From 27dd1a2d55ec85696bd054313b8af56b93c0f6ee Mon Sep 17 00:00:00 2001
From: Alexander Densley <densley.alexander@gmail.com>
Date: Wed, 4 Dec 2024 20:00:05 -0700
Subject: [PATCH] allow optional density for node (#116)

* add support for selecting density in node

* revert styling

* revert stylings part 2

* add height and update args

* remove space

* clarify props

* sort
---
 node-zerox/src/index.ts | 4 ++++
 node-zerox/src/types.ts | 2 ++
 node-zerox/src/utils.ts | 8 ++++++--
 3 files changed, 12 insertions(+), 2 deletions(-)

diff --git a/node-zerox/src/index.ts b/node-zerox/src/index.ts
index d7550d1..a7d59b0 100644
--- a/node-zerox/src/index.ts
+++ b/node-zerox/src/index.ts
@@ -24,6 +24,8 @@ export const zerox = async ({
   concurrency = 10,
   correctOrientation = true,
   filePath,
+  imageDensity = 300,
+  imageHeight = 2048,
   llmParams = {},
   maintainFormat = false,
   maxTesseractWorkers = -1,
@@ -105,6 +107,8 @@ export const zerox = async ({
       // Convert the file to a series of images
       await convertPdfToImages({
         correctOrientation,
+        imageDensity,
+        imageHeight,
         localPath: pdfPath,
         maxTesseractWorkers,
         pagesToConvertAsImages,
diff --git a/node-zerox/src/types.ts b/node-zerox/src/types.ts
index ccd491d..2faf3af 100644
--- a/node-zerox/src/types.ts
+++ b/node-zerox/src/types.ts
@@ -3,6 +3,8 @@ export interface ZeroxArgs {
   concurrency?: number;
   correctOrientation?: boolean;
   filePath: string;
+  imageDensity?: number;
+  imageHeight?: number;
   llmParams?: LLMParams;
   maintainFormat?: boolean;
   maxTesseractWorkers?: number;
diff --git a/node-zerox/src/utils.ts b/node-zerox/src/utils.ts
index 2f92490..7ac774d 100644
--- a/node-zerox/src/utils.ts
+++ b/node-zerox/src/utils.ts
@@ -205,6 +205,8 @@ const determineOptimalRotation = async ({
 // Convert each page to a png, correct orientation, and save that image to tmp
 export const convertPdfToImages = async ({
   correctOrientation,
+  imageDensity,
+  imageHeight,
   localPath,
   maxTesseractWorkers,
   pagesToConvertAsImages,
@@ -213,6 +215,8 @@ export const convertPdfToImages = async ({
   trimEdges,
 }: {
   correctOrientation: boolean;
+  imageDensity: number;
+  imageHeight: number;
   localPath: string;
   maxTesseractWorkers: number;
   pagesToConvertAsImages: number | number[];
@@ -221,9 +225,9 @@ export const convertPdfToImages = async ({
   trimEdges: boolean;
 }) => {
   const options = {
-    density: 300,
+    density: imageDensity,
     format: "png",
-    height: 2048,
+    height: imageHeight,
     preserveAspectRatio: true,
     saveFilename: path.basename(localPath, path.extname(localPath)),
     savePath: tempDir,