I have a table of events (in a sqlite3 database for what it's worth) with a column called "when" that contains a timestamp detailing precisely when the event that particular row denotes is set to occur. Right now, I have
#events = Event.find(:all)
in my controller and I am using template helper methods to calculate where to place each event on my display page based on the day of the week it occurs on. For example:
<% if(event.when.wday == 6) %>
# DO SOMETHING
<% end %>
I want to abstract this logic to the controller however. My idea was to do the following:
#thursday_events = Event.find(:all, :conditions => ["when.wday=4"])
Obviously (I guess?) this didn't work. Throwing the error "SQLite3::SQLException: near "when": syntax error: SELECT * FROM "events" WHERE (when.wday=4)".
I'm assuming this is because I tried to use a helper method within a find condition but I don't know a better way to do this. Any advice? Thanks!
The conditions parameter needs to be a fragment of SQL.
:conditions => ["when.wday=4"]
is a fragment of Ruby code, so no go.
Try
# Model Event has a datetime field named 'when'
Event.find(:all, :conditions => ["strftime('%w', events.when) = 4"])
SQLLite ref: http://www.sqlite.org/lang_datefunc.html
Added:
While more closely reading your post, I think you're planning to send multiple instance variables (one per day of the week) from your controller to your view. That's a good idea--moving logic out of the view. But, don't do more dbms queries!
Each query has significant overhead. Eg:
#Do NOT do it this way (too many db queries)
#sunday_events = Event.find(:all,
:conditions => ["strftime('%w', events.when) = 0"])
#monday_events = Event.find(:all,
:conditions => ["strftime('%w', events.when) = 1"])
#thursday_events = Event.find(:all,
:conditions => ["strftime('%w', events.when) = 4"])
# ... etc
# Better: Just 1 database query--
events = Event.find(:all)
#sunday_events = events.select{|e| e.when.wday == 0}
#monday_events = events.select{|e| e.when.wday == 1}
#thursday_events = events.select{|e| e.when.wday == 4}
# ... etc
Final comment:
Current best practice thinking is to move code into the models from controllers wherever reasonable. This is called "Fat model, skinny controller" In the above example, you could have a class-level method in the model create the individual instance variables. Or perhaps better, one hash that contains 7 values, each being an array of the records. Eg
# in Event model
def Event.find_by_day
events = Event.find(:all)
result = {}
days = [:sun, :mon, :tue, :wed, :thu, :fri, :sat]
(0..6).each{|day_i| result[days[day_i]] =
events.select{|e| e.when.wday == day_i}
}
result
end
# in controller
#events = Event.find_by_day
# in view
# #events[:sun] is array of the Sunday events
# so do something with them...
Related
Simple question
In rails I have an ApplicationHelper method
def sum_customer_yearly_revenue(customer_id, year)
sum_customer_yearly_revenue = Sale.sum(:net_amount, :conditions => ['customer_id = ? AND financial_year = ?', customer_id, year])
end
In my view I then call sum_customer_yearly_revenue(123456, 2014). How do I call the same method but with 'all' years. If I was using SQL it would be sum_customer_yearly_revenue(123456, *) but that returns an error. If it pass in "" it looks for a year that is empty. If I leave it blank it just errors aswell.
def sum_customer_yearly_revenue(customer_id, year="all")
sales = Sale.where(customer_id: customer_id)
sales = sales.where(year: year) if year!="all"
sales.sum(:net_amount)
end
This might be useful to you. In this case if you want to retrieve sum for all year there is no necessary to pass second argument.
Change the method to:
def sum_customer_yearly_revenue(customer_id, year = nil)
conditions = { :customer_id => customer_id }
conditions.merge!(:financial_year => year) if year
Sale.sum(:net_amount, :conditions => conditions)
end
And call it like this:
sum_customer_yearly_revenue(123456)
Can't in this form. Make a new helper, sum_customer_total_revenue. Or introduce way more logic in this function, but that's not as pretty.
Also, not sure those should be any kind of helpers; that kind of stuff belongs right into a model. If you need it, make sure the model passes those information to the view by including it from a controller, not directly in a helper.
(EDIT: upvoted jbmyid; that's a better syntax to do this. However... Still think it doesn't belong in a function that's called the way it's called; still think it doesn't belong in a helper.)
named_scope :with_country, lambad { |country_id| ...}
named_scope :with_language, lambad { |language_id| ...}
named_scope :with_gender, lambad { |gender_id| ...}
if params[:country_id]
Event.with_country(params[:country_id])
elsif params[:langauge_id]
Event.with_state(params[:language_id])
else
......
#so many combinations
end
If I get both country and language then I need to apply both of them. In my real application I have 8 different named_scopes that could be applied depending on the case. How to apply named_scopes incrementally or hold on to named_scopes somewhere and then later apply in one shot.
I tried holding on to values like this
tmp = Event.with_country(1)
but that fires the sql instantly.
