It would be great if we could localize the replacements in convert_misc_characters.
Is this something that's seems reasonable ?

Thanks

Equals signs seems not to be stringified: "hello=world" seems to end up being "hello=world", when it probably could end up being "hello-equals-world".

I would like to suggest adding an option to NOT convert underscores into dashes, something like :convert_underscore => true (which would be the default option). It seems quite easy to do, but I rather let you do it than add a fork myself that would never get integrated into the official gem...

And by the way, I am missing a euro sign conversion.

I have not been able to come up with a 100% reproducible test case, but I have noticed several occasions now where a record is created and somehow stringex/acts_as_url creates a nil url.

even with auto sync set, updating the url attribute for this record does not cause the url to be re-generated.

it may have to do with plurals or creating a record whose url attribute contains a subset of the same url as an existing record like:
"Happy Farm"
"Happy Farms"

Will try some more extensive testing at some point in the near future.

class Article
  include Mongoid::Document
  include Stringex::ActsAsUrl

  field :title, type: String

  embeds_many :comments

  acts_as_url :title, sync_url: true

  # using Stringex gem - https://github.com/rsl/stringex
  def to_param
    url # or whatever you set :url_attribute to
  end
end

class Comment
  include Mongoid::Document
  include Stringex::ActsAsUrl

  embedded_in :article

  field :subject, type: String

  acts_as_url :subject, sync_url: true

  # using Stringex gem - https://github.com/rsl/stringex
  def to_param
    url # or whatever you set :url_attribute to
  end
end

article = Article.create title: 'hello'
Comment.create subject: 'nice one', article: article

$ rake db:seed

Problem:
Problem:
  Document(s) not found for class Article with id(s) all, {:conditions=>["url LIKE ?", "hello%"]}.
Summary:
  When calling Article.find with an id or array of ids, each parameter must match a document in the database or this error will be raised. The search was for the id(s): all, {:conditions=>["url LIKE ?", "hello%"]} ... (2 total) and the following ids were not found: all, {:conditions=>["url LIKE ?", "hello%"]}.
Resolution:
  Search for an id that is in the database or set the Mongoid.raise_not_found_error configuration option to false, which will cause a nil to be returned instead of raising this error when searching for a single id, or only the matched documents when searching for multiples.

Looks like it is hardcoded to use SQL like syntax such as "url LIKE".

From MongoDB documentation...

http://api.mongodb.org/wiki/current/Advanced%20Queries.html

You may use regexes in database query expressions:

db.customers.find( { name : /acme.*corp/i } );

Translating this to Mongoid 3.x

article = Article.create title: 'hello you'
puts Article.where(title: /hello.*/).first.inspect

=>
#<Article _id: 50b88995ff0961046e000001, _type: nil, created_at: 2012-11-30 10:25:25 UTC, updated_at: 2012-11-30 10:25:25 UTC, title: "hello you">

The key is to replace the {:conditions=>["url LIKE ?", "hello%"]} with where(title: /hello.*/).first for mongoid.

It would be nice to abstract the SQL part into an SQL adapter and then also include a Mongoid adapter as well IMO. I will fork the project and see if I can make such a solution, then make a pull request.
For now, please mention that it only works for "SQL based" models.
Thanks!

The new stringex gem shipped today has a dependency on itself.

http://rubygems.org/gems/stringex

I encounter this bug on my machine with ruby 1.9.2dev (2010-04-24 trunk 27474) [x86_64-darwin10.3.0] and ruby-1.9.1-p376

@@@

trung-les-mac-pro:stringex MAC$ gem install stringex-1.1.0.gem
Successfully installed stringex-1.1.0
1 gem installed
Installing ri documentation for stringex-1.1.0...
Installing RDoc documentation for stringex-1.1.0...
ERROR: While executing gem ... (Encoding::CompatibilityError)
incompatible character encodings: UTF-8 and ASCII-8BIT

@@@

If you to_url on an ellipsis you get ... (three dots). If you to_url those dots again you get "dot-dot-dot"

Having dots in a to_param slug messes up a lot of stuff in rails, so I think you should translate an ellipsis directly to dot-dot-dot

Currently duplicates in ActsAsUrl are checked only within the scope of the current class (see https://github.com/rsl/stringex/blob/master/lib/stringex/acts_as_url.rb#L97).

When using Single Table Inheritance this may sometimes be the wanted behaviour but there could be some cases where duplicates should be checked across the differents submodels.

I'd submit a patch but I don't see how to change the current implementation. Maybe adding an option to specify the class name of the model to check for duplicates ?

I'm trying to use stringex and acts_as_url on a model.

