feat: Bump version to 1.4.1

Merge pull request #40 from newtextdoc1111/dev
Merge Dev
2025-10-14 23:54:20 +09:00 · 2025-10-14 23:49:43 +09:00 · 2025-10-14 23:46:41 +09:00 · 2025-08-30 23:40:27 +09:00 · 2025-08-30 23:39:52 +09:00 · 2025-07-28 16:01:58 +09:00
19 changed files with 1611 additions and 402 deletions
@@ -0,0 +1,5 @@
+{
+  "test": {
+    "NODE_OPTIONS": "--experimental-vm-modules"
+  }
+}
@@ -8,11 +8,6 @@

 **ComfyUI-Autocomplete-Plus** is a custom node that provides multiple input assistance features for any text area in [ComfyUI](https://github.com/comfyanonymous/ComfyUI). Currently, it supports Danbooru and e621 tags (e621 does not support some functions).

-## Recent Updates :fire:
- Added automatic and manual update check functionality for CSV files
- Fixed bug where autocomplete suggestions wouldn't display at certain timings with Microsoft IME
- Support for loading and displaying e621 tag CSV
-
 ## Features

 - **:zap:No setup required**: Automatically downloads CSV data optimized for Danbooru tags.
@@ -24,7 +19,6 @@
    - Natural tag insertion that considers cursor position and existing text.
 - **:art:Design**: Supports both light and dark themes of ComfyUI.
 - **:pencil:User CSV**: Allows users to add their own CSV files for autocomplete suggestions.
- **Zero Dependencies**: All input assistance processing works in the browser without external libraries.

 ## Installation

@@ -47,6 +41,7 @@ When you type in a text input area, tags that partially match the text are displ
 - Tags are color-coded by category. The color-coding rules are the same as Danbooru.
 - Tags that have already been entered are displayed grayed out.
 - You can display Danbooru and e621 tags at the same time. You can also change the priority from the settings.
+- Supports autocomplete for Lora and Embedding inputs. You can enable/disable this feature in the settings.

 ## Related Tags

@@ -92,7 +87,12 @@ Also, displaying related tags is not supported.

 ### User CSV

-Users can also use their own CSV files. For example, frequently used meta tags can be added to the autocomplete suggestions by placing a file named `danbooru_tags_meta.csv` in the `data` folder.
+Users can also use their own CSV files. CSV files should be placed in the `data` folder according to the following naming convention:
+
+- **CSV for Autocomplete**: `<danbooru | e621>_tags*.csv`
+- **CSV for Related Tags**: `<danbooru | e621>_tags_cooccurrence*.csv`
+
+For example, you can add frequently used meta tags to the autocomplete suggestions by placing a file named `danbooru_tags_meta.csv` in the `data` folder.
 A header row is not required. A browser reload is necessary to apply the changes.

 **Example of meta tags:**
@@ -106,6 +106,15 @@ low_quality,5,9999999,
 worst_quality,5,9999999,
 ```

+When the browser is reloaded, you can check the list of loaded CSV files in the ComfyUI console log. If a file is not included in the log output, please verify that the file name follows the naming convention.
+
+**Example of ComfyUI console log output:**
+```
+[Autocomplete-Plus] CSV file status:
+  * Danbooru -> base: True, extra: danbooru_tags_meta.csv // If displayed here, meta tags can be autocompleted
+  * E621 -> base: False, extra:
+```
+
 >[!NOTE]
 > If there are multiple user CSV files, they are loaded in alphabetical order. If the same tag exists in multiple files, the one loaded first is retained. The basic CSV is loaded last.

@@ -124,6 +133,8 @@ worst_quality,5,9999999,

 - **Enable Autocomplete**: Enable/disable the autocomplete feature.
 - **Max suggestions**: Maximum number of autocomplete suggestions to display.
+- **Enable Loras and Embeddings**: Display Lora and Embedding in the suggestions.
+- **Use Fast Search**: Switch autocomplete suggestions search to fast processing (see [About Fast Search for Autocomplete](#about-fast-search-for-autocomplete) for details).

 ### Related Tags

@@ -136,6 +147,58 @@ worst_quality,5,9999999,

 - **Check CSV updates**: Click the "Check Now" button to check if new CSV files are available in HuggingFace and download them if necessary.

+## Advanced Settings
+
+### Disabling CSV Update Check on Startup
+
+By default, ComfyUI performs CSV file update checks and downloads at regular intervals during startup.
+When starting in an environment without internet access, startup may be delayed until a timeout occurs.
+
+You can skip the check process during ComfyUI startup by following these steps:
+
+1. Start ComfyUI once with this custom node installed to generate the `csv_meta.json` file.  
+  The `csv_meta.json` file is created directly under this custom node's folder.
+2. Open `csv_meta.json` in a text editor and change the value of `check_updates_on_startup` from `true` to `false` and save.  
+  If `check_updates_on_startup` does not exist, add it under `version`.
+
+**`csv_meta.json` after modification:**
+```json
+{
+  "version": 1,
+  "check_updates_on_startup": false,
+  ...
+}
+```
+
+**Additional notes:**
+- The check process will not be performed until the value of `check_updates_on_startup` is changed back to `true` or the `version` is switched.
+- Even when `check_updates_on_startup` is `false`, manual checking is still possible by pressing the `Check CSV updates` button in the Autocomplete Plus settings.
+
+## More details on how it works
+
+### About Fast Search for Autocomplete
+
+In `v1.3.0`, Added a fast search feature functionality to autocomplete. By enabling this function in the settings screen, tag search processing during text input operates faster, improving responsiveness.  
+While fast operation can be expected in any scenario, I particularly recommend **enabling** it in the following use cases:
+
+- When the loaded CSV files contain a large number of tags or aliases. This is especially useful when the total number of tags exceeds **100,000**.
+- When using natural language instead of comma-separated tags in prompt input.
+
+**Browser startup behavior**
+
+Fast search requires tag index building and is not available immediately after browser startup. The conventional search process is used until building is complete.  
+As of `v1.3.0`, The notification when building is completed is displayed only in the browser's developer tools. This is planned to be improved in future versions.
+
+For example, when index building for approximately 220,000 tags is completed, the following log is recorded:
+
+```
+[Autocomplete-Plus] Building 221787 index for danbooru took 9398.70ms.
+```
+
+> [!NOTE]
+> - Index building occurs in the background even when fast search is disabled in settings
+> - Fast search uses the full-text search library [nextapps-de/flexsearch](https://github.com/nextapps-de/flexsearch)
+
 ## Known Issues

 ### Performance
@@ -150,7 +213,9 @@ worst_quality,5,9999999,

 ## Credits

- [ComfyUI-Custom-Node](https://github.com/pythongosssss/ComfyUI-Custom-Scripts)
+- [pythongosssss/ComfyUI-Custom-Node](https://github.com/pythongosssss/ComfyUI-Custom-Scripts)
  - Referenced for implementing the autocomplete function.
 - [DominikDoom/a1111-sd-webui-tagcomplete](https://github.com/DominikDoom/a1111-sd-webui-tagcomplete)
-  - Referenced for autocomplete function and CSV specifications.
+  - Referenced for autocomplete function and CSV specifications.
+- [nextapps-de/flexsearch](https://github.com/nextapps-de/flexsearch)
+  - Used to implement fast tag search processing for autocomplete.
@@ -1,4 +1,4 @@
-from .modules.api import *  
+from .modules.api import *
 from .modules import downloader

 # check and download necessary csv files
@@ -6,11 +6,6 @@

 **ComfyUI-Autocomplete-Plus** は、[ComfyUI](https://github.com/comfyanonymous/ComfyUI) の任意のテキストエリアに複数の入力支援機能を提供するカスタムノードです。現在はDanbooruとe621のタグに対応しています（e621は一部の機能が未対応です）。

-## 最近の更新 :fire:
- CSV ファイルの自動・手動更新チェック機能の追加
- Microsoft IMEでオートコンプリート候補が表示されないタイミングがある不具合の修正
- e621タグ CSV の読み込みと表示のサポート
-
 ## 特徴

 - **:zap:セットアップ不要**: Danbooruタグに最適化された CSV データを自動でダウンロード
@@ -22,7 +17,6 @@
    - カーソル位置や既存のテキストを考慮した自然なタグ挿入
 - **:art:デザイン**: ComfyUIのライトテーマとダークテーマの両方に対応
 - **:pencil:ユーザーCSV**: ユーザーが用意した CSV をオートコンプリート候補に追加可能
- **依存ライブラリゼロ**: 外部ライブラリを使用せず、すべての入力支援処理がブラウザで動作

 ## インストール

@@ -45,6 +39,7 @@
 - タグのカテゴリ毎に色分けされます。色分けのルールは Danbooru と同じです
 - 入力済みのタグはグレーアウトで表示されます
 - Danbooruとe621のタグを同時に表示出来ます。設定から優先順位を変更できます
+- LoraとEmbeddingの入力補完に対応しています。設定から有効・無効を切り替えられます

 ## 関連タグ

@@ -90,7 +85,12 @@

 ### ユーザーCSV

-ユーザーが自身で用意した CSV を使用することも可能です。例として、よく使うメタタグを `danbooru_tags_meta.csv` の名前で `data` フォルダーに配置することでオートコンプリート候補に追加できます。  
+ユーザーが自身で用意した CSV を使用することも可能です。 CSV ファイルは以下の命名規則に従って `data` フォルダーに配置してください。
+
+- **オートコンプリート用 CSV**: <danbooru | e621>_tags*.csv
+- **関連タグ用 CSV**: <danbooru | e621>_tags_cooccurrence*.csv
+
+例として、よく使うメタタグを `danbooru_tags_meta.csv` の名前で `data` フォルダーに配置することでオートコンプリート候補に追加できます。
 ヘッダー行はなくても構いません。反映にはブラウザのリロードが必要です。

 **メタタグの例**
@@ -104,6 +104,15 @@ low_quality,5,9999999,
 worst_quality,5,9999999,
 ```

+ブラウザリロード時、ロードされる CSV ファイル一覧をComfyUIのコンソールのログで確認出来ます。ログ出力に含まれていない場合はファイル名が命名規則に沿っているか確認してください。
+
+**ComfyUIコンソールログ出力の例:**
+```
+[Autocomplete-Plus] CSV file status:
+  * Danbooru -> base: True, extra: danbooru_tags_meta.csv // ここに表示されていればメタタグを入力補完できます
+  * E621     -> base: False, extra: 
+```
+
 >[!NOTE]
 > ユーザー CSV が複数ある場合アルファベット順に読み込まれます。同じタグが複数のファイルに存在する場合は先に読み込まれた方が保持されます。基本 CSV は最後にロードされます。

@@ -121,7 +130,9 @@ worst_quality,5,9999999,
 ### オートコンプリート

 - **Enable Autocomplete**: オートコンプリート機能の有効化/無効化
- **Max suggestions**: オートコンプリート候補の最大表示件数
+- **Max Suggestions**: オートコンプリート候補の最大表示件数
+- **Enable Loras and Embeddings**: LoraとEmbeddingを候補に表示する
+- **Use Fast Search**: オートコンプリート候補の検索を高速な処理に切り替える（詳細は [オートコンプリートの高速検索について](#オートコンプリートの高速検索について) を確認してください)

 ### 関連タグ

@@ -134,6 +145,57 @@ worst_quality,5,9999999,

 - **Check CSV updates**: 「Check Now」ボタンを押すと新しい CSV ファイルがHuggingFaceにあるか確認し、必要に応じてダウンロードを行います

+## 上級者向け設定
+
+### 起動時のCSV更新チェックを無効化する
+
+デフォルトの動作では、ComfyUI起動時に一定の間隔で CSV ファイルの更新チェックとダウンロード行います。
+インターネットにアクセス出来ない環境で起動した場合、タイムアウトが発生するまで起動が遅延する事があります。
+
+以下の手順を行う事により、ComfyUI起動時のチェック処理をスキップする事が出来ます。
+
+1. このカスタムノードをインストールした状態でComfyUIを一度起動し、 `csv_meta.json` ファイルを生成する  
+  `csv_meta.json` はこのカスタムノードのフォルダー直下に作成されます
+2. `csv_meat.json` をテキストエディターで開き、`check_updates_on_startup` の値を `true` -> `false` に変更し保存する  
+  `check_updates_on_startup` が存在しない場合、 `version` の下に追記してください
+
+**変更後の `csv_meta.json`：**
+```json
+{
+  "version": 1,
+  "check_updates_on_startup": false,
+  ...
+}
+```
+
+**補足事項：**
+- `check_updates_on_startup` の値を再び `true` にするか、 `version` が切り替わるまでチェック処理は行われなくなります
+- `check_updates_on_startup` が `false` でも、Autocompelte Plusの設定から `Check CSV updates` のボタンを押す事で手動チェックが可能です
+
+## 動作に関する詳細
+
+## オートコンプリートの高速検索について
+
+`v1.3.0` にて、オートコンプリートにタグの検索を高速化する機能を追加しました。設定画面から有効にする事でテキスト入力時のタグ検索処理が高速に動作し、応答性が改善されます。  
+どのような場面でも高速な動作が期待できますが、特に以下のユースケースで **有効** にすることをお勧めします。
+
+- 読み込む CSV ファイルに大量のタグやエイリアスが含まれている場合。タグの合計が **10万件** を超える場合は特に有用です
+- プロンプト入力でタグのカンマ区切りの代わりに自然言語を使う場合
+
+**ブラウザ起動時の動作**
+
+高速検索はタグのインデックス構築が必要なため、ブラウザ起動直後は利用できません。構築が完了するまでは従来の検索処理で動作します。  
+`v1.3.0` 時点では構築が完了したタイミングはブラウザーの開発ツールにのみが表示されます。これは将来のバージョンで改善予定です。
+
+例として、約22万件のタグのインデックス構築が完了した際は以下のようなログが記録されます。
+```
+[Autocomplete-Plus] Building 221787 index for danbooru took 9398.70ms.
+```
+
+> [!NOTE]
+> - インデックス構築は設定から高速検索を無効にしてもバックグラウンドで行われます
+> - 高速検索の実現に全文検索ライブラリの [nextapps-de/flexsearch](https://github.com/nextapps-de/flexsearch) を使用しています 
+
 ## 既知の問題

 ### パフォーマンス
@@ -148,7 +210,9 @@ worst_quality,5,9999999,

 ## クレジット

- [ComfyUI-Custom-Node](https://github.com/pythongosssss/ComfyUI-Custom-Scripts)
+- [pythongosssss/ComfyUI-Custom-Node](https://github.com/pythongosssss/ComfyUI-Custom-Scripts)
  - オートコンプリート機能の実装にあたり参考にしました
 - [DominikDoom/a1111-sd-webui-tagcomplete](https://github.com/DominikDoom/a1111-sd-webui-tagcomplete)
  - オートコンプリートの主要な機能や CSV の仕様で参考にしました
+- [nextapps-de/flexsearch](https://github.com/nextapps-de/flexsearch)
+  - オートコンプリートの高速なタグ検索処理の実装に利用させてもらいました
@@ -1,7 +1,10 @@
-import os
 import json
+import os
+
+import folder_paths
 import server
 from aiohttp import web
+
 from . import downloader as dl

 # Get the absolute path to the 'data' directory
@@ -10,11 +13,12 @@ from . import downloader as dl
 # os.path.join(..., '..', 'data') goes up one level and then into 'data'
 DATA_DIR = os.path.normpath(os.path.join(os.path.dirname(__file__), "..", "data"))

-DANBOORU_PREFIX = 'danbooru'
-E621_PREFIX = 'e621'
+DANBOORU_PREFIX = "danbooru"
+E621_PREFIX = "e621"
+
+TAGS_SUFFIX = "tags"
+COOCCURRENCE_SUFFIX = "tags_cooccurrence"

-TAGS_SUFFIX = 'tags'
-COOCCURRENCE_SUFFIX = 'tags_cooccurrence'

 def get_csv_file_status():
    """
@@ -23,17 +27,17 @@ def get_csv_file_status():

    data = {
        DANBOORU_PREFIX: {
-            'base_tags': False,
-            'extra_tags': [],
-            'base_cooccurrence': False,
-            'extra_cooccurrence': [],
+            "base_tags": False,
+            "extra_tags": [],
+            "base_cooccurrence": False,
+            "extra_cooccurrence": [],
        },
        E621_PREFIX: {
-            'base_tags': False,
-            'extra_tags': [],
-            'base_cooccurrence': False,
-            'extra_cooccurrence': [],
-        }
+            "base_tags": False,
+            "extra_tags": [],
+            "base_cooccurrence": False,
+            "extra_cooccurrence": [],
+        },
    }

    for prefix in [DANBOORU_PREFIX, E621_PREFIX]:
@@ -46,27 +50,28 @@ def get_csv_file_status():
        tags_extra_files = []
        cooccurrence_extra_files = []

-        all_csv_files = [f for f in os.listdir(DATA_DIR) if f.startswith(prefix) and f.endswith('.csv')]
+        all_csv_files = [f for f in os.listdir(DATA_DIR) if f.startswith(prefix) and f.endswith(".csv")]

        # Create extra CSV files list
        for filename in all_csv_files:
            if filename in [base_tags_file, base_cooccurrence_file]:
-                continue # Skip base files
+                continue  # Skip base files
            if COOCCURRENCE_SUFFIX in filename.lower():
                cooccurrence_extra_files.append(filename)
            elif TAGS_SUFFIX in filename.lower():
                tags_extra_files.append(filename)

        data[prefix] = {
-            'base_tags': tags_base_exists,
-            'extra_tags': tags_extra_files,
-            'base_cooccurrence': cooccurrence_base_exists,
-            'extra_cooccurrence': cooccurrence_extra_files,
+            "base_tags": tags_base_exists,
+            "extra_tags": tags_extra_files,
+            "base_cooccurrence": cooccurrence_base_exists,
+            "extra_cooccurrence": cooccurrence_extra_files,
        }

    # Return the lists of extra files
    return data

+
 def get_last_check_time_from_metadata():
    """
    Helper function to get the last remote check timestamp from csv_meta.json.
@@ -75,23 +80,25 @@ def get_last_check_time_from_metadata():
    try:
        if not os.path.exists(dl.CSV_META_FILE):
            return None
-        
-        with open(dl.CSV_META_FILE, 'r', encoding='utf-8') as f:
+
+        with open(dl.CSV_META_FILE, "r", encoding="utf-8") as f:
            metadata = json.load(f)
-        
+
        datasets = metadata.get("hf_datasets", [])
        if datasets and len(datasets) > 0:
            return datasets[0].get("last_remote_check_timestamp")
-        
+
        return None
-        
+
    except (IOError, json.JSONDecodeError) as e:
        print(f"[Autocomplete-Plus] Error reading csv_meta.json: {e}")
        return None

+
 # --- API Endpoints ---

-@server.PromptServer.instance.routes.get('/autocomplete-plus/csv')
+
+@server.PromptServer.instance.routes.get("/autocomplete-plus/csv")
 async def get_csv_list(_request):
    """
    Returns CSV file status.
@@ -102,37 +109,48 @@ async def get_csv_list(_request):

    response = {
        DANBOORU_PREFIX: {
-            'base_tags': csv_file_status[DANBOORU_PREFIX]['base_tags'],
-            'extra_tags': len(csv_file_status[DANBOORU_PREFIX]['extra_tags']),
-            'base_cooccurrence': csv_file_status[DANBOORU_PREFIX]['base_cooccurrence'],
-            'extra_cooccurrence': len(csv_file_status[DANBOORU_PREFIX]['extra_cooccurrence']),
+            "base_tags": csv_file_status[DANBOORU_PREFIX]["base_tags"],
+            "extra_tags": csv_file_status[DANBOORU_PREFIX]["extra_tags"],
+            "base_cooccurrence": csv_file_status[DANBOORU_PREFIX]["base_cooccurrence"],
+            "extra_cooccurrence": csv_file_status[DANBOORU_PREFIX]["extra_cooccurrence"],
        },
        E621_PREFIX: {
-            'base_tags': csv_file_status[E621_PREFIX]['base_tags'],
-            'extra_tags': len(csv_file_status[E621_PREFIX]['extra_tags']),
-            'base_cooccurrence': csv_file_status[E621_PREFIX]['base_cooccurrence'],
-            'extra_cooccurrence': len(csv_file_status[E621_PREFIX]['extra_cooccurrence']),
-        }
+            "base_tags": csv_file_status[E621_PREFIX]["base_tags"],
+            "extra_tags": csv_file_status[E621_PREFIX]["extra_tags"],
+            "base_cooccurrence": csv_file_status[E621_PREFIX]["base_cooccurrence"],
+            "extra_cooccurrence": csv_file_status[E621_PREFIX]["extra_cooccurrence"],
+        },
    }
+
+    # Print csv file status to the console for debugging
+    print(f"""[Autocomplete-Plus] CSV file status:
+  * Danbooru -> base: {response[DANBOORU_PREFIX]["base_tags"]}, extra: [{", ".join(response[DANBOORU_PREFIX]["extra_tags"])}]
+  * E621     -> base: {response[E621_PREFIX]["base_tags"]}, extra: [{", ".join(response[E621_PREFIX]["extra_tags"])}]""")
+
    return web.json_response(response)

-@server.PromptServer.instance.routes.get('/autocomplete-plus/csv/{source}/{suffix}/base')
+
+@server.PromptServer.instance.routes.get("/autocomplete-plus/csv/{source}/{suffix}/base")
 async def get_base_tags_file(request):
    """
    Returns the base tags CSV file.
    """
-    source = str(request.match_info['source'])
-    suffix = str(request.match_info['suffix'])
-    if source not in [DANBOORU_PREFIX, E621_PREFIX] or suffix not in [TAGS_SUFFIX, COOCCURRENCE_SUFFIX]:
+    source = str(request.match_info["source"])
+    suffix = str(request.match_info["suffix"])
+    if source not in [DANBOORU_PREFIX, E621_PREFIX] or suffix not in [
+        TAGS_SUFFIX,
+        COOCCURRENCE_SUFFIX,
+    ]:
        return web.json_response({"error": "Invalid tag source or suffix"}, status=400)
-    
+
    file_path = os.path.join(DATA_DIR, f"{source}_{suffix}.csv")
    if not os.path.exists(file_path):
        return web.json_response({"error": "Base tags file not found"}, status=404)

    return web.FileResponse(file_path)

-@server.PromptServer.instance.routes.get('/autocomplete-plus/csv/{source}/{suffix}/extra/{index}')
+
+@server.PromptServer.instance.routes.get("/autocomplete-plus/csv/{source}/{suffix}/extra/{index}")
 async def get_extra_tags_file(request):
    """
    Returns the extra tags CSV file at the specified index.
@@ -140,16 +158,19 @@ async def get_extra_tags_file(request):
    try:
        csv_file_status = get_csv_file_status()

-        source = str(request.match_info['source'])
-        suffix = str(request.match_info['suffix'])
-        if source not in [DANBOORU_PREFIX, E621_PREFIX] or suffix not in [TAGS_SUFFIX, COOCCURRENCE_SUFFIX]:
+        source = str(request.match_info["source"])
+        suffix = str(request.match_info["suffix"])
+        if source not in [DANBOORU_PREFIX, E621_PREFIX] or suffix not in [
+            TAGS_SUFFIX,
+            COOCCURRENCE_SUFFIX,
+        ]:
            return web.json_response({"error": "Invalid tag source or suffix"}, status=400)
-    
-        index = int(request.match_info['index'])
-        if index < 0 or index >= len(csv_file_status[source][f'extra_{suffix}']):
+
+        index = int(request.match_info["index"])
+        if index < 0 or index >= len(csv_file_status[source][f"extra_{suffix}"]):
            return web.json_response({"error": "Invalid index"}, status=404)

