diff --git a/public/scripts/extensions/vectors/index.js b/public/scripts/extensions/vectors/index.js index 36b910f06..30278ddea 100644 --- a/public/scripts/extensions/vectors/index.js +++ b/public/scripts/extensions/vectors/index.js @@ -30,6 +30,8 @@ import { textgen_types, textgenerationwebui_settings } from '../../textgen-setti import { SlashCommandParser } from '../../slash-commands/SlashCommandParser.js'; import { SlashCommand } from '../../slash-commands/SlashCommand.js'; import { ARGUMENT_TYPE, SlashCommandArgument, SlashCommandNamedArgument } from '../../slash-commands/SlashCommandArgument.js'; +import { SlashCommandEnumValue, enumTypes } from '../../slash-commands/SlashCommandEnumValue.js'; +import { slashCommandReturnHelper } from '../../slash-commands/SlashCommandReturnHelper.js'; import { callGenericPopup, POPUP_RESULT, POPUP_TYPE } from '../../popup.js'; import { generateWebLlmChatPrompt, isWebLlmSupported } from '../shared.js'; @@ -1613,25 +1615,55 @@ jQuery(async () => { callback: async (args, query) => { const clamp = (v) => Number.isNaN(v) ? null : Math.min(1, Math.max(0, v)); const threshold = clamp(Number(args?.threshold ?? settings.score_threshold)); + const validateCount = (v) => Number.isNaN(v) || !Number.isInteger(v) || v < 1 ? null : v; + const count = validateCount(Number(args?.count)) ?? settings.chunk_count_db; const source = String(args?.source ?? ''); const attachments = source ? getDataBankAttachmentsForSource(source, false) : getDataBankAttachments(false); const collectionIds = await ingestDataBankAttachments(String(source)); - const queryResults = await queryMultipleCollections(collectionIds, String(query), settings.chunk_count_db, threshold); - - // Map collection IDs to file URLs + const queryResults = await queryMultipleCollections(collectionIds, String(query), count, threshold); + + // Get URLs const urls = Object .keys(queryResults) .map(x => attachments.find(y => getFileCollectionId(y.url) === x)) .filter(x => x) .map(x => x.url); + + // Gets the actual text content of chunks + const getChunksText = () => { + let textResult = ''; + for (const collectionId in queryResults) { + const metadata = queryResults[collectionId].metadata?.filter(x => x.text)?.sort((a, b) => a.index - b.index)?.map(x => x.text)?.filter(onlyUnique) || []; + textResult += metadata.join('\n') + '\n\n'; + } + return textResult; + }; + + if (args.return === 'chunks') { + return getChunksText(); + } - return JSON.stringify(urls); + // @ts-ignore + return slashCommandReturnHelper.doReturn(args.return ?? 'object', urls, { objectToStringFunc: list => list.join('\n') }); + }, aliases: ['databank-search', 'data-bank-search'], helpString: 'Search the Data Bank for a specific query using vector similarity. Returns a list of file URLs with the most relevant content.', namedArgumentList: [ new SlashCommandNamedArgument('threshold', 'Threshold for the similarity score in the [0, 1] range. Uses the global config value if not set.', ARGUMENT_TYPE.NUMBER, false, false, ''), + new SlashCommandNamedArgument('count', 'Maximum number of query results to return.', ARGUMENT_TYPE.NUMBER, false, false, ''), new SlashCommandNamedArgument('source', 'Optional filter for the attachments by source.', ARGUMENT_TYPE.STRING, false, false, '', ['global', 'character', 'chat']), + SlashCommandNamedArgument.fromProps({ + name: 'return', + description: 'How you want the return value to be provided', + typeList: [ARGUMENT_TYPE.STRING], + defaultValue: 'object', + enumList: [ + new SlashCommandEnumValue('chunks', 'Return the actual content chunks', enumTypes.enum, '{}'), + ...slashCommandReturnHelper.enumList({ allowObject: true }) + ], + forceEnum: true, + }) ], unnamedArgumentList: [ new SlashCommandArgument('Query to search by.', ARGUMENT_TYPE.STRING, true, false),