I guess I can write something like
if !params[:country_id].blank? && !params[:language_id].blank? && !params[:gender_id].blank?
Event.with_country(params[:country_id]).with_language(..).with_gender
elsif country && language
elsif country && gender
elsif country && gender
.. you see the problem
Actually, the SQL does not fire instantly. Though I haven't bothered to look up how Rails pulls off this magic (though now I'm curious), the query isn't fired until you actually inspect the result set's contents.
So if you run the following in the console:
wc = Event.with_country(Country.first.id);nil # line returns nil, so wc remains uninspected
wc.with_state(State.first.id)
you'll note that no Event query is fired for the first line, whereas one large Event query is fired for the second. As such, you can safely store Event.with_country(params[:country_id]) as a variable and add more scopes to it later, since the query will only be fired at the end.
To confirm that this is true, try the approach I'm describing, and check your server logs to confirm that only one query is being fired on the page itself for events.
Check Anonymous Scopes.
I had to do something similar, having many filters applied in a view. What I did was create named_scopes with conditions:
named_scope :with_filter, lambda{|filter| { :conditions => {:field => filter}} unless filter.blank?}
In the same class there is a method which receives the params from the action and returns the filtered records:
def self.filter(params)
ClassObject
.with_filter(params[:filter1])
.with_filter2(params[:filter2])
end
Like that you can add all the filters using named_scopes and they are used depending on the params that are sent.
I took the idea from here: http://www.idolhands.com/ruby-on-rails/guides-tips-and-tutorials/add-filters-to-views-using-named-scopes-in-rails
Event.with_country(params[:country_id]).with_state(params[:language_id])
will work and won't fire the SQL until the end (if you try it in the console, it'll happen right away because the console will call to_s on the results. IRL the SQL won't fire until the end).
I suspect you also need to be sure each named_scope tests the existence of what is passed in:
named_scope :with_country, lambda { |country_id| country_id.nil? ? {} : {:conditions=>...} }
This will be easy with Rails 3:
products = Product.where("price = 100").limit(5) # No query executed yet
products = products.order("created_at DESC") # Adding to the query, still no execution
products.each { |product| puts product.price } # That's when the SQL query is actually fired
class Product < ActiveRecord::Base
named_scope :pricey, where("price > 100")
named_scope :latest, order("created_at DESC").limit(10)
end
The short answer is to simply shift the scope as required, narrowing it down depending on what parameters are present:
scope = Example
# Only apply to parameters that are present and not empty
if (!params[:foo].blank?)
scope = scope.with_foo(params[:foo])
end
if (!params[:bar].blank?)
scope = scope.with_bar(params[:bar])
end
results = scope.all
A better approach would be to use something like Searchlogic (http://github.com/binarylogic/searchlogic) which encapsulates all of this for you.
Say I have model 'Car' and controller 'cars', and a method 'display'.
I have multiple attributes like:
in_production, year, make
I can easily do something like this to find cars that match all the parameters passed:
def display
#cars = Car.find(:all, :conditions => { :in_production => #{params[:in_production]}, :year => #{params[:year]}, :make => #{params[:make]} })`
end
So what I'm doing is coding hard links in the menu, so if I wanted to find all Nissan cars from 2009 that were in production, I would pass those values as parameters in my link.
On another page I want to show every car from 2009 that is in_production, only two params instead of three. What's the best way to dynamically alter the conditions so it will work with one, two, or three params, whilst using the same action?
Any ideas?
First of all, using
:conditions => "in_production = '#{params[:in_production]}' AND year = '#{params[:year]}' AND make = '#{params[:make]}'"
is vulnerable to SQL injection. You need to escape the user provided parameters before using them in database conditions.
Something like this should let you add conditions more dynamically depending on whether or not the parameters exist. I did not test it, so I may edit it shortly...
def display
conditions = []
conditions << [ "in_production = ?", params[:in_production] ] if params[:in_production].present?
conditions << [ "year = ?", params[:year] ] if params[:year].present?
conditions << [ "make = ?", params[:make] ] if params[:make].present?
#cars = Car.all(:conditions => conditions )
end
Certainly escape the params and ensure that you only query against fields you want to be exposed. Beyond that, you could use what is built into Rails:
Car.find_all_by_in_production_and_year_and_make(in_production, year, make)
Hand-rolling the conditions may allow for additional logic to be applied (search by year only if the year is between x and y, etc). Using the rails finders (which in turn use method_missing) keeps the API clean and flexible without having to stare at direct SQL conditions.
You could construct a Car#search method that takes the entire params hash as input, where the params are sanitized and stripped of non-exposed fields, and construct the Car#find_all_by* method call using the param names themselves. Adding new conditions to search by is then as simple as passing them in the params.
You might check out searchlogic. It uses some method missing magic to construct named_scopes that would do what you want.
http://github.com/binarylogic/searchlogic
I use SmartTuple for stuff like this. Simple, powerful, designed specifically for the task.