-        file_path = os.path.join(DATA_DIR, csv_file_status[source][f'extra_{suffix}'][index])
+        file_path = os.path.join(DATA_DIR, csv_file_status[source][f"extra_{suffix}"][index])
        if not os.path.exists(file_path):
            return web.json_response({"error": "Extra tags file not found"}, status=404)

@@ -158,7 +179,8 @@ async def get_extra_tags_file(request):
    except ValueError:
        return web.json_response({"error": "Invalid index format"}, status=400)

-@server.PromptServer.instance.routes.post('/autocomplete-plus/csv/force-check-updates')
+
+@server.PromptServer.instance.routes.post("/autocomplete-plus/csv/force-check-updates")
 async def force_check_csv_updates(request):
    """
    Forces a check for CSV file updates from HuggingFace, ignoring cooldown.
@@ -166,29 +188,29 @@ async def force_check_csv_updates(request):
    """
    try:
        print("[Autocomplete-Plus] Starting forced check for CSV updates from HuggingFace...")
-        
+
        downloader = dl.Downloader()
        downloader.run_check_and_download(force_check=True)
-        
+
        print("[Autocomplete-Plus] Forced check completed successfully.")
-        
+
        # Get the updated last check time
        last_check_time = get_last_check_time_from_metadata()
-        
-        return web.json_response({
-            "success": True,
-            "message": "Force check completed successfully",
-            "last_check_time": last_check_time
-        })
-        
+
+        return web.json_response(
+            {
+                "success": True,
+                "message": "Force check completed successfully",
+                "last_check_time": last_check_time,
+            }
+        )
+
    except Exception as e:
        print(f"[Autocomplete-Plus] Error during forced check: {e}")
-        return web.json_response({
-            "success": False,
-            "error": str(e)
-        }, status=500)
+        return web.json_response({"success": False, "error": str(e)}, status=500)

-@server.PromptServer.instance.routes.get('/autocomplete-plus/csv/last-check-time')
+
+@server.PromptServer.instance.routes.get("/autocomplete-plus/csv/last-check-time")
 async def get_last_check_time(_request):
    """
    Returns the last remote check timestamp from csv_meta.json.
@@ -196,26 +218,33 @@ async def get_last_check_time(_request):
    """
    try:
        if not os.path.exists(dl.CSV_META_FILE):
-            return web.json_response({
-                "last_check_time": None,
-                "message": "csv_meta.json file not found"
-            })
-        
+            return web.json_response({"last_check_time": None, "message": "csv_meta.json file not found"})
+
        last_check_time = get_last_check_time_from_metadata()
-        
+
        if last_check_time is not None:
-            return web.json_response({
-                "last_check_time": last_check_time
-            })
+            return web.json_response({"last_check_time": last_check_time})
        else:
-            return web.json_response({
-                "last_check_time": None,
-                "message": "No datasets found in metadata"
-            })
-            
+            return web.json_response({"last_check_time": None, "message": "No datasets found in metadata"})
+
    except (IOError, json.JSONDecodeError) as e:
        print(f"[Autocomplete-Plus] Error reading csv_meta.json: {e}")
-        return web.json_response({
-            "last_check_time": None,
-            "error": str(e)
-        }, status=500)
+        return web.json_response({"last_check_time": None, "error": str(e)}, status=500)
+
+
+@server.PromptServer.instance.routes.get("/autocomplete-plus/embeddings")
+async def get_embeddings(request):
+    """
+    Returns a list of embedding files.
+    """
+    embeddings = folder_paths.get_filename_list("embeddings")
+    return web.json_response(list(map(lambda a: os.path.splitext(a)[0], embeddings)))
+
+
+@server.PromptServer.instance.routes.get("/autocomplete-plus/loras")
+async def get_loras(request):
+    """
+    Returns a list of lora files.
+    """
+    loras = folder_paths.get_filename_list("loras")
+    return web.json_response(list(map(lambda a: os.path.splitext(a)[0], loras)))
@@ -20,7 +20,8 @@ CSV_META_FILE = os.path.normpath(os.path.join(os.path.dirname(__file__), "..", C

 DEFAULT_CSV_METADATA = {
    "version": 1,
-    "hf_datasets" : [
+    "check_updates_on_startup": True,
+    "hf_datasets": [
        {
            "hf_dataset_id": "newtextdoc1111/danbooru-tag-csv",
            "last_remote_check_timestamp": None,
@@ -34,12 +35,15 @@ DEFAULT_CSV_METADATA = {
                    "file_name": "danbooru_tags_cooccurrence.csv",
                    "last_download": None,
                    "last_modified_on_hf": None,
-                }
-            ]
+                },
+            ],
        }
-    ]
+    ],
 }

+# --- HuggingFace Constants ---
+HUGGINGFACE_URL = "https://huggingface.co"
+

 # --- Helper Functions ---
 def get_file_path(file_name: str) -> str:
@@ -82,12 +86,14 @@ class Downloader:
            return default_metadata

        try:
-            with open(CSV_META_FILE, 'r', encoding='utf-8') as f:
+            with open(CSV_META_FILE, "r", encoding="utf-8") as f:
                metadata = json.load(f)

                if not isinstance(metadata, dict) or metadata.get("version") != DEFAULT_CSV_METADATA["version"]:
-                    print(f"[Autocomplete-Plus] Metadata version mismatch. Expected {DEFAULT_CSV_METADATA['version']}, "
-                          f"found {metadata.get('version')}. Using default metadata.")
+                    print(
+                        f"[Autocomplete-Plus] Metadata version mismatch. Expected {DEFAULT_CSV_METADATA['version']}, "
+                        f"found {metadata.get('version')}. Using default metadata."
+                    )
                    return default_metadata
                else:
                    self.csv_meta_file_exists_at_start = True
@@ -101,7 +107,7 @@ class Downloader:
        """Saves metadata to CSV_META_FILE."""
        try:
            os.makedirs(os.path.dirname(CSV_META_FILE), exist_ok=True)
-            with open(CSV_META_FILE, 'w', encoding='utf-8') as f:
+            with open(CSV_META_FILE, "w", encoding="utf-8") as f:
                json.dump(self.metadata, f, indent=2)
        except IOError as e:
            print(f"[Autocomplete-Plus] Error saving metadata to {CSV_META_FILE}: {e}")
@@ -111,12 +117,15 @@ class Downloader:
        Retrieves the Last-Modified header for a file on HuggingFace and returns it as an ISO 8601 string.
        hf_filename should be the full filename, e.g., "danbooru_tags.csv".
        """
-        url = f"https://huggingface.co/datasets/{dataset_repo_id}/resolve/main/{hf_filename}"
+        url = f"{HUGGINGFACE_URL}/datasets/{dataset_repo_id}/resolve/main/{hf_filename}"
        try:
-            req = urllib.request.Request(url, method='HEAD',
-                                         headers={"User-Agent": "ComfyUI-Autocomplete-Plus (Python urllib)"})
+            req = urllib.request.Request(
+                url,
+                method="HEAD",
+                headers={"User-Agent": "ComfyUI-Autocomplete-Plus (Python urllib)"},
+            )
            with urllib.request.urlopen(req, timeout=10) as response:
-                last_modified_http = response.getheader('Last-Modified')
+                last_modified_http = response.getheader("Last-Modified")
                if last_modified_http:
                    dt_object = parsedate_to_datetime(last_modified_http)
                    if dt_object.tzinfo is None or dt_object.tzinfo.utcoffset(dt_object) is None:
@@ -141,7 +150,7 @@ class Downloader:
        """
        Downloads a file synchronously with progress display to a temporary location.
        """
-        download_url = f"https://huggingface.co/datasets/{hf_dataset_id}/resolve/main/{file_name}"
+        download_url = f"{HUGGINGFACE_URL}/datasets/{hf_dataset_id}/resolve/main/{file_name}"
        final_path = get_file_path(file_name)
        temp_path = get_temp_download_path(file_name)
        now_utc = datetime.now(timezone.utc).isoformat()
@@ -155,18 +164,28 @@ class Downloader:
            if os.path.exists(temp_path):
                os.remove(temp_path)

-            req = urllib.request.Request(download_url, headers={"User-Agent": "Mozilla/5.0 (Windows NT 11.0; Win64)"})
+            req = urllib.request.Request(
+                download_url,
+                headers={"User-Agent": "Mozilla/5.0 (Windows NT 11.0; Win64)"},
+            )

            with urllib.request.urlopen(req) as response:
                total_size_str = response.getheader("Content-Length")
                total_size = int(total_size_str) if total_size_str else None
                chunk_size = 8192

-                with open(temp_path, "wb") as f_out, \
-                        tqdm(total=total_size, unit='B', unit_scale=True, unit_divisor=1024,
-                             desc=f"[Autocomplete-Plus] Downloading {file_name}", leave=False,
-                             bar_format='{l_bar}{bar}| {n_fmt}/{total_fmt} [{elapsed}<{remaining}, {rate_fmt}{postfix}]'
-                             ) as pbar:
+                with (
+                    open(temp_path, "wb") as f_out,
+                    tqdm(
+                        total=total_size,
+                        unit="B",
+                        unit_scale=True,
+                        unit_divisor=1024,
+                        desc=f"[Autocomplete-Plus] Downloading {file_name}",
+                        leave=False,
+                        bar_format="{l_bar}{bar}| {n_fmt}/{total_fmt} [{elapsed}<{remaining}, {rate_fmt}{postfix}]",
+                    ) as pbar,
+                ):
                    while True:
                        chunk = response.read(chunk_size)
                        if not chunk:
@@ -197,12 +216,15 @@ class Downloader:
            if os.path.exists(final_path):
                try:
                    file_size_at_final = os.path.getsize(final_path)
-                    if file_size_at_final == 0 or \
-                            (total_size and total_size > 0 and file_size_at_final < total_size) or \
-                            (not total_size and downloaded_size > 0 and file_size_at_final < downloaded_size):
+                    if (
+                        file_size_at_final == 0
+                        or (total_size and total_size > 0 and file_size_at_final < total_size)
+                        or (not total_size and downloaded_size > 0 and file_size_at_final < downloaded_size)
+                    ):
                        os.remove(final_path)
                        print(
-                            f"[Autocomplete-Plus] Removed potentially corrupted file at final destination: {final_path}")
+                            f"[Autocomplete-Plus] Removed potentially corrupted file at final destination: {final_path}"
+                        )
                except OSError as rm_e:
                    print(f"[Autocomplete-Plus] Error removing potentially corrupted file {final_path}: {rm_e}")

@@ -213,15 +235,7 @@ class Downloader:

    def _check_new_csv_from_hf_dataset(self, dataset_meta: dict, now_utc: datetime, force_check: bool = False):
        """Checks HuggingFace for file updates and updates metadata."""
-        perform_hf_check = True
-        if not force_check and dataset_meta.get("last_remote_check_timestamp"):
-            try:
-                last_check_dt = datetime.fromisoformat(dataset_meta["last_remote_check_timestamp"])
-                if now_utc - last_check_dt < timedelta(days=7):
-                    perform_hf_check = False
-            except (ValueError, KeyError, TypeError):
-                print(
-                    "[Autocomplete-Plus] Invalid or missing timestamp for last_remote_check_timestamp. Will perform remote check.")
+        perform_hf_check = force_check

        if perform_hf_check:
            huggingface_dataset_id = dataset_meta["hf_dataset_id"]
@@ -246,7 +260,8 @@ class Downloader:
                dataset_meta["last_remote_check_timestamp"] = now_utc.isoformat()
            else:
                print(
-                    "[Autocomplete-Plus] Could not update all remote timestamps from HuggingFace. Will try again later.")
+                    "[Autocomplete-Plus] Could not update all remote timestamps from HuggingFace. Will try again later."
+                )

    def _download_csv_files_if_needed(self, dataset_meta: dict):
        """Downloads CSV files if they are missing, outdated, or previously failed."""
@@ -279,15 +294,16 @@ class Downloader:
            return f"File {file_name} is missing or empty locally."

        # Check if the last modified date on HuggingFace is newer than the last download date
-        try:
-            last_download_dt = datetime.fromisoformat(file_meta_entry["last_download"])
-            hf_modified_dt = datetime.fromisoformat(file_meta_entry["last_modified_on_hf"])
-            if hf_modified_dt > last_download_dt:
-                return f"Remote file {file_name} is newer (HF: {last_download_dt}, Local Download: {hf_modified_dt})."
-        except (ValueError, TypeError):
-            file_meta_entry["last_download"] = None
-            file_meta_entry["last_modified_on_hf"] = None
-            return f"Invalid timestamp format for {file_name}. Forcing download to ensure integrity."
+        # Temporarily comment out until changes to remote files can be detected
+        # try:
+        #     last_download_dt = datetime.fromisoformat(file_meta_entry["last_download"])
+        #     hf_modified_dt = datetime.fromisoformat(file_meta_entry["last_modified_on_hf"])
+        #     if hf_modified_dt > last_download_dt:
+        #         return f"Remote file {file_name} is newer (HF: {hf_modified_dt}, Local Download: {last_download_dt})."
+        # except (ValueError, TypeError):
+        #     file_meta_entry["last_download"] = None
+        #     file_meta_entry["last_modified_on_hf"] = None
+        #     return f"Invalid timestamp format for {file_name}. Forcing download to ensure integrity."

        # If the file is missing or empty, but the last download timestamp exists, we need to retry.
        if not check_file_valid(local_file_path) and file_meta_entry.get("last_download") is not None:
@@ -299,10 +315,14 @@ class Downloader:
        """
        Orchestrates the process of checking for updates and downloading CSV files.
        This is the main entry point for the downloader logic.
-        
+
        Args:
            force_check: If True, forces a check of HuggingFace regardless of the last check timestamp.
        """
+        # If check_updates_on_startup is False and force_check is not set, skip the check
+        if not (self.metadata.get("check_updates_on_startup", True) or force_check):
+            print('[Autocomplete-Plus] "check_updates_on_startup" is disabled. Skipping CSV update check and download.')
+            return

        now_utc = datetime.now(timezone.utc)

@@ -8,6 +8,7 @@
        "@babel/core": "^7.27.1",
        "@babel/preset-env": "^7.27.2",
        "babel-jest": "^29.7.0",
+        "env-cmd": "^10.1.0",
        "jest": "^29.7.0",
        "stylelint": "^16.19.1",
        "stylelint-config-idiomatic-order": "^10.0.0",
@@ -3018,6 +3019,16 @@
      "dev": true,
      "license": "MIT"
    },
+    "node_modules/commander": {
+      "version": "4.1.1",
+      "resolved": "https://registry.npmjs.org/commander/-/commander-4.1.1.tgz",
+      "integrity": "sha512-NOKm8xhkzAjzFx8B2v5OAHT+u5pRQc2UCa2Vq9jYL/31o2wi9mxBA7LIFs3sV5VSC49z6pEhfbMULvShKj26WA==",
+      "dev": true,
+      "license": "MIT",
+      "engines": {
+        "node": ">= 6"
+      }
+    },
    "node_modules/concat-map": {
      "version": "0.0.1",
      "resolved": "https://registry.npmjs.org/concat-map/-/concat-map-0.0.1.tgz",
@@ -3266,6 +3277,23 @@
      "dev": true,
      "license": "MIT"
    },
+    "node_modules/env-cmd": {
+      "version": "10.1.0",
+      "resolved": "https://registry.npmjs.org/env-cmd/-/env-cmd-10.1.0.tgz",
+      "integrity": "sha512-mMdWTT9XKN7yNth/6N6g2GuKuJTsKMDHlQFUDacb/heQRRWOTIZ42t1rMHnQu4jYxU1ajdTeJM+9eEETlqToMA==",
+      "dev": true,
+      "license": "MIT",
+      "dependencies": {
+        "commander": "^4.0.0",
+        "cross-spawn": "^7.0.0"
+      },
+      "bin": {
+        "env-cmd": "bin/env-cmd.js"
+      },
+      "engines": {
+        "node": ">=8.0.0"
+      }
+    },
    "node_modules/env-paths": {
      "version": "2.2.1",
      "resolved": "https://registry.npmjs.org/env-paths/-/env-paths-2.2.1.tgz",
@@ -1,12 +1,13 @@
 {
  "type": "module",
  "scripts": {
-    "test": "jest"
+    "test": "env-cmd -e test -- jest"
  },
  "devDependencies": {
    "@babel/core": "^7.27.1",
    "@babel/preset-env": "^7.27.2",
    "babel-jest": "^29.7.0",
+    "env-cmd": "^10.1.0",
    "jest": "^29.7.0",
    "stylelint": "^16.19.1",
    "stylelint-config-idiomatic-order": "^10.0.0",
@@ -1,7 +1,7 @@
 [project]
 name = "comfyui-autocomplete-plus"
 description = "Autocomplete and Related Tag display for ComfyUI"
-version = "1.0.1"
+version = "1.4.1"
 license = {file = "LICENSE"}
 dependencies = ["",]

@@ -9,7 +9,16 @@ dependencies = ["",]
 Repository = "https://github.com/newtextdoc1111/ComfyUI-Autocomplete-Plus"
 #  Used by Comfy Registry https://comfyregistry.org