I've got my model set up like this:
acts_as_url :name, :url_attribute => :permalink, :only_when_blank => true

However, there are times where I'd like to to directly set a value of :permalink that's not generated from the :name, but it seems that acts_as_url isn't allowing it to happen.

Is there a way to allow this to happen?

In the current edge rails it seems before_validation_on_create is deprecated in favor of
before_validation :asdf, :on => :create

Acts as url would need to be patched to conform to this.

Use case: I have a model whose title is used to create the url / permalink. I'm using sync_url => true. update_attributes is called with a title value that fails validation (blank value when presence is required, for example).

When this happens, my Rails application re-renders the template with errors so the user can fix the validation problems. However, the url value has changed. Now, route shortcuts are broken because the model instance has an empty url, when the url probably never should have been updated at all.

I created a failing test to demonstrate the scenario. Can work on a patch if this makes sense (note that an after_validation callback won't fire if validation fails):

zapnap@00ada38

in https://github.com/rsl/stringex/blob/master/lib/stringex/acts_as_url.rb#L109 it does a find(:all) on the model, which in case of an empty strings or nil in url_attribute, will be all records (> 70k in my case).

I think you need a special strategy to handle empty strings or nil in url_attribute, for the other cases I would also recommend not using find(:all)

is there any good workaround for edge cases such as an input value of "???" which will be turned into the empty string "" by the to_url method

Those should be avoided, as rails cannot handle it as an id to search by.

best regards,

Lennart

Shouldn't this work?

assert_equal 'utf', 'μ†ℱ'.to_url

It's translated to

• m+f in 1.4
• m-plus-f in 1.5

As far as I know, | symbol is not allowed in urls, but .to_url method doesn't remove it.

I propose that the duplicate urls be distinguished with two dashes instead of one. I think that it provides a better semantic.

ex: /peugeot-206--2/ instead of /peugeot-206-2/

This, I think, means that the line 82 of acts_as_url.rb should be "#{base_url}--#{n}".

Hi

I'm successfully using stringex for URLs like /blog/the-title-of-my-post

Is it possible to achieve something like /blog/2010/08/the-title-of-my-post?

One solution I'm thinking of is to populate a 'year_month_title' field on creation that contains both the year, month and title, then use said field with :acts_as_url. Is this the easiest way or have I overlooked an alternative?

Thanks!

Ben

CAUTION: this is hella nitpicky and probably doesn't matter at all, but i'm an OCD freak, what can i say?

module Stringex
  module StringExtensions
    # Returns string with its UTF-8 characters transliterated to ASCII ones. Example:
    #
    #   "⠋⠗⠁⠝⠉⠑".to_ascii #=> "braille"
    def to_ascii
      Stringex::Unidecoder.decode(self)
    end
  end
end

but when i actually run that code:

[Dev]> "⠋⠗⠁⠝⠉⠑".to_ascii #=> "braille"
=> "france"

The following spec fails:

def test_localized_html_entities_conversion_with_blank_replacement
    Stringex::Localization.backend = :internal
    Stringex::Localization.store_translations :da, :html_entities, {
       :nbsp => ""
    }
    Stringex::Localization.locale = :da

    {
      "Det var blå sort" => "Det var blåsort"
    }.each do |entitied, plain|
      assert_equal plain, entitied.convert_miscellaneous_html_entities
    end
end

"Det var blå sort".convert_miscellaneous_html_entities returns "Det var blå sort" with the default replacement for   still in place.

I understand the reasoning behind something like "google.com" -> "google dot com"

However, I feel that for abbreviations, "D.N.A." would be better converted to "dna"

Yay/nay ? I'd be happy to work on and submit a patch.

The documentation in README.rdoc sais:

In your controllers you will need to call Foo.find_by_url(params[:id]) instead of the regular find. - line 36

However, I simply use .find, and this works, while find_by_url doesn't find the correct object. Is the documentation outdated or is there something else going on?

The transliteration of ǝ produces an @ in the final URL which isn't legal :(

Hello,
Thank you for create very nice module.
When I read a document, I found the example below.
"tell your readers 你好".to_url => "tell-your-readers-ni-hao"

Could I make a slug as real Unicode likes below one?

"tell your readers 你好 bla".to_url => "tell-your-readers-你好-bla"

Have a nice day!

when I test with mixture of Chinese and English. The result is puzzled

1.9.3-p327-falcon :001 > require 'stringex'
 => true 
1.9.3-p327-falcon :002 > "中文yinwen混合".to_url
 => "zhong-wen-yinwenhun-he"

maybe zhong-wen-yinwen-hun-he is better?

