Merge pull request #9 from Zipstack/update-client_for_status_api

Changes to handle status API change

Author: jaseemjaskp

.vscode/launch.json

@@ -0,0 +1,17 @@
+{
+  // Use IntelliSense to learn about possible attributes.
+  // Hover to view descriptions of existing attributes.
+  // For more information, visit: https://go.microsoft.com/fwlink/?linkid=830387
+  "version": "0.2.0",
+  "configurations": [
+    {
+      "type": "node",
+      "request": "launch",
+      "name": "Debug Jest Tests",
+      "program": "${workspaceFolder}/node_modules/jest/bin/jest.js",
+      "args": ["--runInBand"],
+      "console": "integratedTerminal",
+      "internalConsoleOptions": "neverOpen"
+    }
+  ]
+}

index.js

@@ -11,12 +11,11 @@
  * 
 
  */
-
+require("dotenv").config();
 const axios = require("axios");
 const winston = require("winston");
 const fs = require("fs");
 const { register } = require("module");
-
 const BASE_URL = "https://llmwhisperer-api.unstract.com/v1";
 const BASE_URL_V2 = "https://llmwhisperer-api.us-central.unstract.com/api/v2";
 
@@ -374,8 +373,12 @@ class LLMWhispererClientV2 {
 
     this.headers = {
       "unstract-key": this.apiKey,
-      "Subscription-Id": "test", //TODO: Remove this line. For testing only
-      "Start-Date": "9-07-2024", //TODO: Remove this line. For testing only
+      // "Subscription-Id": "jsclient-client",
+      // "Subscription-Name": "jsclient-client",
+      // "User-Id": "jsclient-client-user",
+      // "Product-Id": "jsclient-client-product",
+      // "Product-Name": "jsclient-client-product",
+      // "Start-Date": "2024-07-09",
     };
   }
 
@@ -532,34 +535,30 @@ class LLMWhispererClientV2 {
             message["extraction"] = {};
             message["status_code"] = -1;
             message["message"] = "Whisper client operation timed out";
-            break;
+            return message;
           }
           const whisperStatus = await this.whisperStatus(whisperHash);
+          this.logger.debug(`whisperStatus: ${JSON.stringify(whisperStatus)}`);
+
           if (whisperStatus.statusCode !== 200) {
             message["extraction"] = {};
             message["status_code"] = whisperStatus.statusCode;
             message["message"] = "Whisper client operation failed";
-            break;
+            return message;
           }
-          if (whisperStatus.status === "processing") {
-            this.logger.debug("Status: processing");
-          } else if (whisperStatus.status === "delivered") {
-            this.logger.debug("Status: delivered");
-            throw new LLMWhispererClientException(
-              "Whisper operation already delivered",
-              -1,
+          if (whisperStatus.status === "accepted") {
+            this.logger.debug("Status: accepted...");
+          } else if (whisperStatus.status === "processing") {
+            this.logger.debug("Status: processing...");
+          } else if (whisperStatus.status === "error") {
+            this.logger.debug("Status: error");
+            this.logger.error(
+              "Whisper-hash: ${whisperHash} | STATUS: failed with ${whisperStatus.message}",
             );
-          } else if (whisperStatus.status === "unknown") {
-            this.logger.debug("Status: unknown");
-            throw new LLMWhispererClientException(
-              "Whisper operation status unknown",
-              -1,
-            );
-          } else if (whisperStatus.status === "failed") {
-            this.logger.debug("Status: failed");
             message["extraction"] = {};
             message["status_code"] = -1;
-            message["message"] = "Whisper client operation failed";
+            message["status"] = "error";
+            message["message"] = whisperStatus.message;
             break;
           } else if (whisperStatus.status === "processed") {
             this.logger.debug("Status: processed");
@@ -602,25 +601,30 @@ class LLMWhispererClientV2 {
    * @throws {LLMWhispererClientException} Throws an LLMWhispererClientException if an error occurs during the operation.
    */
   async whisperStatus(whisperHash) {
-    this.logger.debug("whisper_status called");
+    this.logger.debug(`whisper_status called for ${whisperHash}`);
     const url = `${this.baseUrl}/whisper-status`;
     const params = { whisper_hash: whisperHash };
     this.logger.debug(`url: ${url}`);
+    this.logger.debug(`params: ${JSON.stringify(params)}`);
+    delete this.headers["Content-Length"];
+    this.logger.debug(`headers: ${JSON.stringify(this.headers)}`);
 
     try {
       const response = await axios.get(url, {
         headers: this.headers,
         params,
         timeout: this.apiTimeout * 1000,
       });
-
       const message = response.data;
       message.statusCode = response.status;
       return message;
     } catch (error) {
+      this.logger.debug("Hel00000000002");
+      this.logger.debug(`error: ${JSON.stringify(error)}`);
       const err = error.response
         ? error.response.data
         : { message: error.message };
+      this.logger.debug(`error: ${JSON.stringify(err)}`);
       err.statusCode = error.response ? error.response.status : -1;
       throw new LLMWhispererClientException(err.message, err.statusCode);
     }

package-lock.json

@@ -1,6 +1,6 @@
 {
   "name": "llmwhisperer-client",
-  "version": "2.0.1",
+  "version": "2.1.0",
   "description": "LLMWhisper JS Client",
   "main": "index.js",
   "scripts": {
@@ -14,11 +14,13 @@
   "license": "MIT",
   "dependencies": {
     "axios": "~1.7.2",
+    "llmwhisperer-client": "^2.0.1",
+    "string-similarity": "^4.0.4",
     "winston": "~3.13.0"
   },
   "devDependencies": {
     "@eslint/js": "^9.4.0",
-    "dotenv": "^16.4.5",
+    "dotenv": "^16.4.7",
     "eslint": "^9.4.0",
     "eslint-config-prettier": "^9.1.0",
     "eslint-config-turbo": "^1.13.3",

package.json

@@ -0,0 +1,3 @@
+LLMWHISPERER_BASE_URL_V2=https://llmwhisperer-api.us-central.unstract.com/api/v2
+LLMWHISPERER_LOG_LEVEL=DEBUG
+LLMWHISPERER_API_KEY=

sample.env

@@ -0,0 +1 @@
+{"test": "HelloWorld"}

test/data/expected/credit_card.ocr.line-printer.txt renamed to test/data/expected/credit_card.native_text.layout_preserving.txt

