Handle aliases, names in namespace

assets/js/utils/local-autocompleter.js

@@ -35,8 +35,10 @@ export class LocalAutocompleter {
     this.referenceStart = this.view.getUint32(backingStore.byteLength - 8, true);
     /** @type {number} */
     this.formatVersion = this.view.getUint32(backingStore.byteLength - 12, true);
+    /** @type {number} */
+    this.numSecondary = (backingStore.byteLength - this.referenceStart - this.numTags * 8 - 12) / 4;
 
-    if (this.formatVersion !== 1) {
+    if (this.formatVersion !== 2) {
       throw new Error('Incompatible autocomplete format version');
     }
   }
@@ -71,11 +73,41 @@ export class LocalAutocompleter {
   getResultAt(i) {
     const nameLocation = this.view.getUint32(this.referenceStart + i * 8, true);
     const imageCount = this.view.getUint32(this.referenceStart + i * 8 + 4, true);
+
+    if (imageCount >>> 31 & 1) {
+      // This is actually an alias, so follow it
+      return this.getResultAt(imageCount & ~(1 << 31));
+    }
+
     const [ name, associations ] = this.getTagFromLocation(nameLocation);
 
     return { name, imageCount, associations };
   }
 
+  /**
+   * Get a Result object as the ith tag inside the file, secondary ordering.
+   *
+   * @param {number} i
+   * @returns {Result}
+   */
+  getSecondaryResultAt(i) {
+    const referenceIndex = this.view.getUint32(this.referenceStart + this.numTags * 8 + i * 4);
+    return this.getResultAt(referenceIndex);
+  }
+
+  /**
+   * Returns the name of a tag without any namespace component.
+   *
+   * @param {string} s
+   * @returns {string}
+   */
+  nameInNamespace(s) {
+    const v = s.split(':', 2);
+
+    if (v.length === 2) return v[1];
+    return v[0];
+  }
+
   /**
    * Find the top k results by image count which match the given string prefix.
    *
@@ -126,6 +158,37 @@ export class LocalAutocompleter {
       }
     }
 
+    // Binary search again to find in secondary list
+    l = 0;
+    r = this.numSecondary;
+
+    while (l < r - 1) {
+      const m = (l + (r - l) / 2) | 0;
+      const { name } = this.getSecondaryResultAt(m);
+
+      if (this.nameInNamespace(name).slice(0, prefix.length) >= prefix) {
+        // too large, go left
+        r = m;
+      }
+      else {
+        // too small, go right
+        l = m;
+      }
+    }
+
+    // Scan forward until no more matches occur
+    while (l < this.numSecondary - 1) {
+      const result = this.getSecondaryResultAt(++l);
+      if (!this.nameInNamespace(result.name).startsWith(prefix)) {
+        break;
+      }
+
+      // Add if no associations are filtered
+      if (hiddenTags.findIndex(ht => result.associations.includes(ht)) === -1) {
+        results.push(result);
+      }
+    }
+
     // Sort results by image count
     results.sort((a, b) => b.imageCount - a.imageCount);
 

lib/philomena/autocomplete.ex

@@ -10,7 +10,7 @@ defmodule Philomena.Autocomplete do
   alias Philomena.Images.Tagging
   alias Philomena.Autocomplete.Autocomplete
 
-  @type tags_list() :: [{String.t(), number(), number()}]
+  @type tags_list() :: [{String.t(), number(), number(), String.t() | nil}]
   @type assoc_map() :: %{String.t() => [number()]}
 
   @spec get_autocomplete() :: Autocomplete.t() | nil
@@ -34,32 +34,78 @@ defmodule Philomena.Autocomplete do
     #       uint32_t associations[];
     #     };
     #
-    #     struct tag_reference {
-    #       uint32_t tag_location;
-    #       uint32_t num_uses;
-    #     };
-    #
 
-    {ac_file, references} =
-      Enum.reduce(tags, {<<>>, <<>>}, fn {name, images_count, _}, {file, references} ->
+    {ac_file, name_locations} =
+      Enum.reduce(tags, {<<>>, %{}}, fn {name, _, _, _}, {file, name_locations} ->
         pos = byte_size(file)
         assn = Map.get(associations, name, [])
         assn_bin = for id <- assn, into: <<>>, do: <<id::32-little>>
 
         {
           <<file::binary, byte_size(name)::8, name::binary, length(assn)::8, assn_bin::binary>>,
-          <<references::binary, pos::32-little, images_count::32-little>>
+          Map.put(name_locations, name, pos)
         }
       end)
 