+[dependency-groups]
+dev = [
+    "ruff>=0.12.2",
+]
+
 [tool.comfy]
 PublisherId = "newtextdoc1111"
 DisplayName = "ComfyUI-Autocomplete-Plus"
 Icon = ""
+
+[tool.ruff]
+# Allow lines to be as long as 120.
+line-length = 120
@@ -0,0 +1,387 @@
+
+import {
+    createFlexSearchDocument,
+    createFlexSearchDocumentForModel,
+    __test__
+} from "../../web/js/searchengine.js";
+
+const { createTagEncoder, createCJKEncoder, createModelEncoder } = __test__;
+
+function parseCSVLine(line) {
+    const result = [];
+    let current = '';
+    let inQuotes = false;
+
+    for (let i = 0; i < line.length; i++) {
+        const char = line[i];
+
+        if (char === '"') {
+            if (inQuotes && i + 1 < line.length && line[i + 1] === '"') {
+                current += '"';
+                i++;
+            } else {
+                inQuotes = !inQuotes;
+            }
+        } else if (char === ',' && !inQuotes) {
+            result.push(current);
+            current = '';
+        } else {
+            current += char;
+        }
+    }
+
+    result.push(current);
+
+    return result;
+}
+
+describe('FlexSearch Integration', () => {
+
+    const commonCSV = `
+1girl,0,6008644,"1girls,sole_female"
+highres,5,5256195,"high_res,high_resolution,hires"
+solo,0,5000954,"alone,female_solo,single,solo_female,solo_in_panel"
+long_hair,0,4350743,"/lh,longhair,very_long_hair"
+one_two_three,0,29389,
+`;
+
+    const cjkAliasCSV = `
+blue_hair,0,676176,"青髪,青い髪,水色髪"
+red_hair,0,413261,"赤髪,紅髪,红发,빨강머리,빨간머리"
+smile,0,2294308,"笑い,スマイル,笑顔,笑顏,守りたい、この笑顔,笑,笑容,微笑み,微笑,微笑む,미소,守りたいこの笑顔"
+gloves,0,1105296,"手袋,裸手袋,手袋コキ,てぶくろ,장갑,手套"
+dragon_girl,0,37930,"竜娘,ドラゴン娘,龍娘,龙娘,メスドラ,辰娘"
+double_bun,0,103538,"お団子頭,お団子"
+sanshoku_dango,0,2061,"三色団子,三色团子,花見団子,花见团子"
+`;
+
+    const specialCharCSV = `
+:d,0,436700,
+>:),0,11041,
+year:1999,0,1999,
+d.d.,0,1999,
+copyright_(series),2,1298,"copyright,コピーライト (シリーズ),コピーライト名,コピーライト,著作"
+__wildcard__,0,0,
+`;
+
+    const ModelCSV = `
+<lora:my_lora1>,0,0,
+<lora:日本語Lora_v1>,0,0,
+embedding: my_embedding,0,0,
+`;
+
+    const mockCSV = [
+        commonCSV, cjkAliasCSV, specialCharCSV
+    ].map(csv => csv.trim()).join('\n');
+
+    let mockTags, mockModelTags;
+
+    let tagEncoder, cjkEncoder, modelEncoder;
+    let document, modelDocument;
+
+    let performSearch = function (query, limit = 100) {
+        const ids1 = document.search(query, {
+            field: ["tag", "alias"],
+            limit: limit,
+            suggest: false,
+            merge: true,
+        }).map(r => r.id);
+
+        const result1 = mockTags.filter(tag => ids1.includes(tag.id)).map(tag => tag.tag);
+
+        const ids2 = modelDocument.search(query, {
+            field: ["tag", "alias"],
+            limit: limit,
+            suggest: false,
+            merge: true,
+        }).map(r => r.id);
+
+        const result2 = mockModelTags.filter(tag => ids2.includes(tag.id)).map(tag => tag.tag);
+
+        return [...result1, ...result2];
+    }
+
+    beforeEach(() => {
+        mockTags = mockCSV.split('\n').map((line, id) => {
+            const [tag, category, count, alias] = parseCSVLine(line);
+            return { id, tag, category: parseInt(category), count: parseInt(count), alias };
+        });
+
+        tagEncoder = createTagEncoder();
+        cjkEncoder = createCJKEncoder();
+
+        document = createFlexSearchDocument();
+
+        mockTags.forEach(data => document.add(data));
+
+        mockModelTags = ModelCSV.split('\n').map((line, id) => {
+            const [tag, category, count, alias] = parseCSVLine(line);
+            return { id, tag, category: parseInt(category), count: parseInt(count), alias };
+        });
+
+        modelEncoder = createModelEncoder();
+        modelDocument = createFlexSearchDocumentForModel();
+        mockModelTags.forEach(data => modelDocument.add(data));
+    });
+
+    describe('Encoder', () => {
+        test('should split underscore-separated tags', () => {
+            const encoded = tagEncoder.encode('sanshoku_dango');
+            expect(encoded).toEqual(['sanshoku', 'dango']);
+        });
+
+        test('should extract words from parentheses', () => {
+            const encoded = tagEncoder.encode('copyright_(series)');
+            expect(encoded).toEqual(['copyright', 'series']);
+        });
+        test('should preserve colon-separated special tags', () => {
+            const encoded = tagEncoder.encode('year:1234');
+            expect(encoded).toEqual(['year:1234']);
+        });
+        test('should preserve dot-separated tags', () => {
+            const encoded = tagEncoder.encode('d.d.');
+            expect(encoded).toEqual(['d.d.']);
+        });
+        test('should preserve double underscore wildcard tags', () => {
+            const encoded = tagEncoder.encode('__wildcard__');
+            expect(encoded).toEqual(['__wildcard__']);
+        });
+        test('should convert katakana to hiragana', () => {
+            const encoded = cjkEncoder.encode('ガーデン');
+            expect(encoded).toEqual(['がーでん']);
+        });
+        test('should remove trailing underscores when splitting', () => {
+            const encoded = tagEncoder.encode('one_two_');
+            expect(encoded).toEqual(['one', 'two']);
+        });
+        test('should properly encode embedding notation', () => {
+            let encoded = modelEncoder.encode('embedding:path/to/my_embed1');
+            expect(encoded).toEqual(['embedding:', 'path', 'to', 'my', 'embed1']);
+
+            expect(
+                modelEncoder.encode('embedding:path\\to\\my-embed1')
+            ).toEqual(['embedding:', 'path', 'to', 'my', 'embed1']);
+
+            expect(
+                modelEncoder.encode('embedding:path\\to\\this is my embed. my-negative01 (v1)__by me')
+            ).toEqual(['embedding:', 'path', 'to', 'this', 'is', 'my', 'embed', 'my', 'negative01', 'v1', 'by', 'me']);
+        });
+        test('should properly encode lora notation', () => {
+            expect(
+                modelEncoder.encode('<lora:path/to/my_lora1>')
+            ).toEqual(['lora:', 'path', 'to', 'my', 'lora1']);
+
+            expect(
+                modelEncoder.encode('<lora:path\\to\\my-lora1>')
+            ).toEqual(['lora:', 'path', 'to', 'my', 'lora1']);
+
+            expect(
+                modelEncoder.encode('<lora:path\\to\\this is my lora. my-style01 (v1)__by me>')
+            ).toEqual(['lora:', 'path', 'to', 'this', 'is', 'my', 'lora', 'my', 'style01', 'v1', 'by', 'me']);
+        });
+    });
+
+    describe('Basic Search', () => {
+        test('should find a tag by exact match', () => {
+            const results = performSearch('1girl');
+            expect(results.length).toBeGreaterThan(0);
+            expect(results).toContain('1girl');
+        });
+
+        test('should find a tag by partial match (substring)', () => {
+            const results = performSearch('blue');
+            expect(results.length).toBeGreaterThan(0);
+            expect(results).toContain('blue_hair');
+        });
+
+        test('should be case-insensitive', () => {
+            const results = performSearch('BLUE_HAIR');
+            expect(results.length).toBeGreaterThan(0);
+
+            expect(results).toContain('blue_hair');
+        });
+
+        test('should find a tag by backward', () => {
+            const results = performSearch('gon');
+            expect(results.length).toEqual(1);
+
+            expect(results).toContain('dragon_girl');
+        });
+
+        test('should find a tag for terms contain space', () => {
+            const results = performSearch('double ');
+            expect(results.length).toEqual(1);
+
+            expect(results).toContain('double_bun');
+        });
+
+        test('should find a tag for terms contain underscore', () => {
+            const results = performSearch('double_');
+            expect(results.length).toEqual(1);
+
+            expect(results).toContain('double_bun');
+        });
+    });
+
+    describe('Alias Search', () => {
+        test('should find a tag by its Japanese alias', () => {
+            const results = performSearch('髪');
+            expect(results.length).toBeGreaterThan(0);
+
+            expect(results).toContain('blue_hair');
+        });
+
+        test('should find a tag by one of its multiple aliases', () => {
+            const results = performSearch('笑顔');
+            expect(results.length).toBeGreaterThan(0);
+
+            expect(results).toContain('smile');
+        });
+
+        test('should find a tag by its partial Japanese alias', () => {
+            const results = performSearch('青い');
+            expect(results.length).toBeGreaterThan(0);
+
+            expect(results).toContain('blue_hair');
+        });
+
+        test('should find a tag by katakana', () => {
+            const results = performSearch('テブクロ');
+            expect(results.length).toEqual(1);
+
+            expect(results).toContain('gloves');
+        });
+
+        test('should find a tag by hiragana', () => {
+            const results = performSearch('すまいる');
+            expect(results.length).toEqual(1);
+
+            expect(results).toContain('smile');
+        });
+
+        test('should not find a tag by english alias substring', () => {
+            const results = performSearch('meg');
+            expect(results.length).toEqual(0);
+        });
+
+        test('should find a tag by japanese alias substring', () => {
+            const results = performSearch('団子');
+            expect(results.length).toEqual(2);
+
+            expect(results).toContain('sanshoku_dango');
+            expect(results).toContain('double_bun');
+        });
+    });
+
+    describe('Special Characters and Edge Cases', () => {
+        test('should find a tag with parentheses', () => {
+            const results = performSearch('copyright_(series)');
+            expect(results.length).toBeGreaterThan(0);
+
+            expect(results).toContain('copyright_(series)');
+        });
+
+        test('should find a tag by searching for content inside parentheses', () => {
+            const results = performSearch('series');
+            expect(results.length).toBeGreaterThan(0);
+
+            expect(results).toContain('copyright_(series)');
+        });
+
+        test('should find a tag by partial word', () => {
+            const results = performSearch('right');
+            expect(results.length).toBeGreaterThan(0);
+
+            expect(results).toContain('copyright_(series)');
+        });
+
+        test('should return an empty array for a non-existent tag', () => {
+            const results = performSearch('non_existent_tag_xyz');
+            expect(results).toEqual([]);
+        });
+
+        test('should match to special character only tag', () => {
+            const tag = '>:)';
+            const results = performSearch(tag);
+            expect(results.length).toEqual(1);
+
+            expect(results).toContain(tag);
+        });
+
+        test('should match to contain special character tag', () => {
+            const tag = ':d';
+            const results = performSearch(tag);
+            expect(results.length).toEqual(1);
+
+            expect(results).toContain(tag);
+        });
+
+        test('should match to contain special character tag2', () => {
+            const tag = 'year:1999';
+            const results = performSearch(tag);
+            expect(results.length).toEqual(1);
+
+            expect(results).toContain(tag);
+        });
+
+        test('should match to contain special character tag3', () => {
+            const tag = 'd.d.';
+            const results = performSearch(tag);
+            expect(results.length).toEqual(1);
+
+            expect(results).toContain(tag);
+        });
+
+        test('should match to wildcard tag', () => {
+            const tag = '__';
+            const results = performSearch(tag);
+            expect(results.length).toEqual(1);
+
+            expect(results).toContain('__wildcard__');
+        });
+
+        test('should match to lora tag', () => {
+            const tag = '<lora';
+            const results = performSearch(tag);
+            expect(results.length).toEqual(2);
+
+            expect(results).toContain("<lora:my_lora1>");
+            expect(results).toContain("<lora:日本語Lora_v1>");
+        });
+
+        test('should match to lora tag2', () => {
+            const tag = 'lora:';
+            const results = performSearch(tag);
+            expect(results.length).toEqual(2);
+
+            expect(results).toContain("<lora:my_lora1>");
+            expect(results).toContain("<lora:日本語Lora_v1>");
+        });
+
+        test('should match to lora that contain CJK characters', () => {
+            const word = 'lora: 日本語';
+            const results = performSearch(word);
+            expect(results.length).toEqual(1);
+
+            expect(results).toContain("<lora:日本語Lora_v1>");
+        });
+    });
+
+    describe('Search Options', () => {
+        test('should respect the limit option', () => {
+            const results = performSearch('hair', 1);
+
+            expect(results.length).toEqual(1);
+        });
+
+        test('should return all matches when limit is higher than results', () => {
+            const results = performSearch('hair', 5);
+            expect(results.length).toBeGreaterThan(0);
+
+            expect(results).toHaveLength(3);
+            expect(results).toContain('long_hair');
+            expect(results).toContain('blue_hair');
+            expect(results).toContain('red_hair');
+        });
+    });
+});
@@ -37,6 +37,62 @@ body.dark-theme {
  color: var(--input-text);
 }

+#autocomplete-plus-root .danbooru[data-tag-category="general"] {
+  color: var(--autocomplete-plus-text-color-cat-blue);
+}
+
+#autocomplete-plus-root .danbooru[data-tag-category="artist"] {
+  color: var(--autocomplete-plus-text-color-cat-red);
+}
+
+#autocomplete-plus-root .danbooru[data-tag-category="copyright"] {
+  color: var(--autocomplete-plus-text-color-cat-purple);
+}
+
+#autocomplete-plus-root .danbooru[data-tag-category="character"] {
+  color: var(--autocomplete-plus-text-color-cat-green);
+}
+
+#autocomplete-plus-root .danbooru[data-tag-category="meta"] {
+  color: var(--autocomplete-plus-text-color-cat-yellow);
+}
+
+#autocomplete-plus-root .danbooru[data-tag-category="unknown"] {
+  color: var(--autocomplete-plus-text-color-cat-gray);
+}
+
+#autocomplete-plus-root .e621[data-tag-category="general"] {
+  color: var(--autocomplete-plus-text-color-cat-sky);
+}
+
+#autocomplete-plus-root .e621[data-tag-category="artist"] {
+  color: var(--autocomplete-plus-text-color-cat-orange);
+}
+
+#autocomplete-plus-root .e621[data-tag-category="copyright"] {
+  color: var(--autocomplete-plus-text-color-cat-purple);
+}
+
+#autocomplete-plus-root .e621[data-tag-category="character"] {
+  color: var(--autocomplete-plus-text-color-cat-green);
+}
+
+#autocomplete-plus-root .e621[data-tag-category="species"] {
+  color: var(--autocomplete-plus-text-color-cat-red);
+}
+
+#autocomplete-plus-root .e621[data-tag-category="meta"] {
+  color: var(--autocomplete-plus-text-color-cat-white);
+}
+
+#autocomplete-plus-root .e621[data-tag-category="lore"] {
+  color: var(--autocomplete-plus-text-color-cat-red);
+}
+
+#autocomplete-plus-root .e621[data-tag-category="unknown"] {
+  color: var(--autocomplete-plus-text-color-cat-purple);
+}
+
 #autocomplete-plus-list {
  display: grid;
  box-shadow: 0 2px 8px rgb(0 0 0 / 30%);
@@ -52,62 +108,6 @@ body.dark-theme {
  grid-template-columns: subgrid;
 }

-.autocomplete-plus-item.danbooru[data-tag-category="general"] {
-  color: var(--autocomplete-plus-text-color-cat-blue);
-}
-
-.autocomplete-plus-item.danbooru[data-tag-category="artist"] {
-  color: var(--autocomplete-plus-text-color-cat-red);
-}
-
-.autocomplete-plus-item.danbooru[data-tag-category="copyright"] {
-  color: var(--autocomplete-plus-text-color-cat-purple);
-}
-
-.autocomplete-plus-item.danbooru[data-tag-category="character"] {
-  color: var(--autocomplete-plus-text-color-cat-green);
-}
-
-.autocomplete-plus-item.danbooru[data-tag-category="meta"] {
-  color: var(--autocomplete-plus-text-color-cat-yellow);
-}
-
-.autocomplete-plus-item.danbooru[data-tag-category="unknown"] {
-  color: var(--autocomplete-plus-text-color-cat-gray);
-}
-
-.autocomplete-plus-item.e621[data-tag-category="general"] {
-  color: var(--autocomplete-plus-text-color-cat-sky);
-}
-
-.autocomplete-plus-item.e621[data-tag-category="artist"] {
-  color: var(--autocomplete-plus-text-color-cat-orange);
-}
-
-.autocomplete-plus-item.e621[data-tag-category="copyright"] {
-  color: var(--autocomplete-plus-text-color-cat-purple);
-}
-
-.autocomplete-plus-item.e621[data-tag-category="character"] {
-  color: var(--autocomplete-plus-text-color-cat-green);
-}
-
-.autocomplete-plus-item.e621[data-tag-category="species"] {
-  color: var(--autocomplete-plus-text-color-cat-red);
-}
-
-.autocomplete-plus-item.e621[data-tag-category="meta"] {
-  color: var(--autocomplete-plus-text-color-cat-white);
-}
-
-.autocomplete-plus-item.e621[data-tag-category="lore"] {
-  color: var(--autocomplete-plus-text-color-cat-red);
-}
-
-.autocomplete-plus-item.e621[data-tag-category="unknown"] {
-  color: var(--autocomplete-plus-text-color-cat-purple);
-}
-
 .autocomplete-plus-item span {
  align-content: center;
  padding: 4px 8px;
@@ -124,11 +124,7 @@ body.dark-theme {
 }

 .autocomplete-plus-item.selected span {
-  background-color: var(
-    --comfy-menu-bg-selected,
-    var(--comfy-button-bg)
-  );
-  font-weight: bold;
+  background-color: var(--border-color);
 }

 .autocomplete-plus-item:hover span {
@@ -172,26 +168,91 @@ body.dark-theme {
  color: var(--input-text);
 }

+#related-tags-root .danbooru[data-tag-category="general"] {
+  color: var(--autocomplete-plus-text-color-cat-blue);
+}
+
+#related-tags-root .danbooru[data-tag-category="artist"] {
+  color: var(--autocomplete-plus-text-color-cat-red);
+}
+
+#related-tags-root .danbooru[data-tag-category="copyright"] {
+  color: var(--autocomplete-plus-text-color-cat-purple);
+}
+
+#related-tags-root .danbooru[data-tag-category="character"] {
+  color: var(--autocomplete-plus-text-color-cat-green);
+}
+
+#related-tags-root .danbooru[data-tag-category="meta"] {
+  color: var(--autocomplete-plus-text-color-cat-yellow);
+}
+
+#related-tags-root .danbooru[data-tag-category="unknown"] {
+  color: var(--autocomplete-plus-text-color-cat-gray);
+}
+
+#related-tags-root .e621[data-tag-category="general"] {
+  color: var(--autocomplete-plus-text-color-cat-sky);
+}
+
+#related-tags-root .e621[data-tag-category="artist"] {
+  color: var(--autocomplete-plus-text-color-cat-orange);
+}
+
+#related-tags-root .e621[data-tag-category="copyright"] {
+  color: var(--autocomplete-plus-text-color-cat-purple);
+}
+
+#related-tags-root .e621[data-tag-category="character"] {
+  color: var(--autocomplete-plus-text-color-cat-green);
+}
+
+#related-tags-root .e621[data-tag-category="species"] {
+  color: var(--autocomplete-plus-text-color-cat-red);
+}
+
+#related-tags-root .e621[data-tag-category="meta"] {
+  color: var(--autocomplete-plus-text-color-cat-white);
+}
+
+#related-tags-root .e621[data-tag-category="lore"] {
+  color: var(--autocomplete-plus-text-color-cat-red);
+}
+
+#related-tags-root .e621[data-tag-category="unknown"] {
+  color: var(--autocomplete-plus-text-color-cat-gray);
+}
+
 #related-tags-header {
  position: sticky;
  z-index: 1;
  top: 0;
  display: flex;
  align-items: center;
-  justify-content: space-between;
-  padding: 8px 12px;
+  padding: 8px;
  border-bottom: 1px solid var(--border-color);
  background-color: var(--comfy-menu-bg);
  color: var(--descrip-text);
+  gap: 8px;
 }

-.related-tags-header-text {
-  flex-grow: 1;
+.related-tags-header-text-container {
+  display:flex;
+  flex: 1;
+  flex-wrap: wrap;
+  align-items: center;
+  gap: 12px;
+}

+.related-tags-header-tag-text {
+  padding: 2px 0;
+  line-height: 1.25em;
 }

 .related-tags-header-tag-name {
  font-weight: bold;
+  word-break: break-all;
 }

 .related-tags-header-tag-name .autocomplete-plus-tag-icon-svg {
@@ -200,17 +261,22 @@ body.dark-theme {
  vertical-align: middle;
 }

+.related-tags-header-tag-alias {
+    overflow: auto;
+    max-height: 5em;
+    padding: 2px 4px;
+    border-radius: 4px;
+    background-color: var(--bg-color);
+    color: var(--descrip-text);
+    line-height: 1.25em;
+}
+
 .related-tags-header-controls {
-  display: flex;
-  align-items: center;
-  margin-left: 16px;
-  gap: 4px;
+  margin-left: auto;
 }

-.related-tags-layout-toggle {
-  padding: 2px 6px;
+.related-tags-header-controls button {
  border: none;
-  border-radius: 4px;
  background: none;
  color: var(--input-text);
  cursor: pointer;
@@ -218,19 +284,8 @@ body.dark-theme {
  transition: background-color 0.2s;
 }

-.related-tags-layout-toggle:hover {
-  background-color: var(--comfy-button-bg);
-}
-
-.related-tags-pin-toggle {
-  padding: 2px 6px;
-  border: none;
-  border-radius: 4px;
-  background: none;
-  color: var(--input-text);
-  cursor: pointer;
-  font-size: 14px;
-  transition: background-color 0.2s;
+.related-tags-header-controls button:hover {
+  background-color: var(--border-color);
 }

 #related-tags-list {
@@ -248,62 +303,6 @@ body.dark-theme {
  grid-template-columns: subgrid;
 }

-.related-tag-item.danbooru[data-tag-category="general"] {
-  color: var(--autocomplete-plus-text-color-cat-blue);
-}
-
-.related-tag-item.danbooru[data-tag-category="artist"] {
-  color: var(--autocomplete-plus-text-color-cat-red);
-}
-
-.related-tag-item.danbooru[data-tag-category="copyright"] {
-  color: var(--autocomplete-plus-text-color-cat-purple);
-}
-
-.related-tag-item.danbooru[data-tag-category="character"] {
-  color: var(--autocomplete-plus-text-color-cat-green);
-}
-
-.related-tag-item.danbooru[data-tag-category="meta"] {
-  color: var(--autocomplete-plus-text-color-cat-yellow);
-}
-
-.related-tag-item.danbooru[data-tag-category="unknown"] {
-  color: var(--autocomplete-plus-text-color-cat-gray);
-}
-
-.related-tag-item.e621[data-tag-category="general"] {
-  color: var(--autocomplete-plus-text-color-cat-sky);
-}
-
-.related-tag-item.e621[data-tag-category="artist"] {
-  color: var(--autocomplete-plus-text-color-cat-orange);
-}
-
-.related-tag-item.e621[data-tag-category="copyright"] {
-  color: var(--autocomplete-plus-text-color-cat-purple);
-}
-
-.related-tag-item.e621[data-tag-category="character"] {
-  color: var(--autocomplete-plus-text-color-cat-green);
-}
-
-.related-tag-item.e621[data-tag-category="species"] {
-  color: var(--autocomplete-plus-text-color-cat-red);
-}
-
-.related-tag-item.e621[data-tag-category="meta"] {
-  color: var(--autocomplete-plus-text-color-cat-white);
-}
-
-.related-tag-item.e621[data-tag-category="lore"] {
-  color: var(--autocomplete-plus-text-color-cat-red);
-}
-
-.related-tag-item.e621[data-tag-category="unknown"] {
-  color: var(--autocomplete-plus-text-color-cat-gray);
-}
-
 /* stylelint-disable-next-line no-descending-specificity */
 .related-tag-item span {
  align-content: center;
@@ -324,11 +323,7 @@ body.dark-theme {
 }

 .related-tag-item.selected span {
-  background-color: var(
-    --comfy-menu-bg-selected,
-    var(--comfy-button-bg)
-  );
-  font-weight: bold;
+  background-color: var(--border-color);
 }

 .related-tag-item .related-tag-name {
@@ -352,9 +347,11 @@ body.dark-theme {
  white-space: nowrap;
 }

-.related-tags-loading-message {
+.related-tags-message {
+  width: 100%;
  padding: 12px;
  font-style: italic;
+  grid-column: 1 / -1;
  text-align: center;
 }

@@ -1,13 +1,16 @@
 import {
+    ModelTagSource,
    TagCategory,
    TagData,
    autoCompleteData,
    getEnabledTagSourceInPriorityOrder
 } from './data.js';
 import {
-    formatCountHumanReadable,
+    isLongText,
    hiraToKata,
    kataToHira,
+    formatCountHumanReadable,
+    escapeHtml,
    isContainsLetterOrNumber,
    normalizeTagToInsert,
    normalizeTagToSearch,
@@ -78,20 +81,17 @@ function matchWord(target, queries) {
 * @returns {Array<TagData>} The list of matching candidates.
 */
 function searchCompletionCandidates(textareaElement) {
-    const startTime = performance.now(); // Record start time for performance measurement

    const ESCAPE_SEQUENCE = ["#", "/"]; // If the first string is that character, autocomplete will not be displayed.
    const partialTag = getCurrentPartialTag(textareaElement);
-    if (!partialTag || partialTag.length <= 0 || ESCAPE_SEQUENCE.some(seq => partialTag.startsWith(seq))) {
-        return []; // No valid input for autocomplete    
+    if (!partialTag || partialTag.length <= 0 ||
+        ESCAPE_SEQUENCE.some(seq => partialTag.startsWith(seq)) ||
+        isLongText(partialTag)) {
+        return []; // No valid input for autocomplete
    }

-    const exactMatches = [];
-    const partialMatches = [];
-    const addedTags = new Set();
-
    // Generate Hiragana/Katakana variations if applicable
-    const queryVariations = new Set([partialTag, normalizeTagToSearch(partialTag)]);
+    const queryVariations = new Set([partialTag.toLowerCase(), normalizeTagToSearch(partialTag).toLowerCase()]);
    const kataQuery = hiraToKata(partialTag);
    if (kataQuery !== partialTag) {
        queryVariations.add(kataQuery);
@@ -101,6 +101,26 @@ function searchCompletionCandidates(textareaElement) {
        queryVariations.add(hiraQuery);
    }

+    if (settingValues.useFastSearch) {
+        return searchWithFlexSearch(partialTag, queryVariations);
+    } else {
+        return sequentialSearch(partialTag, queryVariations);
+    }
+}
+
+/**
+ * Search completion candidates using sequential search.
+ * @param {string} partialTag 
+ * @param {Set<string>} queryVariations 
+ * @returns 
+ */
+function sequentialSearch(partialTag, queryVariations) {
+    const startTime = performance.now();
+
+    const exactMatches = [];
+    const partialMatches = [];
+    const addedTags = new Set();
+
    const sources = getEnabledTagSourceInPriorityOrder();
    for (const source of sources) {
        // Search in sortedTags (already sorted by count)
@@ -110,15 +130,14 @@ function searchCompletionCandidates(textareaElement) {
            let matchedAlias = null;

            // Check primary tag against all variations for exact/partial match
-            const tagMatch = matchWord(tagData.tag, queryVariations);
+            const tagMatch = matchWord(tagData.tag.toLowerCase(), queryVariations);
            matched = tagMatch.matched;
            isExactMatch = tagMatch.isExactMatch;

            // If primary tag didn't match, check aliases against all variations
            if (!matched && tagData.alias && Array.isArray(tagData.alias) && tagData.alias.length > 0) {
                for (const alias of tagData.alias) {
-                    const lowerAlias = alias.toLowerCase();
-                    const aliasMatch = matchWord(lowerAlias, queryVariations);
+                    const aliasMatch = matchWord(alias.toLowerCase(), queryVariations);
                    if (aliasMatch.matched) {
                        matched = true;
                        isExactMatch = aliasMatch.isExactMatch;
@@ -170,6 +189,69 @@ function searchCompletionCandidates(textareaElement) {
    return candidates;
 }

