Houston Sipes

Houston Sipes

1604088000

How to Find the Stinky Parts of Your Code (Part II)

There are more code smells. Let’s keep changing the aromas. We see several symptoms and situations that make us doubt the quality of our development. Let’s look at some possible solutions.

Most of these smells are just hints of something that might be wrong. They are not rigid rules.

This is part II. Part I can be found here.

Code Smell 06 - Too Clever Programmer

The code is difficult to read, there are tricky with names without semantics. Sometimes using language’s accidental complexity.

_Image Source: NeONBRAND on _Unsplash

Problems

  • Readability
  • Maintainability
  • Code Quality
  • Premature Optimization

Solutions

  1. Refactor the code
  2. Use better names

Examples

  • Optimized loops

Exceptions

  • Optimized code for low-level operations.

Sample Code

Wrong

function primeFactors(n){
	  var f = [],  i = 0, d = 2;  

	  for (i = 0; n >= 2; ) {
	     if(n % d == 0){
	       f[i++]=(d); 
	       n /= d;
	    }
	    else{
	      d++;
	    }     
	  }
	  return f;
	}

Right

function primeFactors(numberToFactor){
	  var factors = [], 
	      divisor = 2,
	      remainder = numberToFactor;

	  while(remainder>=2){
	    if(remainder % divisor === 0){
	       factors.push(divisor); 
	       remainder = remainder/ divisor;
	    }
	    else{
	      divisor++;
	    }     
	  }
	  return factors;
	}

Detection

Automatic detection is possible in some languages. Watch some warnings related to complexity, bad names, post increment variables, etc.

#pixel-face #code-smells #clean-code #stinky-code-parts #refactor-legacy-code #refactoring #stinky-code #common-code-smells

What is GEEK

Buddha Community

How to Find the Stinky Parts of Your Code (Part II)
Houston Sipes

Houston Sipes

1604088000

How to Find the Stinky Parts of Your Code (Part II)

There are more code smells. Let’s keep changing the aromas. We see several symptoms and situations that make us doubt the quality of our development. Let’s look at some possible solutions.

Most of these smells are just hints of something that might be wrong. They are not rigid rules.

This is part II. Part I can be found here.

Code Smell 06 - Too Clever Programmer

The code is difficult to read, there are tricky with names without semantics. Sometimes using language’s accidental complexity.

_Image Source: NeONBRAND on _Unsplash

Problems

  • Readability
  • Maintainability
  • Code Quality
  • Premature Optimization

Solutions

  1. Refactor the code
  2. Use better names

Examples

  • Optimized loops

Exceptions

  • Optimized code for low-level operations.

Sample Code

Wrong

function primeFactors(n){
	  var f = [],  i = 0, d = 2;  

	  for (i = 0; n >= 2; ) {
	     if(n % d == 0){
	       f[i++]=(d); 
	       n /= d;
	    }
	    else{
	      d++;
	    }     
	  }
	  return f;
	}

Right

function primeFactors(numberToFactor){
	  var factors = [], 
	      divisor = 2,
	      remainder = numberToFactor;

	  while(remainder>=2){
	    if(remainder % divisor === 0){
	       factors.push(divisor); 
	       remainder = remainder/ divisor;
	    }
	    else{
	      divisor++;
	    }     
	  }
	  return factors;
	}

Detection

Automatic detection is possible in some languages. Watch some warnings related to complexity, bad names, post increment variables, etc.

#pixel-face #code-smells #clean-code #stinky-code-parts #refactor-legacy-code #refactoring #stinky-code #common-code-smells

Wiley Mayer

Wiley Mayer

1603890000

How to Find the Stinky Parts of Your Code (Part I)

of something that might be wrong. They are not rigid rules.

Code Smell 01 — Anemic Models

Your objects are a bunch of public attributes without behavior.

Photo by Stacey Vandergriff on Unsplash

Protocol is empty (with setters/getters).

If we ask a domain expert to describe an entity he/she would hardly tell it is ‘a bunch of attributes’.

Problems

  • No Encapsulation.
  • No mapping to real world entities.
  • Duplicate Code
  • Coupling

Solutions

    1. Find Responsibilities.
    1. Protect your attributes.
    1. Hide implementations.
    1. Delegate

Examples

  • DTOs

Sample Code

	<?

	Class Window{
	  public height;
	  public width;

	  function getHeight(){
	    return $this->height;
	  }

	  function setHeight($height){
	    $this->height = $height;
	  }

	  function getWidth(){
	    return $this->width;
	  }

	  function setWidth($width){
	    $this->width = $width;
	  }

	}

Wrong

<?

	final Class Window{ 

	  function area(){
	    //...
	  }

	  function open(){
	    //..
	  }

	  function isOpen(){
	    //..
	  }

	}

Right

#code-smells #clean-code #refactoring #refactor-legacy-code #stinky-code #stinky-code-parts #pixel-face #hackernoon-top-story

Johathan McGlynn

Johathan McGlynn

1659283860

Activeinteraction: Manage Application Specific Business Logic Of Ruby

ActiveInteraction

ActiveInteraction manages application-specific business logic. It's an implementation of service objects designed to blend seamlessly into Rails. 

ActiveInteraction gives you a place to put your business logic. It also helps you write safer code by validating that your inputs conform to your expectations. If ActiveModel deals with your nouns, then ActiveInteraction handles your verbs.

API Documentation

Installation

Add it to your Gemfile:

gem 'active_interaction', '~> 5.1'

Or install it manually:

$ gem install active_interaction --version '~> 5.1'

This project uses Semantic Versioning. Check out GitHub releases for a detailed list of changes.

Basic usage

To define an interaction, create a subclass of ActiveInteraction::Base. Then you need to do two things:

Define your inputs. Use class filter methods to define what you expect your inputs to look like. For instance, if you need a boolean flag for pepperoni, use boolean :pepperoni. Check out the filters section for all the available options.

Define your business logic. Do this by implementing the #execute method. Each input you defined will be available as the type you specified. If any of the inputs are invalid, #execute won't be run. Filters are responsible for checking your inputs. Check out the validations section if you need more than that.

That covers the basics. Let's put it all together into a simple example that squares a number.

require 'active_interaction'

class Square < ActiveInteraction::Base
  float :x

  def execute
    x**2
  end
end

Call .run on your interaction to execute it. You must pass a single hash to .run. It will return an instance of your interaction. By convention, we call this an outcome. You can use the #valid? method to ask the outcome if it's valid. If it's invalid, take a look at its errors with #errors. In either case, the value returned from #execute will be stored in #result.

outcome = Square.run(x: 'two point one')
outcome.valid?
# => nil
outcome.errors.messages
# => {:x=>["is not a valid float"]}

outcome = Square.run(x: 2.1)
outcome.valid?
# => true
outcome.result
# => 4.41

You can also use .run! to execute interactions. It's like .run but more dangerous. It doesn't return an outcome. If the outcome would be invalid, it will instead raise an error. But if the outcome would be valid, it simply returns the result.

Square.run!(x: 'two point one')
# ActiveInteraction::InvalidInteractionError: X is not a valid float
Square.run!(x: 2.1)
# => 4.41

Validations

ActiveInteraction checks your inputs. Often you'll want more than that. For instance, you may want an input to be a string with at least one non-whitespace character. Instead of writing your own validation for that, you can use validations from ActiveModel.

