Improve hashtag search with multi word queries

This commit is contained in:
Mark Felder 2025-07-31 17:46:32 -07:00
commit 93c144e397
4 changed files with 92 additions and 6 deletions

View file

@ -0,0 +1 @@
Hashtag searches return real results based on words in your query

View file

@ -135,18 +135,28 @@ defmodule Pleroma.Hashtag do
limit = Keyword.get(options, :limit, 20) limit = Keyword.get(options, :limit, 20)
offset = Keyword.get(options, :offset, 0) offset = Keyword.get(options, :offset, 0)
query search_terms =
|> String.downcase() query
|> String.trim() |> String.downcase()
|> then(fn search_term -> |> String.trim()
|> String.split(~r/\s+/)
|> Enum.filter(&(&1 != ""))
if Enum.empty?(search_terms) do
[]
else
# Use PostgreSQL's ANY operator with array for efficient multi-term search
# This is much more efficient than multiple OR clauses
search_patterns = Enum.map(search_terms, &"%#{&1}%")
from(ht in Hashtag, from(ht in Hashtag,
where: fragment("LOWER(?) LIKE ?", ht.name, ^"%#{search_term}%"), where: fragment("LOWER(?) LIKE ANY(?)", ht.name, ^search_patterns),
order_by: [asc: ht.name], order_by: [asc: ht.name],
limit: ^limit, limit: ^limit,
offset: ^offset offset: ^offset
) )
|> Repo.all() |> Repo.all()
|> Enum.map(& &1.name) |> Enum.map(& &1.name)
end) end
end end
end end

View file

@ -38,6 +38,35 @@ defmodule Pleroma.HashtagTest do
assert results == [] assert results == []
end end
test "searches hashtags by multiple words in query" do
# Create some hashtags
{:ok, _} = Hashtag.get_or_create_by_name("computer")
{:ok, _} = Hashtag.get_or_create_by_name("laptop")
{:ok, _} = Hashtag.get_or_create_by_name("desktop")
{:ok, _} = Hashtag.get_or_create_by_name("phone")
# Search for "new computer" - should return "computer"
results = Hashtag.search("new computer")
assert "computer" in results
refute "laptop" in results
refute "desktop" in results
refute "phone" in results
# Search for "computer laptop" - should return both
results = Hashtag.search("computer laptop")
assert "computer" in results
assert "laptop" in results
refute "desktop" in results
refute "phone" in results
# Search for "new phone" - should return "phone"
results = Hashtag.search("new phone")
assert "phone" in results
refute "computer" in results
refute "laptop" in results
refute "desktop" in results
end
test "supports pagination" do test "supports pagination" do
{:ok, _} = Hashtag.get_or_create_by_name("alpha") {:ok, _} = Hashtag.get_or_create_by_name("alpha")
{:ok, _} = Hashtag.get_or_create_by_name("beta") {:ok, _} = Hashtag.get_or_create_by_name("beta")
@ -50,5 +79,20 @@ defmodule Pleroma.HashtagTest do
results = Hashtag.search("a", limit: 2, offset: 1) results = Hashtag.search("a", limit: 2, offset: 1)
assert length(results) == 2 assert length(results) == 2
end end
test "handles many search terms efficiently" do
# Create hashtags
{:ok, _} = Hashtag.get_or_create_by_name("computer")
{:ok, _} = Hashtag.get_or_create_by_name("laptop")
{:ok, _} = Hashtag.get_or_create_by_name("phone")
{:ok, _} = Hashtag.get_or_create_by_name("tablet")
# Search with many terms - should be efficient with PostgreSQL ANY operator
results = Hashtag.search("new fast computer laptop phone tablet device")
assert "computer" in results
assert "laptop" in results
assert "phone" in results
assert "tablet" in results
end
end end
end end

View file

@ -139,6 +139,37 @@ defmodule Pleroma.Web.MastodonAPI.SearchControllerTest do
assert results["hashtags"] == [] assert results["hashtags"] == []
end end
test "searches hashtags by multiple words in query", %{conn: conn} do
user = insert(:user)
{:ok, _activity1} = CommonAPI.post(user, %{status: "This is my new #computer"})
{:ok, _activity2} = CommonAPI.post(user, %{status: "Check out this #laptop"})
{:ok, _activity3} = CommonAPI.post(user, %{status: "My #desktop setup"})
{:ok, _activity4} = CommonAPI.post(user, %{status: "New #phone arrived"})
results =
conn
|> get("/api/v2/search?#{URI.encode_query(%{q: "new computer"})}")
|> json_response_and_validate_schema(200)
hashtag_names = Enum.map(results["hashtags"], & &1["name"])
assert "computer" in hashtag_names
refute "laptop" in hashtag_names
refute "desktop" in hashtag_names
refute "phone" in hashtag_names
results =
conn
|> get("/api/v2/search?#{URI.encode_query(%{q: "computer laptop"})}")
|> json_response_and_validate_schema(200)
hashtag_names = Enum.map(results["hashtags"], & &1["name"])
assert "computer" in hashtag_names
assert "laptop" in hashtag_names
refute "desktop" in hashtag_names
refute "phone" in hashtag_names
end
test "supports pagination of hashtags search results", %{conn: conn} do test "supports pagination of hashtags search results", %{conn: conn} do
user = insert(:user) user = insert(:user)