I have next Japanease and Russian strings:
"AVアンプ、ホームシアターシステム、スピーカーシステム等".to_url
"У лукоморья дуб зеленый".to_url
This returns:
avanpu,-homusiatasisutemu,-supikasisutemudeng
u-lukomor'ia-dub-zielienyi
I think it's no good to have commas or apostrophes in the url.
Maybe will be fine to add somethig like that to the end of "to_url" method:
.gsub(/[^a-zA-Z0-9-]/, '')
or just add standard
.parameterize
What do you think about that?

Working on configuration changes right now so great time to do it. akradev@1ec9a04 is an example of a user needing to skip loading acts_as_url. Could be a sign of need to separate concerns into separate gems in the future but just looking at configuration concerns for the moment.

Here is the backtrace when I require 'stringex' with Ruby 1.9.1.
Did you plan Ruby 1.9.1 support ? As strings are handled differently, I suppose that it's a big work.

irb(main):001:0> require 'stringex'
SyntaxError: /opt/local/lib/ruby/gems/1.9.1/gems/rsl-stringex-1.0.0/lib/lucky_sneaks/string_extensions.rb:123: invalid multibyte char (US-ASCII)
/opt/local/lib/ruby/gems/1.9.1/gems/rsl-stringex-1.0.0/lib/lucky_sneaks/string_extensions.rb:123: invalid multibyte char (US-ASCII)
/opt/local/lib/ruby/gems/1.9.1/gems/rsl-stringex-1.0.0/lib/lucky_sneaks/string_extensions.rb:123: syntax error, unexpected $end, expecting tASSOC
        /(\s|^)£(\d+)\.(\d+)(\s|$)/u => '\2 pounds \3 pence',
                 ^
    from /opt/local/lib/ruby/gems/1.9.1/gems/rsl-stringex-1.0.0/lib/stringex.rb:1:in `require'
    from /opt/local/lib/ruby/gems/1.9.1/gems/rsl-stringex-1.0.0/lib/stringex.rb:1:in `'
    from (irb):1:in `require'
    from (irb):1
    from /opt/local/bin/irb:12:in `'

When using comma as a hundreds separator, e.g.

"Save $1,000 now".to_url

becomes

"save-1-dollar-000-now"

instead of

"save-1000-dollars-now"

The correct output would be the same output as

"Save $1000 now".to_url

I'll complete it. : )

"カッページ・テラスに日系カフェ＆バー、店内にDJブースも - シンガポール経済新聞".to_url
=> "katupeziterasuniri-xi-kahue&ba,-dian-nei-nidjbusumo-singaporujing-ji-xin-wen"

the grave accent should always be removed

Would be nice to be able to set the global defaults for Config, such as:

# config/initializers/acts_as_url.rb

StringEx::ActsAsUrl.config do |c|
  c.sync_url = true
  ...
end

So I won't have to set this option for all my models if they should all be synced ;) A common use case I would suspect.

This is the best slug library I've come across, but why tie it to ActiveRecord? Lately I've been starting all new projects on Mongoid. Should be easy enough to change, no?

to_url doesn't remove the grave accent symbol despite the fact, that it's not allowed in urls.
Also okina symbol ʻ is replaced with the grave accent instead of being removed.

I use ruby 1.9.1(p376)
gem rdoc stringex --no-ri -V
Installing RDoc documentation for stringex-1.1.0...
ERROR: While executing gem ... (Encoding::CompatibilityError)
incompatible character encodings: UTF-8 and ASCII-8BIT

We have a feature where users can input custom urls, and we allow '' in those urls. Is there a simple way to disable the current conversation from '' to '-' in stringex?

I have the following in my da.yml-file:

stringex:
  characters:
    dot: "-"

When I do

"Spar 1.000 kr".to_url

I get

spar-00-kr

in return, where I should get

spar-1-000-kr

Also, having dot: '' and doing the same conversion yields:

spar-1-dot-000-kr

Meaning that it doesn't respect my choice of the empty string.

Looks like you need to explicitly include the Stringex::ActsAsUrl module in your model in order to add the acts_as_url class method.

Please add this to the top of the README for newbies ;) hehe

I think it would be cool if you could set a flag on the the acts_as_url call to disable the ensure unique behavior altogether. I sometimes just want a failing validation. If this happened then we would need to change ensure_unique_url to something like format_url which then calls the unique check only if the option is selected (which would be the default for compat). I can fork and code this up...

I have next string:
"éç™①".to_ascii => "ec"
So "TM1" is missing.
I propose to add Unicode::normalize_KD to "to_ascii" method.
Like that:
require "unicode"
def to_ascii
LuckySneaks::Unidecoder.decode(Unicode::normalize_KD(self))
end
Than would be:
"éç™①".to_ascii => "ecTM1"
I think this much great!..

