I am trying to achieve what I think will be a fairly complex query using the magic of Rails without having lots of ugly looking SQL in the code.
Since my database is dealing with rather specialised biomedical models I'll translate the following to a more real world scenario.
I have a model Book that
has_many :chapters
and Chapter that
belongs_to :book
Say for instance Chapters have a name attribute and the names can be preface ,introduction and appendix, and a Book could have a Chapter named preface and a Chapter named introduction but no Chapter named appendix. In fact any combination of these
I am looking to find all Books that have both Chapters named preface and introduction.
At present I have a named_scope as follows
Book
named_scope :has_chapter?, lambda { |chapter_names|
condition_string_array = []
chapter_names.size.times{condition_string_array << "chapters.name = ?"}
condition_string = condition_string_array.join(" OR ")
{:joins => [:chapters] , :conditions => [condition_string, * chapter_names]}
}
If I call Book. has_chapter? ["preface", "introduction"] this will find me all books that have either a Chapter named preface or introduction. How could I do a similar thing that would find me isolates that both Chapters preface and introduction?
I am not that familiar with SQL so am not quite sure what kind of join would be needed and whether this could be achieved in a named scope.
Many thanks
Anthony
I am looking to find all Books that
have both Chapters named preface and
introduction.
I use mysql, not postgres, but I assume postgres has support for sub-queries? You could try something like:
class Book
named_scope :has_preface, :conditions => "books.id IN (SELECT book_id FROM chapters WHERE chapters.name = 'preface')"
named_scope :has_introduction, :conditions => "books.id IN (SELECT book_id FROM chapters WHERE chapters.name = 'introduction')"
end
and then:
Book.has_preface.has_introduction
I like the condensing of code to
condition_string = Array.new(chapter_names.size, "chapters.name=?").join(" AND ")
and this does indeed work for a join with "OR".However the AND join will not work since this means that chapters.name in a single row of a join has to be both 'preface' and 'introduction' for example.
An even neater way of doing my original "OR" join is
named_scope :has_chapter?, lambda { | chapter_names |
{:joins => [: chapters] , :conditions => { :chapters => {:name => chapter_names}}}
}
Thanks to Duplex on Rails Forum (Post link)
In order to achieve my original problem though DUPLEX suggested this
named_scope :has_chapters, lambda { |chapter_names|
{:joins => :chapters , :conditions => { :chapters => {:name => chapter_names}},
:select => "books.*, COUNT(chapters.id) AS c_count",
:group => "books.id", :having => "c_count = #{chapter_names.is_a?(Array) ? chapter_names.size : 1}" }
}
This may work in some flavour of SQL but not in PostgreSQL. I had to use the following
named_scope : has_chapter?, lambda { | chapter_names |
{:joins => :chapters , :conditions => { :chapters => {:name => chapter_names}},
:select => "books.* , COUNT(chapters.id)",
:group => Book.column_names.collect{|column_name| "books.#{column_name}"}.join(","), :having => "COUNT(chapters.id) = #{chapter_names.is_a?(Array) ? chapter_names.size : 1}" }
}
The code
Book.column_names.collect{|column_name| "books.#{column_name}"}.join(",")
is required because in PostgreSQL you can't select all columns with books.* and then GROUP BY books.* each column has to be named individually :(
I am looking to find all Books that
have both Chapters named preface and
introduction.
At present I have a named_scope as
follows
You can do this without a named_scope.
class Book < ActiveRecord::Base
has_many :chapters
def self.has_chapter?(chapter_names)
Chapter.find_all_by_name(chapter_names, :include => [:book]).map(&:book)
end
end
Related
I am new to Rail and trying to make a query in ActiveRecord. I am trying to get all of the records with the status of 'Landed', that are over 60 days old. My query works up to the point of getting all of the projects with the status of 'Landed'. When I add in the last condition of "created_at < ? ", then I always get an empty relation. I know that I have projects that fit that description, so I am doing something wrong in my query and dont understand. I believe my error is in the date comparison, but I am not sure.
1. Projects
belongs_to :status
has_many :project_status_histories
2. Status
has_many :projects
has_many :project_status_histories
3. Project_Status_Histories
belongs_to :status
belongs_to :project
Project.find(:all, :joins => [:project_status_histories, :status], :conditions => {:projects => {:status_id => Status.where(:name => 'Landed').first.id }, :project_status_histories => {:created_at => ["created_at < ?", (Date.today - 60.days)]}})
I have tried to build the query, step by step, with the dbconsole and am not having any luck. Thanks for all the help in advance.