#cars = Car.all(:conditions => (SmartTuple.new(" AND ") +
({:in_production => params[:in_production]} if params[:in_production].present?) +
({:year => params[:year]} if params[:year].present?) +
({:make => params[:make]} if params[:make].present?)
).compile)
or
#cars = Car.all(:conditions => [SmartTuple.new(" AND "),
({:in_production => params[:in_production]} if params[:in_production].present?),
({:year => params[:year]} if params[:year].present?),
({:make => params[:make]} if params[:make].present?),
].sum.compile)
or
keys = [:in_production, :year, :make]
#cars = Car.all(:conditions => (SmartTuple.new(" AND ").add_each(keys) do |k|
{k => params[k]} if params[k].present?
end).compile)
Pick the one you like the most. :)
This is a snippet of code from an update method in my application. The method is POSTed an array of user id's in params[:assigned_ users_ list_ id]
The idea is to synchronise the DB associations entries with the ones that were just submitted, by removing the right ones (those that exist in the DB but not the list) and adding the right ones (vise-versa).
#list_assigned_users = User.find(:all, :conditions => { :id => params[:assigned_users_list_id]})
#assigned_users_to_remove = #task.assigned_users - #list_assigned_users
#assigned_users_to_add = #list_assigned_users - #task.assigned_users
#assigned_users_to_add.each do |user|
unless #task.assigned_users.include?(user)
#task.assigned_users << user
end
end
#assigned_users_to_remove.each do |user|
if #task.assigned_users.include?(user)
#task.assigned_users.delete user
end
end
It works - great!
My first questions is, are those 'if' and 'unless' statements totally redundant, or is it prudent to leave them in place?
My next question is, I want to repeat this exact code immediately after this, but with 'subscribed' in place of 'assigned'... To achieve this I just did a find & replace in my text editor, leaving me with almost this code in my app twice. That's hardly in keeping with the DRY principal!
Just to be clear, every instance of the letters 'assigned' becomes 'subscribed'. It is passed params[:subscribed_ users_ list_ id], and uses #task.subscribed_ users.delete user etc...
How can I repeat this code without repeating it?
Thanks as usual
You don't need if and unless statements.
As for the repetition you can make array of hashes representing what you need.
Like this:
[
{ :where_clause => params[:assigned_users_list_id], :user_list => #task.assigned_users} ,
{ :where_clause => params[:subscribed_users_list_id], :user_list => #task.subscribed_users}
] each do |list|
#list_users = User.find(:all, :conditions => { :id => list[:where_clause] })
#users_to_remove = list[:user_list] - #list_users
#users_to_add = #list_users - list[:user_list]
#users_to_add.each do |user|
list[:user_list] << user
end
#users_to_remove.each do |user|
list[:user_list].delete user
end
end
My variable names are not the happiest choice so you can change them to improve readability.
I seem to be missing something here, but aren't you just doing this?
#task.assigned_users = User.find(params[:assigned_users_list_id])
I'm doing this:
#snippets = Snippet.find :all, :conditions => { :user_id => session[:user_id] }
#snippets.each do |snippet|
snippet.tags.each do |tag|
#tags.push tag
end
end
But if a snippets has the same tag two time, it'll push the object twice.
I want to do something like if #tags.in_object(tag)[...]
Would it be possible? Thanks!
I think there are 2 ways to go about it to get a faster result.
1) Add a condition to your find statement ( in MySQL DISTINCT ). This will return only unique result. DBs in general do much better jobs than regular code at getting results.
2) Instead if testing each time with include, why don't you do uniq after you populate your array.
here is example code
ar = []
data = []
#get some radom sample data
100.times do
data << ((rand*10).to_i)
end
# populate your result array
# 3 ways to do it.
# 1) you can modify your original array with
data.uniq!
# 2) you can populate another array with your unique data
# this doesn't modify your original array
ar.flatten << data.uniq
# 3) you can run a loop if you want to do some sort of additional processing
data.each do |i|
i = i.to_s + "some text" # do whatever you need here
ar << i
end
Depending on the situation you may use either.
But running include on each item in the loop is not the fastest thing IMHO
Good luck
Another way would be to simply concat the #tags and snippet.tags arrays and then strip it of duplicates.
#snippets.each do |snippet|
#tags.concat(snippet.tags)
end
#tags.uniq!
I'm assuming #tags is an Array instance.
Array#include? tests if an object is already included in an array. This uses the == operator, which in ActiveRecord tests for the same instance or another instance of the same type having the same id.
Alternatively, you may be able to use a Set instead of an Array. This will guarantee that no duplicates get added, but is unordered.
You can probably add a group to the query:
Snippet.find :all, :conditions => { :user_id => session[:user_id] }, :group => "tag.name"
Group will depend on how your tag data works, of course.
Or use uniq:
#tags << snippet.tags.uniq