+/**
+ * Search completion candidates using FlexSearch for fast matching.
+ * @param {string} partialTag 
+ * @param {Set<string>} queryVariations 
+ * @returns 
+ */
+function searchWithFlexSearch(partialTag, queryVariations) {
+    const startTime = performance.now();
+
+    let mergedResult = [];
+    let totalSearchCount = 0;
+
+    const sources = getEnabledTagSourceInPriorityOrder();
+    for (const source of sources) {
+        if (!autoCompleteData[source].flexSearchDocument) continue;
+        if (mergedResult.length >= settingValues.maxSuggestions) break;
+
+        // Use the FlexSearch Document to search
+        // NOTE: The limit param is reflected separately for "tag" and "alias".
+        let searchResult = autoCompleteData[source].flexSearchDocument.search(partialTag, {
+            field: ["tag", "alias"],
+            limit: Math.min(settingValues.maxSuggestions * 10, 500),
+            merge: true,
+            suggest: false,
+            cache: true,
+        });
+
+        if (!searchResult || searchResult.length <= 0) continue;
+
+        // Sort results based on exact matches and counts
+        searchResult = searchResult
+            .map(r => autoCompleteData[source].sortedTags[r.id])
+            .sort((aTag, bTag) => {
+                if (matchWord(bTag.tag, queryVariations).isExactMatch) {
+                    return 999999999999;
+                }
+                if (matchWord(aTag.tag, queryVariations).isExactMatch) {
+                    return -999999999999;
+                }
+                if (bTag.alias && bTag.alias.some(alias => matchWord(alias, queryVariations).isExactMatch)) {
+                    return 999999999999;
+                }
+                if (aTag.alias && aTag.alias.some(alias => matchWord(alias, queryVariations).isExactMatch)) {
+                    return -999999999999;
+                }
+                return bTag.count - aTag.count;
+            });
+
+        // Merge results into the final array
+        mergedResult = mergedResult.concat(searchResult.slice(0, settingValues.maxSuggestions - mergedResult.length));
+
+        totalSearchCount += searchResult.length;
+    }
+
+    if (settingValues._logprocessingTime) {
+        const endTime = performance.now();
+        const duration = endTime - startTime;
+        console.debug(`[Autocomplete-Plus] Fast Search for "${partialTag}" took ${duration.toFixed(2)}ms.Found ${mergedResult.length} candidates within ${totalSearchCount} searches from flexsearch.`);
+    }
+
+    return mergedResult;
+}
+
 /**
 * Extracts the current tag being typed before the cursor.
 * @param {HTMLTextAreaElement} inputElement
@@ -218,9 +300,9 @@ function getCurrentPartialTag(inputElement) {
 * Inserts the selected tag into the textarea, replacing the partial tag,
 * making the change undoable.
 * @param {HTMLTextAreaElement} inputElement
- * @param {string} tagToInsert The raw tag string to insert.
+ * @param {TagData} tagDataToInsert The raw tag string to insert.
 */
-function insertTagToTextArea(inputElement, tagToInsert) {
+function insertTagToTextArea(inputElement, tagDataToInsert) {
    const text = inputElement.value;
    const cursorPos = inputElement.selectionStart;

@@ -228,7 +310,13 @@ function insertTagToTextArea(inputElement, tagToInsert) {
    const replaceStart = Math.min(cursorPos, tagStart);
    let replaceEnd = cursorPos;

-    const normalizedTag = normalizeTagToInsert(tagToInsert);
+    let normalizedTag;
+    if(Object.values(ModelTagSource).includes(tagDataToInsert.source)){
+        // If the tag is from a model tag source, don't want to normalize it
+        normalizedTag = tagDataToInsert.tag;
+    }else{
+        normalizedTag = normalizeTagToInsert(tagDataToInsert.tag);
+    }

    const currentTagAfterCursor = text.substring(cursorPos, tagEnd).trimEnd();
    if (normalizedTag.lastIndexOf(currentTagAfterCursor) !== -1) {
@@ -297,7 +385,7 @@ class AutocompleteUI {
        this.tagsList.addEventListener('mousedown', (e) => {
            const row = e.target.closest('.autocomplete-plus-item');
            if (row && row.dataset.tag) {
-                this.#insertTag(row.dataset.tag);
+                this.#insertTag(row.dataset);
                e.preventDefault(); // Prevent focus loss from input
                e.stopPropagation();
            }
@@ -362,10 +450,12 @@ class AutocompleteUI {
        this.#highlightItem();
    }

-    /** Selects the currently highlighted item */
+    /** Selects the currently highlighted item
+     * @returns {TagData|null} The selected tag data.
+     */
    getSelectedTag() {
        if (this.selectedIndex >= 0 && this.selectedIndex < this.candidates.length) {
-            return this.candidates[this.selectedIndex].tag;
+            return this.candidates[this.selectedIndex];
        }

        return null; // No valid selection
@@ -411,9 +501,10 @@ class AutocompleteUI {
        if (settingValues.tagSourceIconPosition == 'hidden') {
            tagName.textContent = tagData.tag;
        } else {
+            const escapedTag = escapeHtml(tagData.tag);
            tagName.innerHTML = settingValues.tagSourceIconPosition == 'left'
-                ? `${tagSourceIconHtml} ${tagData.tag}`
-                : `${tagData.tag} ${tagSourceIconHtml}`;
+                ? `${tagSourceIconHtml} ${escapedTag}`
+                : `${escapedTag} ${tagSourceIconHtml}`;
        }

        // grayout tag name if it already exists
@@ -554,10 +645,10 @@ class AutocompleteUI {

    /**
     * Handles the selection of an item
-     * @param {string} selectedTag The tag to insert.
+     * @param {TagData} selectedTag The tag to insert.
     */
    #insertTag(selectedTag) {
-        if (!this.target || !selectedTag || selectedTag.length <= 0) {
+        if (!this.target || !selectedTag) {
            this.hide();
            return;
        }
@@ -918,4 +1009,4 @@ export class AutocompleteEventHandler {
     */
    handleClick(event) {
    }
-}
+}
@@ -1,13 +1,20 @@
-import { settingValues } from "./settings.js";
+import { settingValues, updateMaxTagLength } from "./settings.js";
+import { createFlexSearchDocument, createFlexSearchDocumentForModel } from "./searchengine.js";

 // --- Constants ---

-// Tag data sources
+// Tag sources for booru-like tag data.
 export const TagSource = {
    Danbooru: 'danbooru',
    E621: 'e621',
 }

+// Tag sources for model based tag data.
+export const ModelTagSource = {
+    Embeddings: 'embeddings',
+    Lora: 'lora'
+}
+
 export const TagCategory = {
    'danbooru': [
        'general',
@@ -27,6 +34,12 @@ export const TagCategory = {
        'invalid',
        'meta',
        'lore',
+    ],
+    'embeddings': [
+        'embeddings'
+    ],
+    'lora': [
+        'lora'
    ]
 }

@@ -39,19 +52,19 @@ export class TagData {
    /**
     * Create a tag data object
     * @param {string} tag - The tag name
-     * @param {string[]} [alias=[]] - Array of aliases for the tag
-     * @param {string} [category='general'] - Category of the tag
+     * @param {number} [category] - Category index of the tag
     * @param {number} [count=0] - Frequency count/popularity of the tag
-     * @param {string} [source=TagSources.Danbooru] - The source of the tag data
+     * @param {string[]} [alias=[]] - Array of aliases for the tag
+     * @param {string} [source=TagSource.Danbooru] - The source of the tag data
     */
-    constructor(tag, alias = [], category = 'general', count = 0, source = TagSource.Danbooru) {
+    constructor(tag, category, count = 0, alias = [], source = TagSource.Danbooru) {
        /** @type {string} */
        this.tag = tag;

        /** @type {string[]} */
        this.alias = alias;

-        /** @type {string} */
+        /** @type {number} */
        this.category = category;

        /** @type {number} */
@@ -63,6 +76,9 @@ export class TagData {

 class AutocompleteData {
    constructor() {
+        /** @type {Document} */
+        this.flexSearchDocument = null;
+
        /** @type {TagData[]} */
        this.sortedTags = [];

@@ -80,7 +96,6 @@ class AutocompleteData {

        // Progress of "base" csv loading
        this.baseLoadingProgress = {
-            // tags: 0,
            cooccurrence: 0
        };
    }
@@ -95,24 +110,32 @@ export const autoCompleteData = {};
 const TAGS_CSV_HEADER = 'tag,category,count,alias';
 const TAGS_CSV_HEADER_COLUMNS = TAGS_CSV_HEADER.split(',');
 const TAG_INDEX = TAGS_CSV_HEADER_COLUMNS.indexOf('tag');
-const ALIAS_INDEX = TAGS_CSV_HEADER_COLUMNS.indexOf('alias');
 const CATEGORY_INDEX = TAGS_CSV_HEADER_COLUMNS.indexOf('category');
 const COUNT_INDEX = TAGS_CSV_HEADER_COLUMNS.indexOf('count');
+const ALIAS_INDEX = TAGS_CSV_HEADER_COLUMNS.indexOf('alias');

 // --- Helder Functions ---

+
 /**
 * Get the available tag sources in priority order based on the current settings.
 * @returns {string[]} Array of available tag sources in priority order
 */
 export function getEnabledTagSourceInPriorityOrder() {
-    return Object.values(TagSource)
+    let enabledTagSources = Object.values(TagSource)
        .filter((s) => {
            return settingValues.tagSource === s || settingValues.tagSource === 'all';
        })
        .toSorted((a, b) => {
            return a === settingValues.primaryTagSource ? -1 : 1;
        });
+
+    // Append Loras and Embeddings if enabled
+    if (settingValues.enableModels) {
+        enabledTagSources = [...enabledTagSources, ...Object.values(ModelTagSource)];
+    }
+
+    return enabledTagSources;
 }

 // --- Data Loading Functions ---
@@ -156,7 +179,9 @@ async function loadTags(csvUrl, siteName) {
                const aliases = aliasStr ? aliasStr.split(',').map(a => a.trim()).filter(a => a.length > 0) : [];

                // Create a TagData instance instead of a plain object
-                const tagData = new TagData(tag, aliases, category, count, siteName);
+                const tagData = new TagData(tag, category, count, aliases, siteName);
+
+                updateMaxTagLength(tag.length);

                autoCompleteData[siteName].sortedTags.push(tagData);
            } else {
@@ -187,6 +212,52 @@ async function loadTags(csvUrl, siteName) {
    }
 }

+/**
+ * Build FlexSearch index for the given site name.
+ * @param {string} siteName 
+ */
+async function buildFlexSearchIndex(siteName) {
+    try {
+        if (autoCompleteData[siteName].sortedTags.length === 0) {
+            return;
+        }
+
+        let document = null;
+        if (Object.values(TagSource).includes(siteName)) {
+            document = createFlexSearchDocument();
+        } else if (Object.values(ModelTagSource).includes(siteName)) {
+            document = createFlexSearchDocumentForModel();
+        } else {
+            throw new Error(`[Autocomplete-Plus] Invalid site name: ${siteName}`);
+        }
+
+        let startIdx = 0;
+        const startTime = performance.now();
+        function processChunkTasks() {
+            const chunkSize = 1000;
+            const end = Math.min(startIdx + chunkSize, autoCompleteData[siteName].sortedTags.length);
+            for (; startIdx < end; startIdx++) {
+                const tagData = autoCompleteData[siteName].sortedTags[startIdx];
+                document.add(startIdx, tagData);
+            }
+
+            if (startIdx < autoCompleteData[siteName].sortedTags.length) {
+                setTimeout(processChunkTasks, 0);
+                // console.log(`[Autocomplete-Plus] Current porcess: ${startIdx}`);
+            } else {
+                autoCompleteData[siteName].flexSearchDocument = document;
+
+                const endTime = performance.now();
+                const duration = endTime - startTime;
+                console.info(`[Autocomplete-Plus] Building ${autoCompleteData[siteName].sortedTags.length} index for ${siteName} took ${duration.toFixed(2)}ms.`);
+            }
+        }
+        processChunkTasks();
+    } catch (error) {
+        console.error(`[Autocomplete-Plus] Failed to building flexSearch index`, error);
+    }
+}
+
 /**
 * Loads co-occurrence data from a single CSV file.
 * @param {string} csvUrl - The URL of the CSV file to load.
@@ -297,7 +368,11 @@ function parseCSVLine(line) {
    return result;
 }

-export async function fetchCsvList() {
+/**
+ * Fetch the list of CSV files from the API endpoint
+ * @returns {Promise<void>}
+ */
+async function fetchCsvList() {
    try {
        const response = await fetch('/autocomplete-plus/csv');
        if (!response.ok) {
@@ -313,9 +388,8 @@ export async function fetchCsvList() {

 /**
 * Initializes the autocomplete data by fetching the list of CSV files and loading them.
- * This function is called when the extension is initialized.
 */
-export async function initializeData(csvListData, source) {
+async function initializeDataFromCSV(csvListData, source) {
    if (autoCompleteData.hasOwnProperty(source) === false) {
        autoCompleteData[source] = new AutocompleteData();
    }
@@ -326,7 +400,6 @@ export async function initializeData(csvListData, source) {

    const startTime = performance.now();
    autoCompleteData[source].isInitializing = true;
-    // console.log("[Autocomplete-Plus] Initializing autocomplete data...");

    try {
        // Store functions that return Promises (Promise Factories)
@@ -340,8 +413,8 @@ export async function initializeData(csvListData, source) {
            return;
        }

-        const extraTagsCount = csvListData[source].extra_tags || 0;
-        const extraCooccurrenceCount = csvListData[source].extra_cooccurrence || 0;
+        const extraTagsFileList = csvListData[source].extra_tags || [];
+        const extraCooccurrenceFileList = csvListData[source].extra_cooccurrence || [];

        const tagsUrl = `/autocomplete-plus/csv/${source}/tags`;
        const cooccurrenceUrl = `/autocomplete-plus/csv/${source}/tags_cooccurrence`;
@@ -349,7 +422,7 @@ export async function initializeData(csvListData, source) {
        // Factory for loading tags for the current sourte
        const siteTagsLoaderFactory = async () => {
            let promiseChain = Promise.resolve();
-            for (let i = 0; i < extraTagsCount; i++) {
+            for (let i = 0; i < extraTagsFileList.length; i++) {
                promiseChain = promiseChain.then(() => loadTags(`${tagsUrl}/extra/${i}`, source));
            }
            if (csvListData[source].base_tags) {
@@ -362,7 +435,7 @@ export async function initializeData(csvListData, source) {
        // Factory for loading cooccurrence data for the current sourte
        const siteCooccurrenceLoaderFactory = async () => {
            let promiseChain = Promise.resolve();
-            for (let i = 0; i < extraCooccurrenceCount; i++) {
+            for (let i = 0; i < extraCooccurrenceFileList.length; i++) {
                promiseChain = promiseChain.then(() => loadCooccurrence(`${cooccurrenceUrl}/extra/${i}`, source));
            }
            if (csvListData[source].base_cooccurrence) {
@@ -375,16 +448,22 @@ export async function initializeData(csvListData, source) {
        // Now, execute all promise factories and wait for their completion.
        // The actual loading (fetch calls) will start when the factories are invoked here.
        await Promise.all([
-            Promise.all(tagsLoadPromiseFactories.map(factory => factory())).then(() => {
-                const endTime = performance.now();
-                console.log(`[Autocomplete-Plus] "${source}" Tags loading complete in ${(endTime - startTime).toFixed(2)}ms`);
-            }),
+            Promise.all(tagsLoadPromiseFactories.map(factory => factory()))
+                .then(() => {
+                    // Build FlexSearch index after tags are loaded
+                    return buildFlexSearchIndex(source);
+                })
+                .then(() => {
+                    const endTime = performance.now();
+                    if (csvListData[source].base_tags) {
+                        console.log(`[Autocomplete-Plus] "${source}" Tags loading complete in ${(endTime - startTime).toFixed(2)}ms`);
+                    }
+                }),
            Promise.all(cooccurrenceLoadPromiseFactories.map(factory => factory())).then(() => {
                const endTime = performance.now();
-                // The original log for extraCooccurrenceCount was potentially misleading as it showed
-                // the count for the last processed sourte. Removed for clarity.
-                // If a total count is needed, it should be calculated across all sites.
-                console.log(`[Autocomplete-Plus] "${source}" Co-occurrence loading complete in ${(endTime - startTime).toFixed(2)}ms.`);
+                if (csvListData[source].base_cooccurrence) {
+                    console.log(`[Autocomplete-Plus] "${source}" Co-occurrence loading complete in ${(endTime - startTime).toFixed(2)}ms.`);
+                }
            })
        ]);

@@ -395,3 +474,88 @@ export async function initializeData(csvListData, source) {
        autoCompleteData[source].isInitializing = false;
    }
 }
+
+/**
+ * Load Embeddings data from the API endpoint
+ * @returns {Promise<void>}
+ */
+async function loadEmbeddings() {
+    try {
+        const response = await fetch('/autocomplete-plus/embeddings', { cache: "no-store" });
+        if (!response.ok) {
+            throw new Error(`HTTP error! status: ${response.status}`);
+        }
+        const embeddings = await response.json();
+        const source = ModelTagSource.Embeddings;
+
+        if (autoCompleteData.hasOwnProperty(source) === false) {
+            autoCompleteData[source] = new AutocompleteData();
+        }
+
+        embeddings.forEach(embedding => {
+            if (!autoCompleteData[source].tagMap.has(embedding)) {
+                const tagData = new TagData(`embedding:${embedding}`, 0, 0, [], source);
+                autoCompleteData[source].sortedTags.push(tagData);
+                autoCompleteData[source].tagMap.set(embedding, tagData);
+
+                updateMaxTagLength(embedding.length);
+            }
+        });
+
+        await buildFlexSearchIndex(ModelTagSource.Embeddings);
+
+        console.log(`[Autocomplete-Plus] Loaded ${embeddings.length} Embeddings`);
+    } catch (error) {
+        console.error(`[Autocomplete-Plus] Failed to fetch Embeddings data:`, error);
+    }
+}
+
+/**
+ * Load LoRA data from the API endpoint
+ * @returns {Promise<void>}
+ */
+async function loadLoras() {
+    try {
+        const response = await fetch('/autocomplete-plus/loras', { cache: "no-store" });
+        if (!response.ok) {
+            throw new Error(`HTTP error! status: ${response.status}`);
+        }
+        const loraNames = await response.json();
+        const source = ModelTagSource.Lora;
+
+        if (autoCompleteData.hasOwnProperty(source) === false) {
+            autoCompleteData[source] = new AutocompleteData();
+        }
+
+        loraNames.forEach(loraName => {
+            if (!autoCompleteData[source].tagMap.has(loraName)) {
+                const tagData = new TagData(`<lora:${loraName}>`, 0, 0, [], source);
+                autoCompleteData[source].sortedTags.push(tagData);
+                autoCompleteData[source].tagMap.set(loraName, tagData);
+
+                updateMaxTagLength(loraName.length);
+            }
+        });
+
+        await buildFlexSearchIndex(ModelTagSource.Lora);
+
+        console.log(`[Autocomplete-Plus] Loaded ${loraNames.length} LoRA models`);
+    } catch (error) {
+        console.error(`[Autocomplete-Plus] Failed to fetch LoRA data:`, error);
+    }
+}
+
+/**
+ * Load all data sources asynchronously.
+ */
+export async function loadDataAsync() {
+    return Promise.all([
+        fetchCsvList().then((csvList) => {
+            Object.values(TagSource).forEach((source) => {
+                initializeDataFromCSV(csvList, source);
+            });
+        }),
+        loadEmbeddings(),
+        loadLoras(),
+    ]);
+}
@@ -3,7 +3,7 @@ import { $el } from "/scripts/ui.js";
 import { ComfyWidgets } from "/scripts/widgets.js";
 import { settingValues } from "./settings.js";
 import { loadCSS } from "./utils.js";
-import { TagSource, fetchCsvList, initializeData } from "./data.js";
+import { TagSource, loadDataAsync } from "./data.js";
 import { AutocompleteEventHandler } from "./autocomplete.js";
 import { RelatedTagsEventHandler } from "./related-tags.js";

@@ -190,40 +190,41 @@ async function addExtraSettings() {
    }

    // Add extra setting for checking new CSV updates
-    app.ui.settings.addSetting({
-        id: id + ".check_new_csv",
-        defaultValue: null,
-        name: "Check CSV updates",
-        category: [name, "Misc", "Check new CSV"],
-        type: () => {
-            const lastCheckSpan = $el("span", {
-                textContent: lastCheckTimeText,
-                className: "text-sm text-gray-500",
-                style: {
-                    marginRight: "16px"
-                }
-            });
+    // Note: Temporarily comment out until changes to remote files can be detected
+    // app.ui.settings.addSetting({
+    //     id: id + ".check_new_csv",
+    //     defaultValue: null,
+    //     name: "Check CSV updates",
+    //     category: [name, "Misc", "Check new CSV"],
+    //     type: () => {
+    //         const lastCheckSpan = $el("span", {
+    //             textContent: lastCheckTimeText,
+    //             className: "text-sm text-gray-500",
+    //             style: {
+    //                 marginRight: "16px"
+    //             }
+    //         });

-            const checkButton = $el("button", {
-                textContent: "Check now",
-                className: "p-button p-component p-button-primary",
-                onclick: async () => {
-                    await performUpdateCheck(checkButton, lastCheckSpan);
-                }
-            });
+    //         const checkButton = $el("button", {
+    //             textContent: "Check now",
+    //             className: "p-button p-component p-button-primary",
+    //             onclick: async () => {
+    //                 await performUpdateCheck(checkButton, lastCheckSpan);
+    //             }
+    //         });

-            return $el("div", {
-                className: "flex-row items-center gap-2",
-            }, [
-                $el("div", {
-                    className: "p-component",
-                }, [
-                    lastCheckSpan,
-                    checkButton,
-                ]),
-            ]);
-        }
-    });
+    //         return $el("div", {
+    //             className: "flex-row items-center gap-2",
+    //         }, [
+    //             $el("div", {
+    //                 className: "p-component",
+    //             }, [
+    //                 lastCheckSpan,
+    //                 checkButton,
+    //             ]),
+    //         ]);
+    //     }
+    // });
 }

 /**
@@ -240,11 +241,7 @@ app.registerExtension({
        let rootPath = import.meta.url.replace("js/main.js", "");
        loadCSS(rootPath + "css/autocomplete-plus.css"); // Load CSS for autocomplete

-        fetchCsvList().then((csvList) => {
-            Object.values(TagSource).forEach((source) => {
-                initializeData(csvList, source);
-            });
-        });
+        await loadDataAsync();
    },

    // One the Settings Screen, displays reverse order in same category
@@ -286,6 +283,28 @@ app.registerExtension({
            }
        },
        // --- Autocomplete Settings ---
+        {
+            id: id + ".use_fast_search",
+            name: "Use Fast Search",
+            tooltip: "Tag search processing during text input operates faster, improving responsiveness",
+            type: "boolean",
+            defaultValue: false,
+            category: [name, "Autocompletion", "Use Fast Search"],
+            onChange: (newVal, oldVal) => {
+                settingValues.useFastSearch = newVal;
+            }
+        },
+        {
+            id: id + ".enable_models",
+            name: "Enable Loras and Embeddings",
+            tooltip: "Enable Lora and Embedding suggestions",
+            type: "boolean",
+            defaultValue: true,
+            category: [name, "Autocompletion", "Enable Loras and Embeddings"],
+            onChange: (newVal, oldVal) => {
+                settingValues.enableModels = newVal;
+            }
+        },
        {
            id: id + ".max_suggestions",
            name: "Max suggestions",
@@ -1,13 +1,14 @@
-import { TagSource, TagCategory, TagData, autoCompleteData } from './data.js';
+import { TagCategory, TagData, TagSource, autoCompleteData } from './data.js';
 import { settingValues } from './settings.js';
 import {
    extractTagsFromTextArea,
    findAllTagPositions,
+    getCurrentTagRange,
    getViewportMargin,
+    isLongText,
    isValidTag,
    normalizeTagToInsert,
    normalizeTagToSearch,
-    getCurrentTagRange,
 } from './utils.js';