@@ -1,3 +1,3 @@
-export LLMWHISPERER_BASE_URL=https://llmwhisperer-api.unstract.com/v1
-export LLMWHISPERER_LOG_LEVEL=DEBUG
-export LLMWHISPERER_API_KEY=
+LLMWHISPERER_BASE_URL_V2=https://llmwhisperer-api.us-central.unstract.com/api/v2
+LLMWHISPERER_LOG_LEVEL=DEBUG
+LLMWHISPERER_API_KEY=

test/data/expected/handwritten-form.ocr.line-printer.txt renamed to test/data/expected/handwritten-form.form.layout_preserving.txt

@@ -1,10 +1,12 @@
+
 const fs = require("fs");
 const path = require("path");
-const LLMWhispererClient = require("../index").LLMWhispererClient;
-const client = new LLMWhispererClient({
-  apiKey: process.env.LLMWHISPERER_API_KEY,
-});
-describe("LLMWhispererClient", () => {
+const stringSimilarity = require("string-similarity");
+const LLMWhispererClientV2 = require("../index").LLMWhispererClientV2;
+
+const client = new LLMWhispererClientV2();
+
+describe("LLMWhispererClientV2", () => {
   test("get_usage_info", async () => {
     const usage_info = await client.getUsageInfo();
     console.info(usage_info);
@@ -22,74 +24,44 @@ describe("LLMWhispererClient", () => {
     );
   });
 
+
   const test_cases = [
-    ["ocr", "line-printer", "restaurant_invoice_photo.pdf"],
-    ["ocr", "line-printer", "credit_card.pdf"],
-    ["ocr", "line-printer", "handwritten-form.pdf"],
-    ["ocr", "text", "restaurant_invoice_photo.pdf"],
-    ["text", "line-printer", "restaurant_invoice_photo.pdf"],
-    ["text", "text", "handwritten-form.pdf"],
+    ["high_quality", "layout_preserving", "restaurant_invoice_photo.pdf", 99],
+    ["native_text", "layout_preserving", "credit_card.pdf", 99],
+    ["form", "layout_preserving", "handwritten-form.pdf", 99],
+    ["high_quality", "layout_preserving", "handwritten-form.pdf", 80],
   ];
 
   test.each(test_cases)(
     "whisper(%s, %s, %s)",
-    async (processing_mode, output_mode, input_file) => {
+    async (mode, output_mode, input_file, percent_simlarity) => {
       const data_dir = path.join(__dirname, "data");
       const file_path = path.join(data_dir, input_file);
       const response = await client.whisper({
-        processingMode: processing_mode,
+        mode: mode,
         outputMode: output_mode,
         filePath: file_path,
         timeout: 200,
+        waitForCompletion: true
       });
-      console.debug(response);
 
-      const exp_basename = `${path.parse(input_file).name}.${processing_mode}.${output_mode}.txt`;
+
+      const exp_basename = `${path.parse(input_file).name}.${mode}.${output_mode}.txt`;
       const exp_file = path.join(data_dir, "expected", exp_basename);
-      const exp = await fs.promises.readFile(exp_file, "utf-8");
+      const expected_text = await fs.promises.readFile(exp_file, "utf-8");
 
       expect(typeof response).toBe("object");
-      expect(response.statusCode).toBe(200);
-      // expect(response.extracted_text).toBe(exp);
-    },
-    200000,
-  );
 
-  // TODO: Review and port to Jest based tests
-  test.skip("whisper", () => {
-    // response = client.whisper(
-    //   'https://storage.googleapis.com/pandora-static/samples/bill.jpg.pdf'
-    // );
-    const response = client.whisper("test_files/restaurant_invoice_photo.pdf", {
-      timeout: 200,
-      store_metadata_for_highlighting: true,
-    });
-    console.info(response);
-    // expect(typeof response).toBe('object');
-  });
+      const extracted_text = response.extraction.result_text
 
-  test.skip("whisper_status", () => {
-    const response = client.whisper_status(
-      "7cfa5cbb|5f1d285a7cf18d203de7af1a1abb0a3a",
-    );
-    console.info(response);
-    expect(typeof response).toBe("object");
-  });
+      console.log(`Extracted Text: ${extracted_text}`);
+      expect(response.status_code).toBe(200);
+      const similarity = stringSimilarity.compareTwoStrings(extracted_text, expected_text);
+      console.log(`Similarity: ${(similarity * 100).toFixed(2)}%`);
+      expect(similarity * 100).toBeGreaterThan(percent_simlarity); // Expect at least 80% match
 
-  test.skip("whisper_retrieve", () => {
-    const response = client.whisper_retrieve(
-      "7cfa5cbb|5f1d285a7cf18d203de7af1a1abb0a3a",
-    );
-    console.info(response);
-    expect(typeof response).toBe("object");
-  });
-
-  test.skip("whisper_highlight_data", () => {
-    const response = client.highlight_data(
-      "9924d865|5f1d285a7cf18d203de7af1a1abb0a3a",
-      "Indiranagar",
-    );
-    console.info(response);
-    expect(typeof response).toBe("object");
-  });
+    },
+    200000,
+  );
 });
+