Implement search with PostgreSQL trigrams

Signed-off-by: Thomas Citharel <tcit@tcit.fr>

Rename function to reflect that we only get one result

Signed-off-by: Thomas Citharel <tcit@tcit.fr>

Add loggers and make Ecto call parallels during search

Signed-off-by: Thomas Citharel <tcit@tcit.fr>

Implement trigrams for events & replace pg similarity operator % with <%

Signed-off-by: Thomas Citharel <tcit@tcit.fr>

Fix tests

Signed-off-by: Thomas Citharel <tcit@tcit.fr>
This commit is contained in:
Thomas Citharel
2019-02-21 18:11:49 +01:00
parent 131152abac
commit 4ec40d601b
18 changed files with 422 additions and 141 deletions

View File

@@ -467,6 +467,11 @@ defmodule Mobilizon.Actors do
|> Repo.preload(:organized_events)
end
@doc """
Getting an actor from url, eventually creating it
"""
# TODO: Move this to Mobilizon.Service.ActivityPub
@spec get_or_fetch_by_url(String.t(), bool()) :: {:ok, Actor.t()} | {:error, String.t()}
def get_or_fetch_by_url(url, preload \\ false) do
with {:ok, actor} <- get_actor_by_url(url, preload) do
{:ok, actor}
@@ -498,16 +503,32 @@ defmodule Mobilizon.Actors do
end
@doc """
Find local users by it's username
Find local users by their username
"""
# TODO: This doesn't seem to be used anyway
def find_local_by_username(username) do
actors =
Repo.all(
from(
a in Actor,
where:
(ilike(a.preferred_username, ^like_sanitize(username)) or
ilike(a.name, ^like_sanitize(username))) and is_nil(a.domain)
fragment(
"f_unaccent(?) <% f_unaccent(?) or
f_unaccent(coalesce(?, '')) <% f_unaccent(?)",
a.preferred_username,
^username,
a.name,
^username
),
where: is_nil(a.domain),
order_by:
fragment(
"word_similarity(?, ?) + word_similarity(coalesce(?, ''), ?) desc",
a.preferred_username,
^username,
a.name,
^username
)
)
)
@@ -526,48 +547,31 @@ defmodule Mobilizon.Actors do
from(
a in Actor,
where:
ilike(a.preferred_username, ^like_sanitize(username)) or
ilike(a.name, ^like_sanitize(username))
fragment(
"f_unaccent(?) %> f_unaccent(?) or
f_unaccent(coalesce(?, '')) %> f_unaccent(?)",
a.preferred_username,
^username,
a.name,
^username
),
order_by:
fragment(
"word_similarity(?, ?) + word_similarity(coalesce(?, ''), ?) desc",
a.preferred_username,
^username,
a.name,
^username
)
)
|> paginate(page, limit)
)
end
# Sanitize the LIKE queries
defp like_sanitize(value) do
"%" <> String.replace(value, ~r/([\\%_])/, "\\1") <> "%"
end
@email_regex ~r/^[a-zA-Z0-9.!#$%&'*+\/=?^_`{|}~-]+@[a-zA-Z0-9](?:[a-zA-Z0-9-]{0,61}[a-zA-Z0-9])?(?:\.[a-zA-Z0-9](?:[a-zA-Z0-9-]{0,61}[a-zA-Z0-9])?)*$/
@spec search(String.t()) :: {:ok, list(Actor.t())} | {:ok, []} | {:error, any()}
def search(name) do
# find already saved accounts
case find_actors_by_username_or_name(name) do
[] ->
# no accounts found, let's test if it's an username@domain.tld
with true <- Regex.match?(@email_regex, name),
# creating the actor in that case
{:ok, actor} <- ActivityPub.find_or_make_actor_from_nickname(name) do
{:ok, [actor]}
else
false ->
{:ok, []}
# error fingering the actor
{:error, err} ->
{:error, err}
end
actors = [_ | _] ->
# actors already saved found !
{:ok, actors}
end
end
@doc """
Find a group by its actor id
Get a group by its actor id
"""
def find_group_by_actor_id(actor_id) do
def get_group_by_actor_id(actor_id) do
case Repo.get_by(Actor, id: actor_id, type: :Group) do
nil -> {:error, :group_not_found}
actor -> {:ok, actor}