 // --- RelatedTags Logic ---
@@ -67,7 +68,7 @@ function searchRelatedTags(tag) {

    const tagSource = TagSource.Danbooru; // TODO: Leave the tag source as Danbooru until e621_tags_cooccurrence.csv is ready

-    if (!tag || !autoCompleteData[tagSource].cooccurrenceMap.has(tag)) {
+    if (!tag || !autoCompleteData[tagSource].initialized || !autoCompleteData[tagSource].cooccurrenceMap.has(tag)) {
        return [];
    }

@@ -199,11 +200,20 @@ class RelatedTagsUI {
        this.header = document.createElement('div');
        this.header.id = 'related-tags-header';

+        this.headerTextContainer = document.createElement('div');
+        this.headerTextContainer.className = 'related-tags-header-text-container';
+        this.header.appendChild(this.headerTextContainer);
+
        // Create header text div for the left side
        this.headerText = document.createElement('div');
-        this.headerText.className = 'related-tags-header-text';
+        this.headerText.className = 'related-tags-header-tag-text';
        this.headerText.textContent = 'Related Tags';
-        this.header.appendChild(this.headerText);
+        this.headerTextContainer.appendChild(this.headerText);
+
+        // Create header alias div for the 2nd line
+        this.headerAlias = document.createElement('div');
+        this.headerAlias.className = 'related-tags-header-tag-alias';
+        this.headerTextContainer.appendChild(this.headerAlias);

        // Create header controls for the right side
        this.headerControls = document.createElement('div');
@@ -249,7 +259,7 @@ class RelatedTagsUI {
        this.headerControls.appendChild(this.pinBtn);

        this.header.appendChild(this.headerControls);
-
+        
        this.root.appendChild(this.header);

        // Create a tbody for the tags
@@ -300,7 +310,7 @@ class RelatedTagsUI {
        const currentTag = getTagFromCursorPosition(textareaElement);

        if (!this.isPinned) {
-            if (isValidTag(currentTag)) {
+            if (!isLongText(currentTag) && isValidTag(currentTag)) {
                this.currentTag = currentTag
            } else {
                this.hide();
@@ -391,7 +401,7 @@ class RelatedTagsUI {
     */
    #updateHeader() {
        // Find the tag data for the current tag
-        const tagData = Object.values(TagSource)
+        let tagData = Object.values(TagSource)
            .map((source) => {
                if (source in autoCompleteData && autoCompleteData[source].tagMap.has(this.currentTag)) {
                    return autoCompleteData[source].tagMap.get(this.currentTag);
@@ -399,21 +409,49 @@ class RelatedTagsUI {
            })
            .find((tagData) => tagData !== undefined);

+        if (!tagData) {
+            tagData = new TagData({
+                tag: this.currentTag,
+                source: TagSource.Danbooru,
+                category: 'unknown',
+                count: 0,
+                alias: [],
+            });
+        }
+
+        const tagText = this.currentTag;
+        const categoryText = TagCategory[tagData.source][tagData.category];
+
        // Update header text with current tag
        this.headerText.innerHTML = ''; // Clear previous content
        this.headerText.textContent = 'Tags related to: ';
+
        const tagName = document.createElement('span');
-        tagName.className = 'related-tags-header-tag-name';
-        tagName.textContent = this.currentTag;
-        if (tagData && ['left', 'right'].includes(settingValues.tagSourceIconPosition)) {
+        tagName.classList.add('related-tags-header-tag-name', tagData.source);
+        tagName.title = `Count: ${tagData.count}\nCategory: ${categoryText}`;
+        tagName.dataset.tagCategory = categoryText;
+        if (tagData.source && ['left', 'right'].includes(settingValues.tagSourceIconPosition)) {
            const tagSourceIconHtml = `<svg class="autocomplete-plus-tag-icon-svg"><use xlink:href="#autocomplete-plus-icon-${tagData.source}"></use></svg>`;
            tagName.innerHTML = settingValues.tagSourceIconPosition == 'left'
-                ? `${tagSourceIconHtml} ${tagData.tag}`
-                : `${tagData.tag} ${tagSourceIconHtml}`;
+                ? `${tagSourceIconHtml} ${tagText}`
+                : `${tagText} ${tagSourceIconHtml}`;
+        } else {
+            tagName.textContent += tagText;
        }

+
        this.headerText.appendChild(tagName);

+        // Clear previous alias
+        this.headerAlias.style.display = 'none';
+        this.headerAlias.innerHTML = '';
+
+        // Add alias if available
+        if (tagData.alias?.length > 0) {
+            this.headerAlias.textContent = tagData.alias.join(', ');
+            this.headerAlias.style.display = 'block';
+        }
+
        // Update pin button
        this.pinBtn.textContent = this.isPinned ? '🎯' : '📌';

@@ -432,7 +470,7 @@ class RelatedTagsUI {
        if (!autoCompleteData[TagSource.Danbooru].initialized) {
            // Show loading message
            const messageDiv = document.createElement('div');
-            messageDiv.className = 'related-tags-loading-message';
+            messageDiv.className = 'related-tags-message';
            messageDiv.textContent = `Initializing cooccurrence data... [${autoCompleteData[TagSource.Danbooru].baseLoadingProgress.cooccurrence}%]`;
            this.tagsContainer.appendChild(messageDiv);
            return;
@@ -441,6 +479,7 @@ class RelatedTagsUI {
        if (!this.relatedTags || this.relatedTags.length === 0) {
            // Show no related tags message
            const messageDiv = document.createElement('div');
+            messageDiv.className = 'related-tags-message';
            messageDiv.textContent = 'No related tags found';
            this.tagsContainer.appendChild(messageDiv);
            return;
@@ -530,20 +569,24 @@ class RelatedTagsUI {
        this.root.style.maxWidth = '';
        this.tagsContainer.style.maxHeight = '';
        const rootRect = this.root.getBoundingClientRect();
-        const headerRect = this.header.getBoundingClientRect();
-        // Hide it again after measurement
-        this.root.style.display = 'none';
-        this.root.style.visibility = 'visible';

        // Get the optimal placement area
        const placementArea = this.#getOptimalPlacementArea(rootRect.width, rootRect.height);

-        // Apply Styles
+        // Apply position and size
        this.root.style.left = `${placementArea.x}px`;
        this.root.style.top = `${placementArea.y}px`;
        this.root.style.maxWidth = `${placementArea.width}px`;

-        this.tagsContainer.style.maxHeight = `${placementArea.height - headerRect.height}px`;
+        const newHeaderRect = this.header.getBoundingClientRect();
+
+        if(this.relatedTags.length > 0){
+            this.tagsContainer.style.maxHeight = `${placementArea.height - newHeaderRect.height}px`;
+        }
+
+        // Hide it again after measurement
+        this.root.style.display = 'none';
+        this.root.style.visibility = 'visible';
    }

    /** Highlights the item (row) at the given index */
@@ -0,0 +1,133 @@
+import { Charset, Encoder, Document } from './thirdparty/flexsearch.bundle.module.min.js'
+import { kataToHira } from './utils.js';
+
+/**
+ * Creates an encoder optimized for processing English tag names.
+ * Handles tag formatting like underscores and parentheses commonly used in Danbooru tags.
+ * @returns {Encoder} FlexSearch encoder for English tags
+ */
+function createTagEncoder() {
+    return new Encoder({
+        normalize: true,
+        dedupe: false,
+        numeric: false,
+        cache: true,
+        // filter: new Set(['and', 'to', 'be', 'on']),
+        replacer: [/(?<=[a-zA-Z\)])_$/, ''],  // Remove trailing underscores after letters/parentheses
+        split: /(?<=[a-zA-Z\)])_(?=[a-zA-Z\(])|\((?=[a-zA-Z])|(?<=[a-zA-Z\)])\)|[ \n]/  // Split on underscores between words, parentheses, spaces, and newlines
+    });
+}
+
+/**
+ * Creates an encoder optimized for processing CJK (Chinese, Japanese, Korean) characters.
+ * Uses exact character matching and converts katakana to hiragana for better Japanese search.
+ * @returns {Encoder} FlexSearch encoder for CJK text
+ */
+function createCJKEncoder() {
+    return new Encoder(Charset.Exact, {
+        dedupe: true,
+        numeric: true,
+        cache: true,
+        filter: new Set(['(', ')']),  // Filter out parentheses characters
+        finalize: (term) => {         // Convert katakana to hiragana for better Japanese matching
+            return term.map(str => kataToHira(str));
+        }
+    });
+}
+
+/**
+ * Creates an encoder optimized for processing Embedding or Lora notation.
+ * @returns {Encoder} FlexSearch encoder
+ */
+function createModelEncoder() {
+    return new Encoder({
+        normalize: true,
+        dedupe: false,
+        numeric: true,
+        cache: true,
+        prepare: function (str) {
+            return str.replace(/^<|>$/g, '').split(/(lora:|embedding:|[^\u0000-\u007f]+)/g).filter(Boolean).join(" ").trim();
+        },
+        split: /(?<=lora:.*|embedding:.*)[_./\(\)\-\s\\]+/
+    });
+}
+
+/**
+ * Creates a FlexSearch Document instance optimized for tag searching.
+ * Configures separate encoders for English tags and CJK aliases with appropriate tokenization.
+ * @returns {Document} Configured FlexSearch document for tag indexing
+ */
+export function createFlexSearchDocument() {
+    const tagEncoder = createTagEncoder();
+    const cjkEncoder = createCJKEncoder();
+
+    // Custom encoding function for alias field that handles mixed language content
+    const encodeAlias = function (word) {
+        return word.split(",")
+            .flatMap(str => {
+                if (/[^\u0000-\u007f]/.test(str)) {
+                    // Contains non-ASCII characters (CJK text)
+                    return cjkEncoder.encode(str);
+                } else {
+                    // ASCII characters only (English text)
+                    return tagEncoder.encode(str);
+                }
+            })
+            .filter(Boolean);
+    }
+
+    // Configure the FlexSearch document with optimized indexing settings
+    const document = new Document({
+        document: {
+            id: "id",
+            index: [
+                {
+                    field: "tag",
+                    tokenize: "bidirectional",  // Allow partial matching from both ends
+                    encoder: tagEncoder,        // Use tag-optimized encoder
+                },
+                {
+                    field: "alias",             // Index the alias field for multi-language support
+                    tokenize: "full",           // Full tokenization for complete alias matching
+                    encode: encodeAlias,        // Use custom multi-language encoding function
+                }
+            ]
+        }
+    });
+
+    return document;
+}
+
+/**
+ * Creates a FlexSearch Document instance optimized for lora or embedding searching.
+ * @returns {Document} Configured FlexSearch document
+ */
+export function createFlexSearchDocumentForModel() {
+    const modelEncoder = createModelEncoder();
+
+
+    // Configure the FlexSearch document with optimized indexing settings
+    // Note: alias field is not indexed for lora or embedding search
+    const document = new Document({
+        tokenize: "full",  // Allow partial matching from both ends
+        encoder: modelEncoder,
+        document: {
+            id: "id",
+            index: [
+                {
+                    field: "tag",
+
+                },
+                {
+                    field: "alias",
+                }
+            ]
+        }
+    });
+
+    return document;
+}
+
+// Export functions for testing when in test environment
+const isTestEnvironment = typeof process !== 'undefined' && process.env.NODE_ENV === 'test';
+export const __test__ = isTestEnvironment ? { createTagEncoder, createCJKEncoder, createModelEncoder } : undefined;
@@ -7,6 +7,8 @@ export const settingValues = {
    // Autocomplete feature settings
    enabled: true,
    maxSuggestions: 10,
+    enableModels: true, // Enable Lora and Embedding suggestions
+    useFastSearch: false,

    // Related tags feature settings
    enableRelatedTags: true,
@@ -17,8 +19,19 @@ export const settingValues = {

    // Internal logic settings
    _useFallbackAttachmentForEventListener: false, // Fallback to attach event listener when somthing goes wrong
+    _maxTagLength: 100, // Maximum tag length to prevent performance issues with long text input

