Merge branch 'staging' into wi-processing-refactoring

default/config.yaml

@@ -95,3 +95,8 @@ deepl:
   formality: default
 # -- SERVER PLUGIN CONFIGURATION --
 enableServerPlugins: false
+# User session timeout *in seconds* (defaults to 24 hours).
+## Set to a positive number to expire session after a certain time of inactivity
+## Set to 0 to expire session when the browser is closed
+## Set to a negative number to disable session expiration
+sessionTimeout: 86400

default/content/index.json

@@ -678,5 +678,13 @@
     {
         "filename": "presets/context/ChatML-Names.json",
         "type": "context"
+    },
+    {
+        "filename": "presets/context/Gemma 2.json",
+        "type": "context"
+    },
+    {
+        "filename": "presets/instruct/Gemma 2.json",
+        "type": "instruct"
     }
 ]

default/content/presets/context/Gemma 2.json

@@ -0,0 +1,12 @@
+{
+    "story_string": "<start_of_turn>user\n{{#if system}}{{system}}\n{{/if}}{{#if wiBefore}}{{wiBefore}}\n{{/if}}{{#if description}}{{description}}\n{{/if}}{{#if personality}}{{char}}'s personality: {{personality}}\n{{/if}}{{#if scenario}}Scenario: {{scenario}}\n{{/if}}{{#if wiAfter}}{{wiAfter}}\n{{/if}}{{#if persona}}{{persona}}\n{{/if}}{{trim}}<end_of_turn>",
+    "example_separator": "",
+    "chat_start": "",
+    "use_stop_strings": false,
+    "allow_jailbreak": false,
+    "always_force_name2": true,
+    "trim_sentences": false,
+    "include_newline": false,
+    "single_line": false,
+    "name": "Gemma 2"
+}

default/content/presets/instruct/Gemma 2.json

@@ -0,0 +1,24 @@
+{
+    "system_prompt": "Continue writing this story and portray characters realistically.",
+    "input_sequence": "<start_of_turn>user",
+    "output_sequence": "<start_of_turn>model",
+    "last_output_sequence": "",
+    "system_sequence": "<start_of_turn>system",
+    "stop_sequence": "<end_of_turn>",
+    "wrap": true,
+    "macro": true,
+    "names": false,
+    "names_force_groups": false,
+    "activation_regex": "",
+    "system_sequence_prefix": "",
+    "system_sequence_suffix": "",
+    "first_output_sequence": "",
+    "skip_examples": false,
+    "output_suffix": "<end_of_turn>\n",
+    "input_suffix": "<end_of_turn>\n",
+    "system_suffix": "<end_of_turn>\n",
+    "user_alignment_message": "",
+    "system_same_as_user": true,
+    "last_system_sequence": "",
+    "name": "Gemma 2"
+}

default/content/settings.json

@@ -13,6 +13,7 @@
         },
         "world_info_depth": 2,
         "world_info_budget": 25,
+        "world_info_include_names": true,
         "world_info_recursive": true,
         "world_info_overflow_alert": false,
         "world_info_case_sensitive": false,

public/index.html

@@ -3542,6 +3542,12 @@
                                             </div>
                                         </div>
                                         <div id="wiCheckboxes" class="flex1 flex-container flexFlowColumn">
+                                            <label title="Include names with each message into the context for scanning" data-i18n="[title]Include names with each message into the context for scanning" class="checkbox_label flex1">
+                                                <input id="world_info_include_names" type="checkbox" />
+                                                <small data-i18n="Include Names" class="whitespacenowrap flex1">
+                                                    Include Names
+                                                </small>
+                                            </label>
                                             <label title="Entries can activate other entries by mentioning their keywords" data-i18n="[title]Entries can activate other entries by mentioning their keywords" class="checkbox_label flex1">
                                                 <input id="world_info_recursive" type="checkbox" />
                                                 <small data-i18n="Recursive Scan" class="whitespacenowrap flex1">

