Demos 1171 UI path

spikes/uipath/.gitignore

@@ -1,3 +1,5 @@
 node_modules/
 .env
 .venv/
+uipath.log
+*.pdf

spikes/uipath/extractDoc.js

@@ -1,26 +1,29 @@
-import axios from "axios";
+import {
+  duPost,
+  UIPATH_BASE_URL,
+  getExtractorGuid,
+  getProjectId,
+} from "./uipathClient.js";
 
 export async function extractDoc(token, docId) {
-  const extract = await axios.post(
-    `https://govcloud.uipath.us:443/${process.env.EXTRACTOR_GUID}/du_/api/framework/projects/${process.env.ZERO_PROJECT_ID}/extractors/generative_extractor/extraction/start?api-version=1.0`,
-    {
-      documentId: docId,
-      pageRange: null,
-      prompts: [
-        {
-          id: "State",
-          question: "What state is this 1115 waver for?",
-          fieldType: "Text",
-          multiValued: false,
-        },
-      ],
-      configuration: null,
-    },
-    {
-      headers: {
-        Authorization: `Bearer ${token}`,
-      },
-    }
-  );
+  const extractorGuid = getExtractorGuid(); // NOTE: Zoe might make her own. So we may need to query here to get the right GUID
+  const projectId = getProjectId();
+  const url = `${UIPATH_BASE_URL}:443/${extractorGuid}/du_/api/framework/projects/${projectId}/extractors/generative_extractor/extraction/start`;
+
+  const extract = await duPost(url, token, {
+    documentId: docId,
+    pageRange: null,
+    prompts: activeQuestionBlobs,
+    configuration: null,
+  });
   return extract.data.resultUrl;
 }
+
+const activeQuestionBlobs = [
+  {
+    id: "State",
+    question: "What state is this 1115 waver for?",
+    fieldType: "Text",
+    multiValued: false,
+  },
+];

spikes/uipath/index.js

@@ -4,30 +4,71 @@ import { getToken } from "./getToken.js";
 import { uploadDocument } from "./uploadDocument.js";
 import { extractDoc } from "./extractDoc.js";
 import { fetchExtractionResult } from "./fetchExtractResult.js";
+import { createLogFile, log } from "./logFile.js";
+import { getProjectId, getExtractorGuid } from "./uipathClient.js";
 
 dotenv.config();
 
 function sleep(ms) {
   return new Promise((resolve) => setTimeout(resolve, ms));
 }
 
-const token = await getToken();
-console.log("Got the auth token.");
+// Positional args: [0]=node, [1]=index.js, [2]=input file, [3]=log file
+const inputFile = process.argv[2] || "ak-behavioral-health-demo-pa.pdf";
+const logPath = createLogFile(process.argv[3] || process.env.LOG_FILE || "uipath.log");
 
-const docId = await uploadDocument(token, "ak-behavioral-health-demo-pa.pdf");
-console.log(`Document ID: ${docId}`);
+if (! inputFile) {
+  log("Usage: node index.js <input-file> [log-file]", logPath);
+  throw new Error("Input file is required");
+  process.exit(1);
+}
+
+try {
+  // Validate required env early to avoid undefined IDs in URLs
+  getProjectId();
+  getExtractorGuid();
 
-const resultUrl = await extractDoc(token, docId);
-console.log(`Result URL: ${resultUrl}`);
+  const token = await getToken();
+  if (!token) {
+    throw new Error("No auth token received.");
+  }
+  log("Got the auth token.", logPath);
 
-while (true) {
-  await sleep(1 * 1000);  
-  let status = await fetchExtractionResult(token, resultUrl);
+  log(`Using input file: ${inputFile}`, logPath);
+
+  const docId = await uploadDocument(token, inputFile);
+  if (!docId || typeof docId !== "string") {
+    log(`Upload returned unexpected response: ${util.inspect(docId)}`, logPath);
+    throw new Error("Upload failed or returned invalid document ID.");
+  }
+  log(`Document ID: ${docId}`, logPath);
 
-  if (status.status === "Succeeded") {
-    console.log(util.inspect(status, false, null, true /* enable colors */));
-    break;
+  const resultUrl = await extractDoc(token, docId);
+  if (!resultUrl || typeof resultUrl !== "string") {
+    log(`Extraction start returned unexpected response: ${util.inspect(resultUrl)}`, logPath);
+    throw new Error("Extraction start failed or returned invalid result URL.");
   }
+  log(`Result URL: ${resultUrl}`, logPath);
 
