INSTALLATION
gem install blix-rest
CREATE A SIMPLE WEBSERVICE
put the following in config.ru
require 'blix/rest'
class HomeController < Blix::Rest::Controller
get '/hello', :accept=>[:html,:json], :default=>:html do
if format == :json
{"message"=>"hello world"}
else
"<h1>hello world</h1>"
end
end
end
run Blix::Rest::Server.new
at the command line ..
ruby -S rackup -p3000
now go to your browser and enter ..
http://localhost:3000/hello
or
http://localhost:3000/hello.json
Note on JSON
the default json parser uses multi json. load the specific json library you need
before loading blix/rest
.
when using oj then you may need to set some default options eg:
MultiJson.default_options = {:mode=>:custom, :use_as_json=>true}
NOTE ON PATHS
get '/user/:user_id/list
path_params[:user_id]
contains the content of the path at location :user_id
get '/resource/*wildpath'
path_params[:wildpath]
contains the remainder of the path where the * is.
if there is a more specific path then it will be used first :
get '/resource/aaa'
will be used before get '/resource/*'
get '/*'
will be used as a default path if no other paths match.
all '/mypath'
will accept all http_methods but if a more specific handler
is specified then it will be used first.
Path options
:accept : the format or formats to accept eg: :html or [:png, :jpeg]
:default : default format if not derived through other means.
:force : force response into the given format
:query : derive format from request query (default: false)
:extension : derive format from path extension (default: true)
use :accept=>:*
in combination with :force
to accept all request formats.
APPLICATION MOUNT POINT
this is the path of the mount path of the application
this will be set to the environment variable BLIX_REST_ROOT
if present
otherwise set it manually with:
Blix::Rest.set_path_root( "/myapplication")
GENERATE AN ERROR RESPONSE
send_error(message,status,headers)
or for standard headers and status 406 just ..
send_error "my error message
HEADERS && STATUS
add special headers to your response with eg:
add_headers( "AAA"=>"xxx","BBB"=>"yyyy")
change the status of a success response with eg:
set_status(401)
to specify ALL the headers for a given format of response use eg:
srv = Blix::Rest::Server.new
srv.set_custom_headers(:html, 'Content-Type'=>'text/html; charset=utf-8', 'X-OTHER'=>'')
...
srv.run
remember to always set at least the content type!
BASIC AUTH
in controller..
login,password = get_basic_auth
auth_error( "invalid login or password" ) unless .... # validate login and password
REQUEST FORMAT
you can provide custom responses to a request format by registering a format parser for that format. you can also override the standard html,json or xml behavior.
Note that the format for a non standard (html/json/xml) request is only taken from the extension part ( after the .) of the url ... eg
http://mydomain.com/mypage.jsonp
will give a format of jsonp
you can specify the :default option in your route for a fallback format other than :json
eg :default=>:html
class MyParser < FormatParser
def set_default_headers(headers)
headers[CACHE_CONTROL]= CACHE_NO_STORE
headers[PRAGMA] = NO_CACHE
headers[CONTENT_TYPE] = CONTENT_TYPE_JSONP
end
def format_error(message)
message.to_s
end
def format_response(value,response)
response.content = "<script>load(" +
MultiJson.dump( value) +
")</script>"
end
end
s = Blix::Rest::Server.new
s.register_parser(:jsonp,MyParser.new)
Then in your controller accept that format..
get "/details" :accept=>[:jsonp] do
{"id"=>12}
end
CUSTOM RESPONSE WITHOUT CUSTOM PARSER
to force a response in a certain format use the :force option in your route.
to return a custom response use :force=>:raw
. You will have to specify all the
headers and the body is returned as it is.
use the following to accept requests in a special format ..
get '/custom', :accept=>:xyz, :force=>:raw do
add_headers 'Content-Type'=>'text/xyz'
"xyz"
end
Alternatively it is possible to raise a RawResponse:
add_headers 'Content-Type'=>'text/xyz'
raise RawResponse, 'xyz'
or with status and headers:
raise RawResponse.new('xyz', 123, 'Content-Type'=>'text/xyz')
FORMATS
the format of a request is derived from
-
the
:force
option value if present -
the request query
format
parameter if the:query
option is true -
the url extension unless the
:extension
option is false. -
the accept header format
-
the format specified in the
:default
option -
:json
Controller
Blix::Rest::Controller
base class for controllers. within your block handling a particular route you have access to a number of methods
env : the request environment hash
method : the request method lowercase( 'get'/'post' ..)
