diff options
author | A.J. Shulman <Shulman.aj@gmail.com> | 2024-09-19 09:58:26 -0400 |
---|---|---|
committer | A.J. Shulman <Shulman.aj@gmail.com> | 2024-09-19 09:58:26 -0400 |
commit | 2e67d85a1e7921eeb6487d01f7a36f69d725ab84 (patch) | |
tree | 7392807bc524833e149174346d5d7ec5af9a1b70 /src/server/ApiManagers/AssistantManager.ts | |
parent | 5441df3cb1ff5986f4e2e871dd3d2c7a312a2d21 (diff) |
got rid of unnecessary endpoint on server (chunkDocument)
Diffstat (limited to 'src/server/ApiManagers/AssistantManager.ts')
-rw-r--r-- | src/server/ApiManagers/AssistantManager.ts | 69 |
1 files changed, 1 insertions, 68 deletions
diff --git a/src/server/ApiManagers/AssistantManager.ts b/src/server/ApiManagers/AssistantManager.ts index 063ba193a..d98d32d30 100644 --- a/src/server/ApiManagers/AssistantManager.ts +++ b/src/server/ApiManagers/AssistantManager.ts @@ -13,7 +13,6 @@ import { UnstructuredClient } from 'unstructured-client'; import { PartitionResponse } from 'unstructured-client/sdk/models/operations'; import { ChunkingStrategy, Strategy } from 'unstructured-client/sdk/models/shared'; import * as cheerio from 'cheerio'; -import { ScrapflyClient, ScrapeConfig } from 'scrapfly-sdk'; import { google } from 'googleapis'; import * as puppeteer from 'puppeteer'; import { JSDOM } from 'jsdom'; @@ -72,7 +71,7 @@ const readFileAsync = promisify(fs.readFile); */ export default class AssistantManager extends ApiManager { /** - * Registers all API routes and initializes necessary services like OpenAI, Scrapfly, and UnstructuredClient. + * Registers all API routes and initializes necessary services like OpenAI and Google Custom Search. * @param register The registration method to register routes and handlers. */ protected initialize(register: Registration): void { @@ -82,16 +81,6 @@ export default class AssistantManager extends ApiManager { dangerouslyAllowBrowser: true, }); - // Initialize UnstructuredClient for document processing - const unstructuredClient = new UnstructuredClient({ - security: { - apiKeyAuth: process.env._CLIENT_UNSTRUCTURED_API_KEY!, - }, - }); - - // Initialize ScrapflyClient for scraping purposes - const scrapflyClient = new ScrapflyClient({ key: process.env._CLIENT_SCRAPFLY_API_KEY! }); - // Initialize Google Custom Search API const customsearch = google.customsearch('v1'); @@ -467,61 +456,5 @@ export default class AssistantManager extends ApiManager { } }, }); - - // Register an API route to chunk a document using the UnstructuredClient - register({ - method: Method.POST, - subscription: '/chunkDocument', - secureHandler: async ({ req, res }) => { - const { file_path } = req.body; // Get the file path from the request body - const public_path = path.join(publicDirectory, file_path); // Resolve the full path in the public directory - const file_name = path.basename(file_path); // Extract the file name from the path - - try { - // Read the file content as a Buffer - const file_data = await fs.promises.readFile(public_path); - - try { - // Use UnstructuredClient to partition the document into chunks - const result = await unstructuredClient.general.partition({ - partitionParameters: { - files: { - content: file_data, - fileName: file_name, - }, - strategy: Strategy.Auto, // Automatically determine the chunking strategy - chunkingStrategy: ChunkingStrategy.ByTitle, // Chunk by title - extractImageBlockTypes: ['Image', 'Table'], // Extract images and tables - }, - }); - - if (result.statusCode === 200) { - console.log(result.elements); - const jsonElements = JSON.stringify(result.elements, null, 2); - console.log(jsonElements); // Log the JSON result of the partitioned elements - res.send({ document_json: jsonElements }); // Send the partitioned data as a JSON response - } else { - console.error(`Unexpected status code: ${result.statusCode}`); - res.status(result.statusCode).send({ - error: 'Failed to process the document', - details: result, - }); - } - } catch (e: any) { - console.error('Error during partitioning:', e); - res.status(500).send({ - error: 'Failed to partition the document', - details: e.message, - }); - } - } catch (error: any) { - console.error('Error reading file:', error); - res.status(500).send({ - error: 'Failed to read the file', - details: error.message, - }); - } - }, - }); } } |