66 lines
1.9 KiB
Ruby
66 lines
1.9 KiB
Ruby
class ArticleSuggester
|
|
def initialize(article)
|
|
@article = article
|
|
@total_articles_count = self.class.articles_count
|
|
end
|
|
|
|
def articles(max: 4)
|
|
if cached_tag_list_array.any?
|
|
# avoid loading more data if we don't need to
|
|
tagged_suggestions = suggestions_by_tag(max: max)
|
|
return tagged_suggestions if tagged_suggestions.size == max
|
|
|
|
# if there are not enough tagged articles, load other suggestions
|
|
# ignoring tagged articles that might be relevant twice, hence avoiding duplicates
|
|
num_remaining_needed = max - tagged_suggestions.size
|
|
other_articles = other_suggestions(
|
|
max: num_remaining_needed,
|
|
ids_to_ignore: tagged_suggestions.map(&:id),
|
|
)
|
|
tagged_suggestions.union(other_articles)
|
|
else
|
|
other_suggestions(max: max)
|
|
end
|
|
end
|
|
|
|
def self.articles_count
|
|
Article.published.estimated_count
|
|
end
|
|
|
|
private
|
|
|
|
attr_reader :article
|
|
|
|
def other_suggestions(max: 4, ids_to_ignore: [])
|
|
ids_to_ignore << article.id
|
|
Article.published
|
|
.where.not(id: ids_to_ignore)
|
|
.where.not(user_id: article.user_id)
|
|
.order(hotness_score: :desc)
|
|
.offset(rand(0..offset))
|
|
.first(max)
|
|
end
|
|
|
|
def suggestions_by_tag(max: 4)
|
|
Article.published.tagged_with(cached_tag_list_array, any: true)
|
|
.where.not(user_id: article.user_id)
|
|
.where(tag_suggestion_query)
|
|
.order(hotness_score: :desc)
|
|
.offset(rand(0..offset))
|
|
.first(max)
|
|
end
|
|
|
|
def offset
|
|
@total_articles_count > 1000 ? 200 : (@total_articles_count / 10)
|
|
end
|
|
|
|
def tag_suggestion_query
|
|
# Fore big communities like DEV we can look at organic page views for indicator.
|
|
# For smaller communities, we'll a basic score check.
|
|
@total_articles_count > 1000 ? "organic_page_views_past_month_count > 5" : "score > 1"
|
|
end
|
|
|
|
def cached_tag_list_array
|
|
(article.cached_tag_list || "").split(", ")
|
|
end
|
|
end
|