req : the rack request
body : the request body as a string
path : the request path
query_params : a hash of parameters as passed in the url as parameters
path_params : a hash of parameters constructed from variable parts of the path
post_params : a hash of parameters passed in the body of the request
params : all the params combined
allow_methods : allow the non standard http verbs in the controller. eg `:propfind `
user : the user making this request ( or nil if
format : the format the response should be in :json or :html
before : before hook ( opts ) - remember to add 'super' as first line !!!
after : after hook (opts,response)- remember to add 'super' as first line !!!
proxy : forward the call to another service (service,path, opts={}) , :include_query=>true/false
session : req.session
set_status : set the http response status ( 200 )
redirect : (path, status=302) redirect to another url.
request_ip : the ip of the request
render_erb : (template_name [,:layout=>name])
server_cache : get the server cache object
server_cache_get : retrieve/store value in cache
path_for : (path) give the external path for an internal path
url_for : (path) give the full url for an internal path
h : escape html string to avoid XSS
escape_javascript : escape a javascript string
server_options : options as passed to server at create time
logger : system logger
mode_test? : test mode ?
mode_production? : production mode ?
mode_development? : development mode?
send_data : send raw data (data, options )
[:type=>mimetype]
[:filename=>name]
[:disposition=>inline|attachment]
[:status=>234]
get_session_id(session_name, opts={}) :
refresh_session_id(session_name, opts={}) :
to accept requests other than json then set :accept=>[:json,:html]
as options in the route
eg post '/myform' :accept=>[:html] # this will only accept html requests.
Hooks
a before or after hook can be defined on a controller. Only define the hook once for a given controller per source file. A hook included from another source file is ok though.
class MyController < Blix::Rest::Controller
before do
...
end
after do
...
end
end
manipulate the route path or options
the before_route
hook can be used to modify the path or options of a route.
the verb can not be modified
example:
class MyController < Blix::Rest::Controller
before_route do |route|
route.default_option(:level,:visitor)
route.path_prefix('/app')
end
...
end
the following methods are available on the route:
verb # readonly, the 'GET','POST' etc verb of the route
path # the path of the route
options # the options associated with the route eg :accept
path_prefix('/xx') # ensure the path has the given prefix
default_option(:xxx,'foo') # ensure that option has the given default value
Sessions
the following methods are available in the controller for managing sessions.
get_session_id(session_name, opts={})
this will set up a session and setup the relevant cookie headers forthe browser.
refresh_session_id(session_name, opts={})
this will generate a new session_id and setup the relevant headers
options can include:
:secure => true # secure cookies only
:http = >true # cookies for http only not javascript requests
:samesite =>:strict # use strict x-site policy
:samesite =>:lax # use lax x-site policy
For more complete session management:
require 'blix/utils/redis_store'
require 'blix/rest/session'
class MyController < Blix::Rest::Controller
include Blix::Rest::Session
session_name :xxx # optional default'blix'
session_opts :http=>true # optional
session_manager: MyManager.new # optional , default Blix::Redis::Store
def myroutine
@xxx = session['xxx']
session['yyy'] = true
end
end
options can include:
:secure # false
:http # false
:samesite # lax
:path # '/'
:expire_secs # 30 mins
:cleanup_every_secs # 5 minutes
:max_age # nil
the following methods are available:
reset_session # gen new session id
session # session hash
csrf_token # the session csrf token
route options that affect sessions:
:nosession=>true # no session will be set/retrieved
:cache=>true # no session will be set/retrieved
:csrf=>true # request will be validated for calid csrf token.
# token must be in header X_CSRF_TOKEN field
session configuration is inherited from superclass controllers unless overridden.
Cache
the server has a cache which can also be used for storing your own data.
within a controller access the controller with server_cache
which returns the
cache object.
cache object methods:
get(key) # return value from the cache or nil
set(key,value) # set a value in the cache
key?(key) # is a key present in the cache
delete(key) # delete a key from the cache
clear # delete all keys from the cache.
there is also a server_cache_get
method.
server_cache_get(key){ action }
get the value from the cache. If the key is missing in the cache then perform the action in the provided block and store the result in the cache.
the default cache is just a ruby hash in memory. Pass a custom cache to
when creating a server with the :cache
parameter.
class MyCache < Blix::Rest::Cache
def get(key)
..
end
def set(key,value)
..
end
def key?(key)
..
end
def delete(key)
..
end
def clear
..
end
end
cache = MyCache.new
app = Blix::Rest::Server.new(:cache=>cache)
there is a redis cache already defined:
require 'blix/rest/redis_cache'
cache = Blix::Rest::RedisCache.new(:expire_secs=>60*60*24) # expire after 1 day
run Blix::Rest::Server.new(:cache=>cache)
automatically cache server responses
add :cache=>true
to your route options in order to cache this route.
add :cache_reset=>true
to your route options if the cache should be cleared when
calling this route.
the cache is not used in development/testmode , only in production mode.
