mirror of
https://github.com/discourse/discourse-ai.git
synced 2025-03-06 17:30:20 +00:00
cabecb801e
Embedding search is rate limited due to potentially expensive hyde operation (which require LLM access). Embedding generally is very cheap compared to it. (usually 100x cheaper) This raises the limit to 100 per minute for embedding searches, while keeping the old 4 per minute for HyDE powered search.
90 lines
2.7 KiB
Ruby
90 lines
2.7 KiB
Ruby
# frozen_string_literal: true
|
|
|
|
module DiscourseAi
|
|
module Embeddings
|
|
class EmbeddingsController < ::ApplicationController
|
|
requires_plugin ::DiscourseAi::PLUGIN_NAME
|
|
|
|
SEMANTIC_SEARCH_TYPE = "semantic_search"
|
|
|
|
MAX_HYDE_SEARCHES_PER_MINUTE = 4
|
|
MAX_SEARCHES_PER_MINUTE = 100
|
|
|
|
def search
|
|
query = params[:q].to_s
|
|
skip_hyde = params[:hyde].to_s.downcase == "false" || params[:hyde].to_s == "0"
|
|
|
|
if query.length < SiteSetting.min_search_term_length
|
|
raise Discourse::InvalidParameters.new(:q)
|
|
end
|
|
|
|
grouped_results =
|
|
Search::GroupedSearchResults.new(
|
|
type_filter: SEMANTIC_SEARCH_TYPE,
|
|
term: query,
|
|
search_context: guardian,
|
|
use_pg_headlines_for_excerpt: false,
|
|
can_lazy_load_categories: guardian.can_lazy_load_categories?,
|
|
)
|
|
|
|
semantic_search = DiscourseAi::Embeddings::SemanticSearch.new(guardian)
|
|
|
|
if !skip_hyde && !semantic_search.cached_query?(query)
|
|
RateLimiter.new(
|
|
current_user,
|
|
"semantic-search",
|
|
MAX_HYDE_SEARCHES_PER_MINUTE,
|
|
1.minutes,
|
|
).performed!
|
|
else
|
|
RateLimiter.new(
|
|
current_user,
|
|
"semantic-search-non-hyde",
|
|
MAX_SEARCHES_PER_MINUTE,
|
|
1.minutes,
|
|
).performed!
|
|
end
|
|
|
|
hijack do
|
|
semantic_search
|
|
.search_for_topics(query, _page = 1, hyde: !skip_hyde)
|
|
.each { |topic_post| grouped_results.add(topic_post) }
|
|
|
|
render_serialized(grouped_results, GroupedSearchResultSerializer, result: grouped_results)
|
|
end
|
|
end
|
|
|
|
def quick_search
|
|
# this search function searches posts (vs: topics)
|
|
# it requires post embeddings and a reranker
|
|
# it will not perform a hyde expantion
|
|
query = params[:q].to_s
|
|
|
|
if query.length < SiteSetting.min_search_term_length
|
|
raise Discourse::InvalidParameters.new(:q)
|
|
end
|
|
|
|
grouped_results =
|
|
Search::GroupedSearchResults.new(
|
|
type_filter: SEMANTIC_SEARCH_TYPE,
|
|
term: query,
|
|
search_context: guardian,
|
|
use_pg_headlines_for_excerpt: false,
|
|
)
|
|
|
|
semantic_search = DiscourseAi::Embeddings::SemanticSearch.new(guardian)
|
|
|
|
if !semantic_search.cached_query?(query)
|
|
RateLimiter.new(current_user, "semantic-search", 60, 1.minutes).performed!
|
|
end
|
|
|
|
hijack do
|
|
semantic_search.quick_search(query).each { |topic_post| grouped_results.add(topic_post) }
|
|
|
|
render_serialized(grouped_results, GroupedSearchResultSerializer, result: grouped_results)
|
|
end
|
|
end
|
|
end
|
|
end
|
|
end
|