diff options
author | A.J. Shulman <Shulman.aj@gmail.com> | 2024-09-19 09:53:19 -0400 |
---|---|---|
committer | A.J. Shulman <Shulman.aj@gmail.com> | 2024-09-19 09:53:19 -0400 |
commit | 00b12111c4007ef6ecff645327007a67f6655d8b (patch) | |
tree | b5ecfd18a7a7e5f2cd97b8307b248882d056c7e0 /src/server/ApiManagers/AssistantManager.ts | |
parent | ba0520baaa1f84d9fb08d3b2880c68302d28350a (diff) |
added comments and fixed some styling and error handling
Diffstat (limited to 'src/server/ApiManagers/AssistantManager.ts')
-rw-r--r-- | src/server/ApiManagers/AssistantManager.ts | 224 |
1 files changed, 161 insertions, 63 deletions
diff --git a/src/server/ApiManagers/AssistantManager.ts b/src/server/ApiManagers/AssistantManager.ts index 88a81c5de..063ba193a 100644 --- a/src/server/ApiManagers/AssistantManager.ts +++ b/src/server/ApiManagers/AssistantManager.ts @@ -19,6 +19,7 @@ import * as puppeteer from 'puppeteer'; import { JSDOM } from 'jsdom'; import { Readability } from '@mozilla/readability'; +// Enumeration of directories where different file types are stored export enum Directory { parsed_files = 'parsed_files', images = 'images', @@ -32,41 +33,76 @@ export enum Directory { scrape_images = 'scrape_images', } +/** + * Constructs a normalized path to a file in the server's file system. + * @param directory The directory where the file is stored. + * @param filename The name of the file. + * @returns The full normalized path to the file. + */ export function serverPathToFile(directory: Directory, filename: string) { return path.normalize(`${filesDirectory}/${directory}/${filename}`); } +/** + * Constructs a normalized path to a directory in the server's file system. + * @param directory The directory to access. + * @returns The full normalized path to the directory. + */ export function pathToDirectory(directory: Directory) { return path.normalize(`${filesDirectory}/${directory}`); } +/** + * Constructs the client-accessible URL for a file. + * @param directory The directory where the file is stored. + * @param filename The name of the file. + * @returns The URL path to the file. + */ export function clientPathToFile(directory: Directory, filename: string) { return `/files/${directory}/${filename}`; } +// Promisified versions of filesystem functions const writeFileAsync = promisify(writeFile); const readFileAsync = promisify(fs.readFile); +/** + * Class responsible for handling various API routes related to the Assistant functionality. + * This class extends `ApiManager` and handles registration of routes and secure request handlers. + */ export default class AssistantManager extends ApiManager { + /** + * Registers all API routes and initializes necessary services like OpenAI, Scrapfly, and UnstructuredClient. + * @param register The registration method to register routes and handlers. + */ protected initialize(register: Registration): void { + // Initialize OpenAI API with client key const openai = new OpenAI({ - apiKey: process.env._CLIENT_OPENAI_KEY, // Use client key so don't have to set key seperately for client and server. + apiKey: process.env._CLIENT_OPENAI_KEY, 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'); + // Register Wikipedia summary API route register({ method: Method.POST, subscription: '/getWikipediaSummary', secureHandler: async ({ req, res }) => { const { title } = req.body; try { + // Fetch summary from Wikipedia using axios const response = await axios.get('https://en.wikipedia.org/w/api.php', { params: { action: 'query', @@ -75,25 +111,26 @@ export default class AssistantManager extends ApiManager { format: 'json', }, }); - const summary = response.data.query.search[0].snippet; - if (!summary || summary.length === 0 || summary === '' || summary === ' ') { - res.send({ text: 'No article found with that title.' }); - } else { - res.send({ text: summary }); - } + const summary = response.data.query.search[0]?.snippet || 'No article found with that title.'; + res.send({ text: summary }); } catch (error: any) { - console.error('Error retrieving article summary from Wikipedia:', error); - res.status(500).send({ error: 'Error retrieving article summary from Wikipedia.', details: error.message }); + console.error('Error retrieving Wikipedia summary:', error); + res.status(500).send({ + error: 'Error retrieving article summary from Wikipedia.', + details: error.message, + }); } }, }); + // Register