public/locales/ja-jp.json

@@ -501,7 +501,7 @@
     "Active World(s) for all chats": "すべてのチャットのアクティブなワールド",
     "-- World Info not found --": "-- ワールド情報が見つかりません --",
     "Global World Info/Lorebook activation settings": "グローバルワールド情報/Lorebookの有効化設定",
-    "Click to expand": "クリックして拡大",
+    "Click to expand": "クリックして展開",
     "Scan Depth": "スキャンの深さ",
     "Context %": "コンテキスト％",
     "Budget Cap": "予算キャップ",
@@ -965,13 +965,13 @@
     "After EM": "↓EM",
     "Before AN": "著者のメモの前",
     "After AN": "著者のメモの後",
-    "at Depth System": "笑笑笑笑",
+    "at Depth System": "指定深度(システム)",
-    "at Depth User": "👤 こんばんは",
+    "at Depth User": "指定深度(ユーザー)",
-    "at Depth AI": "🤖 だめだ",
+    "at Depth AI": "指定深度(AI)",
     "Depth": "深さ",
     "Order:": "順序:",
     "Order": "順序：",
-    "Trigger %:": "引き金 ％：",
+    "Trigger %:": "発生 ％：",
     "Probability": "確率",
     "Duplicate world info entry": "重複したワールド情報エントリ",
     "Delete world info entry": "世界情報エントリを削除",
@@ -981,10 +981,10 @@
     "Comma separated list": "カンマ区切りリスト",
     "Switch to plaintext mode": "プレーンテキストモードに切り替える",
     "Logic": "論理",
-    "AND ANY": "任意のAND",
+    "AND ANY": "AND ANY",
-    "AND ALL": "すべてのAND",
+    "AND ALL": "AND ALL",
-    "NOT ALL": "すべてのNOT",
+    "NOT ALL": "NOT ALL",
-    "NOT ANY": "どれかのNOT",
+    "NOT ANY": "NOT ANY",
     "(ignored if empty)": "(空の場合は無視されます)",
     "Optional Filter": "オプションフィルタ",
     "Keywords or Regexes (ignored if empty)": "キーワードまたは正規表現（空の場合は無視されます）",
@@ -1004,7 +1004,7 @@
     "Filter to Character(s)": "キャラクターにフィルター",
     "Character Exclusion": "キャラクターの除外",
     "-- Characters not found --": "-- キャラクターが見つかりません --",
-    "Inclusion Group": "含蓋グループ",
+    "Inclusion Group": "包含グループ",
     "Inclusion Groups ensure only one entry from a group is activated at a time, if multiple are triggered.Documentation: World Info - Inclusion Group": "包含グループにより、複数のエントリがトリガーされた場合、一度に 1 つのグループから 1 つのエントリのみがアクティブ化されます。カンマで区切られた複数のグループをサポートします。ドキュメント: World Info - 包含グループ",
     "Prioritize this entry: When checked, this entry is prioritized out of all selections.If multiple are prioritized, the one with the highest 'Order' is chosen.": "このエントリを優先する: チェックすると、このエントリがすべての選択項目の中で優先されます。複数のエントリを優先する場合は、「順序」が最も高いエントリが選択されます。",
     "Only one entry with the same label will be activated": "同じラベルのエントリが1つだけ有効になります",
@@ -1014,7 +1014,7 @@
     "Use Probability": "確率を使用",
     "Add Memo": "メモを追加",
     "Text or token ids": "テキストまたは[トークンID]",
-    "close": "近い",
+    "close": "閉じる",
     "prompt_manager_edit": "編集",
     "prompt_manager_name": "名前",
     "A name for this prompt.": "このプロンプトの名前。",

public/script.js

@@ -35,6 +35,7 @@ import {
     setWorldInfoButtonClass,
     importWorldInfo,
     wi_anchor_position,
+    world_info_include_names,
 } from './scripts/world-info.js';
 
 import {
@@ -3500,7 +3501,7 @@ export async function Generate(type, { automatic_trigger, force_name2, quiet_pro
     // Add WI to prompt (and also inject WI to AN value via hijack)
     // Make quiet prompt available for WIAN
     setExtensionPrompt('QUIET_PROMPT', quiet_prompt || '', extension_prompt_types.IN_PROMPT, 0, true);
-    const chatForWI = coreChat.map(x => `${x.name}: ${x.mes}`).reverse();
+    const chatForWI = coreChat.map(x => world_info_include_names ? `${x.name}: ${x.mes}` : x.mes).reverse();
     const { worldInfoString, worldInfoBefore, worldInfoAfter, worldInfoExamples, worldInfoDepth } = await getWorldInfoPrompt(chatForWI, this_max_context, dryRun);
     setExtensionPrompt('QUIET_PROMPT', '', extension_prompt_types.IN_PROMPT, 0, true);
 

public/scripts/extensions/stable-diffusion/index.js

@@ -30,7 +30,7 @@ import { SlashCommand } from '../../slash-commands/SlashCommand.js';
 import { ARGUMENT_TYPE, SlashCommandArgument, SlashCommandNamedArgument } from '../../slash-commands/SlashCommandArgument.js';
 import { debounce_timeout } from '../../constants.js';
 import { SlashCommandEnumValue } from '../../slash-commands/SlashCommandEnumValue.js';
-import { POPUP_TYPE, callGenericPopup } from '../../popup.js';
+import { POPUP_TYPE, Popup, callGenericPopup } from '../../popup.js';
 export { MODULE_NAME };
 