    // Debugging settings (use internally)
    _hideWhenOutofFocus: true, // Hide UI when the input is out of focus
    _logprocessingTime: false, // Log processing time for debugging
-}
+}
+
+/**
+ * Update the maximum tag length setting value.
+ * This function ensures that the maximum tag length is always at least as long as the new length provided.
+ * @param {number} newLength 
+ */
+export function updateMaxTagLength(newLength) {
+    if(isNaN(newLength)) return;
+    settingValues._maxTagLength = Math.max(settingValues._maxTagLength, newLength);
+}
@@ -0,0 +1,106 @@
+/**!
+ * FlexSearch.js v0.8.200 (Bundle/Module)
+ * Author and Copyright: Thomas Wilkerling
+ * Licence: Apache-2.0
+ * Hosted by Nextapps GmbH
+ * https://github.com/nextapps-de/flexsearch
+ */
+var w;function H(a,c,b){const e=typeof b,d=typeof a;if("undefined"!==e){if("undefined"!==d){if(b){if("function"===d&&e===d)return function(h){return a(b(h))};c=a.constructor;if(c===b.constructor){if(c===Array)return b.concat(a);if(c===Map){var f=new Map(b);for(var g of a)f.set(g[0],g[1]);return f}if(c===Set){g=new Set(b);for(f of a.values())g.add(f);return g}}}return a}return b}return"undefined"===d?c:a}function aa(a,c){return"undefined"===typeof a?c:a}function I(){return Object.create(null)}
+function M(a){return"string"===typeof a}function ba(a){return"object"===typeof a}function ca(a){const c=[];for(const b of a.keys())c.push(b);return c}function ea(a,c){if(M(c))a=a[c];else for(let b=0;a&&b<c.length;b++)a=a[c[b]];return a};const fa=/[^\p{L}\p{N}]+/u,ha=/(\d{3})/g,ia=/(\D)(\d{3})/g,ja=/(\d{3})(\D)/g,ka=/[\u0300-\u036f]/g;function la(a={}){if(!this||this.constructor!==la)return new la(...arguments);if(arguments.length)for(a=0;a<arguments.length;a++)this.assign(arguments[a]);else this.assign(a)}w=la.prototype;
+w.assign=function(a){this.normalize=H(a.normalize,!0,this.normalize);let c=a.include,b=c||a.exclude||a.split,e;if(b||""===b){if("object"===typeof b&&b.constructor!==RegExp){let d="";e=!c;c||(d+="\\p{Z}");b.letter&&(d+="\\p{L}");b.number&&(d+="\\p{N}",e=!!c);b.symbol&&(d+="\\p{S}");b.punctuation&&(d+="\\p{P}");b.control&&(d+="\\p{C}");if(b=b.char)d+="object"===typeof b?b.join(""):b;try{this.split=new RegExp("["+(c?"^":"")+d+"]+","u")}catch(f){this.split=/\s+/}}else this.split=b,e=!1===b||2>"a1a".split(b).length;
+this.numeric=H(a.numeric,e)}else{try{this.split=H(this.split,fa)}catch(d){this.split=/\s+/}this.numeric=H(a.numeric,H(this.numeric,!0))}this.prepare=H(a.prepare,null,this.prepare);this.finalize=H(a.finalize,null,this.finalize);b=a.filter;this.filter="function"===typeof b?b:H(b&&new Set(b),null,this.filter);this.dedupe=H(a.dedupe,!0,this.dedupe);this.matcher=H((b=a.matcher)&&new Map(b),null,this.matcher);this.mapper=H((b=a.mapper)&&new Map(b),null,this.mapper);this.stemmer=H((b=a.stemmer)&&new Map(b),
+null,this.stemmer);this.replacer=H(a.replacer,null,this.replacer);this.minlength=H(a.minlength,1,this.minlength);this.maxlength=H(a.maxlength,1024,this.maxlength);this.rtl=H(a.rtl,!1,this.rtl);if(this.cache=b=H(a.cache,!0,this.cache))this.I=null,this.R="number"===typeof b?b:2E5,this.B=new Map,this.H=new Map,this.M=this.L=128;this.h="";this.N=null;this.A="";this.O=null;if(this.matcher)for(const d of this.matcher.keys())this.h+=(this.h?"|":"")+d;if(this.stemmer)for(const d of this.stemmer.keys())this.A+=
+(this.A?"|":"")+d;return this};w.addStemmer=function(a,c){this.stemmer||(this.stemmer=new Map);this.stemmer.set(a,c);this.A+=(this.A?"|":"")+a;this.O=null;this.cache&&Q(this);return this};w.addFilter=function(a){"function"===typeof a?this.filter=a:(this.filter||(this.filter=new Set),this.filter.add(a));this.cache&&Q(this);return this};
+w.addMapper=function(a,c){if("object"===typeof a)return this.addReplacer(a,c);if(1<a.length)return this.addMatcher(a,c);this.mapper||(this.mapper=new Map);this.mapper.set(a,c);this.cache&&Q(this);return this};w.addMatcher=function(a,c){if("object"===typeof a)return this.addReplacer(a,c);if(2>a.length&&(this.dedupe||this.mapper))return this.addMapper(a,c);this.matcher||(this.matcher=new Map);this.matcher.set(a,c);this.h+=(this.h?"|":"")+a;this.N=null;this.cache&&Q(this);return this};
+w.addReplacer=function(a,c){if("string"===typeof a)return this.addMatcher(a,c);this.replacer||(this.replacer=[]);this.replacer.push(a,c);this.cache&&Q(this);return this};
+w.encode=function(a,c){if(this.cache&&a.length<=this.L)if(this.I){if(this.B.has(a))return this.B.get(a)}else this.I=setTimeout(Q,50,this);this.normalize&&("function"===typeof this.normalize?a=this.normalize(a):a=ka?a.normalize("NFKD").replace(ka,"").toLowerCase():a.toLowerCase());this.prepare&&(a=this.prepare(a));this.numeric&&3<a.length&&(a=a.replace(ia,"$1 $2").replace(ja,"$1 $2").replace(ha,"$1 "));const b=!(this.dedupe||this.mapper||this.filter||this.matcher||this.stemmer||this.replacer);let e=
+[],d=I(),f,g,h=this.split||""===this.split?a.split(this.split):[a];for(let m=0,l,n;m<h.length;m++)if((l=n=h[m])&&!(l.length<this.minlength||l.length>this.maxlength)){if(c){if(d[l])continue;d[l]=1}else{if(f===l)continue;f=l}if(b)e.push(l);else if(!this.filter||("function"===typeof this.filter?this.filter(l):!this.filter.has(l))){if(this.cache&&l.length<=this.M)if(this.I){var k=this.H.get(l);if(k||""===k){k&&e.push(k);continue}}else this.I=setTimeout(Q,50,this);if(this.stemmer){this.O||(this.O=new RegExp("(?!^)("+
+this.A+")$"));let u;for(;u!==l&&2<l.length;)u=l,l=l.replace(this.O,p=>this.stemmer.get(p))}if(l&&(this.mapper||this.dedupe&&1<l.length)){k="";for(let u=0,p="",t,r;u<l.length;u++)t=l.charAt(u),t===p&&this.dedupe||((r=this.mapper&&this.mapper.get(t))||""===r?r===p&&this.dedupe||!(p=r)||(k+=r):k+=p=t);l=k}this.matcher&&1<l.length&&(this.N||(this.N=new RegExp("("+this.h+")","g")),l=l.replace(this.N,u=>this.matcher.get(u)));if(l&&this.replacer)for(k=0;l&&k<this.replacer.length;k+=2)l=l.replace(this.replacer[k],
+this.replacer[k+1]);this.cache&&n.length<=this.M&&(this.H.set(n,l),this.H.size>this.R&&(this.H.clear(),this.M=this.M/1.1|0));if(l){if(l!==n)if(c){if(d[l])continue;d[l]=1}else{if(g===l)continue;g=l}e.push(l)}}}this.finalize&&(e=this.finalize(e)||e);this.cache&&a.length<=this.L&&(this.B.set(a,e),this.B.size>this.R&&(this.B.clear(),this.L=this.L/1.1|0));return e};function Q(a){a.I=null;a.B.clear();a.H.clear()};function ma(a,c,b){b||(c||"object"!==typeof a?"object"===typeof c&&(b=c,c=0):b=a);b&&(a=b.query||a,c=b.limit||c);let e=""+(c||0);b&&(e+=(b.offset||0)+!!b.context+!!b.suggest+(!1!==b.resolve)+(b.resolution||this.resolution)+(b.boost||0));a=(""+a).toLowerCase();this.cache||(this.cache=new R);let d=this.cache.get(a+e);if(!d){const f=b&&b.cache;f&&(b.cache=!1);d=this.search(a,c,b);f&&(b.cache=f);this.cache.set(a+e,d)}return d}function R(a){this.limit=a&&!0!==a?a:1E3;this.cache=new Map;this.h=""}
+R.prototype.set=function(a,c){this.cache.set(this.h=a,c);this.cache.size>this.limit&&this.cache.delete(this.cache.keys().next().value)};R.prototype.get=function(a){const c=this.cache.get(a);c&&this.h!==a&&(this.cache.delete(a),this.cache.set(this.h=a,c));return c};R.prototype.remove=function(a){for(const c of this.cache){const b=c[0];c[1].includes(a)&&this.cache.delete(b)}};R.prototype.clear=function(){this.cache.clear();this.h=""};const na={normalize:!1,numeric:!1,dedupe:!1};const qa={};const ra=new Map([["b","p"],["v","f"],["w","f"],["z","s"],["x","s"],["d","t"],["n","m"],["c","k"],["g","k"],["j","k"],["q","k"],["i","e"],["y","e"],["u","o"]]);const sa=new Map([["ae","a"],["oe","o"],["sh","s"],["kh","k"],["th","t"],["ph","f"],["pf","f"]]),ta=[/([^aeo])h(.)/g,"$1$2",/([aeo])h([^aeo]|$)/g,"$1$2",/(.)\1+/g,"$1"];const ua={a:"",e:"",i:"",o:"",u:"",y:"",b:1,f:1,p:1,v:1,c:2,g:2,j:2,k:2,q:2,s:2,x:2,z:2,"\u00df":2,d:3,t:3,l:4,m:5,n:5,r:6};var va={Exact:na,Default:qa,Normalize:qa,LatinBalance:{mapper:ra},LatinAdvanced:{mapper:ra,matcher:sa,replacer:ta},LatinExtra:{mapper:ra,replacer:ta.concat([/(?!^)[aeo]/g,""]),matcher:sa},LatinSoundex:{dedupe:!1,include:{letter:!0},finalize:function(a){for(let b=0;b<a.length;b++){var c=a[b];let e=c.charAt(0),d=ua[e];for(let f=1,g;f<c.length&&(g=c.charAt(f),"h"===g||"w"===g||!(g=ua[g])||g===d||(e+=g,d=g,4!==e.length));f++);a[b]=e}}},CJK:{split:""},LatinExact:na,LatinDefault:qa,LatinSimple:qa};function wa(a,c,b,e){let d=[];for(let f=0,g;f<a.index.length;f++)if(g=a.index[f],c>=g.length)c-=g.length;else{c=g[e?"splice":"slice"](c,b);const h=c.length;if(h&&(d=d.length?d.concat(c):c,b-=h,e&&(a.length-=h),!b))break;c=0}return d}
+function za(a){if(!this||this.constructor!==za)return new za(a);this.index=a?[a]:[];this.length=a?a.length:0;const c=this;return new Proxy([],{get(b,e){if("length"===e)return c.length;if("push"===e)return function(d){c.index[c.index.length-1].push(d);c.length++};if("pop"===e)return function(){if(c.length)return c.length--,c.index[c.index.length-1].pop()};if("indexOf"===e)return function(d){let f=0;for(let g=0,h,k;g<c.index.length;g++){h=c.index[g];k=h.indexOf(d);if(0<=k)return f+k;f+=h.length}return-1};
+if("includes"===e)return function(d){for(let f=0;f<c.index.length;f++)if(c.index[f].includes(d))return!0;return!1};if("slice"===e)return function(d,f){return wa(c,d||0,f||c.length,!1)};if("splice"===e)return function(d,f){return wa(c,d||0,f||c.length,!0)};if("constructor"===e)return Array;if("symbol"!==typeof e)return(b=c.index[e/2**31|0])&&b[e]},set(b,e,d){b=e/2**31|0;(c.index[b]||(c.index[b]=[]))[e]=d;c.length++;return!0}})}za.prototype.clear=function(){this.index.length=0};
+za.prototype.destroy=function(){this.proxy=this.index=null};za.prototype.push=function(){};function S(a=8){if(!this||this.constructor!==S)return new S(a);this.index=I();this.h=[];this.size=0;32<a?(this.B=Aa,this.A=BigInt(a)):(this.B=Ba,this.A=a)}S.prototype.get=function(a){const c=this.index[this.B(a)];return c&&c.get(a)};S.prototype.set=function(a,c){var b=this.B(a);let e=this.index[b];e?(b=e.size,e.set(a,c),(b-=e.size)&&this.size++):(this.index[b]=e=new Map([[a,c]]),this.h.push(e),this.size++)};
+function T(a=8){if(!this||this.constructor!==T)return new T(a);this.index=I();this.h=[];this.size=0;32<a?(this.B=Aa,this.A=BigInt(a)):(this.B=Ba,this.A=a)}T.prototype.add=function(a){var c=this.B(a);let b=this.index[c];b?(c=b.size,b.add(a),(c-=b.size)&&this.size++):(this.index[c]=b=new Set([a]),this.h.push(b),this.size++)};w=S.prototype;w.has=T.prototype.has=function(a){const c=this.index[this.B(a)];return c&&c.has(a)};
+w.delete=T.prototype.delete=function(a){const c=this.index[this.B(a)];c&&c.delete(a)&&this.size--};w.clear=T.prototype.clear=function(){this.index=I();this.h=[];this.size=0};w.values=T.prototype.values=function*(){for(let a=0;a<this.h.length;a++)for(let c of this.h[a].values())yield c};w.keys=T.prototype.keys=function*(){for(let a=0;a<this.h.length;a++)for(let c of this.h[a].keys())yield c};w.entries=T.prototype.entries=function*(){for(let a=0;a<this.h.length;a++)for(let c of this.h[a].entries())yield c};
+function Ba(a){let c=2**this.A-1;if("number"==typeof a)return a&c;let b=0,e=this.A+1;for(let d=0;d<a.length;d++)b=(b*e^a.charCodeAt(d))&c;return 32===this.A?b+2**31:b}function Aa(a){let c=BigInt(2)**this.A-BigInt(1);var b=typeof a;if("bigint"===b)return a&c;if("number"===b)return BigInt(a)&c;b=BigInt(0);let e=this.A+BigInt(1);for(let d=0;d<a.length;d++)b=(b*e^BigInt(a.charCodeAt(d)))&c;return b};let Ca,Da;
+async function Ea(a){a=a.data;var c=a.task;const b=a.id;let e=a.args;switch(c){case "init":Da=a.options||{};(c=a.factory)?(Function("return "+c)()(self),Ca=new self.FlexSearch.Index(Da),delete self.FlexSearch):Ca=new V(Da);postMessage({id:b});break;default:let d;"export"===c&&(e[1]?(e[0]=Da.export,e[2]=0,e[3]=1):e=null);"import"===c?e[0]&&(a=await Da.import.call(Ca,e[0]),Ca.import(e[0],a)):((d=e&&Ca[c].apply(Ca,e))&&d.then&&(d=await d),d&&d.await&&(d=await d.await),"search"===c&&d.result&&(d=d.result));
+postMessage("search"===c?{id:b,msg:d}:{id:b})}};function Fa(a){Ga.call(a,"add");Ga.call(a,"append");Ga.call(a,"search");Ga.call(a,"update");Ga.call(a,"remove");Ga.call(a,"searchCache")}let Ha,Ia,Ja;function Ka(){Ha=Ja=0}
+function Ga(a){this[a+"Async"]=function(){const c=arguments;var b=c[c.length-1];let e;"function"===typeof b&&(e=b,delete c[c.length-1]);Ha?Ja||(Ja=Date.now()-Ia>=this.priority*this.priority*3):(Ha=setTimeout(Ka,0),Ia=Date.now());if(Ja){const f=this;return new Promise(g=>{setTimeout(function(){g(f[a+"Async"].apply(f,c))},0)})}const d=this[a].apply(this,c);b=d.then?d:new Promise(f=>f(d));e&&b.then(e);return b}};let W=0;
+function La(a={},c){function b(h){function k(m){m=m.data||m;const l=m.id,n=l&&f.h[l];n&&(n(m.msg),delete f.h[l])}this.worker=h;this.h=I();if(this.worker){d?this.worker.on("message",k):this.worker.onmessage=k;if(a.config)return new Promise(function(m){1E9<W&&(W=0);f.h[++W]=function(){m(f)};f.worker.postMessage({id:W,task:"init",factory:e,options:a})});this.priority=a.priority||4;this.encoder=c||null;this.worker.postMessage({task:"init",factory:e,options:a});return this}}if(!this||this.constructor!==La)return new La(a);
+let e="undefined"!==typeof self?self._factory:"undefined"!==typeof window?window._factory:null;e&&(e=e.toString());const d="undefined"===typeof window,f=this,g=Ma(e,d,a.worker);return g.then?g.then(function(h){return b.call(f,h)}):b.call(this,g)}X("add");X("append");X("search");X("update");X("remove");X("clear");X("export");X("import");La.prototype.searchCache=ma;Fa(La.prototype);
+function X(a){La.prototype[a]=function(){const c=this,b=[].slice.call(arguments);var e=b[b.length-1];let d;"function"===typeof e&&(d=e,b.pop());e=new Promise(function(f){"export"===a&&"function"===typeof b[0]&&(b[0]=null);1E9<W&&(W=0);c.h[++W]=f;c.worker.postMessage({task:a,id:W,args:b})});return d?(e.then(d),this):e}}
+function Ma(a,c,b){return c?"undefined"!==typeof module?new(require("worker_threads")["Worker"])(__dirname+"/worker/node.js"):import("worker_threads").then(function(worker){return new worker["Worker"](import.meta.dirname+"/node/node.mjs")}):a?new window.Worker(URL.createObjectURL(new Blob(["onmessage="+Ea.toString()],{type:"text/javascript"}))):new window.Worker("string"===typeof b?b:import.meta.url.replace("/worker.js","/worker/worker.js").replace("flexsearch.bundle.module.min.js",
+"module/worker/worker.js"),{type:"module"})};Na.prototype.add=function(a,c,b){ba(a)&&(c=a,a=ea(c,this.key));if(c&&(a||0===a)){if(!b&&this.reg.has(a))return this.update(a,c);for(let h=0,k;h<this.field.length;h++){k=this.G[h];var e=this.index.get(this.field[h]);if("function"===typeof k){var d=k(c);d&&e.add(a,d,!1,!0)}else if(d=k.J,!d||d(c))k.constructor===String?k=[""+k]:M(k)&&(k=[k]),Qa(c,k,this.K,0,e,a,k[0],b)}if(this.tag)for(e=0;e<this.F.length;e++){var f=this.F[e];d=this.tag.get(this.P[e]);let h=I();if("function"===typeof f){if(f=f(c),!f)continue}else{var g=
+f.J;if(g&&!g(c))continue;f.constructor===String&&(f=""+f);f=ea(c,f)}if(d&&f){M(f)&&(f=[f]);for(let k=0,m,l;k<f.length;k++)if(m=f[k],!h[m]&&(h[m]=1,(g=d.get(m))?l=g:d.set(m,l=[]),!b||!l.includes(a))){if(l.length===2**31-1){g=new za(l);if(this.fastupdate)for(let n of this.reg.values())n.includes(l)&&(n[n.indexOf(l)]=g);d.set(m,l=g)}l.push(a);this.fastupdate&&((g=this.reg.get(a))?g.push(l):this.reg.set(a,[l]))}}}if(this.store&&(!b||!this.store.has(a))){let h;if(this.D){h=I();for(let k=0,m;k<this.D.length;k++){m=
+this.D[k];if((b=m.J)&&!b(c))continue;let l;if("function"===typeof m){l=m(c);if(!l)continue;m=[m.U]}else if(M(m)||m.constructor===String){h[m]=c[m];continue}Ra(c,h,m,0,m[0],l)}}this.store.set(a,h||c)}this.worker&&(this.fastupdate||this.reg.add(a))}return this};function Ra(a,c,b,e,d,f){a=a[d];if(e===b.length-1)c[d]=f||a;else if(a)if(a.constructor===Array)for(c=c[d]=Array(a.length),d=0;d<a.length;d++)Ra(a,c,b,e,d);else c=c[d]||(c[d]=I()),d=b[++e],Ra(a,c,b,e,d)}
+function Qa(a,c,b,e,d,f,g,h){if(a=a[g])if(e===c.length-1){if(a.constructor===Array){if(b[e]){for(c=0;c<a.length;c++)d.add(f,a[c],!0,!0);return}a=a.join(" ")}d.add(f,a,h,!0)}else if(a.constructor===Array)for(g=0;g<a.length;g++)Qa(a,c,b,e,d,f,g,h);else g=c[++e],Qa(a,c,b,e,d,f,g,h);else d.db&&d.remove(f)};function Sa(a,c,b,e){if(!a.length)return a;if(1===a.length)return a=a[0],a=b||a.length>c?a.slice(b,b+c):a,e?Ta.call(this,a):a;let d=[];for(let f=0,g,h;f<a.length;f++)if((g=a[f])&&(h=g.length)){if(b){if(b>=h){b-=h;continue}g=g.slice(b,b+c);h=g.length;b=0}h>c&&(g=g.slice(0,c),h=c);if(!d.length&&h>=c)return e?Ta.call(this,g):g;d.push(g);c-=h;if(!c)break}d=1<d.length?[].concat.apply([],d):d[0];return e?Ta.call(this,d):d};function Ua(a,c,b,e){var d=e[0];if(d[0]&&d[0].query)return a[c].apply(a,d);if(!("and"!==c&&"not"!==c||a.result.length||a.await||d.suggest))return 1<e.length&&(d=e[e.length-1]),(e=d.resolve)?a.await||a.result:a;let f=[],g=0,h=0,k,m,l;var n;let u;for(c=0;c<e.length;c++)if(d=e[c]){var p=void 0;if(d.constructor===Y)p=d.await||d.result;else if(d.then||d.constructor===Array)p=d;else{g=d.limit||0;h=d.offset||0;l=d.suggest;m=d.resolve;k=(n=d.highlight&&m)||d.enrich&&m;p=d.queue;let t=d.async||p;(n=d.index)?
+a.index||(a.index=n):n=a.index;if(d.query||d.tag){const r=d.field||d.pluck;r&&(n=a.index.index.get(r));if(p&&(u||a.await)){u=1;let q;const y=a.C.length,v=new Promise(function(A){q=A});(function(A,B){v.h=function(){B.index=null;B.resolve=!1;let E=t?A.searchAsync(B):A.search(B);if(E.then)return E.then(function(C){a.C[y]=C=C.result||C;q(C);return C});E=E.result||E;q(E);return E}})(n,Object.assign({},d));a.C.push(v);f[c]=v;continue}else d.resolve=!1,d.index=null,p=t?n.searchAsync(d):n.search(d),d.resolve=
+m,d.index=n}else if(d.and)p=Va(d,"and",n);else if(d.or)p=Va(d,"or",n);else if(d.not)p=Va(d,"not",n);else if(d.xor)p=Va(d,"xor",n);else continue}p.await?(u=1,p=p.await):p.then?(u=1,p=p.then(function(t){return t.result||t})):p=p.result||p;f[c]=p}u&&!a.await&&(a.await=new Promise(function(t){a.return=t}));if(u){const t=Promise.all(f).then(function(r){for(let q=0;q<a.C.length;q++)if(a.C[q]===t){a.C[q]=function(){return b.call(a,r,g,h,k,m,l)};break}Wa(a)});a.C.push(t)}else if(a.await)a.C.push(function(){return b.call(a,
+f,g,h,k,m,l)});else return b.call(a,f,g,h,k,m,l);return m?a.await||a.result:a}function Va(a,c,b){a=a[c];const e=a[0]||a;e.index||(e.index=b);b=new Y(e);1<a.length&&(b=b[c].apply(b,a.slice(1)));return b};Y.prototype.or=function(){return Ua(this,"or",Xa,arguments)};function Xa(a,c,b,e,d){a.length&&(this.result.length&&a.push(this.result),2>a.length?this.result=a[0]:(this.result=Ya(a,c,b,!1,this.h),b=0));d&&(this.await=null);return d?this.resolve(c,b,e):this};Y.prototype.and=function(){return Ua(this,"and",Za,arguments)};function Za(a,c,b,e,d,f){if(!f&&!this.result.length)return d?this.result:this;let g;if(a.length)if(this.result.length&&a.unshift(this.result),2>a.length)this.result=a[0];else{let h=0;for(let k=0,m,l;k<a.length;k++)if((m=a[k])&&(l=m.length))h<l&&(h=l);else if(!f){h=0;break}h?(this.result=$a(a,h,c,b,f,this.h,d),g=!0):this.result=[]}else f||(this.result=a);d&&(this.await=null);return d?this.resolve(c,b,e,g):this};Y.prototype.xor=function(){return Ua(this,"xor",ab,arguments)};
+function ab(a,c,b,e,d,f){if(a.length)if(this.result.length&&a.unshift(this.result),2>a.length)this.result=a[0];else{a:{f=b;var g=this.h;const h=[],k=I();let m=0;for(let l=0,n;l<a.length;l++)if(n=a[l]){m<n.length&&(m=n.length);for(let u=0,p;u<n.length;u++)if(p=n[u])for(let t=0,r;t<p.length;t++)r=p[t],k[r]=k[r]?2:1}for(let l=0,n,u=0;l<m;l++)for(let p=0,t;p<a.length;p++)if(t=a[p])if(n=t[l])for(let r=0,q;r<n.length;r++)if(q=n[r],1===k[q])if(f)f--;else if(d){if(h.push(q),h.length===c){a=h;break a}}else{const y=
+l+(p?g:0);h[y]||(h[y]=[]);h[y].push(q);if(++u===c){a=h;break a}}a=h}this.result=a;g=!0}else f||(this.result=a);d&&(this.await=null);return d?this.resolve(c,b,e,g):this};Y.prototype.not=function(){return Ua(this,"not",bb,arguments)};
+function bb(a,c,b,e,d,f){if(!f&&!this.result.length)return d?this.result:this;if(a.length&&this.result.length){a:{f=b;var g=[];a=new Set(a.flat().flat());for(let h=0,k,m=0;h<this.result.length;h++)if(k=this.result[h])for(let l=0,n;l<k.length;l++)if(n=k[l],!a.has(n))if(f)f--;else if(d){if(g.push(n),g.length===c){a=g;break a}}else if(g[h]||(g[h]=[]),g[h].push(n),++m===c){a=g;break a}a=g}this.result=a;g=!0}d&&(this.await=null);return d?this.resolve(c,b,e,g):this};function Y(a,c){if(!this||this.constructor!==Y)return new Y(a,c);let b=0,e,d,f;if(a&&a.index){const g=a;c=g.index;b=g.boost||0;if(g.query){const h=g.resolve;a=g.async||g.queue;g.resolve=!1;g.index=null;a=a?c.searchAsync(g):c.search(g);g.resolve=h;g.index=c;a=a.result||a}else a=[]}if(a&&a.then){const g=this;a=a.then(function(h){g.C[0]=g.result=h.result||h;Wa(g)});e=[a];a=[];d=new Promise(function(h){f=h})}this.index=c||null;this.result=a||[];this.h=b;this.C=e||[];this.await=d||null;this.return=f||
+null}Y.prototype.limit=function(a){if(this.result.length){const c=[];for(let b=0,e;b<this.result.length;b++)if(e=this.result[b])if(e.length<=a){if(c[b]=e,a-=e.length,!a)break}else{c[b]=e.slice(0,a);break}this.result=c}return this};Y.prototype.offset=function(a){if(this.result.length){const c=[];for(let b=0,e;b<this.result.length;b++)if(e=this.result[b])e.length<=a?a-=e.length:(c[b]=e.slice(a),a=0);this.result=c}return this};Y.prototype.boost=function(a){this.h+=a;return this};
+function Wa(a,c){let b=a.result;for(let d=0,f;d<a.C.length;d++)if(f=a.C[d])if("function"===typeof f)b=f(),a.C[d]=b=b.result||b,d--;else if(f.h)b=f.h(),a.C[d]=b=b.result||b,d--;else if(f.then)return a.await;const e=a.return;a.result=b;a.C=[];a.await=null;a.return=null;c||e(b);return b}
+Y.prototype.resolve=function(a,c,b,e){let d=this.await?Wa(this,!0):this.result;if(d.then){const f=this;return d.then(function(){return f.resolve(a,c,b)})}d.length&&("object"===typeof a&&(b=a.enrich,c=a.offset,a=a.limit),d=e?b?Ta.call(this.index,d):d:Sa.call(this.index,d,a||100,c,b));e=this.return;this.return=this.await=this.C=this.result=this.index=null;e&&e(d);return d};function $a(a,c,b,e,d,f,g){const h=a.length;let k=[],m,l;m=I();for(let n=0,u,p,t,r;n<c;n++)for(let q=0;q<h;q++)if(t=a[q],n<t.length&&(u=t[n]))for(let y=0;y<u.length;y++){p=u[y];(l=m[p])?m[p]++:(l=0,m[p]=1);r=k[l]||(k[l]=[]);if(!g){let v=n+(q||!d?0:f||0);r=r[v]||(r[v]=[])}r.push(p);if(g&&b&&l===h-1&&r.length-e===b)return e?r.slice(e):r}if(a=k.length)if(d)k=1<k.length?Ya(k,b,e,g,f):(k=k[0])&&b&&k.length>b||e?k.slice(e,b+e):k;else{if(a<h)return[];k=k[a-1];if(b||e)if(g){if(k.length>b||e)k=k.slice(e,b+
+e)}else{d=[];for(let n=0,u;n<k.length;n++)if(u=k[n])if(e&&u.length>e)e-=u.length;else{if(b&&u.length>b||e)u=u.slice(e,b+e),b-=u.length,e&&(e-=u.length);d.push(u);if(!b)break}k=d}}return k}
+function Ya(a,c,b,e,d){const f=[],g=I();let h;var k=a.length;let m;if(e)for(d=k-1;0<=d;d--){if(m=(e=a[d])&&e.length)for(k=0;k<m;k++)if(h=e[k],!g[h])if(g[h]=1,b)b--;else if(f.push(h),f.length===c)return f}else for(let l=k-1,n,u=0;0<=l;l--){n=a[l];for(let p=0;p<n.length;p++)if(m=(e=n[p])&&e.length)for(let t=0;t<m;t++)if(h=e[t],!g[h])if(g[h]=1,b)b--;else{let r=(p+(l<k-1?d||0:0))/(l+1)|0;(f[r]||(f[r]=[])).push(h);if(++u===c)return f}}return f}
+function cb(a,c,b){const e=I(),d=[];for(let f=0,g;f<c.length;f++){g=c[f];for(let h=0;h<g.length;h++)e[g[h]]=1}if(b)for(let f=0,g;f<a.length;f++)g=a[f],e[g]&&(d.push(g),e[g]=0);else for(let f=0,g,h;f<a.result.length;f++)for(g=a.result[f],c=0;c<g.length;c++)h=g[c],e[h]&&((d[f]||(d[f]=[])).push(h),e[h]=0);return d};I();function eb(a,c,b,e,d){let f,g,h;"string"===typeof d?(f=d,d=""):f=d.template;g=f.indexOf("$1");h=f.substring(g+2);g=f.substring(0,g);let k=d&&d.boundary,m=!d||!1!==d.clip,l=d&&d.merge&&h&&g&&new RegExp(h+" "+g,"g");d=d&&d.ellipsis;var n=0;if("object"===typeof d){var u=d.template;n=u.length-2;d=d.pattern}"string"!==typeof d&&(d=!1===d?"":"...");n&&(d=u.replace("$1",d));u=d.length-n;let p,t;"object"===typeof k&&(p=k.before,0===p&&(p=-1),t=k.after,0===t&&(t=-1),k=k.total||9E5);n=new Map;for(let Oa=0,
+da,db,oa;Oa<c.length;Oa++){let pa;if(e)pa=c,oa=e;else{var r=c[Oa];oa=r.field;if(!oa)continue;pa=r.result}db=b.get(oa);da=db.encoder;r=n.get(da);"string"!==typeof r&&(r=da.encode(a),n.set(da,r));for(let xa=0;xa<pa.length;xa++){var q=pa[xa].doc;if(!q)continue;q=ea(q,oa);if(!q)continue;var y=q.trim().split(/\s+/);if(!y.length)continue;q="";var v=[];let ya=[];var A=-1,B=-1,E=0;for(var C=0;C<y.length;C++){var F=y[C],z=da.encode(F);z=1<z.length?z.join(" "):z[0];let x;if(z&&F){var D=F.length,K=(da.split?
+F.replace(da.split,""):F).length-z.length,G="",N=0;for(var O=0;O<r.length;O++){var P=r[O];if(P){var L=P.length;L+=K;N&&L<=N||(P=z.indexOf(P),-1<P&&(G=(P?F.substring(0,P):"")+g+F.substring(P,P+L)+h+(P+L<D?F.substring(P+L):""),N=L,x=!0))}}G&&(k&&(0>A&&(A=q.length+(q?1:0)),B=q.length+(q?1:0)+G.length,E+=D,ya.push(v.length),v.push({match:G})),q+=(q?" ":"")+G)}if(!x)F=y[C],q+=(q?" ":"")+F,k&&v.push({text:F});else if(k&&E>=k)break}E=ya.length*(f.length-2);if(p||t||k&&q.length-E>k)if(E=k+E-2*u,C=B-A,0<p&&
+(C+=p),0<t&&(C+=t),C<=E)y=p?A-(0<p?p:0):A-((E-C)/2|0),v=t?B+(0<t?t:0):y+E,m||(0<y&&" "!==q.charAt(y)&&" "!==q.charAt(y-1)&&(y=q.indexOf(" ",y),0>y&&(y=0)),v<q.length&&" "!==q.charAt(v-1)&&" "!==q.charAt(v)&&(v=q.lastIndexOf(" ",v),v<B?v=B:++v)),q=(y?d:"")+q.substring(y,v)+(v<q.length?d:"");else{B=[];A={};E={};C={};F={};z={};G=K=D=0;for(O=N=1;;){var U=void 0;for(let x=0,J;x<ya.length;x++){J=ya[x];if(G)if(K!==G){if(C[x+1])continue;J+=G;if(A[J]){D-=u;E[x+1]=1;C[x+1]=1;continue}if(J>=v.length-1){if(J>=