Google Web Search Results API route register({ method: Method.POST, subscription: '/getWebSearchResults', secureHandler: async ({ req, res }) => { const { query, max_results } = req.body; try { + // Fetch search results using Google Custom Search API const response = await customsearch.cse.list({ q: query, cx: process.env._CLIENT_GOOGLE_SEARCH_ENGINE_ID, @@ -111,25 +148,40 @@ export default class AssistantManager extends ApiManager { res.send({ results }); } catch (error: any) { console.error('Error performing web search:', error); - res.status(500).send({ error: 'Failed to perform web search', details: error.message }); + res.status(500).send({ + error: 'Failed to perform web search', + details: error.message, + }); } }, }); + // Axios instance with custom headers for scraping const axiosInstance = axios.create({ headers: { 'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36', }, }); + /** + * Utility function to introduce delay (used for retries). + * @param ms Delay in milliseconds. + */ const delay = (ms: number) => new Promise(resolve => setTimeout(resolve, ms)); + /** + * Function to fetch a URL with retry logic, handling rate limits. + * Retries a request if it fails due to rate limits (HTTP status 429). + * @param url The URL to fetch. + * @param retries The number of retry attempts. + * @param backoff Initial backoff time in milliseconds. + */ const fetchWithRetry = async (url: string, retries = 3, backoff = 300) => { try { const response = await axiosInstance.get(url); return response.data; } catch (error: any) { - if (retries > 0 && error.response && error.response.status === 429) { + if (retries > 0 && error.response?.status === 429) { console.log(`Rate limited. Retrying in ${backoff}ms...`); await delay(backoff); return fetchWithRetry(url, retries - 1, backoff * 2); @@ -138,6 +190,7 @@ export default class AssistantManager extends ApiManager { } }; + // Register a proxy fetch API route register({ method: Method.POST, subscription: '/proxyFetch', @@ -154,18 +207,22 @@ export default class AssistantManager extends ApiManager { res.send({ data }); } catch (error: any) { console.error('Error fetching the URL:', error); - res.status(500).send({ error: 'Failed to fetch the URL', details: error.message }); + res.status(500).send({ + error: 'Failed to fetch the URL', + details: error.message, + }); } }, }); + // Register an API route to scrape website content using Puppeteer and JSDOM register({ method: Method.POST, subscription: '/scrapeWebsite', secureHandler: async ({ req, res }) => { const { url } = req.body; try { - // Launch Puppeteer to navigate to the webpage + // Launch Puppeteer browser to navigate to the webpage const browser = await puppeteer.launch({ args: ['--no-sandbox', '--disable-setuid-sandbox'], }); @@ -173,45 +230,47 @@ export default class AssistantManager extends ApiManager { await page.setUserAgent('Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36'); await page.goto(url, { waitUntil: 'networkidle2' }); - // Get the HTML content of the page + // Extract HTML content const htmlContent = await page.content(); await browser.close(); - // Use JSDOM to parse the HTML content + // Parse HTML content using JSDOM const dom = new JSDOM(htmlContent, { url }); - // Use Readability to extract the readable content + // Extract readable content using Mozilla's Readability API const reader = new Readability(dom.window.document); const article = reader.parse(); if (article) { - // Extract the plain text from the article content const plainText = article.textContent; - - // Return the plain text content res.send({ website_plain_text: plainText }); } else { res.status(500).send({ error: 'Failed to extract readable content' }); } } catch (error: any) { console.error('Error scraping website:', error); - res.status(500).send({ error: 'Failed to scrape website', details: error.message }); + res.status(500).send({ + error: 'Failed to scrape website', + details: error.message, + }); } }, }); + // Register an API route to create documents by sending files to a chatbot register({ method: Method.POST, subscription: '/createDocument', secureHandler: async ({ req, res }) => { const { file_path } = req.body; - const public_path = path.join(publicDirectory, file_path); - const file_name = path.basename(file_path); + const public_path = path.join(publicDirectory, file_path); // Resolve the file path in the public