These validations aren't provided by ActiveInteraction. They're from ActiveModel. You can also use any custom validations you wrote yourself in your interactions.

class SayHello < ActiveInteraction::Base
  string :name

  validates :name,
    presence: true

  def execute
    "Hello, #{name}!"
  end
end

When you run this interaction, two things will happen. First ActiveInteraction will check your inputs. Then ActiveModel will validate them. If both of those are happy, it will be executed.

SayHello.run!(name: nil)
# ActiveInteraction::InvalidInteractionError: Name is required

SayHello.run!(name: '')
# ActiveInteraction::InvalidInteractionError: Name can't be blank

SayHello.run!(name: 'Taylor')
# => "Hello, Taylor!"

Filters

You can define filters inside an interaction using the appropriate class method. Each method has the same signature:

Some symbolic names. These are the attributes to create.

An optional hash of options. Each filter supports at least these two options:

default is the fallback value to use if nil is given. To make a filter optional, set default: nil.

desc is a human-readable description of the input. This can be useful for generating documentation. For more information about this, read the descriptions section.

An optional block of sub-filters. Only array and hash filters support this. Other filters will ignore blocks when given to them.

Let's take a look at an example filter. It defines three inputs: x, y, and z. Those inputs are optional and they all share the same description ("an example filter").

array :x, :y, :z,
  default: nil,
  desc: 'an example filter' do
    # Some filters support sub-filters here.
  end

In general, filters accept values of the type they correspond to, plus a few alternatives that can be reasonably coerced. Typically the coercions come from Rails, so "1" can be interpreted as the boolean value true, the string "1", or the number 1.

Basic Filters

Array

In addition to accepting arrays, array inputs will convert ActiveRecord::Relations into arrays.

class ArrayInteraction < ActiveInteraction::Base
  array :toppings

  def execute
    toppings.size
  end
end

ArrayInteraction.run!(toppings: 'everything')
# ActiveInteraction::InvalidInteractionError: Toppings is not a valid array
ArrayInteraction.run!(toppings: [:cheese, 'pepperoni'])
# => 2

Use a block to constrain the types of elements an array can contain. Note that you can only have one filter inside an array block, and it must not have a name.

array :birthdays do
  date
end

For interface, object, and record filters, the name of the array filter will be singularized and used to determine the type of value passed. In the example below, the objects passed would need to be of type Cow.

array :cows do
  object
end

You can override this by passing the necessary information to the inner filter.

array :managers do
  object class: People
end

Errors that occur will be indexed based on the Rails configuration setting index_nested_attribute_errors. You can also manually override this setting with the :index_errors option. In this state is is possible to get multiple errors from a single filter.

class ArrayInteraction < ActiveInteraction::Base
  array :favorite_numbers, index_errors: true do
    integer
  end

  def execute
    favorite_numbers
  end
end

ArrayInteraction.run(favorite_numbers: [8, 'bazillion']).errors.details
=> {:"favorite_numbers[1]"=>[{:error=>:invalid_type, :type=>"array"}]}

With :index_errors set to false the error would have been:

{:favorite_numbers=>[{:error=>:invalid_type, :type=>"array"}]}

Boolean

Boolean filters convert the strings "1", "true", and "on" (case-insensitive) into true. They also convert "0", "false", and "off" into false. Blank strings will be treated as nil.

class BooleanInteraction < ActiveInteraction::Base
  boolean :kool_aid

  def execute
    'Oh yeah!' if kool_aid
  end
end

BooleanInteraction.run!(kool_aid: 1)
# ActiveInteraction::InvalidInteractionError: Kool aid is not a valid boolean
BooleanInteraction.run!(kool_aid: true)
# => "Oh yeah!"

File

File filters also accept TempFiles and anything that responds to #rewind. That means that you can pass the params from uploading files via forms in Rails.

class FileInteraction < ActiveInteraction::Base
  file :readme

  def execute
    readme.size
  end
end

FileInteraction.run!(readme: 'README.md')
# ActiveInteraction::InvalidInteractionError: Readme is not a valid file
FileInteraction.run!(readme: File.open('README.md'))
# => 21563

Hash

Hash filters accept hashes. The expected value types are given by passing a block and nesting other filters. You can have any number of filters inside a hash, including other hashes.

class HashInteraction < ActiveInteraction::Base
  hash :preferences do
    boolean :newsletter
    boolean :sweepstakes
  end

  def execute
    puts 'Thanks for joining the newsletter!' if preferences[:newsletter]
    puts 'Good luck in the sweepstakes!' if preferences[:sweepstakes]
  end
end

HashInteraction.run!(preferences: 'yes, no')
# ActiveInteraction::InvalidInteractionError: Preferences is not a valid hash
HashInteraction.run!(preferences: { newsletter: true, 'sweepstakes' => false })
# Thanks for joining the newsletter!
# => nil

Setting default hash values can be tricky. The default value has to be either nil or {}. Use nil to make the hash optional. Use {} if you want to set some defaults for values inside the hash.

hash :optional,
  default: nil
# => {:optional=>nil}

hash :with_defaults,
  default: {} do
    boolean :likes_cookies,
      default: true
  end
# => {:with_defaults=>{:likes_cookies=>true}}

By default, hashes remove any keys that aren't given as nested filters. To allow all hash keys, set strip: false. In general we don't recommend doing this, but it's sometimes necessary.

hash :stuff,
  strip: false

String

String filters define inputs that only accept strings.

class StringInteraction < ActiveInteraction::Base
  string :name

  def execute
    "Hello, #{name}!"
  end
end

StringInteraction.run!(name: 0xDEADBEEF)
# ActiveInteraction::InvalidInteractionError: Name is not a valid string
StringInteraction.run!(name: 'Taylor')
# => "Hello, Taylor!"

String filter strips leading and trailing whitespace by default. To disable it, set the strip option to false.

string :comment,
  strip: false

Symbol

Symbol filters define inputs that accept symbols. Strings will be converted into symbols.

class SymbolInteraction < ActiveInteraction::Base
  symbol :method

  def execute
    method.to_proc
  end
end

SymbolInteraction.run!(method: -> {})
# ActiveInteraction::InvalidInteractionError: Method is not a valid symbol
SymbolInteraction.run!(method: :object_id)
# => #<Proc:0x007fdc9ba94118>

Dates and times

Filters that work with dates and times behave similarly. By default, they all convert strings into their expected data types using .parse. Blank strings will be treated as nil. If you give the format option, they will instead convert strings using .strptime. Note that formats won't work with DateTime and Time filters if a time zone is set.

Date

class DateInteraction < ActiveInteraction::Base
  date :birthday

  def execute
    birthday + (18 * 365)
  end
end

DateInteraction.run!(birthday: 'yesterday')
# ActiveInteraction::InvalidInteractionError: Birthday is not a valid date
DateInteraction.run!(birthday: Date.new(1989, 9, 1))
# => #<Date: 2007-08-28 ((2454341j,0s,0n),+0s,2299161j)>

date :birthday,
  format: '%Y-%m-%d'

DateTime

class DateTimeInteraction < ActiveInteraction::Base
  date_time :now

  def execute
    now.iso8601
  end
end

DateTimeInteraction.run!(now: 'now')
# ActiveInteraction::InvalidInteractionError: Now is not a valid date time
DateTimeInteraction.run!(now: DateTime.now)
# => "2015-03-11T11:04:40-05:00"

date_time :start,
  format: '%Y-%m-%dT%H:%M:%S'

Time

In addition to converting strings with .parse (or .strptime), time filters convert numbers with .at.

class TimeInteraction < ActiveInteraction::Base
  time :epoch

  def execute
    Time.now - epoch
  end
end

TimeInteraction.run!(epoch: 'a long, long time ago')
# ActiveInteraction::InvalidInteractionError: Epoch is not a valid time
TimeInteraction.run!(epoch: Time.new(1970))
# => 1426068362.5136619

time :start,
  format: '%Y-%m-%dT%H:%M:%S'

Numbers

All numeric filters accept numeric input. They will also convert strings using the appropriate method from Kernel (like .Float). Blank strings will be treated as nil.

