:
492
+ def show
493
+ book = Book.new
494
+ book.title = 'Everything, abridged'
245
495
 
246
- - `self.root_key(view:)`: to define the root key for a specific `view`
247
- - `self.wrap(serializer, view: nil)`: to define the wrapper for a specific `view` and/or `serializer`. For example, if
248
- you never want to wrap anything, you could define:
249
- ```ruby
250
- def self.wrap(serializer, view: nil)
251
- serializer
496
+ render media: serialize_media(book), content_type: request.format.to_s
252
497
  end
253
- ```
254
498
 
255
- ### Link header
499
+ def create
500
+ json = deserialize(request, context: self) # does validation for us
501
+ puts json
502
+ end
503
+ end
504
+ ```
256
505
 
257
- You can use `to_link_header` to generate a header value for the `Link` header.
506
+ You can also output custom HTML:
258
507
 
259
508
  ```ruby
260
- entries = @last_media_serializer.to_link_header
261
- if entries.present?
262
- response.header[HEADER_LINK] = entries
509
+ class BookSerializer < MediaTypes::Serialization::Base
510
+ validator BookValidator
511
+
512
+ output versions: [1, 2, 3] do |obj, version, context|
513
+ attribute :book do
514
+ link :self, href: context.book_url(obj) if version >= 3
515
+
516
+ attribute :title, obj.title
517
+ attribute :description, obj.description if version >= 2
518
+ end
519
+ end
520
+
521
+ output_raw view: :html do |obj, context|
522
+ render_view 'book/show', context: context, assigns: {
523
+ title: obj.title,
524
+ description: obj.description
525
+ }
526
+ end
527
+
528
+ output_alias 'text/html', view: :html
263
529
  end
264
530
  ```
265
531
 
266
- If you want the link header to be different from the `_links`, you can implement `header_links(view:)` next to
267
- `extract_links(view:)`. This will be called by the `to_link_header` function.
532
+ #### Errors
268
533
 
269
- ### Validation
270
- If you only have `json`/`xml`/structured data responses and you want to use [`media_types-validation`](https://github.com/XPBytes/media_types-validation) in conjunction with this gem, you can create a concern or add the following two functions to your base controller:
534
+ This library adds support for returning errors to clients using the [`application/problem+json`](https://tools.ietf.org/html/rfc7231) media type. You can catch and transform application errors by adding an `output_error` call before `freeze_io!`:
271
535
 
272
536
  ```ruby
273
- def render_media(resource = @resource, **opts)
274
- serializer = serialize_media(resource)
275
- render media: serializer, content_type: request.format.to_s, **opts
276
- validate_media(serializer)
277
- end
537
+ class BookController < ActionController::API
538
+ include MediaTypes::Serialization
278
539
 
279
- def validate_media(serializer = @last_media_serializer)
280
- media_type = serializer.current_media_type
281
- return true unless media_type && response_body
282
- validate_json_with_media_type(serializer.to_hash, media_type: media_type)
540
+ output_error CanCan::AccessDenied do |p, error|
541
+ p.title 'You do not have enough permissions to perform this action.', lang: 'en'
542
+ p.title 'Je hebt geen toestemming om deze actie uit te voeren.', lang: 'nl-NL'
543
+
544
+ p.status_code :forbidden
545
+ end
546
+
547
+ freeze_io!
548
+
549
+ # ...
283
550
  end
284
551
  ```
285
552
 
286
- As long as the serializer has a `to_json` or `to_hash`, this will work -- but also means that the data will always be validate _as if_ it were json. This covers most use cases.
553
+ The exception you specified will be rescued by the controller and will be displayed to the user along with a link to the shared wiki page for that error type. Feel free to add instructions there on how clients should solve this problem. You can find more information at: [https://docs.delftsolutions.nl/wiki/Error](https://docs.delftsolutions.nl/wiki/Error)
554
+ If you want to override this url you can use the `p.url(href)` function.
555
+
556
+ By default the `message` property of the error is used to fill the `details` field. You can override this by using the `p.override_details(description, lang:)` function.
557
+
558
+ Custom attributes can be added using the `p.attribute(name, value)` function.