+v.length){C[x+1]=1;J>=y.length&&(E[x+1]=1);continue}D-=u}q=v[J].text;if(L=t&&z[x])if(0<L){if(q.length>L)if(C[x+1]=1,m)q=q.substring(0,L);else continue;(L-=q.length)||(L=-1);z[x]=L}else{C[x+1]=1;continue}if(D+q.length+1<=k)q=" "+q,B[x]+=q;else if(m)U=k-D-1,0<U&&(q=" "+q.substring(0,U),B[x]+=q),C[x+1]=1;else{C[x+1]=1;continue}}else{if(C[x])continue;J-=K;if(A[J]){D-=u;C[x]=1;E[x]=1;continue}if(0>=J){if(0>J){C[x]=1;E[x]=1;continue}D-=u}q=v[J].text;if(L=p&&F[x])if(0<L){if(q.length>L)if(C[x]=1,m)q=q.substring(q.length-
+L);else continue;(L-=q.length)||(L=-1);F[x]=L}else{C[x]=1;continue}if(D+q.length+1<=k)q+=" ",B[x]=q+B[x];else if(m)U=q.length+1-(k-D),0<=U&&U<q.length&&(q=q.substring(U)+" ",B[x]=q+B[x]),C[x]=1;else{C[x]=1;continue}}else{q=v[J].match;p&&(F[x]=p);t&&(z[x]=t);x&&D++;let Pa;J?!x&&u&&(D+=u):(E[x]=1,C[x]=1);J>=y.length-1?Pa=1:J<v.length-1&&v[J+1].match?Pa=1:u&&(D+=u);D-=f.length-2;if(!x||D+q.length<=k)B[x]=q;else{U=N=O=E[x]=0;break}Pa&&(E[x+1]=1,C[x+1]=1)}D+=q.length;U=A[J]=1}if(U)K===G?G++:K++;else{K===
+G?N=0:O=0;if(!N&&!O)break;N?(K++,G=K):G++}}q="";for(let x=0,J;x<B.length;x++)J=(x&&E[x]?" ":(x&&!d?" ":"")+d)+B[x],q+=J;d&&!E[B.length]&&(q+=d)}l&&(q=q.replace(l," "));pa[xa].highlight=q}if(e)break}return c};Na.prototype.search=function(a,c,b,e){b||(!c&&ba(a)?(b=a,a=""):ba(c)&&(b=c,c=0));let d=[];var f=[];let g;let h,k,m,l,n;let u=0,p=!0,t;if(b){b.constructor===Array&&(b={index:b});a=b.query||a;g=b.pluck;h=b.merge;m=b.boost;n=g||b.field||(n=b.index)&&(n.index?null:n);var r=this.tag&&b.tag;k=b.suggest;p=!1!==b.resolve;l=b.cache;t=p&&this.store&&b.highlight;var q=!!t||p&&this.store&&b.enrich;c=b.limit||c;var y=b.offset||0;c||(c=p?100:0);if(r&&(!this.db||!e)){r.constructor!==Array&&(r=[r]);var v=[];for(let F=
+0,z;F<r.length;F++)if(z=r[F],z.field&&z.tag){var A=z.tag;if(A.constructor===Array)for(var B=0;B<A.length;B++)v.push(z.field,A[B]);else v.push(z.field,A)}else{A=Object.keys(z);for(let D=0,K,G;D<A.length;D++)if(K=A[D],G=z[K],G.constructor===Array)for(B=0;B<G.length;B++)v.push(K,G[B]);else v.push(K,G)}r=v;if(!a){f=[];if(v.length)for(r=0;r<v.length;r+=2){if(this.db){e=this.index.get(v[r]);if(!e)continue;f.push(e=e.db.tag(v[r+1],c,y,q))}else e=fb.call(this,v[r],v[r+1],c,y,q);d.push(p?{field:v[r],tag:v[r+
+1],result:e}:[e])}if(f.length){const F=this;return Promise.all(f).then(function(z){for(let D=0;D<z.length;D++)p?d[D].result=z[D]:d[D]=z[D];return p?d:new Y(1<d.length?$a(d,1,0,0,k,m):d[0],F)})}return p?d:new Y(1<d.length?$a(d,1,0,0,k,m):d[0],this)}}p||g||!(n=n||this.field)||(M(n)?g=n:(n.constructor===Array&&1===n.length&&(n=n[0]),g=n.field||n.index));n&&n.constructor!==Array&&(n=[n])}n||(n=this.field);let E;v=(this.worker||this.db)&&!e&&[];for(let F=0,z,D,K;F<n.length;F++){D=n[F];if(this.db&&this.tag&&
+!this.G[F])continue;let G;M(D)||(G=D,D=G.field,a=G.query||a,c=aa(G.limit,c),y=aa(G.offset,y),k=aa(G.suggest,k),t=p&&this.store&&aa(G.highlight,t),q=!!t||p&&this.store&&aa(G.enrich,q),l=aa(G.cache,l));if(e)z=e[F];else{A=G||b||{};B=A.enrich;var C=this.index.get(D);r&&(this.db&&(A.tag=r,E=C.db.support_tag_search,A.field=n),!E&&B&&(A.enrich=!1));z=l?C.searchCache(a,c,A):C.search(a,c,A);B&&(A.enrich=B);if(v){v[F]=z;continue}}K=(z=z.result||z)&&z.length;if(r&&K){A=[];B=0;if(this.db&&e){if(!E)for(C=n.length;C<
+e.length;C++){let N=e[C];if(N&&N.length)B++,A.push(N);else if(!k)return p?d:new Y(d,this)}}else for(let N=0,O,P;N<r.length;N+=2){O=this.tag.get(r[N]);if(!O)if(k)continue;else return p?d:new Y(d,this);if(P=(O=O&&O.get(r[N+1]))&&O.length)B++,A.push(O);else if(!k)return p?d:new Y(d,this)}if(B){z=cb(z,A,p);K=z.length;if(!K&&!k)return p?z:new Y(z,this);B--}}if(K)f[u]=D,d.push(z),u++;else if(1===n.length)return p?d:new Y(d,this)}if(v){if(this.db&&r&&r.length&&!E)for(q=0;q<r.length;q+=2){f=this.index.get(r[q]);
+if(!f)if(k)continue;else return p?d:new Y(d,this);v.push(f.db.tag(r[q+1],c,y,!1))}const F=this;return Promise.all(v).then(function(z){b&&(b.resolve=p);z.length&&(z=F.search(a,c,b,z));return z})}if(!u)return p?d:new Y(d,this);if(g&&(!q||!this.store))return d=d[0],p?d:new Y(d,this);v=[];for(y=0;y<f.length;y++){r=d[y];q&&r.length&&"undefined"===typeof r[0].doc&&(this.db?v.push(r=this.index.get(this.field[0]).db.enrich(r)):r=Ta.call(this,r));if(g)return p?t?eb(a,r,this.index,g,t):r:new Y(r,this);d[y]=
+{field:f[y],result:r}}if(q&&this.db&&v.length){const F=this;return Promise.all(v).then(function(z){for(let D=0;D<z.length;D++)d[D].result=z[D];t&&(d=eb(a,d,F.index,g,t));return h?gb(d):d})}t&&(d=eb(a,d,this.index,g,t));return h?gb(d):d};
+function gb(a){const c=[],b=I(),e=I();for(let d=0,f,g,h,k,m,l,n;d<a.length;d++){f=a[d];g=f.field;h=f.result;for(let u=0;u<h.length;u++)if(m=h[u],"object"!==typeof m?m={id:k=m}:k=m.id,(l=b[k])?l.push(g):(m.field=b[k]=[g],c.push(m)),n=m.highlight)l=e[k],l||(e[k]=l={},m.highlight=l),l[g]=n}return c}function fb(a,c,b,e,d){a=this.tag.get(a);if(!a)return[];a=a.get(c);if(!a)return[];c=a.length-e;if(0<c){if(b&&c>b||e)a=a.slice(e,e+b);d&&(a=Ta.call(this,a))}return a}
+function Ta(a){if(!this||!this.store)return a;if(this.db)return this.index.get(this.field[0]).db.enrich(a);const c=Array(a.length);for(let b=0,e;b<a.length;b++)e=a[b],c[b]={id:e,doc:this.store.get(e)};return c};function Na(a){if(!this||this.constructor!==Na)return new Na(a);const c=a.document||a.doc||a;let b,e;this.G=[];this.field=[];this.K=[];this.key=(b=c.key||c.id)&&hb(b,this.K)||"id";(e=a.keystore||0)&&(this.keystore=e);this.fastupdate=!!a.fastupdate;this.reg=!this.fastupdate||a.worker||a.db?e?new T(e):new Set:e?new S(e):new Map;this.D=(b=c.store||null)&&b&&!0!==b&&[];this.store=b&&(e?new S(e):new Map);this.cache=(b=a.cache||null)&&new R(b);a.cache=!1;this.worker=a.worker||!1;this.priority=a.priority||
+4;this.index=ib.call(this,a,c);this.tag=null;if(b=c.tag)if("string"===typeof b&&(b=[b]),b.length){this.tag=new Map;this.F=[];this.P=[];for(let d=0,f,g;d<b.length;d++){f=b[d];g=f.field||f;if(!g)throw Error("The tag field from the document descriptor is undefined.");f.custom?this.F[d]=f.custom:(this.F[d]=hb(g,this.K),f.filter&&("string"===typeof this.F[d]&&(this.F[d]=new String(this.F[d])),this.F[d].J=f.filter));this.P[d]=g;this.tag.set(g,new Map)}}if(this.worker){this.fastupdate=!1;a=[];for(const d of this.index.values())d.then&&
+a.push(d);if(a.length){const d=this;return Promise.all(a).then(function(f){let g=0;for(const h of d.index.entries()){const k=h[0];let m=h[1];m.then&&(m=f[g],d.index.set(k,m),g++)}return d})}}else a.db&&(this.fastupdate=!1,this.mount(a.db))}w=Na.prototype;
+w.mount=function(a){let c=this.field;if(this.tag)for(let f=0,g;f<this.P.length;f++){g=this.P[f];var b=void 0;this.index.set(g,b=new V({},this.reg));c===this.field&&(c=c.slice(0));c.push(g);b.tag=this.tag.get(g)}b=[];const e={db:a.db,type:a.type,fastupdate:a.fastupdate};for(let f=0,g,h;f<c.length;f++){e.field=h=c[f];g=this.index.get(h);const k=new a.constructor(a.id,e);k.id=a.id;b[f]=k.mount(g);g.document=!0;f?g.bypass=!0:g.store=this.store}const d=this;return this.db=Promise.all(b).then(function(){d.db=
+!0})};w.commit=async function(a,c){const b=[];for(const e of this.index.values())b.push(e.commit(a,c));await Promise.all(b);this.reg.clear()};w.destroy=function(){const a=[];for(const c of this.index.values())a.push(c.destroy());return Promise.all(a)};
+function ib(a,c){const b=new Map;let e=c.index||c.field||c;M(e)&&(e=[e]);for(let f=0,g,h;f<e.length;f++){g=e[f];M(g)||(h=g,g=g.field);h=ba(h)?Object.assign({},a,h):a;if(this.worker){var d=void 0;d=(d=h.encoder)&&d.encode?d:new la("string"===typeof d?va[d]:d||{});d=new La(h,d);b.set(g,d)}this.worker||b.set(g,new V(h,this.reg));h.custom?this.G[f]=h.custom:(this.G[f]=hb(g,this.K),h.filter&&("string"===typeof this.G[f]&&(this.G[f]=new String(this.G[f])),this.G[f].J=h.filter));this.field[f]=g}if(this.D){a=
+c.store;M(a)&&(a=[a]);for(let f=0,g,h;f<a.length;f++)g=a[f],h=g.field||g,g.custom?(this.D[f]=g.custom,g.custom.U=h):(this.D[f]=hb(h,this.K),g.filter&&("string"===typeof this.D[f]&&(this.D[f]=new String(this.D[f])),this.D[f].J=g.filter))}return b}function hb(a,c){const b=a.split(":");let e=0;for(let d=0;d<b.length;d++)a=b[d],"]"===a[a.length-1]&&(a=a.substring(0,a.length-2))&&(c[e]=!0),a&&(b[e++]=a);e<b.length&&(b.length=e);return 1<e?b:b[0]}w.append=function(a,c){return this.add(a,c,!0)};
+w.update=function(a,c){return this.remove(a).add(a,c)};w.remove=function(a){ba(a)&&(a=ea(a,this.key));for(var c of this.index.values())c.remove(a,!0);if(this.reg.has(a)){if(this.tag&&!this.fastupdate)for(let b of this.tag.values())for(let e of b){c=e[0];const d=e[1],f=d.indexOf(a);-1<f&&(1<d.length?d.splice(f,1):b.delete(c))}this.store&&this.store.delete(a);this.reg.delete(a)}this.cache&&this.cache.remove(a);return this};
+w.clear=function(){const a=[];for(const c of this.index.values()){const b=c.clear();b.then&&a.push(b)}if(this.tag)for(const c of this.tag.values())c.clear();this.store&&this.store.clear();this.cache&&this.cache.clear();return a.length?Promise.all(a):this};w.contain=function(a){return this.db?this.index.get(this.field[0]).db.has(a):this.reg.has(a)};w.cleanup=function(){for(const a of this.index.values())a.cleanup();return this};
+w.get=function(a){return this.db?this.index.get(this.field[0]).db.enrich(a).then(function(c){return c[0]&&c[0].doc||null}):this.store.get(a)||null};w.set=function(a,c){"object"===typeof a&&(c=a,a=ea(c,this.key));this.store.set(a,c);return this};w.searchCache=ma;w.export=jb;w.import=kb;Fa(Na.prototype);function lb(a,c=0){let b=[],e=[];c&&(c=25E4/c*5E3|0);for(const d of a.entries())e.push(d),e.length===c&&(b.push(e),e=[]);e.length&&b.push(e);return b}function mb(a,c){c||(c=new Map);for(let b=0,e;b<a.length;b++)e=a[b],c.set(e[0],e[1]);return c}function nb(a,c=0){let b=[],e=[];c&&(c=25E4/c*1E3|0);for(const d of a.entries())e.push([d[0],lb(d[1])[0]]),e.length===c&&(b.push(e),e=[]);e.length&&b.push(e);return b}
+function ob(a,c){c||(c=new Map);for(let b=0,e,d;b<a.length;b++)e=a[b],d=c.get(e[0]),c.set(e[0],mb(e[1],d));return c}function pb(a){let c=[],b=[];for(const e of a.keys())b.push(e),25E4===b.length&&(c.push(b),b=[]);b.length&&c.push(b);return c}function qb(a,c){c||(c=new Set);for(let b=0;b<a.length;b++)c.add(a[b]);return c}
+function rb(a,c,b,e,d,f,g=0){const h=e&&e.constructor===Array;var k=h?e.shift():e;if(!k)return this.export(a,c,d,f+1);if((k=a((c?c+".":"")+(g+1)+"."+b,JSON.stringify(k)))&&k.then){const m=this;return k.then(function(){return rb.call(m,a,c,b,h?e:null,d,f,g+1)})}return rb.call(this,a,c,b,h?e:null,d,f,g+1)}
+function jb(a,c,b=0,e=0){if(b<this.field.length){const g=this.field[b];if((c=this.index.get(g).export(a,g,b,e=1))&&c.then){const h=this;return c.then(function(){return h.export(a,g,b+1)})}return this.export(a,g,b+1)}let d,f;switch(e){case 0:d="reg";f=pb(this.reg);c=null;break;case 1:d="tag";f=this.tag&&nb(this.tag,this.reg.size);c=null;break;case 2:d="doc";f=this.store&&lb(this.store);c=null;break;default:return}return rb.call(this,a,c,d,f,b,e)}
+function kb(a,c){var b=a.split(".");"json"===b[b.length-1]&&b.pop();const e=2<b.length?b[0]:"";b=2<b.length?b[2]:b[1];if(this.worker&&e)return this.index.get(e).import(a);if(c){"string"===typeof c&&(c=JSON.parse(c));if(e)return this.index.get(e).import(b,c);switch(b){case "reg":this.fastupdate=!1;this.reg=qb(c,this.reg);for(let d=0,f;d<this.field.length;d++)f=this.index.get(this.field[d]),f.fastupdate=!1,f.reg=this.reg;if(this.worker){c=[];for(const d of this.index.values())c.push(d.import(a));return Promise.all(c)}break;
+case "tag":this.tag=ob(c,this.tag);break;case "doc":this.store=mb(c,this.store)}}}function sb(a,c){let b="";for(const e of a.entries()){a=e[0];const d=e[1];let f="";for(let g=0,h;g<d.length;g++){h=d[g]||[""];let k="";for(let m=0;m<h.length;m++)k+=(k?",":"")+("string"===c?'"'+h[m]+'"':h[m]);k="["+k+"]";f+=(f?",":"")+k}f='["'+a+'",['+f+"]]";b+=(b?",":"")+f}return b};V.prototype.remove=function(a,c){const b=this.reg.size&&(this.fastupdate?this.reg.get(a):this.reg.has(a));if(b){if(this.fastupdate)for(let e=0,d,f;e<b.length;e++){if((d=b[e])&&(f=d.length))if(d[f-1]===a)d.pop();else{const g=d.indexOf(a);0<=g&&d.splice(g,1)}}else tb(this.map,a),this.depth&&tb(this.ctx,a);c||this.reg.delete(a)}this.db&&(this.commit_task.push({del:a}),this.S&&ub(this));this.cache&&this.cache.remove(a);return this};
+function tb(a,c){let b=0;var e="undefined"===typeof c;if(a.constructor===Array)for(let d=0,f,g,h;d<a.length;d++){if((f=a[d])&&f.length){if(e)return 1;g=f.indexOf(c);if(0<=g){if(1<f.length)return f.splice(g,1),1;delete a[d];if(b)return 1;h=1}else{if(h)return 1;b++}}}else for(let d of a.entries())e=d[0],tb(d[1],c)?b++:a.delete(e);return b};const vb={memory:{resolution:1},performance:{resolution:3,fastupdate:!0,context:{depth:1,resolution:1}},match:{tokenize:"forward"},score:{resolution:9,context:{depth:2,resolution:3}}};V.prototype.add=function(a,c,b,e){if(c&&(a||0===a)){if(!e&&!b&&this.reg.has(a))return this.update(a,c);e=this.depth;c=this.encoder.encode(c,!e);const m=c.length;if(m){const l=I(),n=I(),u=this.resolution;for(let p=0;p<m;p++){let t=c[this.rtl?m-1-p:p];var d=t.length;if(d&&(e||!n[t])){var f=this.score?this.score(c,t,p,null,0):wb(u,m,p),g="";switch(this.tokenize){case "full":if(2<d){for(let r=0,q;r<d;r++)for(f=d;f>r;f--){g=t.substring(r,f);q=this.rtl?d-1-r:r;var h=this.score?this.score(c,t,p,g,q):wb(u,
+m,p,d,q);xb(this,n,g,h,a,b)}break}case "bidirectional":case "reverse":if(1<d){for(h=d-1;0<h;h--){g=t[this.rtl?d-1-h:h]+g;var k=this.score?this.score(c,t,p,g,h):wb(u,m,p,d,h);xb(this,n,g,k,a,b)}g=""}case "forward":if(1<d){for(h=0;h<d;h++)g+=t[this.rtl?d-1-h:h],xb(this,n,g,f,a,b);break}default:if(xb(this,n,t,f,a,b),e&&1<m&&p<m-1)for(d=I(),g=this.T,f=t,h=Math.min(e+1,this.rtl?p+1:m-p),d[f]=1,k=1;k<h;k++)if((t=c[this.rtl?m-1-p-k:p+k])&&!d[t]){d[t]=1;const r=this.score?this.score(c,f,p,t,k-1):wb(g+(m/
+2>g?0:1),m,p,h-1,k-1),q=this.bidirectional&&t>f;xb(this,l,q?f:t,r,a,b,q?t:f)}}}}this.fastupdate||this.reg.add(a)}else c=""}this.db&&(c||this.commit_task.push({del:a}),this.S&&ub(this));return this};
+function xb(a,c,b,e,d,f,g){let h=g?a.ctx:a.map,k;if(!c[b]||g&&!(k=c[b])[g])if(g?(c=k||(c[b]=I()),c[g]=1,(k=h.get(g))?h=k:h.set(g,h=new Map)):c[b]=1,(k=h.get(b))?h=k:h.set(b,h=k=[]),h=h[e]||(h[e]=[]),!f||!h.includes(d)){if(h.length===2**31-1){c=new za(h);if(a.fastupdate)for(let m of a.reg.values())m.includes(h)&&(m[m.indexOf(h)]=c);k[e]=h=c}h.push(d);a.fastupdate&&((e=a.reg.get(d))?e.push(h):a.reg.set(d,[h]))}}
+function wb(a,c,b,e,d){return b&&1<a?c+(e||0)<=a?b+(d||0):(a-1)/(c+(e||0))*(b+(d||0))+1|0:0};V.prototype.search=function(a,c,b){b||(c||"object"!==typeof a?"object"===typeof c&&(b=c,c=0):(b=a,a=""));if(b&&b.cache)return b.cache=!1,a=this.searchCache(a,c,b),b.cache=!0,a;let e=[],d,f,g,h=0,k,m,l,n,u;b&&(a=b.query||a,c=b.limit||c,h=b.offset||0,f=b.context,g=b.suggest,u=(k=b.resolve)&&b.enrich,l=b.boost,n=b.resolution,m=this.db&&b.tag);"undefined"===typeof k&&(k=this.resolve);f=this.depth&&!1!==f;let p=this.encoder.encode(a,!f);d=p.length;c=c||(k?100:0);if(1===d)return yb.call(this,p[0],"",c,
+h,k,u,m);if(2===d&&f&&!g)return yb.call(this,p[1],p[0],c,h,k,u,m);let t=I(),r=0,q;f&&(q=p[0],r=1);n||0===n||(n=q?this.T:this.resolution);if(this.db){if(this.db.search&&(b=this.db.search(this,p,c,h,g,k,u,m),!1!==b))return b;const y=this;return async function(){for(let v,A;r<d;r++){if((A=p[r])&&!t[A]){t[A]=1;v=await zb(y,A,q,0,0,!1,!1);if(v=Ab(v,e,g,n)){e=v;break}q&&(g&&v&&e.length||(q=A))}g&&q&&r===d-1&&!e.length&&(n=y.resolution,q="",r=-1,t=I())}return Bb(e,n,c,h,g,l,k)}()}for(let y,v;r<d;r++){if((v=
+p[r])&&!t[v]){t[v]=1;y=zb(this,v,q,0,0,!1,!1);if(y=Ab(y,e,g,n)){e=y;break}q&&(g&&y&&e.length||(q=v))}g&&q&&r===d-1&&!e.length&&(n=this.resolution,q="",r=-1,t=I())}return Bb(e,n,c,h,g,l,k)};function Bb(a,c,b,e,d,f,g){let h=a.length,k=a;if(1<h)k=$a(a,c,b,e,d,f,g);else if(1===h)return g?Sa.call(null,a[0],b,e):new Y(a[0],this);return g?k:new Y(k,this)}
+function yb(a,c,b,e,d,f,g){a=zb(this,a,c,b,e,d,f,g);return this.db?a.then(function(h){return d?h||[]:new Y(h,this)}):a&&a.length?d?Sa.call(this,a,b,e):new Y(a,this):d?[]:new Y([],this)}function Ab(a,c,b,e){let d=[];if(a&&a.length){if(a.length<=e){c.push(a);return}for(let f=0,g;f<e;f++)if(g=a[f])d[f]=g;if(d.length){c.push(d);return}}if(!b)return d}
+function zb(a,c,b,e,d,f,g,h){let k;b&&(k=a.bidirectional&&c>b)&&(k=b,b=c,c=k);if(a.db)return a.db.get(c,b,e,d,f,g,h);a=b?(a=a.ctx.get(b))&&a.get(c):a.map.get(c);return a};function V(a,c){if(!this||this.constructor!==V)return new V(a);if(a){var b=M(a)?a:a.preset;b&&(a=Object.assign({},vb[b],a))}else a={};b=a.context;const e=!0===b?{depth:1}:b||{},d=M(a.encoder)?va[a.encoder]:a.encode||a.encoder||{};this.encoder=d.encode?d:"object"===typeof d?new la(d):{encode:d};this.resolution=a.resolution||9;this.tokenize=b=(b=a.tokenize)&&"default"!==b&&"exact"!==b&&b||"strict";this.depth="strict"===b&&e.depth||0;this.bidirectional=!1!==e.bidirectional;this.fastupdate=!!a.fastupdate;
+this.score=a.score||null;(b=a.keystore||0)&&(this.keystore=b);this.map=b?new S(b):new Map;this.ctx=b?new S(b):new Map;this.reg=c||(this.fastupdate?b?new S(b):new Map:b?new T(b):new Set);this.T=e.resolution||3;this.rtl=d.rtl||a.rtl||!1;this.cache=(b=a.cache||null)&&new R(b);this.resolve=!1!==a.resolve;if(b=a.db)this.db=this.mount(b);this.S=!1!==a.commit;this.commit_task=[];this.commit_timer=null;this.priority=a.priority||4}w=V.prototype;
+w.mount=function(a){this.commit_timer&&(clearTimeout(this.commit_timer),this.commit_timer=null);return a.mount(this)};w.commit=function(a,c){this.commit_timer&&(clearTimeout(this.commit_timer),this.commit_timer=null);return this.db.commit(this,a,c)};w.destroy=function(){this.commit_timer&&(clearTimeout(this.commit_timer),this.commit_timer=null);return this.db.destroy()};function ub(a){a.commit_timer||(a.commit_timer=setTimeout(function(){a.commit_timer=null;a.db.commit(a,void 0,void 0)},1))}
+w.clear=function(){this.map.clear();this.ctx.clear();this.reg.clear();this.cache&&this.cache.clear();this.db&&(this.commit_timer&&clearTimeout(this.commit_timer),this.commit_timer=null,this.commit_task=[{clear:!0}]);return this};w.append=function(a,c){return this.add(a,c,!0)};w.contain=function(a){return this.db?this.db.has(a):this.reg.has(a)};w.update=function(a,c){const b=this,e=this.remove(a);return e&&e.then?e.then(()=>b.add(a,c)):this.add(a,c)};
+w.cleanup=function(){if(!this.fastupdate)return this;tb(this.map);this.depth&&tb(this.ctx);return this};w.searchCache=ma;w.export=function(a,c,b=0,e=0){let d,f;switch(e){case 0:d="reg";f=pb(this.reg);break;case 1:d="cfg";f=null;break;case 2:d="map";f=lb(this.map,this.reg.size);break;case 3:d="ctx";f=nb(this.ctx,this.reg.size);break;default:return}return rb.call(this,a,c,d,f,b,e)};
+w.import=function(a,c){if(c)switch("string"===typeof c&&(c=JSON.parse(c)),a=a.split("."),"json"===a[a.length-1]&&a.pop(),3===a.length&&a.shift(),a=1<a.length?a[1]:a[0],a){case "reg":this.fastupdate=!1;this.reg=qb(c,this.reg);break;case "map":this.map=mb(c,this.map);break;case "ctx":this.ctx=ob(c,this.ctx)}};
+w.serialize=function(a=!0){let c="",b="",e="";if(this.reg.size){let f;for(var d of this.reg.keys())f||(f=typeof d),c+=(c?",":"")+("string"===f?'"'+d+'"':d);c="index.reg=new Set(["+c+"]);";b=sb(this.map,f);b="index.map=new Map(["+b+"]);";for(const g of this.ctx.entries()){d=g[0];let h=sb(g[1],f);h="new Map(["+h+"])";h='["'+d+'",'+h+"]";e+=(e?",":"")+h}e="index.ctx=new Map(["+e+"]);"}return a?"function inject(index){"+c+b+e+"}":c+b+e};Fa(V.prototype);const Cb="undefined"!==typeof window&&(window.indexedDB||window.mozIndexedDB||window.webkitIndexedDB||window.msIndexedDB),Db=["map","ctx","tag","reg","cfg"],Eb=I();
+function Fb(a,c={}){if(!this||this.constructor!==Fb)return new Fb(a,c);"object"===typeof a&&(c=a,a=a.name);a||console.info("Default storage space was used, because a name was not passed.");this.id="flexsearch"+(a?":"+a.toLowerCase().replace(/[^a-z0-9_\-]/g,""):"");this.field=c.field?c.field.toLowerCase().replace(/[^a-z0-9_\-]/g,""):"";this.type=c.type;this.fastupdate=this.support_tag_search=!1;this.db=null;this.h={}}w=Fb.prototype;w.mount=function(a){if(a.index)return a.mount(this);a.db=this;return this.open()};
+w.open=function(){if(this.db)return this.db;let a=this;navigator.storage&&navigator.storage.persist();Eb[a.id]||(Eb[a.id]=[]);Eb[a.id].push(a.field);const c=Cb.open(a.id,1);c.onupgradeneeded=function(){const b=a.db=this.result;for(let e=0,d;e<Db.length;e++){d=Db[e];for(let f=0,g;f<Eb[a.id].length;f++)g=Eb[a.id][f],b.objectStoreNames.contains(d+("reg"!==d?g?":"+g:"":""))||b.createObjectStore(d+("reg"!==d?g?":"+g:"":""))}};return a.db=Z(c,function(b){a.db=b;a.db.onversionchange=function(){a.close()}})};
+w.close=function(){this.db&&this.db.close();this.db=null};w.destroy=function(){const a=Cb.deleteDatabase(this.id);return Z(a)};w.clear=function(){const a=[];for(let b=0,e;b<Db.length;b++){e=Db[b];for(let d=0,f;d<Eb[this.id].length;d++)f=Eb[this.id][d],a.push(e+("reg"!==e?f?":"+f:"":""))}const c=this.db.transaction(a,"readwrite");for(let b=0;b<a.length;b++)c.objectStore(a[b]).clear();return Z(c)};
+w.get=function(a,c,b=0,e=0,d=!0,f=!1){a=this.db.transaction((c?"ctx":"map")+(this.field?":"+this.field:""),"readonly").objectStore((c?"ctx":"map")+(this.field?":"+this.field:"")).get(c?c+":"+a:a);const g=this;return Z(a).then(function(h){let k=[];if(!h||!h.length)return k;if(d){if(!b&&!e&&1===h.length)return h[0];for(let m=0,l;m<h.length;m++)if((l=h[m])&&l.length){if(e>=l.length){e-=l.length;continue}const n=b?e+Math.min(l.length-e,b):l.length;for(let u=e;u<n;u++)k.push(l[u]);e=0;if(k.length===b)break}return f?
+g.enrich(k):k}return h})};w.tag=function(a,c=0,b=0,e=!1){a=this.db.transaction("tag"+(this.field?":"+this.field:""),"readonly").objectStore("tag"+(this.field?":"+this.field:"")).get(a);const d=this;return Z(a).then(function(f){if(!f||!f.length||b>=f.length)return[];if(!c&&!b)return f;f=f.slice(b,b+c);return e?d.enrich(f):f})};
+w.enrich=function(a){"object"!==typeof a&&(a=[a]);const c=this.db.transaction("reg","readonly").objectStore("reg"),b=[];for(let e=0;e<a.length;e++)b[e]=Z(c.get(a[e]));return Promise.all(b).then(function(e){for(let d=0;d<e.length;d++)e[d]={id:a[d],doc:e[d]?JSON.parse(e[d]):null};return e})};w.has=function(a){a=this.db.transaction("reg","readonly").objectStore("reg").getKey(a);return Z(a).then(function(c){return!!c})};w.search=null;w.info=function(){};
+w.transaction=function(a,c,b){a+="reg"!==a?this.field?":"+this.field:"":"";let e=this.h[a+":"+c];if(e)return b.call(this,e);let d=this.db.transaction(a,c);this.h[a+":"+c]=e=d.objectStore(a);const f=b.call(this,e);this.h[a+":"+c]=null;return Z(d).finally(function(){d=e=null;return f})};
+w.commit=async function(a,c,b){if(c)await this.clear(),a.commit_task=[];else{let e=a.commit_task;a.commit_task=[];for(let d=0,f;d<e.length;d++)if(f=e[d],f.clear){await this.clear();c=!0;break}else e[d]=f.del;c||(b||(e=e.concat(ca(a.reg))),e.length&&await this.remove(e))}a.reg.size&&(await this.transaction("map","readwrite",function(e){for(const d of a.map){const f=d[0],g=d[1];g.length&&(c?e.put(g,f):e.get(f).onsuccess=function(){let h=this.result;var k;if(h&&h.length){const m=Math.max(h.length,g.length);
+for(let l=0,n,u;l<m;l++)if((u=g[l])&&u.length){if((n=h[l])&&n.length)for(k=0;k<u.length;k++)n.push(u[k]);else h[l]=u;k=1}}else h=g,k=1;k&&e.put(h,f)})}}),await this.transaction("ctx","readwrite",function(e){for(const d of a.ctx){const f=d[0],g=d[1];for(const h of g){const k=h[0],m=h[1];m.length&&(c?e.put(m,f+":"+k):e.get(f+":"+k).onsuccess=function(){let l=this.result;var n;if(l&&l.length){const u=Math.max(l.length,m.length);for(let p=0,t,r;p<u;p++)if((r=m[p])&&r.length){if((t=l[p])&&t.length)for(n=
+0;n<r.length;n++)t.push(r[n]);else l[p]=r;n=1}}else l=m,n=1;n&&e.put(l,f+":"+k)})}}}),a.store?await this.transaction("reg","readwrite",function(e){for(const d of a.store){const f=d[0],g=d[1];e.put("object"===typeof g?JSON.stringify(g):1,f)}}):a.bypass||await this.transaction("reg","readwrite",function(e){for(const d of a.reg.keys())e.put(1,d)}),a.tag&&await this.transaction("tag","readwrite",function(e){for(const d of a.tag){const f=d[0],g=d[1];g.length&&(e.get(f).onsuccess=function(){let h=this.result;
+h=h&&h.length?h.concat(g):g;e.put(h,f)})}}),a.map.clear(),a.ctx.clear(),a.tag&&a.tag.clear(),a.store&&a.store.clear(),a.document||a.reg.clear())};function Gb(a,c,b){const e=a.value;let d,f=0;for(let g=0,h;g<e.length;g++){if(h=b?e:e[g]){for(let k=0,m,l;k<c.length;k++)if(l=c[k],m=h.indexOf(l),0<=m)if(d=1,1<h.length)h.splice(m,1);else{e[g]=[];break}f+=h.length}if(b)break}f?d&&a.update(e):a.delete();a.continue()}
+w.remove=function(a){"object"!==typeof a&&(a=[a]);return Promise.all([this.transaction("map","readwrite",function(c){c.openCursor().onsuccess=function(){const b=this.result;b&&Gb(b,a)}}),this.transaction("ctx","readwrite",function(c){c.openCursor().onsuccess=function(){const b=this.result;b&&Gb(b,a)}}),this.transaction("tag","readwrite",function(c){c.openCursor().onsuccess=function(){const b=this.result;b&&Gb(b,a,!0)}}),this.transaction("reg","readwrite",function(c){for(let b=0;b<a.length;b++)c.delete(a[b])})])};
+function Z(a,c){return new Promise((b,e)=>{a.onsuccess=a.oncomplete=function(){c&&c(this.result);c=null;b(this.result)};a.onerror=a.onblocked=e;a=null})};export default {Index:V,Charset:va,Encoder:la,Document:Na,Worker:La,Resolver:Y,IndexedDB:Fb,Language:{}};
+export const Index=V;export const Charset=va;export const Encoder=la;export const Document=Na;export const Worker=La;export const Resolver=Y;export const IndexedDB=Fb;export const Language={};
@@ -1,3 +1,5 @@
+import { settingValues } from './settings.js';
+
 // --- Html String constants ---