 const MODULE_NAME = 'sd';
@@ -3150,7 +3150,12 @@ async function onComfyOpenWorkflowEditorClick() {
         }),
     })).json();
     const editorHtml = $(await $.get('scripts/extensions/stable-diffusion/comfyWorkflowEditor.html'));
-    const popupResult = callGenericPopup(editorHtml, POPUP_TYPE.CONFIRM, '', { okButton: 'Save', cancelButton: 'Cancel', wide: true, large: true });
+    const saveValue = (/** @type {Popup} */ _popup) => {
+        workflow = $('#sd_comfy_workflow_editor_workflow').val().toString();
+        return true;
+    };
+    const popup = new Popup(editorHtml, POPUP_TYPE.CONFIRM, '', { okButton: 'Save', cancelButton: 'Cancel', wide: true, large: true, onClosing: saveValue });
+    const popupResult = popup.show();
     const checkPlaceholders = () => {
         workflow = $('#sd_comfy_workflow_editor_workflow').val().toString();
         $('.sd_comfy_workflow_editor_placeholder_list > li[data-placeholder]').each(function (idx) {
@@ -3219,7 +3224,7 @@ async function onComfyOpenWorkflowEditorClick() {
             headers: getRequestHeaders(),
             body: JSON.stringify({
                 file_name: extension_settings.sd.comfy_workflow,
-                workflow: $('#sd_comfy_workflow_editor_workflow').val().toString(),
+                workflow: workflow,
             }),
         });
         if (!response.ok) {

public/scripts/tokenizers.js

@@ -7,7 +7,7 @@ import { kai_flags } from './kai-settings.js';
 import { textgen_types, textgenerationwebui_settings as textgen_settings, getTextGenServer, getTextGenModel } from './textgen-settings.js';
 import { getCurrentDreamGenModelTokenizer, getCurrentOpenRouterModelTokenizer, openRouterModels } from './textgen-models.js';
 
-const { OOBA, TABBY, KOBOLDCPP, APHRODITE, LLAMACPP, OPENROUTER, DREAMGEN } = textgen_types;
+const { OOBA, TABBY, KOBOLDCPP, VLLM, APHRODITE, LLAMACPP, OPENROUTER, DREAMGEN } = textgen_types;
 
 export const CHARACTERS_PER_TOKEN_RATIO = 3.35;
 const TOKENIZER_WARNING_KEY = 'tokenizationWarningShown';
@@ -39,7 +39,7 @@ export const SENTENCEPIECE_TOKENIZERS = [
     //tokenizers.NERD2,
 ];
 
-export const TEXTGEN_TOKENIZERS = [OOBA, TABBY, KOBOLDCPP, LLAMACPP, APHRODITE];
+export const TEXTGEN_TOKENIZERS = [OOBA, TABBY, KOBOLDCPP, LLAMACPP, VLLM, APHRODITE];
 
 const TOKENIZER_URLS = {
     [tokenizers.GPT2]: {
@@ -769,6 +769,7 @@ function getTextgenAPITokenizationParams(str) {
         api_type: textgen_settings.type,
         url: getTextGenServer(),
         legacy_api: textgen_settings.legacy_api && (textgen_settings.type === OOBA || textgen_settings.type === APHRODITE),
+        vllm_model: textgen_settings.vllm_model,
     };
 }
 

public/scripts/world-info.js

@@ -24,6 +24,7 @@ export {
     world_info_depth,
     world_info_min_activations,
     world_info_min_activations_depth_max,
+    world_info_include_names,
     world_info_recursive,
     world_info_overflow_alert,
     world_info_case_sensitive,
@@ -83,6 +84,7 @@ let world_info_min_activations = 0; // if > 0, will continue seeking chat until
 let world_info_min_activations_depth_max = 0; // used when (world_info_min_activations > 0)
 
 let world_info_budget = 25;
+let world_info_include_names = true;
 let world_info_recursive = false;
 let world_info_overflow_alert = false;
 let world_info_case_sensitive = false;
@@ -718,6 +720,7 @@ export function getWorldInfoSettings() {
         world_info_min_activations,
         world_info_min_activations_depth_max,
         world_info_budget,
+        world_info_include_names,
         world_info_recursive,
         world_info_overflow_alert,
         world_info_case_sensitive,
@@ -747,7 +750,7 @@ const worldInfoCache = new Map();
 