I don't think it's the date arithmetic. One nice way to do this would be with named scopes. Add the following to project.rb:
scope :landed, joins(:status).where('statuses.name' => 'Landed')
scope :recent, lambda \
{ joins(:project_status_histories) \
.where('project_status_histories.created_at < ?', Date.today - 60.days) }
Then you can retrieve the relevant records/objects with:
Project.landed.recent
This worked for me in my test. You should also check out the rails guide, from which I stole most of this:
http://guides.rubyonrails.org/active_record_querying.html#scopes
Your query is a little bit complicated...
I would rather do it like this:
Project.where("status.name = ? AND project_status_histories.created_at < ?", "Landed", Time.now.day - 60.days)
I think this should work better. Let me know if it doesn't, maybe I've wrote something wrong, unfortunately I can't test it right now...
[Edit]
You also might want to see what is the generated SQL, use the "explain" method for that, just add it to the end of your query and print the result, for instance with your query:
Project.find(:all, :joins => [:project_status_histories, :status], :conditions => {:projects => {:status_id => Status.where(:name => 'Landed').first.id }, :project_status_histories => {:created_at => ["created_at < ?", (Date.today - 60.days)]}}).explain
I have posts which are sent by users to other users. There are two models - :post and :user, and :post has the following named associations:
belongs_to :from_user, :class_name => "User", :foreign_key => "from_user_id"
belongs_to :to_user, :class_name => "User", :foreign_key => "to_user_id"
Both :user and :post have "is_public" column, indicating that either a single post and/or the entire user profile can be public or private.
My goal is to fetch a list of posts which are public AND whose recipients have public profiles. At the same time, I would like to "include" both sender and recipient info to minimize the # of db calls. The challenge is that I am effectively "including" the same table twice via the named associations, but in my "conditions" I need to make sure that I only filter by the recipient's "is_public" column.
I can't do the following because "conditions" does not accept an association name as a parameter:
Post.find(:all, :include => [ :to_user, :from_user ],
:conditions => { :is_public => true, :to_user => { :is_public => true }})
So, one way I can accomplish this is by doing an additional "join" on the "users" table:
Post.find(:all, :include => [ :to_user, :from_user ],
:joins => "inner join users toalias on posts.to_user_id = toalias.id",
:conditions => { :is_public => true, 'toalias.is_public' => true })
Is there a better, perhaps cleaner, way to do this?
Thanks in advance
I was facing the same problem and found a solution after watching sql query generated from rails query, sql query automatically generates an alias
try this,
Post.find(:all, :include => [ :to_user, :from_user ],
:conditions => { :is_public => true, 'to_users_posts.is_public' => true })
It worked for me :)
I have not been able to find a better solution than the one originally stated in my question. This one doesn't depend on how Rails names/aliases tables when compiling a query and therefore appears to be cleaner than the alternatives (short of using 3rd party gems or plugins):
Post.find(:all, :include => [ :to_user, :from_user ],
:joins => "inner join users toalias on posts.to_user_id = toalias.id",
:conditions => { :is_public => true, 'toalias.is_public' => true })
If you are on Rails 3, take a look at squeel gem, esp if you are doing these kind of complex joins often. Or if you dont want to add a extra gem, take look at the Arel table in Rails 3.
I very pleasant for that question, because I've tried to find proper solution about 2h, so after using few tips above, I've found the proper, in my case, solution. My case:
I need filter instances by created_by_id/updated_by_id fields, that fields are in every table, so...what I did
In concern 'Filterable' I wrote the method and when I needed filter by that fields I used that ->
key = "#{key.pluralize}_#{name.pluralize.downcase}.email" if %w(created_by updated_by).include?(key)
# in case with invoices key = 'updated_bies_invoices.email'
results = results.eager_load(:created_by, :updated_by).where("#{key} = ?", value)
I just got referred to stackoverflow by a friend here to help with a problem I am having. I am fairly new to ruby on rails and I am working on a collaborative project where we have a script (medal_worker.rb) that is scheduled to run at a fixed intervals to award people various medals based on various participation and success on our website. One of the new medals I am working on rewards people for "milestones". For the purpose of this problem, let's say we want to give them medals when they make 100, 1000, and 10000 comments. I would like to do this by using named_scopes from the User model (user.rb) to give me filtered lists of the users I am looking for.
My question is: How do I find the users who do not have the respective medals for the respective milestone comment level (preferably using the named_scopes from the User model)?