 /**
@@ -44,6 +46,15 @@ const REG_STRIP_TRAILING_PAREN = /(.*)(?<!\\)\)$/s;
 const REG_WILDCARD_WEIGHTED_TAG = /(\d+)[_\s]*::(.*?)(?=\||$)/g;
 const REG_WILDCARD_SIMPLE_WORD = /[^{}_|]+/g;

+/**
+ * Checks if the input text is longer than the maximum allowed tag length.
+ * @param {string} text The text to check.
+ * @return {boolean} True if the text is longer than the maximum tag length, false otherwise.
+ */
+export function isLongText(text) {
+    return text?.length > settingValues._maxTagLength;
+}
+
 /**
 * Converts Hiragana to Katakana.
 * @param {string} str Input string.
@@ -98,6 +109,30 @@ export function formatCountHumanReadable(num) {
    return (num / si[i].value).toFixed(1).replace(rx, "$1") + si[i].symbol;
 }

+/**
+ * Escapes HTML special characters in a string.
+ * @param {string} str The input string.
+ * @returns {string} The escaped string.
+ */
+export function escapeHtml(str) {
+    if (typeof str !== 'string') {
+        return str;
+    }
+
+    const escapeMap = {
+        '&': '&amp;',
+        '<': '&lt;',
+        '>': '&gt;',
+        '"': '&quot;',
+        "'": '&#x27;',
+        '`': '&#x60;',
+        '/': '&#x2F;'
+    };
+
+    return str.replace(/[&<>"'`/]/g, match =>
+        escapeMap[match]);
+}
+
 /**
 * Escapes parentheses in a string for use in prompts.
 * Replaces '(' with '\(' and ')' with '\)'.
Author	SHA1	Message	Date
newtextdoc1111	c4d5eee89d	feat: Bump version to 1.4.1	2025-10-14 23:54:20 +09:00
newtextdoc1111	fa4d87ee77	Merge pull request #40 from newtextdoc1111/dev Merge Dev	2025-10-14 23:49:43 +09:00
newtextdoc1111	3cb5a41cc8	fix: Temporarily comment out CSV update check button until remote file detection is properly implemented	2025-10-14 23:46:41 +09:00
newtextdoc1111	16a325b798	fix: Temporarily comment out remote file modification checks in the downloader	2025-08-30 23:40:27 +09:00
newtextdoc1111	e3171940fb	fix: Removed the process that periodically checks for CSV file updates.	2025-08-30 23:39:52 +09:00
newtextdoc1111	8db37967f8	fix: Bump version to 1.4.0	2025-07-28 16:01:58 +09:00
newtextdoc1111	e43e51b48a	Merge pull request #31 from newtextdoc1111/feature/#26_embedding_and_lora_autocompletion Feature/#26 embedding and lora autocompletion	2025-07-28 16:00:37 +09:00
newtextdoc1111	7b89683bfc	fix: Change to `toLowerCase` to simplify string normalization	2025-07-28 15:45:23 +09:00
newtextdoc1111	72b9bd94ee	fix: Improved the encoder to provide more accurate suggestions	2025-07-28 14:05:49 +09:00
newtextdoc1111	8a4a94bc7e	docs: Added Lora and Emb suggestions to README	2025-07-28 05:36:53 +09:00
newtextdoc1111	67b7b334b8	feat: Add embeddings and loras autocompletion	2025-07-28 05:21:57 +09:00
newtextdoc1111	9cdc18e566	Merge pull request #30 from newtextdoc1111/feature/improved_fast_search_behavior Feature/improved fast search behavior	2025-07-24 03:59:49 +09:00
newtextdoc1111	3bbdb0312d	bump: Update version to 1.3.1	2025-07-24 03:56:37 +09:00
newtextdoc1111	5b6fb8f490	refactor: Separates autocomplete logic into two methods: "sequentialSearch" and "searchWithFlexSearch"	2025-07-23 12:39:52 +09:00
newtextdoc1111	83dc2585ed	fix: Correct formatting	2025-07-23 11:14:23 +09:00
newtextdoc1111	7e10c106d4	feat: Add env-cmd for managing environment variables in test scripts	2025-07-23 11:13:43 +09:00
newtextdoc1111	bf7823e985	fix: Increase the search result limit for FlexSearch to provide more accurate suggestions.	2025-07-23 11:12:23 +09:00
newtextdoc1111	6c29336435	feat: Optimized FlexSearch document and encoder settings to improve autocompletion	2025-07-23 07:47:52 +09:00
newtextdoc1111	083c1f5acc	feat: add FlexSearch integration tests	2025-07-23 07:47:52 +09:00
newtextdoc1111	bdd86ab04d	Merge pull request #29 from newtextdoc1111/dev Merge Dev	2025-07-19 09:30:49 +09:00
newtextdoc1111	bafce3ca47	fix: Remove unused Charset import from FlexSearch module	2025-07-19 09:29:14 +09:00
newtextdoc1111	1729d5c4f7	docs: Fix link format	2025-07-19 09:11:12 +09:00
newtextdoc1111	d0dfcc031a	fix: Bump version to 1.3.0	2025-07-19 09:03:43 +09:00
newtextdoc1111	ab2a4bedb3	docs: Add 'Use Fast Search' documentation	2025-07-19 08:59:45 +09:00
newtextdoc1111	6e7ebdae0a	feat: Add tooltip for 'Use Fast Search' setting	2025-07-19 08:41:48 +09:00
newtextdoc1111	bd12dd933f	docs: Add comment to fast search code	2025-07-19 08:39:07 +09:00
newtextdoc1111	784f7019ce	Merge pull request #28 from IceSandwich/fast_search Improve the performance	2025-07-17 22:30:53 +09:00
IceSandwich	6b34f3876f	change `flexSearchLimit` to `flexSearchLimitMultiplier`	2025-07-17 20:51:59 +08:00
IceSandwich	59938c08d0	calculate the search limit from CSV files	2025-07-17 11:12:24 +08:00
IceSandwich	9c6215139b	display actual candidate count when using fast search	2025-07-17 10:12:15 +08:00
IceSandwich	cfda834141	Merge branch 'dev' into fast_search	2025-07-17 09:59:44 +08:00
IceSandwich	b4175e700b	fix the issue when input in tag and alias at the same time	2025-07-16 15:46:22 +08:00
IceSandwich	b6c03233ba	fix the behavior of matching words	2025-07-16 14:20:51 +08:00
IceSandwich	385f9d1c4f	remove period handling completely	2025-07-16 10:58:03 +08:00
newtextdoc1111	f019960d34	fix: Remove unneccesary period handling in findAllTagPositions	2025-07-16 08:14:33 +09:00
newtextdoc1111	1d85f3e284	feat: Add 'Use Fast Search' setting to autocomplete options	2025-07-16 08:01:45 +09:00
newtextdoc1111	e072832216	fix: Improve sorting logic in searchCompletionCandidates for better tag matching	2025-07-16 08:01:11 +09:00
newtextdoc1111	a374ca6c11	feat: Refactor autocomplete data handling to build FlexSearch index after tags are loaded	2025-07-16 08:00:24 +09:00
IceSandwich	8750b5dcac	fix indents	2025-07-14 14:48:20 +08:00
IceSandwich	456bc59240	improve performance by changing the search engine to flexsearch	2025-07-14 14:36:01 +08:00
newtextdoc1111	c3e661439f	Merge pull request #27 from newtextdoc1111/bugfix/#25_fix_natural_language_performance Bugfix/#25 fix natural language performance	2025-07-13 22:57:01 +09:00
newtextdoc1111	ea2ff76fab	fix: Bump version to 1.2.1	2025-07-13 22:48:48 +09:00
newtextdoc1111	8b38ade1b9	fix: Add optional chaining to prevent errors in isLongText function	2025-07-13 22:48:36 +09:00
newtextdoc1111	944ecff9ce	fix: Implement maximum tag length check for performance issue	2025-07-13 22:21:21 +09:00
newtextdoc1111	a9273fb74d	docs: Update README to clarify steps for disabling CSV update check on startup	2025-07-06 20:01:09 +09:00
newtextdoc1111	cf30c94faa	Merge pull request #24 from newtextdoc1111/feature/#23_add_check_csv_updates_on_startup_settings	2025-07-06 19:17:09 +09:00
newtextdoc1111	828c08df0c	fix: Bump version to 1.2.0 in pyproject.toml	2025-07-06 19:11:12 +09:00
newtextdoc1111	3dbfd29274	docs: Added disable CSV update check on startup section	2025-07-06 19:10:06 +09:00
newtextdoc1111	86527e35b4	feat: Add check_updates_on_startup option to control CSV update checks	2025-07-06 19:08:24 +09:00
newtextdoc1111	1bc609893b	refactor: Use constant for HuggingFace URL	2025-07-06 19:07:55 +09:00
newtextdoc1111	a536d58cf2	docs: Remove recent updates section from README files	2025-07-06 18:23:04 +09:00
newtextdoc1111	4c921da6cb	fix: Update CSV file status logging format for better readability	2025-07-05 14:29:30 +09:00
newtextdoc1111	98bf626ec1	refactor: Format code with ruff and update configuration	2025-07-05 13:40:17 +09:00
newtextdoc1111	57f80f0991	Merge pull request #22 from newtextdoc1111/feature/#21-add_tag_alias_to_header Feature/#21 add tag alias to related tags display header	2025-06-14 17:48:14 +09:00
newtextdoc1111	31f8fe4517	fix: Update version to 1.1.0 in pyproject.toml	2025-06-14 17:44:32 +09:00
newtextdoc1111	292c99ae17	fix: Remove unused CSS classes for related tags layout	2025-06-10 04:10:08 +09:00
newtextdoc1111	86b63047c2	feat: Add tag aliases to related tags display	2025-06-10 04:05:22 +09:00
newtextdoc1111	5e4b4fae04	Merge pull request #20 from newtextdoc1111/dev Merge Dev	2025-05-29 18:07:45 +09:00
newtextdoc1111	7a73a5b1b3	fix: Bump version to 1.0.2 in pyproject.toml	2025-05-29 17:59:07 +09:00
newtextdoc1111	47635d5aea	docs: Update user CSV instructions with naming conventions and console log details	2025-05-29 17:52:09 +09:00
newtextdoc1111	4415e22744	fix: Log CSV file status in ComfyUI console and browser network tab	2025-05-29 17:51:50 +09:00