Decimal

class DecimalInteraction < ActiveInteraction::Base
  decimal :price

  def execute
    price * 1.0825
  end
end

DecimalInteraction.run!(price: 'one ninety-nine')
# ActiveInteraction::InvalidInteractionError: Price is not a valid decimal
DecimalInteraction.run!(price: BigDecimal(1.99, 2))
# => #<BigDecimal:7fe792a42028,'0.2165E1',18(45)>

To specify the number of significant digits, use the digits option.

decimal :dollars,
  digits: 2

Float

class FloatInteraction < ActiveInteraction::Base
  float :x

  def execute
    x**2
  end
end

FloatInteraction.run!(x: 'two point one')
# ActiveInteraction::InvalidInteractionError: X is not a valid float
FloatInteraction.run!(x: 2.1)
# => 4.41

Integer

class IntegerInteraction < ActiveInteraction::Base
  integer :limit

  def execute
    limit.downto(0).to_a
  end
end

IntegerInteraction.run!(limit: 'ten')
# ActiveInteraction::InvalidInteractionError: Limit is not a valid integer
IntegerInteraction.run!(limit: 10)
# => [10, 9, 8, 7, 6, 5, 4, 3, 2, 1, 0]

When a String is passed into an integer input, the value will be coerced. A default base of 10 is used though it may be overridden with the base option. If a base of 0 is provided, the coercion will respect radix indicators present in the string.

class IntegerInteraction < ActiveInteraction::Base
  integer :limit1
  integer :limit2, base: 8
  integer :limit3, base: 0

  def execute
    [limit1, limit2, limit3]
  end
end

IntegerInteraction.run!(limit1: 71, limit2: 71, limit3: 71)
# => [71, 71, 71]
IntegerInteraction.run!(limit1: "071", limit2: "071", limit3: "0x71")
# => [71, 57, 113]
IntegerInteraction.run!(limit1: "08", limit2: "08", limit3: "08")
ActiveInteraction::InvalidInteractionError: Limit2 is not a valid integer, Limit3 is not a valid integer

Advanced Filters

Interface

Interface filters allow you to specify an interface that the passed value must meet in order to pass. The name of the interface is used to look for a constant inside the ancestor listing for the passed value. This allows for a variety of checks depending on what's passed. Class instances are checked for an included module or an inherited ancestor class. Classes are checked for an extended module or an inherited ancestor class. Modules are checked for an extended module.

class InterfaceInteraction < ActiveInteraction::Base
  interface :exception

  def execute
    exception
  end
end

InterfaceInteraction.run!(exception: Exception)
# ActiveInteraction::InvalidInteractionError: Exception is not a valid interface
InterfaceInteraction.run!(exception: NameError) # a subclass of Exception
# => NameError

You can use :from to specify a class or module. This would be the equivalent of what's above.

class InterfaceInteraction < ActiveInteraction::Base
  interface :error,
    from: Exception

  def execute
    error
  end
end

You can also create an anonymous interface on the fly by passing the methods option.

class InterfaceInteraction < ActiveInteraction::Base
  interface :serializer,
    methods: %i[dump load]

  def execute
    input = '{ "is_json" : true }'
    object = serializer.load(input)
    output = serializer.dump(object)

    output
  end
end

require 'json'

InterfaceInteraction.run!(serializer: Object.new)
# ActiveInteraction::InvalidInteractionError: Serializer is not a valid interface
InterfaceInteraction.run!(serializer: JSON)
# => "{\"is_json\":true}"

Object

Object filters allow you to require an instance of a particular class or one of its subclasses.

class Cow
  def moo
    'Moo!'
  end
end

class ObjectInteraction < ActiveInteraction::Base
  object :cow

  def execute
    cow.moo
  end
end

ObjectInteraction.run!(cow: Object.new)
# ActiveInteraction::InvalidInteractionError: Cow is not a valid object
ObjectInteraction.run!(cow: Cow.new)
# => "Moo!"

The class name is automatically determined by the filter name. If your filter name is different than your class name, use the class option. It can be either the class, a string, or a symbol.

object :dolly1,
  class: Sheep
object :dolly2,
  class: 'Sheep'
object :dolly3,
  class: :Sheep

If you have value objects or you would like to build one object from another, you can use the converter option. It is only called if the value provided is not an instance of the class or one of its subclasses. The converter option accepts a symbol that specifies a class method on the object class or a proc. Both will be passed the value and any errors thrown inside the converter will cause the value to be considered invalid. Any returned value that is not the correct class will also be treated as invalid. Any default that is not an instance of the class or subclass and is not nil will also be converted.

class ObjectInteraction < ActiveInteraction::Base
  object :ip_address,
    class: IPAddr,
    converter: :new

  def execute
    ip_address
  end
end

ObjectInteraction.run!(ip_address: '192.168.1.1')
# #<IPAddr: IPv4:192.168.1.1/255.255.255.255>

ObjectInteraction.run!(ip_address: 1)
# ActiveInteraction::InvalidInteractionError: Ip address is not a valid object

Record

Record filters allow you to require an instance of a particular class (or one of its subclasses) or a value that can be used to locate an instance of the object. If the value does not match, it will call find on the class of the record. This is particularly useful when working with ActiveRecord objects. Like an object filter, the class is derived from the name passed but can be specified with the class option. Any default that is not an instance of the class or subclass and is not nil will also be found. Blank strings passed in will be treated as nil.

class RecordInteraction < ActiveInteraction::Base
  record :encoding

  def execute
    encoding
  end
end

> RecordInteraction.run!(encoding: Encoding::US_ASCII)
=> #<Encoding:US-ASCII>

> RecordInteraction.run!(encoding: 'ascii')
=> #<Encoding:US-ASCII>

A different method can be specified by providing a symbol to the finder option.

Rails

ActiveInteraction plays nicely with Rails. You can use interactions to handle your business logic instead of models or controllers. To see how it all works, let's take a look at a complete example of a controller with the typical resourceful actions.

Setup

We recommend putting your interactions in app/interactions. It's also very helpful to group them by model. That way you can look in app/interactions/accounts for all the ways you can interact with accounts.

- app/
  - controllers/
    - accounts_controller.rb
  - interactions/
    - accounts/
      - create_account.rb
      - destroy_account.rb
      - find_account.rb
      - list_accounts.rb
      - update_account.rb
  - models/
    - account.rb
  - views/
    - account/
      - edit.html.erb
      - index.html.erb
      - new.html.erb
      - show.html.erb

Controller

Index

# GET /accounts
def index
  @accounts = ListAccounts.run!
end

Since we're not passing any inputs to ListAccounts, it makes sense to use .run! instead of .run. If it failed, that would mean we probably messed up writing the interaction.

class ListAccounts < ActiveInteraction::Base
  def execute
    Account.not_deleted.order(last_name: :asc, first_name: :asc)
  end
end

Show

Up next is the show action. For this one we'll define a helper method to handle raising the correct errors. We have to do this because calling .run! would raise an ActiveInteraction::InvalidInteractionError instead of an ActiveRecord::RecordNotFound. That means Rails would render a 500 instead of a 404.