287
559
 
288
560
  ### Related
289
561
 
290
- - [`MediaTypes`](https://github.com/SleeplessByte/media-types-ruby): :gem: Library to create media type definitions, schemes and validations
291
- - [`MediaTypes::Deserialization`](https://github.com/XPBytes/media_types-deserialization): :cyclone: Add media types supported deserialization using your favourite parser, and media type validation.
292
- - [`MediaTypes::Validation`](https://github.com/XPBytes/media_types-validation): :heavy_exclamation_mark: Response validations according to a media-type
562
+ - [`MediaTypes`](https://github.com/SleeplessByte/media-types-ruby): :gem: Library to create media type validators.
563
+
564
+ ## API
565
+
566
+ ### Serializer definition
567
+
568
+ These methods become available during class definition if you inherit from `MediaTypes::Serialization::Base`.
569
+
570
+ #### `unvalidated( prefix )`
571
+
572
+ Disabled validation for this serializer. Prefix is of the form `application/vnd.<organisation>.<name>`.
573
+
574
+ Either unvalidated or validator must be used while defining a serializer.
575
+
576
+ #### `validator( media_type_validator )`
577
+
578
+ Enabled validation for this serializer using a [Media Type Validator](https://github.com/SleeplessByte/media-types-ruby).
579
+
580
+ Either validator or unvalidated must be used while defining a serializer.
581
+
582
+ #### `output( view:, version:, versions: ) do |obj, version, context|`
583
+
584
+ Defines a serialization block. Either version or versions can be set. View should be a symbol or unset.
585
+
586
+ Obj is the object to be serialized, version is the negotiated version and context is the context passed in from the serialize function. When using the controller integration, context is the current controller.
587
+
588
+ The block should return an object to convert into JSON.
589
+
590
+ #### `output_raw( view:, version:, versions: ) do |obj, version, context|`
591
+
592
+ This has the same behavior as `output` but should return a string instead of an object. Output is not validated.
593
+
594
+ #### `output_alias( media_type_identifier, view:, hide_variant: false )`
595
+
596
+ Defines a legacy mapping. This will make the deserializer parse the media type `media_type_identifier` as if it was version `nil` of the specified view. If view is undefined it will use the output serializer without a view defined.
597
+
598
+ Response will have a content type equal to `[media_type_identifier]; variant=[mapped_media_type_identifier]`. If `hide_variant:` is true, the content type emitted will only be `[media_type_identifier]`.
599
+
600
+ #### `output_alias_optional( media_type_identifier, view:, hide_variant: false )`
601
+
602
+ Has the same behavior as `output_alias` but can be used by multiple serializers. The serializer that is loaded last in the controller 'wins' control over this media type identifier. If any of the serializers have an `output_alias` defined with the same media type identifier that one will win instead.
603
+
604
+ Response will have a content type equal to `[media_type_identifier]; variant=[mapped_media_type_identifier]`. If `hide_variant:` is true, the content type emitted will only be `[media_type_identifier]`.
605
+
606
+ #### `input( view:, version:, versions: ) do |obj, version, context|`
607
+
608
+ Defines a deserialization block. Either version or versions can be set. View should be a symbol or unset.
609
+
610
+ Obj is the object to be serialized, version is the negotiated version and context is the context passed in from the serialize function. When using the controller integration, context is the current controller.
611
+
612
+ The block should return the internal representation of the object. Best practise is to make sure not to change state in this function but to leave that up to the controller.
613
+
614
+ #### `input_raw( view:, version:, versions: ) do |bytes, version, context|`
615
+
616
+ This has the same behavior as `input` but takes in raw data. Input is not validated.
617
+
618
+ #### `input_alias( media_type_identifier, view: )`
619
+
620
+ Defines a legacy mapping. This will make the serializer parse the media type `media_type_identifier` as if it was version 1 of the specified view. If view is undefined it will use the input serializer without a view defined.
621
+
622
+ #### `input_alias_optional( media_type_identifier, view: )`
623
+
624
+ Has the same behavior as `input_alias` but can be used by multiple serializers. The serializer that is loaded last in the controller 'wins' control over this media type identifier. If any of the serializers have an `input_alias` defined with the same media type identifier that one will win instead.