 /**
  * Gets the world info based on chat messages.
- * @param {string[]} chat The chat messages to scan.
+ * @param {string[]} chat The chat messages to scan, in reverse order.
  * @param {number} maxContext The maximum context size of the generation.
  * @param {boolean} isDryRun If true, the function will not emit any events.
  * @typedef {{worldInfoString: string, worldInfoBefore: string, worldInfoAfter: string, worldInfoExamples: any[], worldInfoDepth: any[]}} WIPromptResult
@@ -784,6 +787,8 @@ function setWorldInfoSettings(settings, data) {
         world_info_min_activations_depth_max = Number(settings.world_info_min_activations_depth_max);
     if (settings.world_info_budget !== undefined)
         world_info_budget = Number(settings.world_info_budget);
+    if (settings.world_info_include_names !== undefined)
+        world_info_include_names = Boolean(settings.world_info_include_names);
     if (settings.world_info_recursive !== undefined)
         world_info_recursive = Boolean(settings.world_info_recursive);
     if (settings.world_info_overflow_alert !== undefined)
@@ -833,6 +838,7 @@ function setWorldInfoSettings(settings, data) {
     $('#world_info_budget_counter').val(world_info_budget);
     $('#world_info_budget').val(world_info_budget);
 
+    $('#world_info_include_names').prop('checked', world_info_include_names);
     $('#world_info_recursive').prop('checked', world_info_recursive);
     $('#world_info_overflow_alert').prop('checked', world_info_overflow_alert);
     $('#world_info_case_sensitive').prop('checked', world_info_case_sensitive);
@@ -3542,7 +3548,7 @@ export async function getSortedEntries() {
 
 /**
  * Performs a scan on the chat and returns the world info activated.
- * @param {string[]} chat The chat messages to scan.
+ * @param {string[]} chat The chat messages to scan, in reverse order.
  * @param {number} maxContext The maximum context size of the generation.
  * @param {boolean} isDryRun Whether to perform a dry run.
  * @typedef {{ worldInfoBefore: string, worldInfoAfter: string, EMEntries: any[], WIDepthEntries: any[], allActivatedEntries: Set<any> }} WIActivated
@@ -4637,6 +4643,11 @@ jQuery(() => {
         saveSettings();
     });
 
+    $('#world_info_include_names').on('input', function () {
+        world_info_include_names = !!$(this).prop('checked');
+        saveSettings();
+    });
+
     $('#world_info_recursive').on('input', function () {
         world_info_recursive = !!$(this).prop('checked');
         saveSettings();

server.js

@@ -200,11 +200,30 @@ if (enableCorsProxy) {
     });
 }
 
+function getSessionCookieAge() {
+    // Defaults to 24 hours in seconds if not set
+    const configValue = getConfigValue('sessionTimeout', 24 * 60 * 60);
+
+    // Convert to milliseconds
+    if (configValue > 0) {
+        return configValue * 1000;
+    }
+
+    // "No expiration" is just 400 days as per RFC 6265
+    if (configValue < 0) {
+        return 400 * 24 * 60 * 60 * 1000;
+    }
+
+    // 0 means session cookie is deleted when the browser session ends
+    // (depends on the implementation of the browser)
+    return undefined;
+}
+
 app.use(cookieSession({
     name: userModule.getCookieSessionName(),
     sameSite: 'strict',
     httpOnly: true,
-    maxAge: 24 * 60 * 60 * 1000, // 24 hours
+    maxAge: getSessionCookieAge(),
     secret: userModule.getCookieSecret(),
 }));
 

src/endpoints/tokenizers.js

@@ -784,6 +784,7 @@ router.post('/remote/textgenerationwebui/encode', jsonParser, async function (re
     const text = String(request.body.text) || '';
     const baseUrl = String(request.body.url);
     const legacyApi = Boolean(request.body.legacy_api);
+    const vllmModel = String(request.body.vllm_model) || '';
 
     try {
         const args = {
@@ -814,7 +815,9 @@ router.post('/remote/textgenerationwebui/encode', jsonParser, async function (re
                     args.body = JSON.stringify({ 'content': text });
                     break;
                 case TEXTGEN_TYPES.VLLM:
-                    return response.send({ error: true });
+                    url += '/tokenize';
+                    args.body = JSON.stringify({ 'model': vllmModel, 'prompt': text });
+                    break;
                 case TEXTGEN_TYPES.APHRODITE:
                     url += '/v1/tokenize';
                     args.body = JSON.stringify({ 'prompt': text });
@@ -834,7 +837,7 @@ router.post('/remote/textgenerationwebui/encode', jsonParser, async function (re
         }
 
         const data = await result.json();
-        const count = legacyApi ? data?.results[0]?.tokens : (data?.length ?? data?.value ?? data?.tokens?.length);
+        const count = legacyApi ? data?.results[0]?.tokens : (data?.length ?? data?.count ?? data?.value ?? data?.tokens?.length);
         const ids = legacyApi ? [] : (data?.tokens ?? data?.ids ?? []);
 
         return response.send({ count, ids });