# GET /accounts/:id
def show
  @account = find_account!
end

private

def find_account!
  outcome = FindAccount.run(params)

  if outcome.valid?
    outcome.result
  else
    fail ActiveRecord::RecordNotFound, outcome.errors.full_messages.to_sentence
  end
end

This probably looks a little different than you're used to. Rails commonly handles this with a before_filter that sets the @account instance variable. Why is all this interaction code better? Two reasons: One, you can reuse the FindAccount interaction in other places, like your API controller or a Resque task. And two, if you want to change how accounts are found, you only have to change one place.

Inside the interaction, we could use #find instead of #find_by_id. That way we wouldn't need the #find_account! helper method in the controller because the error would bubble all the way up. However, you should try to avoid raising errors from interactions. If you do, you'll have to deal with raised exceptions as well as the validity of the outcome.

class FindAccount < ActiveInteraction::Base
  integer :id

  def execute
    account = Account.not_deleted.find_by_id(id)

    if account
      account
    else
      errors.add(:id, 'does not exist')
    end
  end
end

Note that it's perfectly fine to add errors during execution. Not all errors have to come from checking or validation.

New

The new action will be a little different than the ones we've looked at so far. Instead of calling .run or .run!, it's going to initialize a new interaction. This is possible because interactions behave like ActiveModels.

# GET /accounts/new
def new
  @account = CreateAccount.new
end

Since interactions behave like ActiveModels, we can use ActiveModel validations with them. We'll use validations here to make sure that the first and last names are not blank. The validations section goes into more detail about this.

class CreateAccount < ActiveInteraction::Base
  string :first_name, :last_name

  validates :first_name, :last_name,
    presence: true

  def to_model
    Account.new
  end

  def execute
    account = Account.new(inputs)

    unless account.save
      errors.merge!(account.errors)
    end

    account
  end
end

We used a couple of advanced features here. The #to_model method helps determine the correct form to use in the view. Check out the section on forms for more about that. Inside #execute, we merge errors. This is a convenient way to move errors from one object to another. Read more about it in the errors section.

Create

The create action has a lot in common with the new action. Both of them use the CreateAccount interaction. And if creating the account fails, this action falls back to rendering the new action.

# POST /accounts
def create
  outcome = CreateAccount.run(params.fetch(:account, {}))

  if outcome.valid?
    redirect_to(outcome.result)
  else
    @account = outcome
    render(:new)
  end
end

Note that we have to pass a hash to .run. Passing nil is an error.

Since we're using an interaction, we don't need strong parameters. The interaction will ignore any inputs that weren't defined by filters. So you can forget about params.require and params.permit because interactions handle that for you.

Destroy

The destroy action will reuse the #find_account! helper method we wrote earlier.

# DELETE /accounts/:id
def destroy
  DestroyAccount.run!(account: find_account!)
  redirect_to(accounts_url)
end

In this simple example, the destroy interaction doesn't do much. It's not clear that you gain anything by putting it in an interaction. But in the future, when you need to do more than account.destroy, you'll only have to update one spot.

class DestroyAccount < ActiveInteraction::Base
  object :account

  def execute
    account.destroy
  end
end

Edit

Just like the destroy action, editing uses the #find_account! helper. Then it creates a new interaction instance to use as a form object.

# GET /accounts/:id/edit
def edit
  account = find_account!
  @account = UpdateAccount.new(
    account: account,
    first_name: account.first_name,
    last_name: account.last_name)
end

The interaction that updates accounts is more complicated than the others. It requires an account to update, but the other inputs are optional. If they're missing, it'll ignore those attributes. If they're present, it'll update them.

class UpdateAccount < ActiveInteraction::Base
  object :account

  string :first_name, :last_name,
    default: nil

  validates :first_name,
    presence: true,
    unless: -> { first_name.nil? }
  validates :last_name,
    presence: true,
    unless: -> { last_name.nil? }

  def execute
    account.first_name = first_name if first_name.present?
    account.last_name = last_name if last_name.present?

    unless account.save
      errors.merge!(account.errors)
    end

    account
  end
end

Update

Hopefully you've gotten the hang of this by now. We'll use #find_account! to get the account. Then we'll build up the inputs for UpdateAccount. Then we'll run the interaction and either redirect to the updated account or back to the edit page.

# PUT /accounts/:id
def update
  inputs = { account: find_account! }.reverse_merge(params[:account])
  outcome = UpdateAccount.run(inputs)

  if outcome.valid?
    redirect_to(outcome.result)
  else
    @account = outcome
    render(:edit)
  end
end

Advanced usage

Callbacks

ActiveSupport::Callbacks provides a powerful framework for defining callbacks. ActiveInteraction uses that framework to allow hooking into various parts of an interaction's lifecycle.

class Increment < ActiveInteraction::Base
  set_callback :filter, :before, -> { puts 'before filter' }

  integer :x

  set_callback :validate, :after, -> { puts 'after validate' }

  validates :x,
    numericality: { greater_than_or_equal_to: 0 }

  set_callback :execute, :around, lambda { |_interaction, block|
    puts '>>>'
    block.call
    puts '<<<'
  }

  def execute
    puts 'executing'
    x + 1
  end
end

Increment.run!(x: 1)
# before filter
# after validate
# >>>
# executing
# <<<
# => 2

In order, the available callbacks are filter, validate, and execute. You can set before, after, or around on any of them.

Composition

You can run interactions from within other interactions with #compose. If the interaction is successful, it'll return the result (just like if you had called it with .run!). If something went wrong, execution will halt immediately and the errors will be moved onto the caller.

class Add < ActiveInteraction::Base
  integer :x, :y

  def execute
    x + y
  end
end

class AddThree < ActiveInteraction::Base
  integer :x

  def execute
    compose(Add, x: x, y: 3)
  end
end

AddThree.run!(x: 5)
# => 8

To bring in filters from another interaction, use .import_filters. Combined with inputs, delegating to another interaction is a piece of cake.

class AddAndDouble < ActiveInteraction::Base
  import_filters Add

  def execute
    compose(Add, inputs) * 2
  end
end

Note that errors in composed interactions have a few tricky cases. See the errors section for more information about them.

Defaults

The default value for an input can take on many different forms. Setting the default to nil makes the input optional. Setting it to some value makes that the default value for that input. Setting it to a lambda will lazily set the default value for that input. That means the value will be computed when the interaction is run, as opposed to when it is defined.

Lambda defaults are evaluated in the context of the interaction, so you can use the values of other inputs in them.

