Tyler/add ignore warnings (#100)

* add function to ignore punycode warnings

* 1.0.39

* Also increase max tokens

Author: tylermaran

node-zerox/src/handleWarnings.ts

@@ -0,0 +1,15 @@
+// Tesseract relies on node-fetch v2, which has a deprecated version of punycode
+// Suppress the warning for now. Check in when teseract updates to node-fetch v3
+// https://github.com/naptha/tesseract.js/issues/876
+if (process.stderr.write === process.stderr.constructor.prototype.write) {
+  const stdErrWrite = process.stderr.write;
+  process.stderr.write = function (chunk: any, ...args: any[]) {
+    const str = Buffer.isBuffer(chunk) ? chunk.toString() : chunk;
+
+    // Filter out the punycode deprecation warning
+    if (str.includes("punycode")) {
+      return true;
+    }
+    return stdErrWrite.apply(process.stderr, [chunk]);
+  };
+}

node-zerox/src/index.ts

@@ -1,3 +1,4 @@
+import "./handleWarnings";
 import {
   addWorkersToTesseractScheduler,
   cleanupImage,

node-zerox/src/utils.ts

@@ -16,7 +16,7 @@ const convertAsync = promisify(convert);
 
 const defaultLLMParams: LLMParams = {
   frequencyPenalty: 0, // OpenAI defaults to 0
-  maxTokens: 2000,
+  maxTokens: 4000,
   presencePenalty: 0, // OpenAI defaults to 0
   temperature: 0,
   topP: 1, // OpenAI defaults to 1