625
+
626
+ #### `disable_wildcards`
627
+
628
+ Disables registering wildcard media types.
629
+
630
+ ### Serializer definition
631
+
632
+ The following methods are available within an `output ... do` block.
633
+
634
+ #### `attribute( key, value = {} ) do`
635
+
636
+ Sets a value for the given key. If a block is given, any `attribute`, `link`, `collection` and `index` statements are run in context of `value`.
637
+
638
+ Returns the built up context so far.
639
+
640
+ #### `link( rel, href:, emit_header: true, **attributes )`
641
+
642
+ Adds a `_link` block to the current context. Also adds the specified link to the HTTP Link header. `attributes` allows passing in custom attributes.
643
+
644
+ If `emit_header` is `true` the link will also be emitted as a http header.
645
+
646
+ Returns the built up context so far.
647
+
648
+ #### `index( array, serializer, version:, view: nil )`
649
+
650
+ Adds an `_index` block to the current context. Uses the self links of the specified view to construct an index of urls to the child objects.
651
+
652
+ Returns the built up context so far.
653
+
654
+ #### `collection( array, serializer, version:, view: nil )`
655
+
656
+ Adds an `_embedded` block to the current context. Uses the specified serializer to embed the child objects.
657
+ Optionally a block can be used to modify the output from the child serializer.
658
+
659
+ Returns the built up context so far.
660
+
661
+ #### `hidden do`
662
+
663
+ Sometimes you want to add links without actually modifying the object. Calls to `attribute`, `link`, `index`, `collection` made inside this block won't modify the context. Any calls to link will only set the HTTP Link header.
664
+
665
+ Returns the unmodified context.
666
+
667
+ #### `emit`
668
+
669
+ Can be added to the end of a block to fix up the return value to return the built up context so far.
670
+
671
+ Returns the built up context so far.
672
+
673
+ #### `object do`
674
+
675
+ Runs a block in a new context and returns the result
676
+
677
+ #### `render_view( view, context:, **args)`
678
+
679
+ Can be used to render a view. You can set local variables in the view by assigning a hash to the `assigns:` parameter.
680
+
681
+ ### Controller definition
682
+
683
+ These functions are available during the controller definition if you add `include MediaTypes::Serialization`.
684
+
685
+ #### `allow_output_serializer( serializer, views: nil, **filters )`
686
+
687
+ Configure the controller to allow the client to request responses emitted by the specified serializer. Optionally allows you to specify which views to allow by passing an array in the views parameter.
688
+
689
+ Accepts the same filters as `before_action`.
690
+
691
+ #### `allow_output_html( as: nil, layout: nil, **filters )`
692
+
693
+ Allows falling back to the default Rails view rendering when the client asks for the media type in the `as:` parameter or `text/html` if `as:` is unset.
694
+
695
+ The `Content-Type` of the response will be `text/html` if the `as:` parameter is unset. If the `as:` parameter is set, it will include it in the variant parameter: `text/html; variant=application/vnd.xpbytes.borderless`.
696
+
697
+ Accepts the same filters as `before_action`.
698
+
699
+ #### `allow_output_docs( description, **filters )`
700
+
701
+ Outputs the specified description as help information.
702
+
703
+ Accepts the same filters as `before_action`.
704
+
705
+ #### `allow_input_serializer( serializer, views: nil, **filters )`
706
+
707
+ Configure the controller to allow the client to send bodies with a `Content-Type` that can be deserialized using the specified serializer. Optionally allows you to specify which views to allow by passing an array in the views parameter.
708
+
709
+ Accepts the same filters as `before_action`.
710
+
711
+ #### `allow_all_input( **filters )`
712
+
713
+ Disables input deserialization. Running `deserialize` while allowing all input will result in an error being thrown.
714
+
715
+ #### `not_acceptable_serializer( serializer )`
716
+
717
+ Replaces the serializer used to render the error page when no media type could be negotiated using the `Accept` header.
718
+
719
+ #### `unsupported_media_type_serializer( serializer )`
720
+
721
+ Adds a serializer that can be used to render the error page when the client submits a body with a `Content-Type` that was not added to the whitelist using `allow_input_serialization`.