# This input is optional.
time :a, default: nil
# This input defaults to `Time.at(123)`.
time :b, default: Time.at(123)
# This input lazily defaults to `Time.now`.
time :c, default: -> { Time.now }
# This input defaults to the value of `c` plus 10 seconds.
time :d, default: -> { c + 10 }

Descriptions

Use the desc option to provide human-readable descriptions of filters. You should prefer these to comments because they can be used to generate documentation. The interaction class has a .filters method that returns a hash of filters. Each filter has a #desc method that returns the description.

class Descriptive < ActiveInteraction::Base
  string :first_name,
    desc: 'your first name'
  string :last_name,
    desc: 'your last name'
end

Descriptive.filters.each do |name, filter|
  puts "#{name}: #{filter.desc}"
end
# first_name: your first name
# last_name: your last name

Errors

ActiveInteraction provides detailed errors for easier introspection and testing of errors. Detailed errors improve on regular errors by adding a symbol that represents the type of error that has occurred. Let's look at an example where an item is purchased using a credit card.

class BuyItem < ActiveInteraction::Base
  object :credit_card, :item
  hash :options do
    boolean :gift_wrapped
  end

  def execute
    order = credit_card.purchase(item)
    notify(credit_card.account)
    order
  end

  private def notify(account)
    # ...
  end
end

Having missing or invalid inputs causes the interaction to fail and return errors.

outcome = BuyItem.run(item: 'Thing', options: { gift_wrapped: 'yes' })
outcome.errors.messages
# => {:credit_card=>["is required"], :item=>["is not a valid object"], :"options.gift_wrapped"=>["is not a valid boolean"]}

Determining the type of error based on the string is difficult if not impossible. Calling #details instead of #messages on errors gives you the same list of errors with a testable label representing the error.

outcome.errors.details
# => {:credit_card=>[{:error=>:missing}], :item=>[{:error=>:invalid_type, :type=>"object"}], :"options.gift_wrapped"=>[{:error=>:invalid_type, :type=>"boolean"}]}

Detailed errors can also be manually added during the execute call by passing a symbol to #add instead of a string.

def execute
  errors.add(:monster, :no_passage)
end

ActiveInteraction also supports merging errors. This is useful if you want to delegate validation to some other object. For example, if you have an interaction that updates a record, you might want that record to validate itself. By using the #merge! helper on errors, you can do exactly that.

class UpdateThing < ActiveInteraction::Base
  object :thing

  def execute
    unless thing.save
      errors.merge!(thing.errors)
    end

    thing
  end
end

When a composed interaction fails, its errors are merged onto the caller. This generally produces good error messages, but there are a few cases to look out for.

class Inner < ActiveInteraction::Base
  boolean :x, :y
end

class Outer < ActiveInteraction::Base
  string :x
  boolean :z, default: nil

  def execute
    compose(Inner, x: x, y: z)
  end
end

outcome = Outer.run(x: 'yes')
outcome.errors.details
# => { :x    => [{ :error => :invalid_type, :type => "boolean" }],
#      :base => [{ :error => "Y is required" }] }
outcome.errors.full_messages.join(' and ')
# => "X is not a valid boolean and Y is required"

Since both interactions have an input called x, the inner error for that input is moved to the x error on the outer interaction. This results in a misleading error that claims the input x is not a valid boolean even though it's a string on the outer interaction.

Since only the inner interaction has an input called y, the inner error for that input is moved to the base error on the outer interaction. This results in a confusing error that claims the input y is required even though it's not present on the outer interaction.

Forms

The outcome returned by .run can be used in forms as though it were an ActiveModel object. You can also create a form object by calling .new on the interaction.

Given an application with an Account model we'll create a new Account using the CreateAccount interaction.

# GET /accounts/new
def new
  @account = CreateAccount.new
end

# POST /accounts
def create
  outcome = CreateAccount.run(params.fetch(:account, {}))

  if outcome.valid?
    redirect_to(outcome.result)
  else
    @account = outcome
    render(:new)
  end
end

The form used to create a new Account has slightly more information on the form_for call than you might expect.

<%= form_for @account, as: :account, url: accounts_path do |f| %>
  <%= f.text_field :first_name %>
  <%= f.text_field :last_name %>
  <%= f.submit 'Create' %>
<% end %>

This is necessary because we want the form to act like it is creating a new Account. Defining to_model on the CreateAccount interaction tells the form to treat our interaction like an Account.

class CreateAccount < ActiveInteraction::Base
  # ...

  def to_model
    Account.new
  end
end

Now our form_for call knows how to generate the correct URL and param name (i.e. params[:account]).

# app/views/accounts/new.html.erb
<%= form_for @account do |f| %>
  <%# ... %>
<% end %>

If you have an interaction that updates an Account, you can define to_model to return the object you're updating.

class UpdateAccount < ActiveInteraction::Base
  # ...

  object :account

  def to_model
    account
  end
end

ActiveInteraction also supports formtastic and simple_form. The filters used to define the inputs on your interaction will relay type information to these gems. As a result, form fields will automatically use the appropriate input type.

Shared input options

It can be convenient to apply the same options to a bunch of inputs. One common use case is making many inputs optional. Instead of setting default: nil on each one of them, you can use with_options to reduce duplication.

with_options default: nil do
  date :birthday
  string :name
  boolean :wants_cake
end

Optional inputs

Optional inputs can be defined by using the :default option as described in the filters section. Within the interaction, provided and default values are merged to create inputs. There are times where it is useful to know whether a value was passed to run or the result of a filter default. In particular, it is useful when nil is an acceptable value. For example, you may optionally track your users' birthdays. You can use the inputs.given? predicate to see if an input was even passed to run. With inputs.given? you can also check the input of a hash or array filter by passing a series of keys or indexes to check.

class UpdateUser < ActiveInteraction::Base
  object :user
  date :birthday,
    default: nil

  def execute
    user.birthday = birthday if inputs.given?(:birthday)
    errors.merge!(user.errors) unless user.save
    user
  end
end

Now you have a few options. If you don't want to update their birthday, leave it out of the hash. If you want to remove their birthday, set birthday: nil. And if you want to update it, pass in the new value as usual.

user = User.find(...)

# Don't update their birthday.
UpdateUser.run!(user: user)

# Remove their birthday.
UpdateUser.run!(user: user, birthday: nil)

# Update their birthday.
UpdateUser.run!(user: user, birthday: Date.new(2000, 1, 2))

Translations

ActiveInteraction is i18n aware out of the box! All you have to do is add translations to your project. In Rails, these typically go into config/locales. For example, let's say that for some reason you want to print everything out backwards. Simply add translations for ActiveInteraction to your hsilgne locale.

# config/locales/hsilgne.yml
hsilgne:
  active_interaction:
    types:
      array: yarra
      boolean: naeloob
      date: etad
      date_time: emit etad
      decimal: lamiced
      file: elif
      float: taolf
      hash: hsah
      integer: regetni
      interface: ecafretni
      object: tcejbo
      string: gnirts
      symbol: lobmys
      time: emit
    errors:
      messages:
        invalid: dilavni si
        invalid_type: '%{type} dilav a ton si'
        missing: deriuqer si

Then set your locale and run interactions like normal.

class I18nInteraction < ActiveInteraction::Base
  string :name
end

I18nInteraction.run(name: false).errors.messages[:name]
# => ["is not a valid string"]

I18n.locale = :hsilgne
I18nInteraction.run(name: false).errors.messages[:name]
# => ["gnirts dilav a ton si"]

Everything else works like an activerecord entry. For example, to rename an attribute you can use attributes.

Here we'll rename the num attribute on an interaction named product:

en:
  active_interaction:
    attributes:
      product:
        num: 'Number'

Credits

ActiveInteraction is brought to you by Aaron Lasseigne. Along with Aaron, Taylor Fausak helped create and maintain ActiveInteraction but has since moved on.

If you want to contribute to ActiveInteraction, please read our contribution guidelines. A complete list of contributors is available on GitHub.

