ActiveRecord query to return related Documents using RelatedDocuments join table

问题

In my app, users can collect favorite Documents. I want to return a list of documents most relevant to those they have favorited, using a related_documents join table that has a relevance_score:integer for the relationship between a document and a related document.

TLDR; see the update at the bottom for an almost-working query.

Without success, I have tried various queries like:

some_user.favorited_documents.joins(:related_to_documents)
  .select("docid_id FROM documents.related_documents")
  .order("related_documents.relevance_score DESC")
  .limit(5 * some_user.favorited_documentss.count)

Ideally I'd like to return the top 5 most relevant related documents for each favorited document, but I'd compromise with just the top (5*favorited_documents) overall, as the relevance scores are relatively comparable.

My models look like:

class User
  has_many :favorited_events
  has_many :favorited_documents, through: :favorited_events, source: :document

class Document
  # id
  # docid  ← this is a unique organisational doc identifier
  has_many :related_documents, order: "relevance_score DESC"
  has_many :related_to_documents, through: :related_documents, source: :docid,
            order: "related_documents.relevance_score DESC"
  has_many :related_by_documents, class_name: "RelatedDocument", foreign_key: :docid_id,
            primary_key: :docid  

class RelatedDocument
  # id              : integer
  # document_id     : integer
  # docid_id        : integer
  # relevance_score : integer
  belongs_to :document
  belongs_to :docid, class_name: "Document", primary_key: :docid

Notes:

The Document model has both an id and a docid, the later is a unique organisational identifier for our documents. dicid is used by RelatedDocument to track related_documents as these are not always already in the documents table in this database / instead they're retrieved as needed from another system.

For complex business reasons, related_documents is a one-way relationship / hence use of different primary keys (document.id vs document.docid).

Thanks for any help!

UPDATE:

Okay, I have a mostly-working query now: (uniq is ignored??)

 Document.joins("INNER JOIN related_documents ON 
                 documents.docid = related_documents.docid_id")
         .select("documents.*, related_documents.relevance_score")
         .where("related_documents.document_id IN (?)",    
                 some_user.favorited_documents)
         .order("related_documents.relevance_score DESC")
         .uniq
         .limit(10)

Which returns the top 10 most 'relevant' documents related to some_user's favorited documents.

The only part that isn't working is the 'uniq'. Although SELECT DISTINCT appears in the SQL the returned documents include duplicates. Any ideas?

来源：https://stackoverflow.com/questions/34014637/activerecord-query-to-return-related-documents-using-relateddocuments-join-table

标签

sql

ruby-on-rails

postgresql

activerecord