View File

@@ -45,7 +45,7 @@ defmodule Mobilizon.Actors.User do
:password,
min: 6,
max: 100,
message: "The choosen password is too short."
message: "The chosen password is too short."
)
if Map.has_key?(attrs, :default_actor) do

View File

@@ -248,7 +248,19 @@ defmodule Mobilizon.Events do
query =
from(e in Event,
where: e.visibility == ^:public and ilike(e.title, ^like_sanitize(name)),
where:
e.visibility == ^:public and
fragment(
"f_unaccent(?) %> f_unaccent(?)",
e.title,
^name
),
order_by:
fragment(
"word_similarity(?, ?) desc",
e.title,
^name
),
preload: [:organizer_actor]
)
|> paginate(page, limit)
@@ -256,11 +268,6 @@ defmodule Mobilizon.Events do
Repo.all(query)
end
# Sanitize the LIKE queries
defp like_sanitize(value) do
"%" <> String.replace(value, ~r/([\\%_])/, "\\1") <> "%"
end
@doc """
Creates a event.

View File

@@ -0,0 +1,99 @@
defmodule MobilizonWeb.API.Search do
alias Mobilizon.Service.ActivityPub
alias Mobilizon.Actors
alias Mobilizon.Actors.Actor
alias Mobilizon.Events
alias Mobilizon.Events.{Event, Comment}
require Logger
@doc """
Search
"""
@spec search(String.t(), integer(), integer()) ::
{:ok, list(Actor.t())} | {:ok, []} | {:error, any()}
def search(search, page \\ 1, limit \\ 10) do
do_search(search, page, limit, %{events: true, actors: true})
end
@doc """
Not used at the moment
"""
# TODO: Use me
@spec search_actors(String.t(), integer(), integer()) ::
{:ok, list(Actor.t())} | {:ok, []} | {:error, any()}
def search_actors(search, page \\ 1, limit \\ 10) do
do_search(search, page, limit, %{actors: true})
end
@doc """
Not used at the moment
"""
# TODO: Use me
@spec search_events(String.t(), integer(), integer()) ::
{:ok, list(Event.t())} | {:ok, []} | {:error, any()}
def search_events(search, page \\ 1, limit \\ 10) do
do_search(search, page, limit, %{events: true})
end
# Do the actual search
@spec do_search(String.t(), integer(), integer(), map()) :: {:ok, list(any())}
defp do_search(search, page, limit, opts) do
search = String.trim(search)
cond do
search == "" ->
{:error, "Search can't be empty"}
String.match?(search, ~r/@/) ->
{:ok, process_from_username(search)}
String.starts_with?(search, "https://") ->
{:ok, process_from_url(search)}
String.starts_with?(search, "http://") ->
{:ok, process_from_url(search)}
true ->
events =
Task.async(fn ->
if Map.get(opts, :events, false),
do: Events.find_events_by_name(search, page, limit),
else: []
end)
actors =
Task.async(fn ->
if Map.get(opts, :actors, false),
do: Actors.find_actors_by_username_or_name(search, page, limit),
else: []
end)
{:ok, Task.await(events) ++ Task.await(actors)}
end
end
# If the search string is an username
@spec process_from_username(String.t()) :: Actor.t() | nil
defp process_from_username(search) do
with {:ok, actor} <- ActivityPub.find_or_make_actor_from_nickname(search) do
actor
else
{:error, _err} ->
Logger.debug("Unable to find or make actor '#{search}'")
nil
end
end
# If the search string is an URL
@spec process_from_url(String.t()) :: Actor.t() | Event.t() | Comment.t() | nil
defp process_from_url(search) do
with {:ok, object} <- ActivityPub.fetch_object_from_url(search) do
object
else
{:error, _err} ->
Logger.debug("Unable to find or make object from URL '#{search}'")
nil
end
end
end

