Merge pull request #2683 from eugencowie/tokenizer-slash-command

Add slash command to set tokenizer

public/scripts/slash-commands.js

@@ -52,7 +52,7 @@ import { chat_completion_sources, oai_settings, setupChatCompletionPromptManager
 import { autoSelectPersona, retriggerFirstMessageOnEmptyChat, setPersonaLockState, togglePersonaLock, user_avatar } from './personas.js';
 import { addEphemeralStoppingString, chat_styles, flushEphemeralStoppingStrings, power_user } from './power-user.js';
 import { SERVER_INPUTS, textgen_types, textgenerationwebui_settings } from './textgen-settings.js';
-import { decodeTextTokens, getFriendlyTokenizerName, getTextTokens, getTokenCountAsync } from './tokenizers.js';
+import { decodeTextTokens, getAvailableTokenizers, getFriendlyTokenizerName, getTextTokens, getTokenCountAsync, selectTokenizer } from './tokenizers.js';
 import { debounce, delay, isFalseBoolean, isTrueBoolean, showFontAwesomePicker, stringToRange, trimToEndSentence, trimToStartSentence, waitUntilCondition } from './utils.js';
 import { registerVariableCommands, resolveVariable } from './variables.js';
 import { background_settings } from './backgrounds.js';
@@ -1558,6 +1558,28 @@ export function initDefaultSlashCommands() {
             </div>
         `,
     }));
+    SlashCommandParser.addCommandObject(SlashCommand.fromProps({
+        name: 'tokenizer',
+        callback: selectTokenizerCallback,
+        returns: 'current tokenizer',
+        unnamedArgumentList: [
+            SlashCommandArgument.fromProps({
+                description: 'tokenizer name',
+                typeList: [ARGUMENT_TYPE.STRING],
+                enumList: getAvailableTokenizers().map(tokenizer =>
+                    new SlashCommandEnumValue(tokenizer.tokenizerKey, tokenizer.tokenizerName, enumTypes.enum, enumIcons.default)),
+            }),
+        ],
+        helpString: `
+            <div>
+                Selects tokenizer by name. Gets the current tokenizer if no name is provided.
+            </div>
+            <div>
+                <strong>Available tokenizers:</strong>
+                <pre><code>${getAvailableTokenizers().map(t => t.tokenizerKey).join(', ')}</code></pre>
+            </div>
+        `,
+    }));
 
     registerVariableCommands();
 }
@@ -3562,6 +3584,27 @@ async function setApiUrlCallback({ api = null, connect = 'true' }, url) {
     return textgenerationwebui_settings.server_urls[type] ?? '';
 }
 
+async function selectTokenizerCallback(_, name) {
+    if (!name) {
+        return getAvailableTokenizers().find(tokenizer => tokenizer.tokenizerId === power_user.tokenizer)?.tokenizerKey ?? '';
+    }
+
+    const tokenizers = getAvailableTokenizers();
+    const fuse = new Fuse(tokenizers, { keys: ['tokenizerKey', 'tokenizerName'] });
+    const result = fuse.search(name);
+
+    if (result.length === 0) {
+        toastr.warning(`Tokenizer "${name}" not found`);
+        return '';
+    }
+
+    /** @type {import('./tokenizers.js').Tokenizer} */
+    const foundTokenizer = result[0].item;
+    selectTokenizer(foundTokenizer.tokenizerId);
+
+    return foundTokenizer.tokenizerKey;
+}
+
 export let isExecutingCommandsFromChatInput = false;
 export let commandsFromChatInputAbortController;
 

public/scripts/tokenizers.js

@@ -147,10 +147,46 @@ async function resetTokenCache() {
     }
 }
 
+/**
+ * @typedef {object} Tokenizer
+ * @property {number} tokenizerId - The id of the tokenizer option
+ * @property {string} tokenizerKey - Internal name/key of the tokenizer
+ * @property {string} tokenizerName - Human-readable detailed name of the tokenizer (as displayed in the UI)
+ */
+
+/**
+ * Gets all tokenizers available to the user.
+ * @returns {Tokenizer[]} Tokenizer info.
+ */
+export function getAvailableTokenizers() {
+    const tokenizerOptions = $('#tokenizer').find('option').toArray();
+    return tokenizerOptions.map(tokenizerOption => ({
+        tokenizerId: Number(tokenizerOption.value),
+        tokenizerKey: Object.entries(tokenizers).find(([_, value]) => value === Number(tokenizerOption.value))[0].toLocaleLowerCase(),
+        tokenizerName: tokenizerOption.text,
+    }))
+}
+
+/**
+ * Selects tokenizer if not already selected.
+ * @param {number} tokenizerId Tokenizer ID.
+ */
+export function selectTokenizer(tokenizerId) {
+    if (tokenizerId !== power_user.tokenizer) {
+        const tokenizer = getAvailableTokenizers().find(tokenizer => tokenizer.tokenizerId === tokenizerId);
+        if (!tokenizer) {
+            console.warn('Failed to find tokenizer with id', tokenizerId);
+            return;
+        }
+        $('#tokenizer').val(tokenizer.tokenizerId).trigger('change');
+        toastr.info(`Tokenizer: "${tokenizer.tokenizerName}" selected`);
+    }
+}
+
 /**
  * Gets the friendly name of the current tokenizer.
  * @param {string} forApi API to get the tokenizer for. Defaults to the main API.
- * @returns { { tokenizerName: string, tokenizerId: number } } Tokenizer info
+ * @returns {Tokenizer} Tokenizer info
  */
 export function getFriendlyTokenizerName(forApi) {
     if (!forApi) {
@@ -185,7 +221,9 @@ export function getFriendlyTokenizerName(forApi) {
         ? tokenizers.OPENAI
         : tokenizerId;
 
-    return { tokenizerName, tokenizerId };
+    const tokenizerKey = Object.entries(tokenizers).find(([_, value]) => value === tokenizerId)[0].toLocaleLowerCase();
+
+    return { tokenizerName, tokenizerKey, tokenizerId };
 }
 
 /**