-  console.log(status);
+  let delayMs = 5_000; // start at 5s to reduce polling cost
+  const maxDelayMs = 30_000; // cap backoff at 30s
+
+  while (true) {
+    await sleep(delayMs);
+    let status = await fetchExtractionResult(token, resultUrl);
+
+    if (status.status === "Succeeded") {
+      log(util.inspect(status, false, null, true /* enable colors */), logPath);
+      break;
+    }
+
+    log(status, logPath);
+    delayMs = Math.min(delayMs * 2, maxDelayMs); // exponential backoff to limit requests/credits
+  }
+} catch (error) {
+  log(`Fatal error: ${error.message}`, logPath);
+  if (error.response?.data) {
+    log(`Response data: ${JSON.stringify(error.response.data)}`, logPath);
+  }
+  throw error;
 }

spikes/uipath/logFile.js

@@ -0,0 +1,30 @@
+import fs from "fs";
+import path from "path";
+
+const DEFAULT_LOG_PATH = "uipath.log";
+
+function ensureParentDir(logPath) {
+  const dir = path.dirname(logPath);
+  if (dir && dir !== ".") {
+    fs.mkdirSync(dir, { recursive: true });
+  }
+}
+
+export function createLogFile(logPath = DEFAULT_LOG_PATH, { overwrite = false } = {}) {
+  ensureParentDir(logPath);
+  if (overwrite || !fs.existsSync(logPath)) {
+    fs.writeFileSync(logPath, "");
+  }
+  return logPath;
+}
+
+export function appendToLog(message, logPath = DEFAULT_LOG_PATH) {
+  ensureParentDir(logPath);
+  const output = typeof message === "string" ? message : JSON.stringify(message);
+  fs.appendFileSync(logPath, output + "\n");
+}
+
+export function log(message, logPath = DEFAULT_LOG_PATH) {
+  console.log(message);
+  appendToLog(message, logPath);
+}

spikes/uipath/uipathClient.js

@@ -0,0 +1,40 @@
+import axios from "axios";
+
+export const UIPATH_BASE_URL = "https://govcloud.uipath.us";
+export const UIPATH_TENANT = "globalalliant/Dev";
+export const UIPATH_API_VERSION = "1.0";
+
+export function getProjectId() {
+  const projectId = process.env.ZERO_PROJECT_ID;
+  if (!projectId) {
+    throw new Error("Missing ZERO_PROJECT_ID in environment.");
+  }
+  return projectId;
+}
+
+// Basically the model it's using, may want different models for differnt files.
+// May need to request to checj the differeent extractors.
+export function getExtractorGuid() {
+  const extractorGuid = process.env.EXTRACTOR_GUID;
+  if (!extractorGuid) {
+    throw new Error("Missing EXTRACTOR_GUID in environment.");
+  }
+  return extractorGuid;
+}
+
+// duPost == Document Understanding POST
+export function duPost(url, token, data, options = {}) {
+  const { params = {}, headers = {}, ...rest } = options;
+
+  return axios.post(url, data, {
+    headers: {
+      Authorization: `Bearer ${token}`,
+      ...headers,
+    },
+    params: {
+      "api-version": UIPATH_API_VERSION,
+      ...params,
+    },
+    ...rest,
+  });
+}

spikes/uipath/uploadDocument.js

@@ -1,33 +1,36 @@
-import axios from "axios";
 import fs from "fs";
 import FormData from "form-data";
+import { log } from "./logFile.js";
+import { duPost, UIPATH_BASE_URL, UIPATH_TENANT, getProjectId } from "./uipathClient.js";
+
+export async function uploadDocument(token, fileName) {
+  const projectId = getProjectId();
+  const url = `${UIPATH_BASE_URL}/${UIPATH_TENANT}/du_/api/framework/projects/${projectId}/digitization/start`;
 
-export async function uploadDocument(token,fileName) {
   const formData = new FormData();
-  formData.append(
-    "file",
-    fs.createReadStream(fileName),
-    fileName
-  );
 
-  const doc = await axios.post(
-    `https://govcloud.uipath.us/globalalliant/Dev/du_/api/framework/projects/${process.env.ZERO_PROJECT_ID}/digitization/start?api-version=1.0`,
-    formData,
-    {
+  formData.append("file", fs.createReadStream(fileName), fileName);
+
+  try {
+    const doc = await duPost(url, token, formData, {
       headers: {
-        "Content-Type": "multipart/form-data", // Important for file uploads
+        // form-data requires its own headers so axios can set boundaries
+        ...formData.getHeaders(),
         "x-uipath-page-range": "All",
-        Authorization: `Bearer ${token}`,
       },
       onUploadProgress: (progressEvent) => {
         // Optional: Track upload progress
-        const percentCompleted = Math.round(
-          (progressEvent.loaded * 100) / progressEvent.total
-        );
-        console.log(`Upload progress: ${percentCompleted}%`);
+        const percentCompleted = Math.round((progressEvent.loaded * 100) / progressEvent.total);
+        log(`Upload progress: ${percentCompleted}%`);
       },
-    }
-  );
+    });
 
-  return doc.data.documentId;
+    return doc?.data?.documentId || doc?.data;
+  } catch (error) {
+    log(`Error uploading document: ${error.message}`);
+    if (error.response?.data) {
+      log(`Upload error response: ${JSON.stringify(error.response.data)}`);
+    }
+    throw error;
+  }
 }