It is not mentioned in README. For example, if I initialize a hashids instance at startup, and have several concurrent threads encoding and decoding with the instance, would there be race condition? In other words, will the #encode, #decode, #encode_hex and #decode_hex modify the hashids instance?

Most people here probably know these gems:
https://github.com/namick/obfuscate_id
https://github.com/mguymon/obfuscate
They are quite easy to integrate in existing models by simply adding one line so i tried to port this to this hashids gem. While doing this i stumbled across a problem which you will see in step 2.

Step 1:
Create a file called 'obfuscate_id.rb' in 'app/models/concerns' and put this code inside:

module ObfuscateId
  require 'hashids'
  extend ActiveSupport::Concern

  module ClassMethods
    def hashids
      Hashids.new(table_name, 8)
    end

    def encode_id(id)
      hashids.encode(id)
    end

    def decode_id(id)
      hashids.decode(id).first
    end
  end

  included do
    define_method :to_param do
      self.class.encode_id(id)
    end

    define_singleton_method :find do |*args|
      super(decode_id(args.first))
    end

    define_method :is_obfuscated do
    end
  end
end

Now you can add this line to your model if you want to obfuscate its id:

include ObfuscateId

At this state you have the functionality of the 2 Gems i mentioned it the beginning, but are using hashids

Step 2:
The problem now is that this will only work if you call the find directly on the model
Because if you call something like for example: User.all.find(...) the find will be called on ActiveRecord::AssociationRelation and not on the model, and we didn't overwrite this.
Same problem with current_user.company.users.find(...) but this would be ActiveRecord::Associations::CollectionProxy
This also causes for example the cancancan gem to not work anymore.

I have a solution, but it's ugly and was the reason for me to open this issue, so i hope others can contribute on how to solve this in a better way.

Solution: Create an initializer and put this code inside:

require 'hashids'

class ActiveRecord::Associations::CollectionProxy
  def find(*args)
    if first.class.method_defined? :is_obfuscated
      args = hashids.decode(args.first).first
    end
    super(args)
  end
end

class ActiveRecord::AssociationRelation
  def find(*args)
    if first.class.method_defined? :is_obfuscated
      args = hashids.decode(args.first).first
    end
    super(args)
  end
end

What's is wrong with my code ?

Thanks

2.1.1 :007 > require 'hashids'
 => true 
2.1.1 :008 > h = Hashids.new
 => #<Hashids:0x000000016506c0 @salt="", @min_hash_length=0, @alphabet="gjklmnopqrvwxyzABDEGJKLMNOPQRVWXYZ1234567890", @seps="cfhistuCFHISTU", @guards="abde"> 
2.1.1 :009 > h.encode(111)
NoMethodError: protected method `encode' called for #<Hashids:0x000000016506c0>
    from (irb):9
    from /home/user/.rvm/rubies/ruby-2.1.1/bin/irb:11:in `<main>'

I would like to be able to encrypt negative numbers too.

Hi @peterhellberg

I saw you fixed to support Ruby 2.4.0 on master branch, but did not ship a new version on rubygems.

Could you do that, plz?

Many use UUID as primary keys, because of the many advantages this brings (with UUIDv7, they become even more interesting).

Thus, it would be awesome to be able to use your gem in these situations, too.

The salt appears to have a maximum length after which it has no longer impacts the generated hash.

Here some examples:

Hashids.new('a' * 41).encode(1)
=> "pn" 
Hashids.new('a' * 42).encode(1)
=> "nd" 
Hashids.new('a' * 43).encode(1)
=> "yQ" 
Hashids.new('a' * 44).encode(1)
=> "yQ" 
Hashids.new('a' * 45).encode(1)
=> "yQ" 

If i change the alphabet it also affects this maximum length:

Hashids.new('a' * 22, 0, 'abcdefghijkmnopqrstuvwxyz023456789').encode(1)
=> "b0" 
Hashids.new('a' * 23, 0, 'abcdefghijkmnopqrstuvwxyz023456789').encode(1)
=> "bd" 
Hashids.new('a' * 24, 0, 'abcdefghijkmnopqrstuvwxyz023456789').encode(1)
=> "bd" 
Hashids.new('a' * 25, 0, 'abcdefghijkmnopqrstuvwxyz023456789').encode(1)
=> "bd" 

Looks like something around (alphabet.length * 0.7).floor is the maximum length?