directory + const file_name = path.basename(file_path); // Extract the file name from the path try { - // Read file data and convert to base64 + // Read the file data and encode it as base64 const file_data: string = fs.readFileSync(public_path, { encoding: 'base64' }); + // Send the file data to a local chatbot API for document creation const response = await axios.post( 'http://localhost:8080/createDocument', { @@ -225,65 +284,79 @@ export default class AssistantManager extends ApiManager { } ); + // Retrieve the job ID from the response const jobId = response.data['job_id']; console.log('Job ID:', jobId); + // Send the job ID back to the client res.send({ jobId }); } catch (error: any) { console.error('Error communicating with chatbot:', error); - res.status(500).send({ error: 'Failed to communicate with the chatbot', details: error.message }); + res.status(500).send({ + error: 'Failed to communicate with the chatbot', + details: error.message, + }); } }, }); + // Register an API route to check the progress of a document creation job register({ method: Method.GET, subscription: '/getProgress/:jobId', secureHandler: async ({ req, res }) => { - const { jobId } = req.params; + const { jobId } = req.params; // Get the job ID from the URL parameters try { + // Query the local API to get the progress of the job const progressResponse = await axios.get(`http://localhost:8080/getProgress/${jobId}`); console.log(`Current step: ${progressResponse.data.step}, Progress within step: ${progressResponse.data.progress}%`); - res.json(progressResponse.data); + res.json(progressResponse.data); // Send the progress data back to the client } catch (error) { console.error('Error getting progress:', error); - res.status(500).send({ error: 'Failed to get progress', details: JSON.parse(error as string).message }); + res.status(500).send({ + error: 'Failed to get progress', + details: error, + }); } }, }); + // Register an API route to get the final result of a document creation job register({ method: Method.GET, subscription: '/getResult/:jobId', secureHandler: async ({ req, res }) => { - const { jobId } = req.params; + const { jobId } = req.params; // Get the job ID from the URL parameters try { + // Query the local API to get the final result of the job const finalResponse = await axios.get(`http://localhost:8080/getResult/${jobId}`); console.log('Result:', finalResponse.data); const result = finalResponse.data; + // If the result contains image or table chunks, save the base64 data as image files if (result.chunks && Array.isArray(result.chunks)) { for (const chunk of result.chunks) { if (chunk.metadata && (chunk.metadata.type === 'image' || chunk.metadata.type === 'table')) { let files_directory = '/files/chunk_images/'; const directory = path.join(publicDirectory, files_directory); + // Ensure the directory exists or create it if (!fs.existsSync(directory)) { fs.mkdirSync(directory); } - const fileName = path.basename(chunk.metadata.file_path); - const filePath = path.join(directory, fileName); + const fileName = path.basename(chunk.metadata.file_path); // Get the file name from the path + const filePath = path.join(directory, fileName); // Create the full file path - // Check if base64_data exists + // Check if the chunk contains base64 encoded data if (chunk.metadata.base64_data) { - // Decode Base64 and save as file + // Decode the base64 data and write it to a file const buffer = Buffer.from(chunk.metadata.base64_data, 'base64'); await fs.promises.writeFile(filePath, buffer); - // Update the file path in the chunk + // Update the file path in the chunk's metadata chunk.metadata.file_path = path.join(files_directory, fileName); - chunk.metadata.base64_data = undefined; + chunk.metadata.base64_data = undefined; // Remove the base64 data from the metadata } else { console.warn(`No base64_data found for chunk: ${fileName}`); } @@ -294,32 +367,42 @@ export default class AssistantManager extends ApiManager { console.warn('Not ready'); result.status = 'pending'; } - res.json(result); + res.json(result); // Send the result back to the client } catch (error) { - console.error('Error getting progress:', error); - res.status(500).send({ error: 'Failed to get progress', details: error }); + console.error('Error getting result:', error); + res.status(500).send({ + error: 'Failed to get result', + details: error, + }); } }, }); + // Register an API