ActiveInteraction is licensed under the MIT License.

Author: AaronLasseigne
Source code: https://github.com/AaronLasseigne/active_interaction
License: MIT license

#ruby 

Tyrique Littel

Tyrique Littel

1604008800

Static Code Analysis: What It Is? How to Use It?

Static code analysis refers to the technique of approximating the runtime behavior of a program. In other words, it is the process of predicting the output of a program without actually executing it.

Lately, however, the term “Static Code Analysis” is more commonly used to refer to one of the applications of this technique rather than the technique itself — program comprehension — understanding the program and detecting issues in it (anything from syntax errors to type mismatches, performance hogs likely bugs, security loopholes, etc.). This is the usage we’d be referring to throughout this post.

“The refinement of techniques for the prompt discovery of error serves as well as any other as a hallmark of what we mean by science.”

  • J. Robert Oppenheimer

Outline

We cover a lot of ground in this post. The aim is to build an understanding of static code analysis and to equip you with the basic theory, and the right tools so that you can write analyzers on your own.

We start our journey with laying down the essential parts of the pipeline which a compiler follows to understand what a piece of code does. We learn where to tap points in this pipeline to plug in our analyzers and extract meaningful information. In the latter half, we get our feet wet, and write four such static analyzers, completely from scratch, in Python.

Note that although the ideas here are discussed in light of Python, static code analyzers across all programming languages are carved out along similar lines. We chose Python because of the availability of an easy to use ast module, and wide adoption of the language itself.

How does it all work?

Before a computer can finally “understand” and execute a piece of code, it goes through a series of complicated transformations:

static analysis workflow

As you can see in the diagram (go ahead, zoom it!), the static analyzers feed on the output of these stages. To be able to better understand the static analysis techniques, let’s look at each of these steps in some more detail:

Scanning

The first thing that a compiler does when trying to understand a piece of code is to break it down into smaller chunks, also known as tokens. Tokens are akin to what words are in a language.

A token might consist of either a single character, like (, or literals (like integers, strings, e.g., 7, Bob, etc.), or reserved keywords of that language (e.g, def in Python). Characters which do not contribute towards the semantics of a program, like trailing whitespace, comments, etc. are often discarded by the scanner.

Python provides the tokenize module in its standard library to let you play around with tokens:

Python

1

import io

2

import tokenize

3

4

code = b"color = input('Enter your favourite color: ')"

5

6

for token in tokenize.tokenize(io.BytesIO(code).readline):

7

    print(token)

Python

1

TokenInfo(type=62 (ENCODING),  string='utf-8')

2

TokenInfo(type=1  (NAME),      string='color')

3

TokenInfo(type=54 (OP),        string='=')

4

TokenInfo(type=1  (NAME),      string='input')

5

TokenInfo(type=54 (OP),        string='(')

6

TokenInfo(type=3  (STRING),    string="'Enter your favourite color: '")

7

TokenInfo(type=54 (OP),        string=')')

8

TokenInfo(type=4  (NEWLINE),   string='')

9

TokenInfo(type=0  (ENDMARKER), string='')

(Note that for the sake of readability, I’ve omitted a few columns from the result above — metadata like starting index, ending index, a copy of the line on which a token occurs, etc.)

#code quality #code review #static analysis #static code analysis #code analysis #static analysis tools #code review tips #static code analyzer #static code analysis tool #static analyzer

Muhammad Price

Muhammad Price

1659511140

Roadie: Making HTML Emails Comfortable for The Ruby Rockstars

Roadie 

  
:warning:This gem is now in [passive maintenance mode][passive]. [(more)][passive]

Making HTML emails comfortable for the Ruby rockstars

Roadie tries to make sending HTML emails a little less painful by inlining stylesheets and rewriting relative URLs for you inside your emails.

How does it work?

Email clients have bad support for stylesheets, and some of them blocks stylesheets from downloading. The easiest way to handle this is to work with inline styles (style="..."), but that is error prone and hard to work with as you cannot use classes and/or reuse styling over your HTML.

This gem makes this easier by automatically inlining stylesheets into the document. You give Roadie your CSS, or let it find it by itself from the <link> and <style> tags in the markup, and it will go through all of the selectors assigning the styles to the matching elements. Careful attention has been put into selectors being applied in the correct order, so it should behave just like in the browser.

"Dynamic" selectors (:hover, :visited, :focus, etc.), or selectors not understood by Nokogiri will be inlined into a single <style> element for those email clients that support it. This changes specificity a great deal for these rules, so it might not work 100% out of the box. (See more about this below)

Roadie also rewrites all relative URLs in the email to an absolute counterpart, making images you insert and those referenced in your stylesheets work. No more headaches about how to write the stylesheets while still having them work with emails from your acceptance environments. You can disable this on specific elements using a data-roadie-ignore marker.

Features

  • Writes CSS styles inline.
    • Respects !important styles.
    • Does not overwrite styles already present in the style attribute of tags.
    • Supports the same CSS selectors as Nokogiri; use CSS3 selectors in your emails!
    • Keeps :hover, @media { ... } and friends around in a separate <style> element.
  • Makes image urls absolute.
    • Hostname and port configurable on a per-environment basis.
    • Can be disabled on individual elements.
  • Makes link hrefs and img srcs absolute.
  • Automatically adds proper HTML skeleton when missing; you don't have to create a layout for emails.
    • Also supports HTML fragments / partial documents, where layout is not added.
  • Allows you to inject stylesheets in a number of ways, at runtime.
  • Removes data-roadie-ignore markers before finishing the HTML.

Install & Usage

Add this gem to your Gemfile as recommended by Rubygems and run bundle install.

gem 'roadie', '~> 4.0'

Your document instance can be configured with several options:

  • url_options - Dictates how absolute URLs should be built.
  • keep_uninlinable_css - Set to false to skip CSS that cannot be inlined.
  • merge_media_queries - Set to false to not group media queries. Some users might prefer to not group rules within media queries because it will result in rules getting reordered. e.g.
@media(max-width: 600px) { .col-6 { display: block; } }
@media(max-width: 400px) { .col-12 { display: inline-block; } }
@media(max-width: 600px) { .col-12 { display: block; } }
  • will become
@media(max-width: 600px) { .col-6 { display: block; } .col-12 { display: block; } }
@media(max-width: 400px) { .col-12 { display: inline-block; } }
  • asset_providers - A list of asset providers that are invoked when CSS files are referenced. See below.
  • external_asset_providers - A list of asset providers that are invoked when absolute CSS URLs are referenced. See below.
  • before_transformation - A callback run before transformation starts.
  • after_transformation - A callback run after transformation is completed.

Making URLs absolute

In order to make URLs absolute you need to first configure the URL options of the document.

html = '... <a href="/about-us">Read more!</a> ...'
document = Roadie::Document.new html
document.url_options = {host: "myapp.com", protocol: "https"}
document.transform
  # => "... <a href=\"https://myapp.com/about-us\">Read more!</a> ..."

The following URLs will be rewritten for you:

  • a[href] (HTML)
  • img[src] (HTML)
  • url() (CSS)

You can disable individual elements by adding an data-roadie-ignore marker on them. CSS will still be inlined on those elements, but URLs will not be rewritten.

<a href="|UNSUBSCRIBE_URL|" data-roadie-ignore>Unsubscribe</a>

Referenced stylesheets

By default, style and link elements in the email document's head are processed along with the stylesheets and removed from the head.