+    # Link reference list; self-referential, so must be preprocessed to deal with aliases
+    #
+    #     struct tag_reference {
+    #       uint32_t tag_location;
+    #       uint8_t is_aliased : 1;
+    #       union {
+    #         uint32_t num_uses : 31;
+    #         uint32_t alias_index : 31;
+    #       };
+    #     };
+    #
+
     ac_file = int32_align(ac_file)
     reference_start = byte_size(ac_file)
+    size_of_reference = 8
+
+    reference_locations =
+      tags
+      |> Enum.with_index()
+      |> Enum.map(fn {name, index} -> {name, index} end)
+      |> Map.new()
+
+    references =
+      Enum.reduce(tags, <<>>, fn {name, images_count, _, alias_target}, references ->
+        pos = Map.fetch!(name_locations, name)
+
+        if not is_nil(alias_target) do
+          target = Map.fetch!(reference_locations, alias_target)
+
+          <<references::binary, pos::32-little, 1::1, target::31-little>>
+        else
+          <<references::binary, pos::32-little, 0::1, images_count::31-little>>
+        end
+      end)
+
+    # Reorder tags by name in their namespace to provide a secondary ordering
+    #
+    #     struct secondary_reference {
+    #         uint32_t primary_location;
+    #     };
+    #
+
+    secondary_references =
+      tags
+      |> Enum.map(&{name_in_namespace(elem(&1, 0)), &1})
+      |> Enum.uniq_by(fn {k, _v} -> k end)
+      |> Enum.sort()
+      |> Enum.reduce(<<>>, fn {_k, v}, secondary_references ->
+        target = Map.fetch!(reference_locations, v)
+
+        <<secondary_references::binary, target::32-little>>
+      end)
 
     # Finally add the reference start and number of tags in the footer
     #
     #     struct autocomplete_file {
     #       struct tag tags[];
-    #       struct tag_reference references[];
+    #       struct tag_reference primary_references[];
+    #       struct secondary_reference secondary_references[];
     #       uint32_t format_version;
     #       uint32_t reference_start;
     #       uint32_t num_tags;
@@ -67,8 +113,14 @@ defmodule Philomena.Autocomplete do
     #
 
     ac_file =
-      <<ac_file::binary, references::binary, 1::32-little, reference_start::32-little,
-        length(tags)::32-little>>
+      <<
+        ac_file::binary,
+        references::binary,
+        secondary_references::binary,
+        2::32-little,
+        reference_start::32-little,
+        length(tags)::32-little
+      >>
 
     # Insert the autocomplete binary
     new_ac =
@@ -88,13 +140,23 @@ defmodule Philomena.Autocomplete do
   #
   @spec get_tags() :: tags_list()
   defp get_tags do
+    top_tags =
       Tag
-    |> select([t], {t.name, t.images_count, t.id})
+      |> select([t], {t.name, t.images_count, t.id, nil})
       |> where([t], t.images_count > 0)
       |> order_by(desc: :images_count)
-    |> limit(65_535)
+      |> limit(50_000)
       |> Repo.all()
-    |> Enum.filter(fn {name, _, _} -> byte_size(name) < 255 end)
+
+    aliases_of_top_tags =
+      Tag
+      |> where([t], t.aliased_tag_id in ^Enum.map(top_tags, fn {_, _, id, _} -> id end))
+      |> join(:inner, [t], _ in assoc(t, :aliased_tag))
+      |> select([t, a], {t.name, 0, 0, a.name})
+      |> Repo.all()
+
+    (aliases_of_top_tags ++ top_tags)
+    |> Enum.filter(fn {name, _, _, _} -> byte_size(name) < 255 end)
     |> Enum.sort()
   end
 
@@ -104,7 +166,7 @@ defmodule Philomena.Autocomplete do
   @spec get_associations(tags_list()) :: assoc_map()
   defp get_associations(tags) do
     tags
-    |> Enum.map(fn {name, images_count, id} ->
+    |> Enum.map(fn {name, images_count, id, _} ->
       # Randomly sample 100 images with this tag
       image_sample =
         Tagging
@@ -141,4 +203,22 @@ defmodule Philomena.Autocomplete do
 
     <<bin::binary, 0::size(pad_bits)>>
   end
+
+  #
+  # Remove the artist:, oc: etc. prefix from a tag name,
+  # if one is present.
+  #
+  @spec name_in_namespace(String.t()) :: String.t()
+  defp name_in_namespace(s) do
+    case String.split(s, ":", parts: 2, trim: true) do
+      [_namespace, name] ->
+        name
+
+      [name] ->
+        name
+
+      _unknown ->
+        s
+    end
+  end
 end