feat: add pdf image renderer

packages/lib/server-only/ai/envelope/detect-fields/index.ts

@@ -162,8 +162,8 @@ export const detectFieldsFromPdf = async ({
       // Mask existing fields on the image
       const maskedImage = await maskFieldsOnImage({
         image: page.image,
-        width: page.width,
-        height: page.height,
+        width: page.scaledWidth,
+        height: page.scaledHeight,
         fields: fieldsOnPage,
       });
 

packages/lib/server-only/ai/pdf-to-images.ts

@@ -1,7 +1,11 @@
 import pMap from 'p-map';
 import * as pdfjsLib from 'pdfjs-dist/legacy/build/pdf.mjs';
+import type { PDFDocumentProxy } from 'pdfjs-dist/types/src/display/api';
+import type { ExportFormat } from 'skia-canvas';
 import { Canvas, Image, Path2D } from 'skia-canvas';
 
+import { PDF_IMAGE_RENDER_SCALE } from '../../constants/pdf-viewer';
+
 // @ts-expect-error napi-rs/canvas satisfies the requirements
 globalThis.Path2D = Path2D;
 // @ts-expect-error napi-rs/canvas satisfies the requirements
@@ -42,10 +46,17 @@ class SkiaCanvasFactory {
 
 export type PdfToImagesOptions = {
   scale?: number;
+
+  /**
+   * The format of the images to return.
+   *
+   * Defaults to 'jpeg'.
+   */
+  imageFormat?: ExportFormat;
 };
 
 export const pdfToImages = async (pdfBytes: Uint8Array, options: PdfToImagesOptions = {}) => {
-  const { scale = 2 } = options;
+  const { scale = PDF_IMAGE_RENDER_SCALE, imageFormat = 'jpeg' } = options;
 
   const task = await pdfjsLib.getDocument({
     data: pdfBytes,
@@ -56,37 +67,7 @@ export const pdfToImages = async (pdfBytes: Uint8Array, options: PdfToImagesOpti
 
   const images = await pMap(
     Array.from({ length: pdf.numPages }),
-    async (_, index) => {
-      const pageNumber = index + 1;
-      const page = await pdf.getPage(pageNumber);
-
-      const viewport = page.getViewport({ scale });
-
-      const canvas = new Canvas(viewport.width, viewport.height);
-      canvas.gpu = false;
-
-      const canvasContext = canvas.getContext('2d');
-
-      await page.render({
-        // @ts-expect-error napi-rs/canvas satifies the requirements
-        canvas,
-        // @ts-expect-error napi-rs/canvas satifies the requirements
-        canvasContext,
-        viewport,
-      }).promise;
-
-      const result = {
-        pageNumber,
-        image: await canvas.toBuffer('jpeg'),
-        width: Math.floor(viewport.width),
-        height: Math.floor(viewport.height),
-        mimeType: 'image/jpeg',
-      };
-
-      void page.cleanup();
-
-      return result;
-    },
+    async (_, pageIndex) => getPdfPageAsImage({ pdf, pageIndex, scale, imageFormat }),
     { concurrency: 10 },
   );
 
@@ -95,3 +76,85 @@ export const pdfToImages = async (pdfBytes: Uint8Array, options: PdfToImagesOpti
 
   return images;
 };
+
+export type PdfToImageOptions = {
+  scale?: number;
+  pageIndex: number;
+
+  /**
+   * The format of the image to return.
+   * Defaults to 'jpeg'.
+   */
+  imageFormat?: ExportFormat;
+};
+
+export const pdfToImage = async (pdfBytes: Uint8Array, options: PdfToImageOptions) => {
+  const { scale = PDF_IMAGE_RENDER_SCALE, pageIndex, imageFormat = 'jpeg' } = options;
+
+  if (pageIndex !== undefined && pageIndex < 0) {
+    throw new Error('Page index must be greater than 0');
+  }
+
+  const task = await pdfjsLib.getDocument({
+    data: pdfBytes,
+    CanvasFactory: SkiaCanvasFactory,
+  });
+
+  const pdf = await task.promise;
+
+  const image = await getPdfPageAsImage({ pdf, pageIndex, scale, imageFormat });
+
+  void pdf.destroy().catch((e) => console.error(e));
+  void task.destroy().catch((e) => console.error(e));
+
+  return image;
+};
+
+type GetPdfPageAsImageOptions = {
+  pdf: PDFDocumentProxy;
+  pageIndex: number;
+  scale: number;
+  imageFormat: ExportFormat;
+};
+
+const getPdfPageAsImage = async ({
+  pdf,
+  pageIndex,
+  scale,
+  imageFormat,
+}: GetPdfPageAsImageOptions) => {
+  const page = await pdf.getPage(pageIndex + 1);
+
+  const viewport = page.getViewport({ scale });
+
+  const canvas = new Canvas(viewport.width, viewport.height);
+  canvas.gpu = false;
+
+  const canvasContext = canvas.getContext('2d');
+
+  await page.render({
+    // @ts-expect-error napi-rs/canvas satifies the requirements
+    canvas,
+    // @ts-expect-error napi-rs/canvas satifies the requirements
+    canvasContext,
+    viewport,
+  }).promise;
+
+  const originalViewport = page.getViewport({ scale: 1 });
+
+  const result = {
+    pageIndex,
+    pageNumber: pageIndex + 1,
+    image: await canvas.toBuffer(imageFormat),
+    originalWidth: originalViewport.width,
+    originalHeight: originalViewport.height,
+    scale,
+    scaledWidth: Math.floor(viewport.width),
+    scaledHeight: Math.floor(viewport.height),
+    mimeType: 'image/jpeg',
+  };
+
+  void page.cleanup();
+
+  return result;
+};

packages/lib/server-only/document/send-document.ts

@@ -310,7 +310,7 @@ export const sendDocument = async ({
 
 const injectFormValuesIntoDocument = async (
   envelope: Envelope,
-  envelopeItem: Pick<EnvelopeItem, 'id'> & { documentData: DocumentData },
+  envelopeItem: Pick<EnvelopeItem, 'id'> & { documentData: Omit<DocumentData, 'metadata'> },
 ) => {
   const file = await getFileServerSide(envelopeItem.documentData);