Here is an exerpt from my model_worker.rb file:
def award_comment_milestone(comments)
users = Users.frequent_comment_club_members(comments).not_awarded_medal(Medal.find_by_id(medal.id))
for user in users do
award_medal(medal, nil, user) if #award
end
end
Here is where I am at with the named_scopes in the user model (user.rb):
named_scope :frequent_comment_club_members, lambda { |*args|
{:include => comment_records, :conditions => ['comment_records.comment_type = ? and comment_records.comments >= ?', 'User', (args.first || 0)]}
}
named_scope :not_awarded_medal, lambda { |medal|
{:include => :awards, :conditions => ['awards.medal_id not in (select awards.medal_id from awards where awards.medal_id = ?)", medal.id] }
}
This is not working as I would like, but I don't know if the problem is in the named_scopes or how I am passing arguements or what. Thanks.
Your named_scopes look fine. Except you are starting with a single apostrophe and ending with a double apostrophe in the not_awarded_medal condition statement.
EDIT:
Take it back. Your not_awarded_medal named_scope is off.
Try something like this:
named_scope :not_awarded_medal, lambda { |medal_id|
{ :include => :awards,
:conditions => [
"? not in (select awards.id from awards where awards.user_id = users.id)", medal_id
]
}
}
untested
Now this is assuming that you have the following relationships:
User: has_many :awards
Award: belongs_to :user
If you are using has_many :through then you are going to have to change the SQL to look at the join (users_awards) table.
--
But I do see a couple of things in the award_comment_milestone function.
What is the parameter coming into award_comment_milestone? Is it an array of comments or is it a count of comments? Also where is medal defined?
If comments is an array then you need to pass comments.length into frequent_comment_club_members. If it's the count then I would rename it to comments_count so the next person can understand the logic more quickly.
Some general observations:
not_awarded_medal should just take a medal_id and not the whole object (no need to do multiple queries)
Why are you doing Medal.find_by_id(medal.id)? You already have the medal object.
I really suck at Rails' finders besides the most obvious. I always resort to SQL when things get more advanced than
Model.find(:all, :conditions => ['field>? and field<? and id in (select id from table)', 1,2])
I have this method:
def self.get_first_validation_answer(id)
a=find_by_sql("
select answers.*, answers_registrations.answer_text
from answers_registrations left join answers on answers_registrations.answer_id=answers.id
where
(answers_registrations.question_id in (select id from questions where validation_question=true))
and
(sale_registration_id=#{id})
limit 1
").first
a.answer_text || a.text if a
end
Can someone create a find method that gets me what I want?
Regards,
Jacob
class AnswersRegistration < ActiveRecord::Base
has_many :answers
end
id = 123
the_reg = AnswersRegistration.first(
:joins => :answers,
:conditions => '(question_id in (select id from questions where validation_question = true)) and (sale_registration_id = ?)', id)
(untested)
Just use binarylogic's Searchlogic gem if that satisfies your need.
Here you go: http://github.com/binarylogic/searchlogic
Sometimes AR will choke on complicated nested conditions, but in theory you should be able to do this:
AnswersRegistration.first(:conditions => { :question => { :validation_question => true },
:sale_registration_id => id },
:include => :answer)
Say I have a model Taggable has_many tags, how may I find all taggables by their associated tag's taggable_id field?
Taggable.find(:all, :joins => :tags, :conditions => {:tags => {:taggable_id => [1,2,3]}})
results in this:
SELECT `taggables`.* FROM `taggables` INNER JOIN `tags` ON tags.taggable_id = taggables.id WHERE (`tag`.`taggable_id` IN (1,2,3))
The syntax is incredible but does not fit my needs in that the resulting sql returns any taggable that has any, some or all of the tags.
How can I find taggables with related tags of field taggable_id valued 1, 2 and 3?
Thanks for any advice. :)
UPDATE:
I have found a solution which I'll post for others, should they end up here. Also though for the attention of others whose suggestions for improvement I'd happily receive. :)
Taggable.find(:all, :joins => :tags, :select => "taggables.*, tags.count tag_count", :conditions => {:tags => {:taggable_id => array_of_tag_ids}}, :group => "taggables.id having tag_count = #{array_of_tag_ids.count}"))
Your question is a bit confusing ('tags' seemed to be used quite a bit :)), but I think you want the same thing I needed here:
Taggable.find([1,2,3], :include => :tags).tags.map { |t| t.taggables }
or (if you want the results to be unique, which you probably do):
Taggable.find([1,2,3], :include => :tags).tags.map { |t| t.taggables }.flatten.uniq
This gets at all the taggables that have the same tags as the taggables that have ids 1,2, and 3. Is that what you wanted?