When LuckySneaks::Unidecoder.decode is used in JRuby 1.6 under --1.9 mode, it returns strings that are completely different from --1.8 mode (the default), MRI and Rubinius.

See the attached test case: https://gist.github.com/886474

I looked into this a bit and it appears as if the psych YAML parser, being used in --1.9 mode (and slated for future use by MRI), is failing to parse all the unidecoder_data files. I couldn't see this until I commented out the rescue block in the code and let the exceptions propagate up the stack.

I'm not sure whether the YAML files are invalid, or psych is raising an exception when it shouldn't. I was thinking that maybe if the YAML files could be regenerated from the original source using psych that it might escape things in a way that is closer to the current YAML standard, and easier for psych to parse.

What do we think of allowing acts_as_url to allow a limit to be specified? The reason I ask is because I set a limit on my slug fields in my database to keep index key lengths sane.

If i set the url to nil to clear the url and save the Page it dosen't regenerate the url.
Model:
acts_as_url :title, :only_when_blank => true
console: http://pastie.org/537638

code

# encoding: utf-8

require 'stringex'

puts "a.b".to_url
puts "a。b".to_url

result

a-dot-b
a.-b

When using stringex with PSYCH as a YAML backend, .to_url fails throwing a : Psych::SyntaxError
alex@1:~/diveboard-web$ rails console
Loading development_alex environment (Rails 3.0.6)
irb(main):001:0> require 'stringex'
=> nil
irb(main):002:0> "Sébastien".to_url
Psych::SyntaxError: couldn't parse YAML at line 2 column 3
from /usr/lib/ruby/1.9.1/psych.rb:148:in parse' from /usr/lib/ruby/1.9.1/psych.rb:148:inparse_stream'
from /usr/lib/ruby/1.9.1/psych.rb:119:in parse' from /usr/lib/ruby/1.9.1/psych.rb:106:inload'
from /usr/lib/ruby/1.9.1/psych.rb:205:in load_file' from /var/lib/gems/1.9.1/gems/stringex-1.2.1/lib/lucky_sneaks/unidecoder.rb:8:inblock in module:Unidecoder'
from /var/lib/gems/1.9.1/gems/stringex-1.2.1/lib/lucky_sneaks/unidecoder.rb:19:in yield' from /var/lib/gems/1.9.1/gems/stringex-1.2.1/lib/lucky_sneaks/unidecoder.rb:19:indefault'
from /var/lib/gems/1.9.1/gems/stringex-1.2.1/lib/lucky_sneaks/unidecoder.rb:19:in block in decode' from /var/lib/gems/1.9.1/gems/stringex-1.2.1/lib/lucky_sneaks/unidecoder.rb:16:ingsub'
from /var/lib/gems/1.9.1/gems/stringex-1.2.1/lib/lucky_sneaks/unidecoder.rb:16:in decode' from /var/lib/gems/1.9.1/gems/stringex-1.2.1/lib/lucky_sneaks/unidecoder.rb:59:into_ascii'
from /var/lib/gems/1.9.1/gems/stringex-1.2.1/lib/lucky_sneaks/string_extensions.rb:45:in remove_formatting' from /var/lib/gems/1.9.1/gems/stringex-1.2.1/lib/lucky_sneaks/string_extensions.rb:39:into_url'
from (irb):2
from /var/lib/gems/1.9.1/gems/railties-3.0.6/lib/rails/commands/console.rb:44:in start' from /var/lib/gems/1.9.1/gems/railties-3.0.6/lib/rails/commands/console.rb:8:instart'
from /var/lib/gems/1.9.1/gems/railties-3.0.6/lib/rails/commands.rb:23:in <top (required)>' from script/rails:6:inrequire'
from script/rails:6:in `

A quick temporary fix is to force the use of old syck parser :
in : config/boot.rb

require 'yaml'
YAML::ENGINE.yamler= 'syck'

yet ultimately this issue should be properly addressed :) (syck is not maintained anymore)

Rails 3.2.3 , Ruby 1.9.3-p125 , Stringex 1.4.0

Klass.initialize_urls is just hanging... no progress for hours

Klass is the base-class of an STI class hierarchy.

example:

'Îò÷åò'.to_url
 => "io/ao"

'Îò÷åò'.to_url.to_url
 => "io-slash-ao"

The method to_ascii should goes first.

Options should enable passing an array of strings to exclude from to_url. This allows handling of special cases where it's not desirable to alter the input. Example use.