Mysterious Ruby Block behavior: &block vs. {block.call} - ruby-on-rails

When writing a helper for printing javascript that can be used from both other helpers and views, I stumbled upon the following problem:
def javascript(print_tag = false, &block)
content_for(:javascript) do
if print_tag
javascript_tag(&block) # does not work
javascript_tag { block.call } # does work
else
capture(&block)
end
end
end
This helper should be called with javascript { "alert('hurray'); }.
In the first alternative - which I expected to work - the Rails javascript_tag helper renders an empty <script type="text/javascript"> //<![CDATA[ //]]> </script> tag.
The second alternative, however, works as expected.
What's going on there? How can that be different?

You say you are doing this on your views, right?
<%= javascript { "alert('hurray');" } %>
But for content_tag(&block) to work, you should call javascript the way content_tag is intended to be used in views, which is:
<% javascript do %>
alert('hurray');
<% end %>
content_tag's behavior is different depending on where it's called from, see the function block_called_from_erb? in the source code. In the first case this function returns true because the block does come from an erb (and then it's concated, you don't want that!), in the second returns false (you re-created the block from scratch) and content_tag simply returns the string content, which is what you want.
# ./action_view/helpers/javascript_helper.rb
tag = content_tag(:script, javascript_cdata_section(content), html_options.merge(:type => Mime::JS))
if block_called_from_erb?(block)
concat(tag)
else
tag
end

Related

how to passing data with javascript in rails [duplicate]

In Rails 3.1 it is not possible to access controller instance variables in an asset js.erb or coffee.erb file using syntax such as <%= #foo %>, where #foo is set in the controller. So then the question is what are the best ways for passing controller variables to CoffeeScript or JavaScript assets.
This question has kind of been asked in multiple convoluted forms on the forum, but my point in asking it again is to have a place where all recommendations are gathered together, and the code supplied is simple and readable. Also note that I'm specifically referring to assets and not view response files.
a couple of ways I have done this in the past
put the data in hidden fields, access the data in js/coffee
# single value
<%= hidden_field_tag "foo_name", #foo.name, { :id => "foo-name" } %>
$('#foo-name').val();
# when the 'value' has multiple attributes
<%= hidden_field_tag "foo", #foo.id, { :id => "foo", "data-first-name" => #foo.first_name, "data-last-name" => #foo.last_name } %>
$foo = $('#foo')
console.log $foo.val()
console.log $foo.data("firstName")
console.log $foo.data("lastName")
another option: load data into js data structure in erb, access it from js/coffee
<% content_for(:head) do %>
<script>
window.App = window.App || {};
window.App.Data = window.App.Data || {};
window.App.Data.fooList = [
<% #list.each do |foo| %>
<%= foo.to_json %>,
<% end %>
];
</script>
<% end %>
# coffee
for foo in window.App.Data.fooList
console.log "#{foo.id}, #{foo.first_name} #{foo.last_name}"
I am not a big fan of constructing javascript data from ruby in erb like this, something about it just feels wrong - it can be effective though
and another option: make an ajax call and get the data on-demand from the server
I am also interested in other ideas and approaches
There is a really nice rail cast and quite recent (feb. 2012) about this specific topic:
#324 Passing Data to JavaScript
It shows 3 ways: a script tag, a data attribute, and the Gon gem.
I think house covered all the available techniques. I would only mention that using an AJAX call is best used when you have a large volume of data, dynamic data or combination of both.
Rather than use a hidden field I chose to add a data attribute to the container div which jquery can pick up.
<div class="searchResults" data-query="<%= #q %>"></div>
then the jquery to access it
url: "/search/get_results?search[q]=" + $(".searchResults").data("query") + "&page=" + p
I feel this is the cleanest way to pass data to javascript. After having found no way to pass a variable to a coffee script file with the rails asset pipeline from a controller. This is the method I now use. Can't wait till someone does set up the controller way with rails that will be the best.
In the controller:
#foo_attr = { "data-foo-1" => 1, "data-foo-2" => 2 }
In the view (HAML):
#foo{#foo_attr}
In the CoffeeScript asset:
$("#foo").data("foo-1")
$("#foo").data("foo-2")
In situations where your javascript data gets out of hand, using the gon gem is still the preferred way to go in rails, even in 2015. After setting up gon, you are able to pass data to your javascript files by simply assigning the data to the gon object in rails.
(Gemfile)
gem 'gon'
(controller)
def index
gon.products = Product.all
(layouts)
<%= include_gon %>
(public/javascripts/your_js_can_be_here.js)
alert(gon.products[0]['id');
(html source automatically produced)
<script>
window.gon = {};
gon.products = [{"created_at":"2015", "updated_at":"2015, "id":1, "etc":"etc"}];
You can read more verbose implementation details on Gon or the two other rails-javascript channels from Ryan Bate's screencast.
http://railscasts.com/episodes/324-passing-data-to-javascript
You can edit and add variables to the params array in the controller then access them in the response.js.erb. Here's an example with params[:value]:
def vote
value = params[:type] == "up" ? 1 : -1
params[:value] = value
#public_comment = PublicComment.find(params[:id])
have_voted = #public_comment.evaluators_for(:pub_votes_up) << #public_comment.evaluators_for(:pub_votes_down)
unless have_voted.include?(#current_user) # vote
#public_comment.add_or_update_evaluation(:"pub_votes_#{params[:type]}", value, #current_user)
else # unvote
#public_comment.delete_evaluation(:"pub_votes_#{params[:type]}", #current_user)
params[:value] = 0
end
respond_to do |format|
format.js # vote.js.erb
end
end
And here's an example accompanying response.js.erb
button = $('<%= ".pub#{params[:type]}_#{params[:id]}" %>')
label = button.find('strong')
<% comment = PublicComment.find(params[:id]) %>
label.html('<%= comment.reputation_for(:"pub_votes_#{params[:type]}").to_i %>')
<% if params[:value] == 1 %>
button.addClass('btn-success')
<% elsif params[:value] == -1 %>
button.addClass('btn-danger')
<% else %>
if button.hasClass('btn-success') { button.removeClass('btn-success') }
if button.hasClass('btn-danger') { button.removeClass('btn-danger') }
<% end %>