Should this be validated and raise a SaltError when the salt is too long?

Hey!
Now we can use only string of chars as a alphabet for generating HashID. WDYT if gem will allow to use array of chars as a alphabet too?

Something like:

alphabet = ['a1', 'b2']
hashids = Hashids.new("this is my salt", 0, alphabet)
hash = hashids.encode(1) # => "b2a1b2"

It will be helpful for generating hashes without any bad words or generating hashes with specific chars combinations. WDYT?

Hello and thanks for this gem!

If i'm creating hashids for different ids it gave me duplicate hashes:

hashids = Hashids.new('Whoops is when you accidentally douche with Drano!', 12)
hashids.encode(16137)
hashids.encode(16181)

hashids gem 1.0.3 rails 4.2.9 ruby-2.1.10

Can you reproduce it?

Thanks and best wishes.

I discovered a hash that decodes to 2 values in my codebase. My understanding from the documentation is there should be no collisions. Is this a bug?

hex = "xaFG9d"
alphabet = "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ1234567890"
h1 = Hashids.new("developments", 6, alphabet)
h1.decode(hex)
# => [29, 76161]
h2 = Hashids.new("tenants", 6, alphabet)
h2.decode(hex)
# => [42, 28612]

The following piece of code:

hashids = Hashids.new('this is my salt', 0, '0123456789abcdef')
id = hashids.encode(18446744073709551615)
numbers = hashids.decode(id)

raises a Hashids::AlphabetError with message Alphabet must contain at least 16 unique characters.

The validate_alphabet call needs to be moved after @alphabet assignment.

Hi @peterhellberg

I would like to setup my app to have frozen strings by default.
Example description: https://www.mikeperham.com/2018/02/28/ruby-optimization-with-one-magic-comment/

Using hashids is not possible with globally enabled frozen string. This can be seen by:

#> RUBYOPT="--enable=frozen-string-literal" irb
irb(main):001:0> require 'hashids'
=> true
irb(main):002:0> Hashids.new('secret')
/home/wojtek/.asdf/installs/ruby/3.0.2/lib/ruby/gems/3.0.0/gems/hashids-1.0.5/lib/hashids.rb:217:in `delete!': can't modify frozen String: "cfhistuCFHISTU" (FrozenError)
	from /home/wojtek/.asdf/installs/ruby/3.0.2/lib/ruby/gems/3.0.0/gems/hashids-1.0.5/lib/hashids.rb:217:in `setup_seps'
	from /home/wojtek/.asdf/installs/ruby/3.0.2/lib/ruby/gems/3.0.0/gems/hashids-1.0.5/lib/hashids.rb:199:in `setup_alphabet'
	from /home/wojtek/.asdf/installs/ruby/3.0.2/lib/ruby/gems/3.0.0/gems/hashids-1.0.5/lib/hashids.rb:23:in `initialize'

I could prepare a PR if you're interested in making it fixed.

var hashids = new Hashids("a5ifxlwfM6qCpOfvlgvW", 8, "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ1234567890");

var a = hashids.encode(2713);
var b = hashids.encode(3057);

// a : "MrQkaLzd"
// b : "MrQkaLzd"

Using latest version.

I need to generate hashids from Python, but can't figure out what salt is used by default in the Ruby on Rails deploy.

This would be an awesome detail to add to the README!

Presuming that default alphabet is used:

Hashids.new('salt').decode('asdf-')
NoMethodError: undefined method '*' for nil:NilClass
from /gems/hashids-1.0.1/lib/hashids.rb:183:in `block in unhash'

Edit: Ok, taking it further:

Hashids.new('endpoints').decode('-')
=> []

I’ve started working on the support for 0.3.0 but it is slow going.

I was wondering because there is https://github.com/sqids/sqids-ruby repo...