View File

@@ -2,7 +2,6 @@ defmodule MobilizonWeb.Resolvers.Event do
@moduledoc """
Handles the event-related GraphQL calls
"""
alias Mobilizon.Service.ActivityPub
alias Mobilizon.Activity
alias Mobilizon.Events.{Event, Participant}
alias Mobilizon.Actors.User
@@ -122,39 +121,6 @@ defmodule MobilizonWeb.Resolvers.Event do
{:error, "You need to be logged-in to leave an event"}
end
@doc """
Search events by title
"""
def search_events(_parent, %{search: search, page: page, limit: limit}, _resolution) do
{:ok, Mobilizon.Events.find_events_by_name(search, page, limit)}
end
@doc """
Search events and actors by title
"""
def search_events_and_actors(_parent, %{search: search, page: page, limit: limit}, _resolution) do
search = String.trim(search)
found =
case String.contains?(search, "@") do
true ->
with {:ok, actor} <- ActivityPub.find_or_make_actor_from_nickname(search) do
actor
else
{:error, _err} ->
nil
end
_ ->
Mobilizon.Events.find_events_by_name(search, page, limit) ++
Mobilizon.Actors.find_actors_by_username_or_name(search, page, limit)
end
require Logger
Logger.debug(inspect(found))
{:ok, found}
end
@doc """
Create an event
"""

View File

@@ -81,7 +81,7 @@ defmodule MobilizonWeb.Resolvers.Group do
}
}
) do
with {:ok, %Actor{} = group} <- Actors.find_group_by_actor_id(group_id),
with {:ok, %Actor{} = group} <- Actors.get_group_by_actor_id(group_id),
{:is_owned, true, _} <- User.owns_actor(user, actor_id),
{:ok, %Member{} = member} <- Member.get_member(actor_id, group.id),
{:is_admin, true} <- Member.is_administrator(member),

View File

@@ -0,0 +1,13 @@
defmodule MobilizonWeb.Resolvers.Search do
@moduledoc """
Handles the event-related GraphQL calls
"""
alias MobilizonWeb.API.Search
@doc """
Search events and actors by title
"""
def search_events_and_actors(_parent, %{search: search, page: page, limit: limit}, _resolution) do
Search.search(search, page, limit)
end
end

View File

@@ -122,7 +122,7 @@ defmodule MobilizonWeb.Schema do
arg(:search, non_null(:string))
arg(:page, :integer, default_value: 1)
arg(:limit, :integer, default_value: 10)
resolve(&Resolvers.Event.search_events_and_actors/3)
resolve(&Resolvers.Search.search_events_and_actors/3)
end
import_fields(:user_queries)

View File

@@ -66,6 +66,7 @@ defmodule Mobilizon.Service.ActivityPub do
@doc """
Fetch an object from an URL, from our local database of events and comments, then eventually remote
"""
# TODO: Make database calls parallel
@spec fetch_object_from_url(String.t()) :: {:ok, %Event{}} | {:ok, %Comment{}} | {:error, any()}
def fetch_object_from_url(url) do
Logger.info("Fetching object from url #{url}")
@@ -73,6 +74,7 @@ defmodule Mobilizon.Service.ActivityPub do
with true <- String.starts_with?(url, "http"),
nil <- Events.get_event_by_url(url),
nil <- Events.get_comment_from_url(url),
{:error, :actor_not_found} <- Actors.get_actor_by_url(url),
{:ok, %{body: body, status_code: code}} when code in 200..299 <-
HTTPoison.get(
url,
@@ -97,12 +99,16 @@ defmodule Mobilizon.Service.ActivityPub do
"Note" ->
{:ok, Events.get_comment_full_from_url!(activity.data["object"]["id"])}
"Actor" ->
{:ok, Actors.get_actor_by_url!(activity.data["object"]["id"], true)}
other ->
{:error, other}
end
else
%Event{url: event_url} -> {:ok, Events.get_event_by_url!(event_url)}
%Comment{url: comment_url} -> {:ok, Events.get_comment_full_from_url!(comment_url)}
%Actor{url: actor_url} -> {:ok, Actors.get_actor_by_url!(actor_url, true)}
e -> {:error, e}
end
end