Rails template set value to array of strings

I have an angular module where I set a few strings. I'd also like to set an array of strings in the template as well.
application.html.erb
<script type="text/javascript">
angular.module('userFromServer', [])
.service('currentUser', function() {
<% if logged_in %>
this.name = '<%= #User.name %>';
this.friends = '<%= #User.profile.friends_by_uuid %>';
<% end %>
})
</script>
controller
def friends_by_uuid
self.friends.map{|x| "puser_#{x.uuid}"} # also tried adding .to_json
end
However the output appears to have some escaping issues.
"["puser_589b07ee-b8f1-4214-941d-0ce0b7a6703b", "puser_ec7d2918-d514-4c42-91fc-641ed2958fcd"]"
var desired_output = "["puser_589b07ee-b8f1-4214-941d-0ce0b7a6703b", "puser_ec7d2918-d514-4c42-91fc-641ed2958fcd"]
How can I render an array of strings in a rails template?
You should not put logics in views or controllers: Is it bad to have fat views in rails? Fat models are expected. So the solution is chunk the array in models / using helper functions, feed it to controller, and finally use in views. Since no idea what exactly you want, this is the most I can do.
For example:
# controller
def index
#ids = MyModel.split(params[:ids]) # suppose params[:ids] are the data coming in
end
# view
<% #ids.each do |id| %>
# ...
<% end %>
For ruby to process strings, many posts may help. For example, here.

Using Mustache lambda functions in ERB

I'm trying to share my Mustache templates in Rails across the server and the client on the lines of this Railscast.
All is well, except that I'm unable to figure out where and how to put the definition of a lambda function on the server side.
Let's say my html.erb looks like:
<% if params['client_side'].nil? %>
<%= render 'template', :mustache => #post %>
<% else %>
<script type="text/template" id="template">
<%= render 'template' %>
</script>
<% end %>
The Mustache handler looks like this (exactly as in the Railscast):
module MustacheTemplateHandler
def self.call(template)
if template.locals.include? 'mustache'
"Mustache.render(#{template.source.inspect}, mustache).html_safe"
else
"#{template.source.inspect}.html_safe"
end
end
end
ActionView::Template.register_template_handler(:mustache, MustacheTemplateHandler)
Now for the following template:
<h1>{{title}}</h1>
<div>
{{#marked}}{{content}}{{/marked}}
</div>
the lambda marked is easy to handle in JavaScript, but how can I define it in my Rails code to render content using Redcarpet?
Update
Since posting this, I have tried to expand on the idea of helper functions in the screencast. I now have
<% if params['client_side'].nil? %>
<%= render 'template', :mustache => process(#post) %>
<% else %>
...
The process is defined in ApplicationHelper as
def process(obj)
{
marked: lambda {|text| markdown(Mustache.render(text))}
}
end
This has two problems:
text inside the (Ruby) lambda function is indeed '{{content}}', but Mustache.render(text) fails to do anything with it — it's returning an empty string.
The above code will now only render the marked field and I haven't been able to find a way to retain the other (unprocessed) attributes of the author object (e.g. title). obj.attributes seems like a promising start, but I don't know how to combine the processed response for marked with the other attributes even if #1 above worked.
I got this working myself. The process method in ApplicationHelper now looks like this (using the new lambda syntax):
def process(obj)
obj['marked'] = ->(text) { markdown(Mustache.render(text, obj)) }
obj.attributes
end
This will now catch all invocations of marked in any template.