IMPORTANT - DO NOT CACHE SESSION KEYS AND OTHER SPECIFIC DATA IN CACHE
only cache pages with HEADERS and CONTENT that is not user specific.
CORS
cross origin site requests
MyController < Blix::Rest::Controller
get '/info/crosssite' do
set_accept_cors
{:data=>'foo'}
end
options '/info/' crosssite' do
set_accept_cors, :headers=>'Content-Type',
end
end
within an options
response the following parameters can be passes.
:origin=>'www.othersite.com' # specify specific allowed origin.
:methods => [:get] # allowed methods
:max_age => 86400 # maximum age this auth is valid
:headers => ['Content-Type','X-OTHER'] # allow additional headers
:credentials => true # allow requests with credentials
Views
the location of your views defaults to app/views
otherwise set it manually with:
globally eg:
Blix::Rest.set_erb_root ::File.expand_path('../lib/myapp/views', __FILE__)
or per controller eg:
class MyController < Blix::Rest::Controller
erb_dir ::File.expand_path('../..', __FILE__)
end
then within a controller render your view with.
render_erb( "users/index", :layout=>'layouts/main', :locals=>{:name=>"charles"})
( locals work from ruby 2.1 )
directory structure
views
-----
users
-----
index.html.erb
layouts
-------
main.html.erb
Logging
Blix::Rest.logger = Logger.new('/var/log/myapp.log')
Testing a Service with cucumber
in features/support/setup.rb
require 'blix/rest/cucumber'
and setup your database connections etc
in features/support/hooks.rb
reset your database
now you can use the following in scenarios ........
Given user guest gets "/info"
Given the following users exist:
| name | level |
| anon | guest |
| bob | user |
| mary | provider |
| paul | user |
| admin | admin |
Given user mary posts "/worlds" with {"name":"narnia"} [..or gets/puts/deletes]
Then store the "id" as "world_id"
Given user bob posts "/worlds/:world_id" with {"the_world_id"::world_id }
Then the status should be 200
Then the data type should be "r_type"
Then the data length should be 3
Then there should be an error
Then the error message should include "unique"
Then the data "name" should == "bob"
Then the data should include "name"
And explain
NOTE : if you need to set up your database with users then you can use the following hook ..
in features/support/world.rb .........
require 'blix/rest/cucumber'
class RestWorld
# add a hook to create the user in the database -
#
def before_user_create(user,hash)
name = hash["name"]
u = MyUser.new
u.set(:user_wid, name)
u.set(:name,name)
u.set(:is_super,true) if hash["level"] == "super"
u.save
store["myuser_#{name}_id"] = u.id.to_s
end
end
now you can also use eg :myuser_foo_id
within a request path/json.
Manage Assets
require 'blix/assets'
The asset manager stores a hash of the asset data and the current unique file suffix for each asset in its own file. This config file is stored in a config directory. The default is 'config/assets' but another location can be specified.
Blix::AssetManager.config_dir = "myassets/config/location" # defaults to `"config/assets"`
Compile your assets
require 'blix/assets'
......
......
ASSETS = ['admin.js', 'admin.css', 'standard.js']
ASSETS.each do |name|
compiled_asset = environment[name].to_s
Blix::AssetManager.if_modified(name,compiled_asset,:rewrite=>true) do |a|
filename = File.join(ROOT,"public","assets",a.newname)
puts "writing #{name} to #{filename}"
File.write filename,compiled_asset
File.unlink File.join(ROOT,"public","assets",a.oldname) if a.oldname
end
end
In your erb view
eg:
require 'blix/assets'
........
<script src="<%= asset_path('assets/standard.js') %>" type="text/javascript"></script>
or
<%= asset_tag('/assets/standard.js') %>
or in your controller
eg:
require 'blix/assets'
........
path = asset_path('assets/standard.js')
NOTE ON ASSETS!!
In production mode the compiled version of the assets will be used which will have a unique file name.
In production the expiry date of your assets can be set to far in the future to take advantage of cacheing.
In development or test mode the standard name will be used which then will make use of your asset pipeline ( eg sprockets )
Asset names can contain only one extension. if there are more extensions eg: 'myfile.extra.css' then only the last extension will be used: in this case the name will be simplified to 'myfile.css' !!!