route to format chunks (e.g., text or image chunks) for display register({ method: Method.POST, subscription: '/formatChunks', secureHandler: async ({ req, res }) => { - const { relevantChunks } = req.body; + const { relevantChunks } = req.body; // Get the relevant chunks from the request body + + // Initialize an array to hold the formatted content const content: { type: string; text?: string; image_url?: { url: string } }[] = [{ type: 'text', text: '<chunks>' }]; for (const chunk of relevantChunks) { + // Format each chunk by adding its metadata and content content.push({ type: 'text', - text: `<chunk chunk_id=${chunk.id} chunk_type=${chunk.metadata.type}>`, + text: `<chunk chunk_id=${chunk.id} chunk_type="${chunk.metadata.type}">`, }); + // If the chunk is an image or table, read the corresponding file and encode it as base64 if (chunk.metadata.type === 'image' || chunk.metadata.type === 'table') { try { - const filePath = serverPathToFile(Directory.chunk_images, chunk.metadata.file_path); - const imageBuffer = await readFileAsync(filePath); - const base64Image = imageBuffer.toString('base64'); + const filePath = serverPathToFile(Directory.chunk_images, chunk.metadata.file_path); // Get the file path + const imageBuffer = await readFileAsync(filePath); // Read the image file + const base64Image = imageBuffer.toString('base64'); // Convert the image to base64 + + // Add the base64-encoded image to the content array if (base64Image) { content.push({ type: 'image_url', @@ -334,33 +417,35 @@ export default class AssistantManager extends ApiManager { console.error(`Error reading image file for chunk ${chunk.id}:`, error); } } + + // Add the chunk's text content to the formatted content content.push({ type: 'text', text: `${chunk.metadata.text}\n</chunk>\n` }); } content.push({ type: 'text', text: '</chunks>' }); + // Send the formatted content back to the client res.send({ formattedChunks: content }); }, }); + // Register an API route to create and save a CSV file on the server register({ method: Method.POST, subscription: '/createCSV', secureHandler: async ({ req, res }) => { const { filename, data } = req.body; - // Validate input + // Validate that both the filename and data are provided if (!filename || !data) { res.status(400).send({ error: 'Filename and data fields are required.' }); return; } try { - // Generate a UUID for the file + // Generate a UUID for the file to ensure unique naming const uuidv4 = uuid.v4(); - - // Construct the full filename with the UUID prefix - const fullFilename = `${uuidv4}-${filename}`; + const fullFilename = `${uuidv4}-${filename}`; // Prefix the file name with the UUID // Get the full server path where the file will be saved const serverFilePath = serverPathToFile(Directory.csv, fullFilename); @@ -368,60 +453,73 @@ export default class AssistantManager extends ApiManager { // Write the CSV data (which is a raw string) to the file await writeFileAsync(serverFilePath, data, 'utf8'); - // Construct the full client URL for accessing the file + // Construct the client-accessible URL for the file const fileUrl = clientPathToFile(Directory.csv, fullFilename); - // Return the file URL and UUID to the client + // Send the file URL and UUID back to the client res.send({ fileUrl, id: uuidv4 }); } catch (error: any) { console.error('Error creating CSV file:', error); - res.status(500).send({ error: 'Failed to create CSV file.', details: error.message }); + res.status(500).send({ + error: 'Failed to create CSV file.', + details: error.message, + }); } }, }); + // 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; - const public_path = path.join(publicDirectory, file_path); - const file_name = path.basename(file_path); + 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 file data and convert to base64 + // 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, - chunkingStrategy: ChunkingStrategy.ByTitle, - extractImageBlockTypes: ['Image', 'Table'], + 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); - // Print the processed data. - console.log(jsonElements); - res.send({ document_json: jsonElements }); + 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 }); + 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 }); + 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 }); + res.status(500).send({ + error: 'Failed to read the file', + details: error.message, + }); } }, }); |