You can set a special data-roadie-ignore attribute on style and link tags that you want to ignore (the attribute will be removed, however). This is the place to put things like :hover selectors that you want to have for email clients allowing them.

Style and link elements with media="print" are also ignored.

<head>
  <link rel="stylesheet" type="text/css" href="/assets/emails/rock.css">         <!-- Will be inlined with normal providers -->
  <link rel="stylesheet" type="text/css" href="http://www.metal.org/metal.css">  <!-- Will be inlined with external providers, *IF* specified; otherwise ignored. -->
  <link rel="stylesheet" type="text/css" href="/assets/jazz.css" media="print">  <!-- Will NOT be inlined; print style -->
  <link rel="stylesheet" type="text/css" href="/ambient.css" data-roadie-ignore> <!-- Will NOT be inlined; ignored -->
  <style></style>                    <!-- Will be inlined -->
  <style data-roadie-ignore></style> <!-- Will NOT be inlined; ignored -->
</head>

Roadie will use the given asset providers to look for the actual CSS that is referenced. If you don't change the default, it will use the Roadie::FilesystemProvider which looks for stylesheets on the filesystem, relative to the current working directory.

Example:

# /home/user/foo/stylesheets/primary.css
body { color: green; }

# /home/user/foo/script.rb
html = <<-HTML
<html>
  <head>
  <link rel="stylesheet" type="text/css" href="/stylesheets/primary.css">
  </head>
  <body>
  </body>
</html>
HTML

Dir.pwd # => "/home/user/foo"
document = Roadie::Document.new html
document.transform # =>
                   # <!DOCTYPE html>
                   # <html>
                   #   <head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"></head>
                   #   <body style="color:green;"></body>
                   # </html>

If a referenced stylesheet cannot be found, the #transform method will raise an Roadie::CssNotFound error. If you instead want to ignore missing stylesheets, you can use the NullProvider.

Configuring providers

You can write your own providers if you need very specific behavior for your app, or you can use the built-in providers. Providers come in two groups: normal and external. Normal providers handle paths without host information (/style/foo.css) while external providers handle URLs with host information (//example.com/foo.css, localhost:3001/bar.css, and so on).

The default configuration is to not have any external providers configured, which will cause those referenced stylesheets to be ignored. Adding one or more providers for external assets causes all of them to be searched and inlined, so if you only want this to happen to specific stylesheets you need to add ignore markers to every other styleshheet (see above).

Included providers:

  • FilesystemProvider – Looks for files on the filesystem, relative to the given directory unless otherwise specified.
  • ProviderList – Wraps a list of other providers and searches them in order. The asset_providers setting is an instance of this. It behaves a lot like an array, so you can push, pop, shift and unshift to it.
  • NullProvider – Does not actually provide anything, it always finds empty stylesheets. Use this in tests or if you want to ignore stylesheets that cannot be found by your other providers (or if you want to force the other providers to never run).
  • NetHttpProvider – Downloads stylesheets using Net::HTTP. Can be given a whitelist of hosts to download from.
  • CachedProvider – Wraps another provider (or ProviderList) and caches responses inside the provided cache store.
  • PathRewriterProvider – Rewrites the passed path and then passes it on to another provider (or ProviderList).

If you want to search several locations on the filesystem, you can declare that:

document.asset_providers = [
  Roadie::FilesystemProvider.new(App.root.join("resources", "stylesheets")),
  Roadie::FilesystemProvider.new(App.root.join("system", "uploads", "stylesheets")),
]

NullProvider

If you want to ignore stylesheets that cannot be found instead of crashing, push the NullProvider to the end:

# Don't crash on missing assets
document.asset_providers << Roadie::NullProvider.new

# Don't download assets in tests
document.external_asset_providers.unshift Roadie::NullProvider.new

Note: This will cause the referenced stylesheet to be removed from the source code, so email client will never see it either.

NetHttpProvider

The NetHttpProvider will download the URLs that is is given using Ruby's standard Net::HTTP library.

You can give it a whitelist of hosts that downloads are allowed from:

document.external_asset_providers << Roadie::NetHttpProvider.new(
  whitelist: ["myapp.com", "assets.myapp.com", "cdn.cdnnetwork.co.jp"],
)
document.external_asset_providers << Roadie::NetHttpProvider.new # Allows every host

CachedProvider

You might want to cache providers from working several times. If you are sending several emails quickly from the same process, this might also save a lot of time on parsing the stylesheets if you use in-memory storage such as a hash.

You can wrap any other kind of providers with it, even a ProviderList:

document.external_asset_providers = Roadie::CachedProvider.new(document.external_asset_providers, my_cache)

If you don't pass a cache backend, it will use a normal Hash. The cache store must follow this protocol:

my_cache["key"] = some_stylesheet_instance # => #<Roadie::Stylesheet instance>
my_cache["key"]                            # => #<Roadie::Stylesheet instance>
my_cache["missing"]                        # => nil

Warning: The default Hash store will never be cleared, so make sure you don't allow the number of unique asset paths to grow too large in a single run. This is especially important if you run Roadie in a daemon that accepts arbritary documents, and/or if you use hash digests in your filenames. Making a new instance of CachedProvider will use a new Hash instance.

You can implement your own custom cache store by implementing the [] and []= methods.

class MyRoadieMemcacheStore
  def initialize(memcache)
    @memcache = memcache
  end

  def [](path)
    css = memcache.read("assets/#{path}/css")
    if css
      name = memcache.read("assets/#{path}/name") || "cached #{path}"
      Roadie::Stylesheet.new(name, css)
    end
  end

  def []=(path, stylesheet)
    memcache.write("assets/#{path}/css", stylesheet.to_s)
    memcache.write("assets/#{path}/name", stylesheet.name)
    stylesheet # You need to return the set Stylesheet
  end
end

document.external_asset_providers = Roadie::CachedProvider.new(
  document.external_asset_providers,
  MyRoadieMemcacheStore.new(MemcacheClient.instance)
)

If you are using Rspec, you can test your implementation by using the shared examples for the "roadie cache store" role:

require "roadie/rspec"

describe MyRoadieMemcacheStore do
  let(:memcache_client) { MemcacheClient.instance }
  subject { MyRoadieMemcacheStore.new(memcache_client) }

  it_behaves_like "roadie cache store" do
    before { memcache_client.clear }
  end
end

PathRewriterProvider

With this provider, you can rewrite the paths that are searched in order to more easily support another provider. Examples could include rewriting absolute URLs into something that can be found on the filesystem, or to access internal hosts instead of external ones.

filesystem = Roadie::FilesystemProvider.new("assets")
document.asset_providers << Roadie::PathRewriterProvider.new(filesystem) do |path|
  path.sub('stylesheets', 'css').downcase
end

document.external_asset_providers = Roadie::PathRewriterProvider.new(filesystem) do |url|
  if url =~ /myapp\.com/
    URI.parse(url).path.sub(%r{^/assets}, '')
  else
    url
  end
end

You can also wrap a list, for example to implement external_asset_providers by composing the normal asset_providers:

document.external_asset_providers =
  Roadie::PathRewriterProvider.new(document.asset_providers) do |url|
    URI.parse(url).path
  end

Writing your own provider

Writing your own provider is also easy. You need to provide:

  • #find_stylesheet(name), returning either a Roadie::Stylesheet or nil.
  • #find_stylesheet!(name), returning either a Roadie::Stylesheet or raising Roadie::CssNotFound.
class UserAssetsProvider
  def initialize(user_collection)
    @user_collection = user_collection
  end

  def find_stylesheet(name)
    if name =~ %r{^/users/(\d+)\.css$}
      user = @user_collection.find_user($1)
      Roadie::Stylesheet.new("user #{user.id} stylesheet", user.stylesheet)
    end
  end

  def find_stylesheet!(name)
    find_stylesheet(name) or
      raise Roadie::CssNotFound.new(
        css_name: name, message: "does not match a user stylesheet", provider: self
      )
  end

  # Instead of implementing #find_stylesheet!, you could also:
  #     include Roadie::AssetProvider
  # That will give you a default implementation without any error message. If
  # you have multiple error cases, it's recommended that you implement
  # #find_stylesheet! without #find_stylesheet and raise with an explanatory
  # error message.
end

# Try to look for a user stylesheet first, then fall back to normal filesystem lookup.
document.asset_providers = [
  UserAssetsProvider.new(app),
  Roadie::FilesystemProvider.new('./stylesheets'),
]

You can test for compliance by using the built-in RSpec examples:

require 'spec_helper'
require 'roadie/rspec'

describe MyOwnProvider do
  # Will use the default `subject` (MyOwnProvider.new)
  it_behaves_like "roadie asset provider", valid_name: "found.css", invalid_name: "does_not_exist.css"

  # Extra setup just for these tests:
  it_behaves_like "roadie asset provider", valid_name: "found.css", invalid_name: "does_not_exist.css" do
    subject { MyOwnProvider.new(...) }
    before { stub_dependencies }
  end
end

Keeping CSS that is impossible to inline

Some CSS is impossible to inline properly. :hover and ::after comes to mind. Roadie tries its best to keep these around by injecting them inside a new <style> element in the <head> (or at the beginning of the partial if transforming a partial document).

The problem here is that Roadie cannot possible adjust the specificity for you, so they will not apply the same way as they did before the styles were inlined.

Another caveat is that a lot of email clients does not support this (which is the entire point of inlining in the first place), so don't put anything important in here. Always handle the case of these selectors not being part of the email.

Specificity problems

Inlined styles will have much higher specificity than styles in a <style>. Here's an example:

<style>p:hover { color: blue; }</style>
<p style="color: green;">Hello world</p>

When hovering over this <p>, the color will not change as the color: green rule takes precedence. You can get it to work by adding !important to the :hover rule.

It would be foolish to try to automatically inject !important on every rule automatically, so this is a manual process.

Turning it off

If you'd rather skip this and have the styles not possible to inline disappear, you can turn off this feature by setting the keep_uninlinable_css option to false.

document.keep_uninlinable_css = false

Callbacks

Callbacks allow you to do custom work on documents before they are transformed. The Nokogiri document tree is passed to the callable along with the Roadie::Document instance:

class TrackNewsletterLinks
  def call(dom, document)
    dom.css("a").each { |link| fix_link(link) }
  end

  def fix_link(link)
    divider = (link['href'] =~ /?/ ? '&' : '?')
    link['href'] = link['href'] + divider + 'source=newsletter'
  end
end

document.before_transformation = ->(dom, document) {
  logger.debug "Inlining document with title #{dom.at_css('head > title').try(:text)}"
}
document.after_transformation = TrackNewsletterLinks.new

XHTML vs HTML

You can configure the underlying HTML/XML engine to output XHTML or HTML (which is the default). One usecase for this is that { tokens usually gets escaped to &#123;, which would be a problem if you then pass the resulting HTML on to some other templating engine that uses those tokens (like Handlebars or Mustache).

document.mode = :xhtml

This will also affect the emitted <!DOCTYPE> if transforming a full document. Partial documents does not have a <!DOCTYPE>.

Build Status

Tested with Github CI using:

  • MRI 2.6
  • MRI 2.7
  • MRI 3.0
  • MRI 3.1

Let me know if you want any other runtime supported officially.

Versioning

This project follows Semantic Versioning and has been since version 1.0.0.

FAQ

Why is my markup changed in subtle ways?

Roadie uses Nokogiri to parse and regenerate the HTML of your email, which means that some unintentional changes might show up.

One example would be that Nokogiri might remove your &nbsp;s in some cases.

Another example is Nokogiri's lack of HTML5 support, so certain new element might have spaces removed. I recommend you don't use HTML5 in emails anyway because of bad email client support (that includes web mail!).

I'm getting segmentation faults (or other C-like problems)! What should I do?

Roadie uses Nokogiri to parse the HTML of your email, so any C-like problems like segfaults are likely in that end. The best way to fix this is to first upgrade libxml2 on your system and then reinstall Nokogiri. Instructions on how to do this on most platforms, see Nokogiri's official install guide.

What happened to my @keyframes?

The CSS Parser used in Roadie does not handle keyframes. I don't think any email clients do either, but if you want to keep on trying you can add them manually to a <style> element (or a separate referenced stylesheet) and tell Roadie not to touch them.

My @media queries are reordered, how can I fix this?

Different @media query blocks with the same conditions are merged by default, which will change the order in some cases. You can disable this by setting merge_media_queries to false. (See Install & Usage section above).

How do I get rid of the <body> elements that are added?

It sounds like you want to transform a partial document. Maybe you are building partials or template fragments to later place in other documents. Use Document#transform_partial instead of Document#transform in order to treat the HTML as a partial document.

Can I skip URL rewriting on a specific element?

If you add the data-roadie-ignore attribute on an element, URL rewriting will not be performed on that element. This could be really useful for you if you intend to send the email through some other rendering pipeline that replaces some placeholders/variables.

<a href="/about-us">About us</a>
<a href="|UNSUBSCRIBE_URL|" data-roadie-ignore>Unsubscribe</a>

Note that this will not skip CSS inlining on the element; it will still get the correct styles applied.

What should I do about "Invalid URL" errors?

If the URL is invalid on purpose, see Can I skip URL rewriting on a specific element? above. Otherwise, you can try to parse it yourself using Ruby's URI class and see if you can figure it out.

require "uri"
URI.parse("https://example.com/best image.jpg") # raises
URI.parse("https://example.com/best%20image.jpg") # Works!

Documentation

Running specs

bundle install
rake

Security

Roadie is set up with the assumption that all CSS and HTML passing through it is under your control. It is not recommended to run arbritary HTML with the default settings.

Care has been given to try to secure all file system accesses, but it is never guaranteed that someone cannot access something they should not be able to access.

In order to secure Roadie against file system access, only use your own asset providers that you yourself can secure against your particular environment.

If you have found any security vulnerability, please email me at magnus.bergmark+security@gmail.com to disclose it. For very sensitive issues, please use my public GPG key. You can also encrypt your message with my public key and open an issue if you do not want to email me directly. Thank you.

History and contributors

This gem was previously tied to Rails. It is now framework-agnostic and supports any type of HTML documents. If you want to use it with Rails, check out roadie-rails.

Major contributors to Roadie:

You can see all contributors on GitHub.

License

(The MIT License)

Copyright (c) 2009-2022 Magnus Bergmark, Jim Neath / Purify, and contributors.

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the ‘Software’), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED ‘AS IS’, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

Author: Mange
Source code: https://github.com/Mange/roadie
License: MIT license

#ruby   #ruby-on-rails #html