Render ERB as HTML and ERB from a Rails View

I'm making a style guide where I output the code on the right that is displayed on the left.
I know that adding %% escapes ERB
I have written a helper that takes the contents of a block and renders the code in two places one showing the html and I want the other to show the source ERB that created the html.
The problem is I get back HTML where I wanted ERB.
The View Code
<%= display_code do %>
<%= link_to "Button", "/style_guide, class: "btn" %>
<% end %>
The Helper Code
module StyleGuideHelper
def display_code(&block)
content = with_output_buffer(&block)
html = ""
html << content_tag(:div, content, class: "rendered-code")
html << content_tag(:div, escape_erb(content), class: "source-code-preview")
html.html_safe
end
def escape_erb(code)
code = code.gsub("%=", "%%=")
end
end
Expected Result
Button <%= link_to "Button", "/style_guide, class: "btn" %>
Actual Result
Button Button
Cheers
The issue is that this helper runs the block (link_to "Button", ...) -- it never sees the source code inside the block, just its output. You could replace escape_erb with h to capture the resulting HTML, but that won't pop back up to the ERB that generated it.
As I see it, your options are:
Break out examples into partials, then make a helper that a) renders the partial and b) displays the underlying file.
Specify your ERB fragments as strings (heredocs?), pass the string into the helper, and have the helper a) evaluate it via ERB.new(string).result(binding) to render the result and b) display the string.
Make the helper determine what part of the view invoked it, then parse the .erb well enough to find the block. Catch is, the precise format of what you see in callers is subject to change without notice due to the way views are compiled.
Make a helper that uses crazy metaprogramming juju to evaluate the block in both an ERB context as well as your own special context that intercepts the code being evaluated and turns it back into markup.
...sorted in approximate order of complexity and odds of success.
This code below will allow you to retrieve the code for a given block.
class ERBSource
ERB = ::ActionView::Template::Handlers::ERB
def self.for(block)
new(block).source
end
attr_reader :block, :file, :line_number
def initialize(block)
#block = block
#file, #line_number = *block.source_location
end
def source
lines = File.readlines(file)
relevant_lines = lines[(line_number - 1)..-1] || []
extract_first_expression(relevant_lines)
end
private
def extract_first_expression(lines)
code = lines.slice[0,1].join # add the first two lines so it has to iterate less
lines.each do |line|
code << line
return code if correct_syntax?(compile_erb(code))
end
raise SyntaxError, "unexpected $end"
end
def correct_syntax?(code)
stderr = $stderr
$stderr.reopen(IO::NULL)
RubyVM::InstructionSequence.compile(code)
$stderr.reopen(stderr)
true
rescue Exception
$stderr.reopen(stderr)
false
end
def compile_erb(code)
ERB.erb_implementation.new(
code,
:escape => false,
:trim => (ERB.erb_trim_mode == "-")
).src
end
end
This is what the helper looks like
module StyleGuideHelper
def render_example(name, &block)
code = ERBSource.for(block)
content_tag(:h2, name) +
content_tag(:div, &block) +
content_tag(:pre, content_tag(:code, code))
end
end

How to run some code only once in view

I have a partial view called '_comment.erb', and it may be called by parent many times(e.g. in a loop). The '_comment.erb' looks like:
<script>
function aaa() {}
</script>
<%= comment.content %>
<%=link_to_function 'Do', 'aaa()' %>
You can see if the '_comment.erb' be called many times, that the javascript function 'aaa' will be re-defined many times. I hope it can be define only once, but I don't want to move it to parent view.
I hope there is a method, say 'run_once', and I can use it like this:
<%= run_once do %>
<script>
function aaa() {}
</script>
<% end %>
<%= comment.content %>
<%=link_to_function 'Do', 'aaa()' %>
No matter how many time I call the '_comment.erb', the code inside 'run_once' will be run only once. What shall I do?
This is probably not the best solution to your problem, but creating a helper that will yield only once isn't too much trouble:
Helper
def output_once(name, &block)
#output_once_blocks ||= []
unless #output_once_blocks.include?(name)
#output_once_blocks << name
concat(capture(block), block.binding)
end
end
Usage
<% output_once :define_aaa do %>
Your stuff here
<% end %>
<% if !#once_flag && #once_flag=1 %>
<script>
function aaa() {}
</script>
<% end %>
Test to see if the function has already been defined. If not, define/run it
<script type="text/javascript">
/*<![CDATA[*/
if (!aaa) {
function aaa() {}
}
/*]]>*/
</script>
Try something like this
In your layout create a hidden textfield called 'flag' and make it default value as 0
(make sure you dont include this in your partial)
in your partial script
check that value and if it is != 1 execute your function
This is not a solid solution, but since your logic needs some flag to trace the number of executions this will be a workaround
cheers,
sameera
Hmm, if you want to define stuff once, You can just define another partail.
Define javascript in and render _comment.erb partial multiple times here. It should take care of both issues.
As far as I know, there no function like "run_once" available so you either have to write it yourself or handle your code another way. For javascript just define it in your application.js file and then you can call it whenever you want. If it is ruby code then put it in your application helpers and for HTML it really should be a partial.

Resources