722
+
723
+ #### `clear_unsupported_media_type_serializers!`
724
+
725
+ Clears the list of serializers used to render the error when the client supplies non-valid input.
726
+
727
+ #### `input_validation_failed_serializer( serializer )`
728
+
729
+ Adds a serializer that can be used to render the error page when input validation fails.
730
+
731
+ #### `clear_input_validation_failed_serializers!`
732
+
733
+ Clears the list of serializers used to render the error when the client supplies non-valid input.
734
+
735
+ #### `allow_api_viewer(serializer: MediaTypes::Serialization::Serializers::ApiViewer, **filter_opts)`
736
+
737
+ Enables rendering the api viewer when adding the `api_viewer=last` query parameter to the url.
738
+
739
+ #### `freeze_io!(**filter_opts)`
740
+
741
+ Registers serialization and deserialization in the controller. This function must be called before using the controller.
742
+
743
+ ### Controller usage
744
+
745
+ These functions are available during method execution in the controller.
746
+
747
+ #### `render_media( obj, serializers: nil, not_acceptable_serializer: nil, **options ) do`
748
+
749
+ Serializes an object and renders it using the appropriate content type. Options are passed through to the controller `render` function. Allows you to specify different objects to different serializers using a block:
750
+
751
+ ```ruby
752
+ render_media do
753
+ serializer BookSerializer, book
754
+ serializer BooksSerializer do
755
+ [ book ]
756
+ end
757
+ end
758
+ ```
759
+
760
+ Warning: this block can be called multiple times when used together with recursive serializers like the API viewer. Try to avoid changing state in this block.
761
+
762
+ If you want to render with different serializers than defined in the controller you can pass an array of serializers in the `serializers` property.
763
+
764
+ If you want to override the serializer that is used to render the response when no acceptable Content-Type could be negotiated you can pass the desired serializer in the `not_acceptable_serializer` property.
765
+
766
+ This method throws a `MediaTypes::Serialization::OutputValidationFailedError` error if the output does not conform to the format defined by the configured validator. Best practise is to return a 500 error to the client.
767
+
768
+ If no acceptable Content-Type could be negotiated the response will be rendered using the serialized defined by the class `not_acceptable_serializer` function or by the `not_acceptable_serializer` property.
769
+
770
+ Due to the way this gem is implemented it is not possible to use instance variables (`@variable`) in the `render_media` do block.
771
+
772
+ #### `deserialize( request )`
773
+
774
+ Deserializes the request body using the configured input serializers and returns the deserialized object.
775
+
776
+ Returns nil if no input body was given by the client.
777
+
778
+ This method throws a `MediaTypes::Serialization::InputValidationFailedError` error if the incoming data does not conform to the specified schema.
779
+
780
+ #### `deserialize!( request )`
781
+
782
+ Does the same as `deserialize( request )` but gives the client an error page if no input was supplied.
783
+
784
+ #### `resolve_serializer(request, identifier = nil, registration = @serialization_output_registration)`
785
+
786
+ Returns the serializer class that will handle the given request.
787
+
788
+ ## Customization
789
+
790
+ The easiest way to customize the look and feel of the built in pages is to provide your own logo and background in an initializer:
791
+
792
+ ```ruby
793
+ # config/initializers/serialization.rb
794
+
795
+ MediaTypes::Serialization::Serializers::CommonCSS.background = 'linear-gradient(245deg, #3a2f28 0%, #201a16 100%)'
796
+ MediaTypes::Serialization::Serializers::CommonCSS.logo_width = 12
797
+ MediaTypes::Serialization::Serializers::CommonCSS.logo_data = <<-HERE
798
+ <svg height="150" width="500">
799
+ <ellipse cx="240" cy="100" rx="220" ry="30" style="fill:purple" />
800
+ <ellipse cx="220" cy="70" rx="190" ry="20" style="fill:lime" />
801
+ <ellipse cx="210" cy="45" rx="170" ry="15" style="fill:yellow" />
802
+ </svg>
803
+ HERE
804
+ ```
293
805
 
294
806
  ## Development
295
807